EEVblog Electronics Community Forum
Electronics => Manufacturing & Assembly => Topic started by: jmelson on August 04, 2020, 12:11:13 am
-
Well, at long last, my Quad QSA-30A is supposed to arrive on a truck tomorrow morning! I bought this at an auction in Austin TX for $500 (I was the only bidder). Hopefully, it is in working shape. I will end up paying $5000 total, with shipping, feeders, fork lift rental, plywood (to prevent forklift sinking into back yard), etc. etc.
This is a 2250 Lb machine, 5 x 5.5 feet, 3 nozzles with automatic nozzle exchange, supposedly good for 13K components/hour.
It has flying vision on all 3 nozzles (with line-scan cameras) and is supposed to be reck-solid for 0.65 mm lead pitch. I'm going to try to add an upward camera to get more accurate placement of 0.5 mm lead pitch parts.
Here's hoping it is in good condition.
Jon
-
What a beast! How many 8mm feeders can it take?
It's crazy how old machines become worth such a low amount that the most expensive part is freighting them within your own country!
-
Bahahaha!!! :-DD
Sorry, I just had to laugh!!
I bought my pnp early this year and it was summer, I had no problems getting it into my shed.
In the mean time I decided it would be a great idea to buy another house, the winter rain came and my backyard became non forklift friendly!
With only a couple of weeks left to move it(I bought a forklift as it was going to cost too much to hire a forklift at both ends)(and every man needs a forklift!) I have had to install a brick path to get the forklift to my shed to remove it.. Finally I have the forklift in my shed and pnp firmly attached to its pallet. Next for getting it all back across that dodgy brick path. Hopefully it doesn't all sink half way across.... :-DD
Nice score on the machine! I wish I lived in the US, second had gear like that here in Australia is rarer than hens teeth.
-
What a beast! How many 8mm feeders can it take?
Looks like about 75. Way more than I will ever need.
It's crazy how old machines become worth such a low amount that the most expensive part is freighting them within your own country!
Well, it may have something to do with the current pandemic and economic mess, a good time to be buying capital equipment.
But, of course, I have to get it running. Right now, it seems to be having an electronic fault, so I may have to buy some VME boards.
This could get expensive.
Jon
-
Make sure all of the boards are seated in properly if you haven't already. Things rattle around a bit whilst on the back of a truck.
-
Make sure all of the boards are seated in properly if you haven't already. Things rattle around a bit whilst on the back of a truck.
Even worse than that! The original hard drive went bad, and they just laid another drive on top of it, not screwed down to anything.
So, that drive was flopping and bouncing around for 900+ miles from Austin to here. But, it boots up and loads the software.
Then, I had problems with the PC (Celeron 566 MHz) communicating with the VME computer (68020, I think). The ISA-bus interface
card is totally passive, but had 2 50-pin cables crimped to the board. I hate those things. One of them was coming apart, so I unsoldered the wire termination and soldered in a 50-pin standard header, and crimped a 50-pin female to the cable.
Then, it wouldn't go to the "ready" state, with servos powered. I eventually found a blown fuse.
Now, I can get it totally powered up, but there is a bad theta motor (nozzle rotation) on one head. This prevents it from homing the axes. The X, Y and each Z are brushless servos, but the rotation is done with 5-phase steppers. One of them seems to be bad, but with a weird symptom. Three of the wires show continuity with each other, the other two wires have continuity just to themselves.
That actually seems like a really weird failure for a 5-phase stepper. It just seems odd that it could have two isolated groups of coils.
(On the other motors, all 5 wires have continuity.)
I don't find obviously blown transistors on the stepper driver for that axis.
There's a huge amount of software functionality in their program, it will take me some time to learn my way around it.
Jon
-
Our used UP2000 printer was like that - the HDD had been replaced and the new one was just floating in the drive bay, but at least they had packed it in with some pink static foam.
Do you have wiring diagrams for the machine? Repairs would be immensely more difficult without them.
-
Do you have wiring diagrams for the machine? Repairs would be immensely more difficult without them.
The manuals are very hard to get. I found an outfit that has both the service and operator's manuals, and he offered to throw them in for free when I bought $23000 of feeders from him. Well, when I said I wasn't going to give him 23K, he won't return my emails anymore. The only solution I've found so far is to have a guy who is running these machines to make a copy of his manuals. He knows he has the operator's manual, he wasn't so sure about the service one. So, I'm trying to get the manuals, and hopefully will have them in a week or so.
But, this thing is just about posessed! The computer boots up just fine, but the machine will only go fully operational, with all servos turned on, ONCE a day! The first time I turn it on, it comes up and will go "ready" on the first try, then all the rest of the day, I can get the computer running, but it will not go "ready", meaning that the contactors that power the servos will not pull in. This is obviously where I need the manuals to trace things out.
Anyway, I was getting a "backup battery error" every time I loaded the program, so I replaced a lithium battery on the VME CPU board, and this error message has now disappeared. But, presumably, the info in the battery-backed memory has been lost, and I have no idea how to get it put back. I have no idea if that is related to the issue of only going "ready" one time per day.
And, the reason I'm powering it up and down a lot is that one of the rotation motors appears to be bad. These are hollow-shaft 5-phase motors, so are going to be scarce as hen's teeth.
Jon
-
Wow!! 23000 for some free manuals!!!
I'm sure you will be able to find some somewhere, have you tried asking on SMTnet.com?
For the faulty motor, I would try and disassemble it and see if you can see where the fault has occurred. There must be a common point where all the windings join. Might be easier than finding a replacement. What have you got to loose?
For the machine not going fully operational, you could try some freeze spray about the place and see if your able to get it to go a second time, or try heating up a section with the hot air gun and see it it won't come online after heating a certain area, try it in halves and see if you are able to narrow it down.
When I bought my KE-730 it had had a fault that caused it to reboot after an hour or so when it was in use.
By the time I got it it was rebooting every few minutes or less. Turned out to be a cracked solder on a fet that was mounted to the case in one of the power supply modules, so as it would heat up, power would drop out to the computer.
Take note of any LED's in your power supplies and make sure they are the same when working/not working and perhaps check all of the voltage outputs from them.
-
Wow $23k is extortion!
You probably already know this but with intermittent faults make sure the root cause isn't bad relay or contactor contacts. My UP2000 was powering up cantankerously and it turned out that the main contactor contacts were worn out. New contactor and she has been running sweet since.
-
Wow!! 23000 for some free manuals!!!
Ah, well, he was going to sell me about 25 Quad feeders for the 23K and throw in the manuals. That is not crazy for refurbished and guarateed feeders, but it is way out of my budget.
I'm sure you will be able to find some somewhere, have you tried asking on SMTnet.com?
Yes, even before I got the machine. Some people mentioned who to contact, but I didn't get much of a response from that.
For the faulty motor, I would try and disassemble it and see if you can see where the fault has occurred. There must be a common point where all the windings join. Might be easier than finding a replacement. What have you got to loose?
these are 5-phase motors, they have 5 windings wired in a ring arrangement, electrically. there is no common point. These are also quite small motors, single-stack NEMA 23, there isn't going to be a lot of space inside. And, to display the two separate groups of windings, it must have TWO breaks in the ring.
For the machine not going fully operational, you could try some freeze spray about the place
PLACE? This machine has more boards and wiring in it than a Tesla! Until I get the manual, I really can't trace out what happens when you click these buttons. It appears that the E-stop and ready button may be handled in hardware, with a permission input from the computer. The fact it needs a 10+ hour shutdown to come alive again starts me thinking down a road to a software issue.
The VME CPU does have one of those Dallas Semi Timekeeper modules with an added-on external battery. I replaced the dead battery with a couple of alkaline cells, and it got rid of the "backup battery error" message, so that was poking into the right area. But, presumably, the info the battery was supposed to be preserving is lost, and I don't know how to put it back. The 2nd part of the massage seems to indicate it was preserving what nozzle was last on each head, which would be a really small amount of data, like 3 bytes.
When I bought my KE-730 it had had a fault that caused it to reboot after an hour or so when it was in use.
By the time I got it it was rebooting every few minutes or less. Turned out to be a cracked solder on a fet that was mounted to the case in one of the power supply modules, so as it would heat up, power would drop out to the computer.
Take note of any LED's in your power supplies and make sure they are the same when working/not working and perhaps check all of the voltage outputs from them.
Yes, I have checked a few power supplies, there are racks of them, and several have been bypassed due to failures. The ones that are still hooked up seem to all be well within tolerance.
Thanks for the suggestions,
Jon
-
Ok, to test if it is a timed battery thing, disconnect the battery and reconnect it after 10 hours and see if the machine has to wait a further 10 hours to boot.
I know it would have a lot of cards and power supplies, but just start on one side and see if you can make a difference to the restart time. It would be a real pain having to wait 10 hours to know if it may have made any difference. Do you have a portable air conditioner you could sit in front of the machine to cool the electronics?
If there are power supplies not working and bypassed, I'd be looking into these first(fix all the obvious problems first) There probably isn't much wrong with them, could just be cracked solder on through hole parts. The machines do see a lot of vibration.
-
Ok, to test if it is a timed battery thing, disconnect the battery and reconnect it after 10 hours and see if the machine has to wait a further 10 hours to boot.
I know it would have a lot of cards and power supplies, but just start on one side and see if you can make a difference to the restart time. It would be a real pain having to wait 10 hours to know if it may have made any difference. Do you have a portable air conditioner you could sit in front of the machine to cool the electronics?
If there are power supplies not working and bypassed, I'd be looking into these first(fix all the obvious problems first) There probably isn't much wrong with them, could just be cracked solder on through hole parts. The machines do see a lot of vibration.
I could probably just disconnect the battery and short the terminals on the computer side for a few seconds and get the same result.
I may try this, but right now the machine is quite torn apart.
Some IDIOT replaced a 5-phase stepper motor on the rotation axis with a 2-phase motor. No WONDER it was vibrating and not moving smoothly! Geez! I have the whole head apart now, but need some help form an old Quad tech on how to get the motor completely off the head. Then , I may have quite a problem sourcing a 5-phase, hollow-shaft motor. The same motor is used on several Samsung-built machines, so I may get lucky.
The power supply issue is all related to the feeder power supplies, which are mostly separate from the rest of the machine. So, I don't think that is the issue.
Thanks,
Jon
-
I would think the motor is quite easy to obtain, if you google Samsung theta motor you get quite a few results which at least point you to some Chinese supplies of new, refurb or clones of this kind of kit.
-
I would think the motor is quite easy to obtain, if you google Samsung theta motor you get quite a few results which at least point you to some Chinese supplies of new, refurb or clones of this kind of kit.
Yes, and I have sent inquiries. But, if I can buy a brand-new motor from the manufacturer's authorized distributor in the US for $150 or a used motor from China for $110 plus shipping, the new motor sounds like a better deal. I've made a call to their Chicago branch, haven't gotten a reply yet.
But, at least there are options.
Jon
-
Not so simple. Sanyo Denki confirms this is a custom motor made for Samsung, and they can't even give me specs on it, or if there is a standard product that would be a replacement.
Contacting anybody at Samsung for anything not related to phones or home appliances seems very difficult. The division that made these P&P machines apparently is no longer part of Samsung, it was spun off when Samsung Aerospace went kaput.
But, there are tons of these machines in the far East, so used parts are plentiful. I will hopefully be able to place an order for a used motor today (working around the time difference).
------------------------
Next issue: The machine doesn't want to go to the "ready" state with the servos enabled, except very rarely. There is an I/O page that is supposed to show all the input and output points. But, I finally discovered this page only works in 640 x 480 screen mode.
The second picture is what you are supposed to see, the red and green dots show the state of the input and output points.
But, the 1st picture shows what you get with any higher screen resolution, I think the boxes with the signal names are drawn OVER the indicator dots. But, now that I know how to make that page work right, I can see the ready button is not being sensed by the computer. I verified that the switch contacts are actually working right, so it is the wiring or interface board where it is not getting through.
Hopefully the guy who is still using one of these machines in the US will find a place to have his manuals duplicated, and I will be able to trace the wires when I get the manuals.
This is just kind of a progress report.
Jon
-
Spun off at least twice in fact. I've not noticed the Samsung/Hanwha guys* over here refurb and sell a Quad but they certainly still do CP45V & that generation etc when they come their way. This would suggest parts remain available to some extent unless they are cannibalizing others...
*I suspect the Quad repair guy does tho' but they are not the kind of outfit that splashes that on their website/social, just pages like this https://www.tronteq.co.uk/product/quad-qsa-30v/ (https://www.tronteq.co.uk/product/quad-qsa-30v/)
-
Spun off at least twice in fact. I've not noticed the Samsung/Hanwha guys* over here refurb and sell a Quad but they certainly still do CP45V & that generation etc when they come their way. This would suggest parts remain available to some extent unless they are cannibalizing others...
*I suspect the Quad repair guy does tho' but they are not the kind of outfit that splashes that on their website/social, just pages like this https://www.tronteq.co.uk/product/quad-qsa-30v/ (https://www.tronteq.co.uk/product/quad-qsa-30v/)
Yes, it looks like the Samsung CP33, CP40 and CP45 all use the same 5-phase stepper motors as the Quad QSA-30A and V. The CP30 and and QSA-30
apparently used a 2-phase stepper motor for the rotation (theta) axis. Anyway, I am always giving people the Sanyo Denki motor part number, and not telling them what machine it is for, to avoid confusion. That is apparently how this machine got fouled up, they put a QSA-30 motor on a QSA-30A without looking at the part number. The Qtips documents have a list of part differences between the machines. (Qtips is the Quad internal service notes.)
Jon
-
Next issue: The machine doesn't want to go to the "ready" state with the servos enabled, except very rarely. There is an I/O page that is supposed to show all the input and output points. But, I finally discovered this page only works in 640 x 480 screen mode.
Are you running Windows with a non-default font size (i.e. DPI) setting? It looks like the indicator dots are being drawn at a fixed pixel offset in the window, but all the other window elements are larger than normal, so the dots are being obscured.
-
Are you running Windows with a non-default font size (i.e. DPI) setting? It looks like the indicator dots are being drawn at a fixed pixel offset in the window, but all the other window elements are larger than normal, so the dots are being obscured.
Yes, this seems to be the problem, or part of it. I did not set up this system. In 800x600 and above screen resolutions, the diagnostic window appears INSIDE the main program window. In 600x480 resolution, the diagnostic window shows as full-screen, and displays right. Until I get a complete disk image copy of the original disk, I'm a little worried about making tricky changes to the system. I know I can change the scren resolution down and back pretty safely.
I used to be pretty competent with Win 95, but that was some 20 years ago. How do you change the font size? Is it in control panel/display? or, somewhere else?
Thanks,
Jon
-
IIRC, when you go into Display Settings (where you change resolution), you can click the 'Advanced' button at the bottom, and then the first tab has the font size setting, which is really also a DPI setting. I believe the normal setting is 'Small', at 96 DPI.
-
IIRC, when you go into Display Settings (where you change resolution), you can click the 'Advanced' button at the bottom, and then the first tab has the font size setting, which is really also a DPI setting. I believe the normal setting is 'Small', at 96 DPI.
Great, thanks, I will try that tomorrow! 640x480 is an AWFUL screen, and the rest of the program doesn't fit well.
Jon
-
IIRC, when you go into Display Settings (where you change resolution), you can click the 'Advanced' button at the bottom, and then the first tab has the font size setting, which is really also a DPI setting. I believe the normal setting is 'Small', at 96 DPI.
Ahh, couldn't wait ... So, I tried with 1024 x 768 and 75% normal, small and large fonts, didn't change the bad display. So, it seems like the definition for that page is just bad. I'm surprised nobody ever complained. I think I'm running the last version of the software for this machine. Well, at least I know how to make the page function when I need to diagnose sensors, etc. on the machine. Then, I have to reboot to go back to reasonable screens.
Thanks for the help.
Jon
-
could you take a picture of the vision modules (especially how they designed the illumination)?
I wonder how they solved this topic, some stray-light is required in order to make items be captured properly.
-
could you take a picture of the vision modules (especially how they designed the illumination)?
I wonder how they solved this topic, some stray-light is required in order to make items be captured properly.
Pictures won't tell you a whole lot. CyberOptics and QuadAlign are pretty similar. They have a red diode laser in a housing with some kind of lens or slit assembly that creates a fan beam, and then a lens that collimates that so it projects a parallel line of light. Then, on the other side of the nozzle, there is a line-scan camera chip just bare with no optics. These are similar to what was used in older FAX machine scanners. On my machine, it is a 4000-element linear array of photodiodes with a CMOS charge-coupled analog shift register.
The system is that after picking up the part, it is lifted to a Z height such that the best feature of the part is in the linear light beam.
The part projects a shadow on the linear camera chip. The nozzle is rotated until the shadow is narrowest. This has the part square to the beam. The centroid of the shadow is recorded, then the nozzle is turned 90 degrees and the centroid is taken again. This gives the rotation alignment and the X-Y offset of the part on the nozzle. This is then used to correct the part placement.
You can Google quadalign and get a page from goppm (the people who bought the rights to Quad technology) about it.
Later Samsung machines used a 2-D system. The CP45 has 6 nozzles in a row. They have a mirror that clicks in at 45 degrees under the nozzles. Cameras and lighting arrays give six separate chip cameras a 2D view of the underside of the part. The advantage here is that the solder pads of QFN parts and the balls on BGAs can be examined and aligned.
Jon
-
Hi,
pictures still say more than thousand words ;-)
We're working on improving the optical vision system in our DIY machine, we have started to build a new machine for a second location a while ago (we already have a commercial machine from a vendor from Poland who cannot even count until two in the mechanical area, it was a good inspiration how to not make a pick and place machine).
I'm just interested in the LED positions of those cameras, how many angles they use for flooding the objects with light. I've seen quite a few on youtube already.
Our final system will use a USB 3.0 based camera system with 5 sensors (4 with small resolution, one with high resolution) + fpga (for multiplexing and triggering).
-
OK, this is NOT a standard camera, it is a LINE SCAN camera, exactly one line of pixels. The light source is a red diode laser with optics to project a narrow line of collimated light across the nozzle. The shadow of the part falls on the line scan photodiode array with no lenses.
The first picture is the assembly on the machine head, the 2nd picture is the light source/camera assembly off the machine. In both cases, the light source is on the right, the camera on the left.
They do not flood the part with light, they see the shadow of the part while staring right into the light source. When I next have the machine up enough to do so, I will take a picture of the camera diagnositc screen, it shows a graph of light intensity vs. pixel position across the scan line.
Jon
-
Seems like you're on a big project... maybe mission.
-
Seems like you're on a big project... maybe mission.
Oh, YES! But, I got a really good deal on the machine, $500 + auctioneer's premium. There were a whole bunch of little things that I'm taking care of. Cracked air filter, feeder power supplies have been replaced twice, leaving the old supplies in place and adding new ones, a sensor on the conveyor broken off by the movers, stuff like that.
Then, two real issues. Somebody replaced one of the rotation motors with the wrong type, and the machine mostly will not go to the "ready" state with the servos enabled. I have the right motor on order from China, and another user of this model has duplicated a set of manuals for me and they are on the way. Once I ge the manuals, I can try to trace out the ready botton and see why it is not getting to the computer. So, by sometime late next week, I home to have it up and running.
THEN, I have to learn their software! My current machine (Philips CSM84) is simple by comparison. You create a file that says "take a part from feeder # 12 and place it at coordinates 123.45 234.56 rotated 90 degrees on head 2". Really simple. This machine has automatic nozzle changers, and vision templates for each component type. it has a built-in program to take a CSV board stuffing file and convert to their internal format. I will probably have to add new parts to the vision templates, but that looks fairly easy.
One of the issues is communication. I mucked up the ethernet configuration and now it recognizes the ethernet card but doesn't bind any services to it. I tried to put a CDROM drive on it, but it needed drivers to be loaded to talk to it. It does have serial ports, but I'll need the manuals to know the protocol. I'm not even going to TRY to use the 3.5" floppy drive on it, where would I get media?
Jon
-
I'm not even going to TRY to use the 3.5" floppy drive on it, where would I get media?
I have seen flash drive -> floppy drive emulators so that could be an option if you don't get ethernet going.
-
Arrgh! I got the manuals today, but there's no schematic anywhere. This thing has at least 25 boards and modules scattered all through the base, with mazes of cables running everywhere. The "ready" button is not seen by the computer, while the other front panel buttons are seen. I have no idea where the wires for the ready button go. Usually, the maintenance manual has at least an overall schematic so you know what bundle of wires goes from one place to another.
Just venting, I guess!
Jon
-
That's upsetting! You could try reaching out to the spare parts suppliers in China and see if they have manuals?
-
Well, it was a real BEAR to track the wires from the switch to the logic, but I finally did it. So, the ready pushbutton goes to an interconnect board, then to another cable, then to the DPRAM board that is the interface between the PC computer and the VME computer. I was able to see the switch signal go through an opto-coupler and then a Schmitt trigger chip and come out the other side as good TTL level signals and then it goes into a single chip micro. So, I can't really track the signal into that. Possibly that micro implements some kind of interlock with some other signal and is preventing the button push from getting through to the PC, which displays that the button is NOT being pushed.
This is pretty critical, without it, all the servo and stepper drives do not get powered on.
I've found some of these DPRAM boards on eBay for a reasonable price, but I have no idea if there is actually anything wrong with the board.
Jon
-
OK, some new info. In the I/O diagnostics page, there is an input to the computer labeled "power switch". It is always indicating false. It makes some sense that this would sense the condition of the "main start" contactor, as this powers up all the safety sensors (door switches, travel limit sensors, etc.), and prevent the ready button from powering up the servo amps until the power to the sensors is on. So, I have asked somebody with a similar machine to confirm that is what the "power switch" input point is actually responding to.
If so, then I need to see why the contactor turning on does NOT show up at the computer. Could be a wiring issue or an optocoupler that has gone bad.
And, I'm STILL waiting for the schematics... GoPPM is having the manual scanned for me.
Jon
-
As I said earlier in the thread - I had a burner out auxiliary contact on my UP2000 printer's main power contactor, which prevented it from latching on upon power up. Contactors are cheap so given the age I would just swap it out anyway.
-
As I said earlier in the thread - I had a burner out auxiliary contact on my UP2000 printer's main power contactor, which prevented it from latching on upon power up. Contactors are cheap so given the age I would just swap it out anyway.
CR #1 is a 5-pole contactor, and pretty compact as well. So, I'd have to do some study to find a drop-in replacement. Also, the power contacts SEEM to be working fine. I saw something on one of the auxiliary contacts that looked a little odd, I didn't get exactly 0.00 V across it when it was supposed to be closed.
But, I don't know what is on those particular contacts. I should be getting the schematics shortly, and hopefully all this wiring will start to make more sense once I understand what does what.
Jon
-
Well, I now have the schematics, they are horrible, and don't exactly match my machine. But, mostly they are close. I only got machine wiring schematics, NONE of the board schematics. Ugh, I was hoping for more detail.
I had an idea that the mysterious input point the computer shows as "power switch" might be used to interlock going to ready mode, but it is not hooked up to anything. I was able to ground that optoisolator input, and the input display does show "power switch" going on, but it had no effect on anything else.
So, I guess that is a feature unused in this model. DARN!
So, I'm still stuck with not being able to turn the servos on, which makes it pretty hard to do anything on the machine. It COULD be a defect in the IO_DPRAM board, where all of this action appears to connect, but it could be a software issue. Geez, you'd think there's be a diagnostic tree in the
maintenance manual, but no sign of such a thing.
Jon
-
Well, a long wait to make any progress. I ordered an IO_DPRAM board from Korea, but it never came, so I had to order another, at higher price from a different outfit. This one came very quickly, I plugged it in, and the machine fired up
immediately! A great relief! I'm still waiting for some ball bearings to reload the head 3 ball spline to arrive, then I'll put head 3 back in and should have the machine fully operational. I still have to come up with some way to mount my vibratory feeder on the feeder rails, and also make a set of support pins to hold the board up on the conveyor.
I've ordered some magnets to place on the backup plate, and have to make some spring-loaded pins to go on them.
So, progress at last.
Jon
-
Awesome! That's great to hear!
-
And, the saga just keeps going on! Arrghhh!
So, I replaced the IO_DPRAM board, and the machine goes to "ready", enabling servos and all the features of the software. I rebuilt the 3rd head's ball spline, and got it all together again, put the top back on the machine and
put the video monitors back on. Now, I see that the downward camera image is rolling sideways. After checking a bunch of connectors, etc. with no change, I checked the video out of the camera by BNC, and then probed the camera cable.
I figured out the controller sends out sync and expects the camera to provide video synch'ed to that signal, but the camera is not synching to it. I took the camera apart, it has a bunch of small boards in it with tiny connectors. I reseated all the connectors and cycled all the switches, but it made no difference. So, I had to order a replacement camera from the Philipines.
Maybe that will be the last breakdown? I hope!
Jon
-
Bugger!
At least at the end of it though you will have spent less money than a neoden 4 and have a much much better machine as well as know it inside out allowing you to quickly deal with breakdowns in the future if needed.
-
Bugger!
At least at the end of it though you will have spent less money than a neoden 4 and have a much much better machine as well as know it inside out allowing you to quickly deal with breakdowns in the future if needed.
Well, I sure hope so. The list of things that WERE working at first but then stopped is starting to scare me. Some just required connectors to be reseated, but there are now two major modules that outright failed after initially working fine.
I do note a heck of a lot of pure tin connectors in this thing, which should have been gold on a $100K machine.
I can expect some issues with these connectors from time to time.
Jon
-
Well, I got the camera in from the Philippines, and it worked and solved the problem. So, I took it apart and was able to trace the open PCB trace that caused the original camera to not sync. I put in a wire to bring the signal from the camera connector to a board-to-board connector, put it all together again and the old camera now works. Too bad I had to spend $150 to buy a camera JUST to trace out a wire that had gone open, but now I do have a spare.
I spent the day tinkering with vision fiducial recognition, and made a bit of sense of the procedure. Still a lot to learn before I can be placing parts.
Jon
-
If it's not too much to ask I'd love to see more photos of these parts you're working on. It's always good to see how things are done and compare with my new Kayo machine :)
-
If it's not too much to ask I'd love to see more photos of these parts you're working on. It's always good to see how things are done and compare with my new Kayo machine :)
Do you mean photos of the boards I make? There are a fair number of photos on pico-systems.com, just click on the web store link at top of the page.
So, I'm still learning about the machine. A number of my older boards don't have traditional fiducials on them, I have been using mounting holes for fiducials, and that has worked fin on my older Philips machine. The software on the Quad has an option for "polarity=black" fiducials, but just doesn't work worth a darn on the big holes (~ 3 mm ID). The centroid it detected would be +/- over 1 mm off. I tried to pick up the OD of the plated-through mounting hole (6mm) and that didn't register at all. So, in desperation, I tried to pick up a connector PTH ID, and it worked fine. The only issue is the connectors have a grid of holes, and depending on how accurately the conveyor and stops fix the board, it could maybe pick up the WRONG hole.
I will have to tinker with the options to make sure it either ID's the right hole or rejects the fiducial. It has plenty of settings to do that.
I also tried a board with regular (positive or "white") fiducials, and it seemed to do really well with those. So, the boards with those should be just fine.
Next step is to make up a dummy program that mounts one resistor onto double-sticky tape on a board, and see how well that all works.
Another option for the non-fiducial boards is to use the beam sensor (very old-school, basically a one-pixel camera that is dragged around by the head to locate fiducials and bad marks). Not sure if you have to hand-edit the configuration file or you can select between vision and beam sensor with a GUI option. I need to try this out next time I have the machine on. My old Philips just had the beam sensor.
Jon
-
Do you mean photos of the boards I make? There are a fair number of photos on pico-systems.com, just click on the web store link at top of the page.
I was referring to the parts of the machine you are working on, but it's good to see some examples of what you're making too ;)
It's pretty standard for CMs in China to not even use specific fiducials - even the user manuals for our Kayo machines simply show using SMT component pads. Still, for the best automation through out entire process (including the automatic stencil printer) we're using proper 1mm round fiducials.
-
I was referring to the parts of the machine you are working on, but it's good to see some examples of what you're making too ;)
OK, I really don't have any pictures of the specific parts. I had to replace a VME circuit board that is the interface between the VME conputer (68040) and the ordinary PC motherboard. It also has the safety logic on it which was bad in some way. Somebody replaced the entire head # 3 with one from an earlier version of the machine. The old ones had standard 2-phase steppers, the new one has a 5-phase stepper. So, the rotation stepper on that head didn't work at all. I looked at the part numbers and immediately had a strong suspicion what was wrong. But, changing the motor is quite a detailed operation. in the process, i found out that somebody had really messed up the balll spline that transmits rotation while allowing the Z rod to move up and down. They lost at least 20 balls out of it. I've reloaded it, but it is still a little stiff. Probably the plastic ball carrier has been damaged.
Then, the down camera would no longer sync to the vision system. It turned out to be an open PCB trace inside the camera, but I could not access both sides of the board without desoldering a BUNCH of connections, so I ordered another camera. I was able to ring out the connection on the new camera, and then patched the old one with a jumper wire.
It's pretty standard for CMs in China to not even use specific fiducials - even the user manuals for our Kayo machines simply show using SMT component pads. Still, for the best automation through out entire process (including the automatic stencil printer) we're using proper 1mm round fiducials.
Yes, for low-density stuff like 0805 passives and SOIC, the fixturing is usually repeatable enough to be fine. But, I am going to be working my way up to 0.5 mm TQFP parts, so I have to get precise fiducial pickup working. Those boards DO have good fiducials, though.
One advantage of the down camera is I can actually WATCH the fiducial capture and abort if it doesn't look good. With the beam sensor, I just ahve to trust it.
You can't use SMT component pads becuase they will have solder paste on them.
Jon
-
I'm making EXTREMELY slow progress, mostly because the manuals are ASTONISHINGLY awful! I have figured out how to teach board coordinates and found that large mounting holes don't work for fiducials, but small holes do (the first board I'm trying doesn't have "real" fiducials.)
But, I don't want to teach the board. So, now I've started trying to make ASCII import work. I found it wants tabs for field separators, but without any guide I'm trying to construct a file to import. One issue is the "placement origin". I see in the manual how to specify "PF-1" and then the first fiducial location becomes the origin for the board. But, I don't want to do that! I want the origin of the board to be the lower left corner, and then the CAD system generates all the coordinates relative to that. Does anybody know how to set it up like that? Does anyone have a sample .qsa file they can share?
These files should be quite similar for Quad QSA-30, Samsung CP-30 and CP-40. If you put Samsung in a search string, all you get is cell phone hits.
Thanks for any help anyone can offer,
Jon
-
Have you looked at PCB Synergy? It converts placement files for many different machines.
I couldn't get it to do what I expected for my Juki, but I do need to spend more time with it as I can see that it should make life much easier than using the Juki program.
-
Have you looked at PCB Synergy? It converts placement files for many different machines.
I couldn't get it to do what I expected for my Juki, but I do need to spend more time with it as I can see that it should make life much easier than using the Juki program.
I think this should be the answer, clearly quite a lot of work has gone into it. Arguably tho' it is designed very much inline with how one guys mind works and how he runs/ran would run an SMT line. For instance I've had a fiddle and its almost completely useless to me. I don't start with design files and I don't want a program that allocates parts to feeders. So what I use is Essemtecs built in tools, Libre Calc (for formatting mostly), SMT-Maestro and a centroid file, or if pushed a gerber-> centroid converter.
-
jmelson, are the pneumatic switches installed in the head or are they in the base station of the machine?
Our pneumatic switches are currently in the head, but I think I'll migrate them into the base station.
-
jmelson, are the pneumatic switches installed in the head or are they in the base station of the machine?
Our pneumatic switches are currently in the head, but I think I'll migrate them into the base station.
There are 6 solenoid valves, 3 vacuum generators and 3 vacuum sensors in the head. I think the idea is to keep the tubes from vacuum generator to nozzle as short as possible so the vacuum comes on and stops quickly. The 6 valves are for vacuum on/off and blow on/off for each of the 3 nozzles.
Jon
-
Every nozzle has its own vacuum generator? Does anyone know is that a common thing?
I'm sharing one vac generator for 2 nozzles at the moment, the solenoids are controlled with 24v.
Oh by the way could you make a photo of those modules?
-
Every nozzle has its own vacuum generator? Does anyone know is that a common thing?
I'm sharing one vac generator for 2 nozzles at the moment, the solenoids are controlled with 24v.
My Philips CSM84 (made by Yamaha) did exactly the same thing.
Oh by the way could you make a photo of those modules?
Umm, kind of hard to do, they are buried in a big jumble of hoses and wires. I'll see what I can do.
Jon
-
Further updates. I finally figured out how the machine is set up, the corner that the board locates to is the front-right corner, so all the coordinates grow positive from there. (I think the prior owners changed something in the machine setup to do this.) Anyway, now that I finally understand the coordinate system, I was able to program a board with board dimensions, two fiducials and a component, and get it to place a part. The placment wasn't horribly bad, but consistently off to one side. So, I figured out that I could apply spotting dye to a nozzle, drive that down onto the feeder rail so that it left a ring of dye, note coordinates and then drive the camera directly over the spot and note the coordinates. I then used the difference of those numbers to adjust the camera and nozzle offsets, and ran the placement program again. This time, the part was placed EXACTLY dead center on the component pads!
Now, I need to make my CAD -> P&P converter program do all these manipulations on the pick and place file to get the numbers the way the machine wants them. I'm getting real close.
Jon
-
So, the ASCII import doesn't work. It reads XY rotation part number, etc into a screen, but when it gets imported to the "step program" all rotations are zero.
CAD import is a different function, it DOES bring in the rotation. But, I get a bunch of error messages. The "optimize" button reassigns parts to different steps, and to nozzles and heads and feeder locations. But, it seems to scramble and duplicate some of the parts.
This is just getting to be such a mess -- 2 whole months so far and still not a single board placed.
Jon
-
Why not hand teach the board? Sounds quicker.
I heard Quads are good beginner machines but personally I prefer Samsungs. Since Quad is no longer in business I wouldn't recommend them even though they use Samsung hardware. I believe Samsungs are the best starter machines, good hardware and simple software. Easier to find support, Quads seem like everyone is out to screw you and since they didn't design the hardware I don't feel as confident in their product.
-
Why not hand teach the board? Sounds quicker.
The accuracy and the correctness of teaching a board with 800 components sounds really iffy! I've had my Philips CSM84 for 13 years, and have NEVER taught a board. The VERY FIRST board I did, I hand-edited the CAD system's placement file into the format needed by the Philips, downloaded it and ran the board.
Then, I wrote a small C program to convert the file format. This is what I'm working on now for the QSA. I have over 15 standard boards that I make, and having to hand-teach all of the placements would be an insane horror.
Actually, I could just enter all the numbers from a printout, but it just seems incredibly STUPID to have a bunch of computers and have to HAND-ENTER data from one to the other. This is just crazy that this stuff doesn't work. They have this incredibly "good looking" software to import ASCII files with the placement data, but it seems to be horribly buggy. I'm seriously thinking that if this is as good as it gets, I may just scrap this machine. This will be a GIANT step backwards if this import function just can't be made to work.
I heard Quads are good beginner machines but personally I prefer Samsungs. Since Quad is no longer in business I wouldn't recommend them even though they use Samsung hardware. I believe Samsungs are the best starter machines, good hardware and simple software. Easier to find support, Quads seem like everyone is out to screw you and since they didn't design the hardware I don't feel as confident in their product.
Well, the Quad QSA-30A is actually a pretty standard Samsung CP30 modified to take Quad feeders instead of Samsung. Instead of a rack of air cylinders to advance the all-mechanical Samsung feeders, the QSA-30 supplies power to the Quad feeders, and when the nozzle drops down and raises up with the part, the Quad electronic feeder advances the tape to the next part. Otherwise, it is VERY similar to a Samsung, and actually uses exactly the same software. There is an option in the setup file that tells it what machine it is running on.
Jon
-
It has been suggested before, but just how different is the file format for a Samsung CP40 to a CP30? PCBSynergy http://members.iinet.net.au/~sarason/ (http://members.iinet.net.au/~sarason/) supports that format and maybe even the author could add your quad as a supported format if you gave him a simple of a simple taught PCB.
I've never been near one but is it possible the import functions don't work correctly until they have a library of parts (so like a broken relational database) were there any files left on the machine from its previous life you can snoop at, or even files they think they deleted but didn't...?
-
Why not hand teach the board? Sounds quicker.
The accuracy and the correctness of teaching a board with 800 components sounds really iffy! I've had my Philips CSM84 for 13 years, and have NEVER taught a board. The VERY FIRST board I did, I hand-edited the CAD system's placement file into the format needed by the Philips, downloaded it and ran the board.
Then, I wrote a small C program to convert the file format. This is what I'm working on now for the QSA. I have over 15 standard boards that I make, and having to hand-teach all of the placements would be an insane horror.
Actually, I could just enter all the numbers from a printout, but it just seems incredibly STUPID to have a bunch of computers and have to HAND-ENTER data from one to the other. This is just crazy that this stuff doesn't work. They have this incredibly "good looking" software to import ASCII files with the placement data, but it seems to be horribly buggy. I'm seriously thinking that if this is as good as it gets, I may just scrap this machine. This will be a GIANT step backwards if this import function just can't be made to work.
I heard Quads are good beginner machines but personally I prefer Samsungs. Since Quad is no longer in business I wouldn't recommend them even though they use Samsung hardware. I believe Samsungs are the best starter machines, good hardware and simple software. Easier to find support, Quads seem like everyone is out to screw you and since they didn't design the hardware I don't feel as confident in their product.
Well, the Quad QSA-30A is actually a pretty standard Samsung CP30 modified to take Quad feeders instead of Samsung. Instead of a rack of air cylinders to advance the all-mechanical Samsung feeders, the QSA-30 supplies power to the Quad feeders, and when the nozzle drops down and raises up with the part, the Quad electronic feeder advances the tape to the next part. Otherwise, it is VERY similar to a Samsung, and actually uses exactly the same software. There is an option in the setup file that tells it what machine it is running on.
Jon
800 components is not too bad. We have hand taught 3000. I do agree pcbsynergy is a great tool. I have used for a Juki in the past.
Nice to know Quad uses Samsung software. I thought they had their own custom software. Personally I prefer mechanical feeders as they are cheaper and easier to maintain.
-
It has been suggested before, but just how different is the file format for a Samsung CP40 to a CP30? PCBSynergy http://members.iinet.net.au/~sarason/ (http://members.iinet.net.au/~sarason/) supports that format and maybe even the author could add your quad as a supported format if you gave him a simple of a simple taught PCB.
I've never been near one but is it possible the import functions don't work correctly until they have a library of parts (so like a broken relational database) were there any files left on the machine from its previous life you can snoop at, or even files they think they deleted but didn't...?
I think the format should be identical. But, the horrible issue is the import program in the QSA-30 has some real problems. ASCII import throws away all the rotation info. CAD import handles that properly, but you have to manually enter the fiducial coordinates. That isn't so bad, just two points.
I have set up the library parts for this first board, that was quite easy.
There are THOUSANDS of files on the machine, but they are pretty inscrutable. All binary, although the export feature will pull some of the info out and make it human-readable.
I'm still having trouble figuring out how to deal with the automatic nozzle changer. I can program it so it can put the nozzles in the changer and then get them back out, when I do it manually. But, when you try to run a program, if the nozzles are in the changer, it gives one message, if the nozzles are on the head and the system KNOWS that, it gives another message.
Jon
-
800 components is not too bad. We have hand taught 3000. I do agree pcbsynergy is a great tool. I have used for a Juki in the past.
Nice to know Quad uses Samsung software. I thought they had their own custom software. Personally I prefer mechanical feeders as they are cheaper and easier to maintain.
Yes, I was pretty comfortable with my Philips (Yamaha) feeders, and could thread them in under a minute. The Samsung mechanical feeders were quite similar.
The Quad electronic feeders are quite a bit different, and I'm still getting used to loading them and getting them set up right.
I did have a lot of jams, failure to peel the cover tape and such stuff with the Philips feeders. I'm hoping the Quad feeders do better in that regard.
Jon
-
OK, more progress! I still find the CAD -> P&P import process quite a bit more cumbersome that I think it ought to be. But, I have now loaded a data file with designator, X/Y/R and part type into the machine, and with a LOT of hand fiddling, got it loaded in. They have an import function that brings in the raw data only. Then, there's an "optimizer" that rescrambles everything to try to maximize assembly rate. It reassigns tape feeders and nozzles to the parts found in the imported data. There are a bunch of options in the optimizer, and I'm pretty sure I'm not making the right settings there. But, by hand, mostly, I got it all assigned. I don't know if the machine is scrambling the file, if the CPU has bad memory, or it is just me clicking on the wrong things that is causing the scrambling.
But, through all that, and a number of other issues, I did manage to get 14 parts put onto double-stick tape on a test board. One of the QuadAlign cameras has some debris in it that causes alignment issues, I'm going to have to pull that out again and try to clean it. And, now, I have to get some more nozzles.
But, I'm getting really close, here!
Jon
-
Nice work, it is very satisfying to get these machines to work and pump panels out faster than the reflow oven can bake them, especially with little to no help!
Next thing I need to figure out is if I can copy feeder info from one job to the next. That will save a whole heap of manual setting up when setting up a new job.
-
Nice work, it is very satisfying to get these machines to work and pump panels out faster than the reflow oven can bake them, especially with little to no help!
Next thing I need to figure out is if I can copy feeder info from one job to the next. That will save a whole heap of manual setting up when setting up a new job.
Yes, I am a bit disappointed that this is taking so MUCH fiddling. On my Philips, I could compose the complete placement file on my desktop computer and just download it to the machine. On the Quad, it seems I will have to do a lot of hand-editing of the placement file (step program) after the optimizer has scrambled things. I only have the 3 nozzles that came with the machine, I have so far not found out how to tell the optimizer that that is all I have. Maybe as I gain more familiarity, I will get it down to a routine.
I still have some kind of dirt on one of the QuadAlign cameras, and the 3rd head that was extensively messed with by the previous owner still have a very "gritty" ball spline. I may have to swap out that whole head to take care of those issues.
But, what it assembled a few days ago looked to be SO MUCH more accurate than my old Philips that I'm getting pretty excited about doing some more challenging boards. Anyway, I'm going to try an actual production run this weekend with solder paste and reflow, and see how it goes.
Jon
-
Despite some lingering issues, I populated the back side of 20 boards today. There were just 2 part types there, an 0805 capacitor and an SOD-323 diode.
I did have a few mis-picks where the cover tape on the 0.1 uF 0805 caps didn't peel well enough. This specific part (from a variety of manufacturers) has ALWAYS given me fits on my old machine with the cover tape not peeling and causing the feeder to get stuck. But, it seems the Quad feeders did better with it. I may need some new rollers with more grip on them.
The fiducial camera seems to be blinking, I think the connector board on the back of the camera has another failing connection. I'm going to swap out that board shortly. Now, I have to learn how to program and teach the SOIC parts on the vibratory feeder, so I can do the front side of the board.
Jon
-
So, over the weekend I did a bunch of the front sides of those boards until I ran out of parts. I had a major fit trying to get the QuadAlign vision system to accept the SOIC-16 parts, although it accepted the SOIC-8 parts with the factory stock settings. I still don't understand what was going on there, but I just kept fiddling with parameters until it accepted the parts.
My old Philips CSM84 was certainly fast enough for me, this machine, running at 50% speed, is about twice as fast as the Philips, even when the Philips didn't have to go use the mechanical alignment station.
I still don't have the automatic nozzle changer working, it is clearly a software setup issue. Every time I start the system, I have to reset all the ANC settings and hole positions, and then it forgets them the next time.
The fiducial camera flaked out, but by swapping boards with a spare, I was able to get it running. Hopefully, I have gotten past the issue with boards going bad. I need to get a few more feeders and make some repairs on the ones I have, to have enough for the next boards I will be doing.
As for teaching boards, I won't have to do that. I have now developed a flow from the CAD system, through a C program I wrote, into the CAD import function of the QSA-30, but then NOT using their built-in "optimizer" which totally scrambles the placement data. It takes some additional hand editing of the placement data to assign parts to feeders, etc. but it works. I have to manually enter the board dimensions and fiducial location and settings, but then all the placement data is brought in from the CAD system. (correctly!)
I'm still getting used to the Quad feeders. At first, I was thinking they were going to take a LOT longer to thread and set up, but it now looks like it only takes a little longer than simple mechanical feeders. And, I had a LOT of trouble with my old Philips/Yamaha feeders not being able to peel the cover tape reliably. Once threaded, the Quad feeders seem to have enough pull to peel the tape reliably. This might actually be a BIG improvement that I was not expecting. On the old Philips, I had to constantly be leaning in there to mess with cover tapes and unjam them.
I still have to make a waffle tray holder or find an affordable one. And, I still need to calibrate the QuadAlign cameras. Placement is already better than my Philips most of the time, but I know that I have knocked these out of alignment when I disassembled things.
Just keeping the list updated on progress.
Jon
-
Keep the updates coming - they're good to read! Your progress thus far is great :)
-
So, I got the front side of 20 boards done, and am now moving on to the second board to assemble. This one has a bunch of different parts
(crystals, opto-couplers, quad flat packs, big capacitors and inductors, etc.) so I had to program the shapes of these for the alignment system.
Then, I had problems getting the system to recognize the fiducials. Not sure what is going on there, but it probably needs some parameters to be tweaked.
Then, I have to buy some more 8mm feeders and build a holder for the waffle tray. These seem to be unobtanium.
Jon
-
Wow, it just keeps getting crazier. I was having a LOT of difficulty getting fiducials to be recognized reliably. I did get 20 boards done, and started programming the next board. I just couldn't get it to recognize fiducials better than about 5% of the time. I tinkered with parameters, cleaned the fiducials, changed the camera aperture, tweaked the focus, nothing seemed to help. I finally pulled the DSP board that does the vision processing, and found seven bad solder joints on one of the FPGAs! I couldn't believe it. I fixed them, and now the machine recognizes fiducials well off-center and puts the recognition crosshairs perfectly on center, 100% of the time.
This is pretty crazy, as the machine had apparently been that way since it was built.
Well, I'm taking small steps, but things are getting fixed a little at a time.
Jon
-
That's crazy! Surely it would have been working properly for the initial stages of its life? I'm guessing the poor solder joints were OK until the machine got trucked across the country to you.
There's a really high level of satisfaction watching a machine recognise fiducials :)
-
That would have been frustrating!
I have had some problems, but that was all due to HASL and the blob left on the fiducial.
Solderwick and some fine sandpaper fixed that for me. Note to self, all boards to be ENIG in the future!
-
That's crazy! Surely it would have been working properly for the initial stages of its life? I'm guessing the poor solder joints were OK until the machine got trucked across the country to you.
There's a really high level of satisfaction watching a machine recognise fiducials :)
Yeah, crazy is right! No, the only way it would have worked is for the lead to make contact with the solder lump on the pad. My experience is that can work fine for a functional test right after reflow, but after a couple weeks a little oxide develops and the contact goes flaky. I've seen this lots of times on the equipment I manufacture. Now, there were a few leads that still had weak solder attachment, but I could tell, looking at it under a microscope, that at least five of them NEVER had any attachment to the pad.
This machine may be a Frankenstein combination of parts from several machines, so they might have put in a board that was in sombody's used/unknown condition stack, and it worked well enough that they moved on to other issues. The big killer was that they swapped an entire head from an incompatible machine version and didn't know why it woudln't home. It didn't take me too long to figure out they'd put a 2-phase stepper motor where a 5-phase motor needed to be.
I'm somewhat concerned that two (other) boards went bad AFTER I got the machine. A VME interface board stopped allowing the machine to go to the "ready" state, and the fiducial camera had two failures. I'm guessing that after the last ill-fated "repairs" on the machine in 2014, it was pushed into an un-conditioned space in Austin TX and absorbed a LOT of moisture over 6 years. (I know when it was last used or attempted to be used from the file time stamps on the disk.) I hope these delayed failures settle down!
I'm making parts for a waffle tray holder to go on the back feeder rail. I need that for the next board to be made.
Jon
-
I made the back side of 26 boards today. I had a few small issues, one board needed to have the fiducials cleaned with a pencil eraser, and one of the feeders was not consistently pulling the cover tape, but that cleared up after a while, I guess the roller was a little slick at first.
Also, I was placing 2 0603 parts per board without the right nozzle, so some of them were going on a little tilted. Otherwise, it just hummed along.
I have mostly completed a waffle tray holder, picture attached.
I'm waiting for some parts from Digi-Key to do the front side of the board.
Jon
-
I did a dozen front sides of these boards today. I learned a bunch about how to set up a waffle tray with several different parts on it, and also set up the vision profiles for about a dozen different parts (1117 regulator, oxicap, inductor, TQFP44, quartz crystal, 4-pin opto-coupler, etc.) I am waiting on a loan of a calibration nozzle to calibrate the QuadAlign cameras. Until I get that done the placement is not as good as this machine should be able to do, but it is definitely coming along.
I got some experience with using the partial assembly feature of the Quad/Samsung to put on the missed parts ONLY. That was pretty cool!
Jon
-
I got the calibration tools from the guy who used to support these machines, and did the on-head alignment camera calibration, and then did 14 more boards, tweaking the camera alignment offsets until the placement of the QFP-44 chips were as good as I could get them. I would guess they were coming in with +/- 0.05 mm accuracy, or maybe just a little worse. Plenty good for a 0.8mm lead pitch. I will have to see how well it does on higher density parts.
I had a few feeder issues, I will have to go through these feeders and deal with rollers that don't grip anymore, etc. But, I'm getting the process down pretty well. The tray holder and setup for waffle trays works like a charm, now that I understand how to set it up.
Jon
-
I am jumping the wagon a bit late, but could not resist - your painfull path seems so similar to mine. Only I am dealing with Universal GSM's. However, it was well worth the effort. Alhough machines
I have run on OS/2 (yes, OS/2), software is actually pretty good. Most of the software issues you faced simply do not exist on GSM. ASCII file importer is extremely flexible and you can set-up
the board in a few minutes, if the parts used are already in machine's library. Also, UIC has pretty good web site with documentation and even some troubleshooting advices. They require
500$ sw. registration fee if you bought used machine, but it's well worth since you get access to that web site. I am running machines with 4-spindle Flex heads and now I am almost done with
bringing-up two-beam machine with 7-spindle heads. A lot of joy...
-
I am jumping the wagon a bit late, but could not resist - your painfull path seems so similar to mine. Only I am dealing with Universal GSM's. However, it was well worth the effort. Alhough machines
I have run on OS/2 (yes, OS/2), software is actually pretty good. Most of the software issues you faced simply do not exist on GSM. ASCII file importer is extremely flexible and you can set-up
the board in a few minutes, if the parts used are already in machine's library. Also, UIC has pretty good web site with documentation and even some troubleshooting advices. They require
500$ sw. registration fee if you bought used machine, but it's well worth since you get access to that web site. I am running machines with 4-spindle Flex heads and now I am almost done with
bringing-up two-beam machine with 7-spindle heads. A lot of joy...
The original hard drive died, somebody put in a replacement drive, which is ALSO 20 years old, and was not even bolted into the machine, just hanging by the cables! I was moderately amazed when the machine booted up! But, I think there could be software rot or a silently damaged sector there. One of the data import modes loses rotation data, and the "optimizer" that is supposed to order the placements for fastest assembly of a board totally scrambles the placement file. It is like a spreadsheet where it sorts on one column without moving the rest of the row with it.
At some point, I may try to build a new software system, with a new(er) OS and see if I can get a complete set of files for the machine software.
I did manage to clone the old, but working, hard drive to a new drive, and there were no errors reported in the process.
I am still learning how to use it, but have gotten two batches of boards done, so that is programs for 4 sides. It is getting easier. I am setting up my own parts library, and entering my own vision alignment settings, and they work! Without their optimizer, which really is supposed to do a lot of the data import task, setting up is quite cumbersome. And, I'm still dealing with some flaky feeders, and learning the ins and outs of those, too.
Jon
-
Well, I'm a convert!!! I was a bit skeptical about the Quad feeders. I have extensive experience with the old Yamaha/Philips feeders on my old CSM84. They needed a LOT of babysitting, as the cover tape peeling was marginal. Also, some tall parts in plastic tape would lift the hold-down plate and the sprocket would jump teeth.
Well, I had a few issues with the Quad feeders on my last batch of a board run, and fiddled with them a bit. I just ran 9 boards with 292 parts/board, and had ZERO pickup errors for the entire run. That's a total of almost 3000 parts. So, I am now a believer that these Quad feeders are a huge improvement over the Yamaha all-mechanical feeders that advance ther tape on the return stroke of the plunger that actuates them.
The only issue I've seen with the Quad feeders is if the tip of the plastic tape catches on an edge in the track inside the feeder, the drive sprocket chews the tape up pretty badly. Maybe snipping the corner of the tape would prevent it from catching, I will have to try that.
As for speed, the machine did 292 parts in 4 minutes 17 seconds, so that is about 4100 parts/hour. Plenty good enough for me!
As for the "bit rot" that I was worried about, it seems that if you set everything up properly (nozzle assignment, part profiles, feeders, etc. ) before running the optimizer, it DOES work correctly.
Jon
-
Great news Jon and congratulations! I can imagine the satisfaction in getting to this point!
Interesting to hear about the quad feeders. The Yamaha CL feeders, while ubiquitous, do have their quirks and over the past 6 months we've learned all of the tricks to help ensure they run smoothly. As you mention, the biggest issue is the various conditions which allow the sprocket to misalign.
4100 CPH is a nice speed. It's the average of what we achieve on our new Kayo 6 head machine and while of course faster is always better it's proving to be an adequate speed.
-
Great news Jon and congratulations! I can imagine the satisfaction in getting to this point!
Interesting to hear about the quad feeders. The Yamaha CL feeders, while ubiquitous, do have their quirks and over the past 6 months we've learned all of the tricks to help ensure they run smoothly. As you mention, the biggest issue is the various conditions which allow the sprocket to misalign.
4100 CPH is a nice speed. It's the average of what we achieve on our new Kayo 6 head machine and while of course faster is always better it's proving to be an adequate speed.
Yup, the problem I had with the optimizer was just doing things in the wrong order. You have to define all the parts and vision profiles FIRST (should have been obvious, really!) before running the optimizer, since that searches the part database to link placements on the board with components and available nozzles. When it is done in the right order, then the optimizer assigns heads and nozzles to the parts and organizes the sequence.
Then, I had to replace the peel rollers on the feeders to get good pull of the cover tapes. After dealing with those issues, it all really started to run well!
It took 6 years or so to get really proficient with my Philips machine, so I'm doing pretty well already. As for the speed, I have been running the machine at 60%, that seems to be some kind of default setting. I have not played with that, but sometime I'll try turning it up just to see what it does. But, I may keep it at 60% just to save wear and tear on the gantry.
Jon
-
Let's see how long before you crank up the speed further. I know I wouldn't be able to help myself ;)
-
Placement speed is a curious fish and the such a moveable metric that promo literature and how it is measured is full of cheating even when they try and apply standards to it. I haven't fully optimised the feeder location on the current run on our Essemtec and the screen currently has a live cph on screen of 5400, this will drop off radically as it starts to place things like ICs at the end of the panel and so uses fewer nozzles and slower feeders. It's also a lie because it doesn't start that count until its located the PCB and loaded its first 4 nozzles, and then over a run you have the overhead of manually securing each panel as its standalone. Even with a very densely populated board and parts loaded in feeders very close to the camera, I have never managed to push that number past 6200 and yet this machine has a theoretical speed of 15k and an IPC one of somewhere between 9k and 10k. I lose up to 25% by not having the fastest feeders, I will grant you that but that still leaves me some way off even its "IPC" rate in a best case scenario.
Luckily 4-6k is still a really good speed for a lot of product types and certainly radically faster than conventional assembly, where it doesn't work is for any kind of volume where a client might start expecting to see a low price. Its also a window a lot of machines seem to operate in and was IIRC a number a lot of entry level commercial machines hit in the real world about 10 years ago. These days we seem to have 6heads plus even on the entry level specs and a headline rate about 2-4 times higher than it was back then.
Interesting that 60% seems to be some kind of default, I'd be suprised if cranking it up caused any issues unless there was a physical issue it was hiding.
-
Yup, my Quad QSA-30A has an advertised speed of 17K PPH, which now that I've seen it run is obviously a prepared special case to get the highest number possible. (I have seen videos of some insane machines with a wheel of nozzles. It goes to the feeder area and picks up 20 parts, a camera at the top of the wheel checks the rotation and centering of the parts, and then it goes to the board and places the 20 parts. it places 20 parts in about 2-3 seconds, from pick-up to
placement. That was QUITE impressive, but such a machine is a chip shooter, and can't handle a wide range of part sizes.)
Anyway, my guess is turning up the speed on a small board with all the feeders next to it will have very little effect on speed. It is all in the acceleration. If you are placing huge panels with feeders far from the board, then it might make a difference.
Jon
-
Literally a wheel is a Universal Lightning head I believe and its even 30 parts at a time, you can have a machine with one of those and a 7 at a time flex head tho http://www.uic.com/wp-content/uploads/2014/01/Fuzion_Platform_brochure.pdf. (http://www.uic.com/wp-content/uploads/2014/01/Fuzion_Platform_brochure.pdf.) But its just as big and expensive as two machines...
Plenty of others have a rotary turret ASM and Europlacer manage to do that and be flexible, although its 12parts at a time and a fair bit slower (still impressive to watch). Interestingly when Europlacer targeted higher speeds they dropped the turrets and went back to a row of nozzles, but even they admit its a compromise as each head is limited to 1/4 of the machine so you have more work to balance the parts.
The two sides seem to have endless arguments about which is better, a line of nozzles can gang pick but that in itself makes some huge assumptions about how well you can optimise your loadout or indeed how many reels of the same thing you want to buy.
-
So, the QSA-30 is now my only P&P machine (hope it doesn't get lonely.) I sold my old Philips CSM84 to another eevblog member and he picked it up last week. Lots of ruts in my back yard, but we got it out with some help from towing straps and 4WD trucks.
And, I now have a bunch more space in my shop.
Jon
-
That CSM84 sure have 9 life just like a cat!
-
That CSM84 sure have 9 life just like a cat!
Well, I think all the old-school P&P that were very well made just keep going. The Yamaha/Philips CSM series were built like tanks, and used very rugged consumables. I made my own nozzles on my lathe and mill, and made a few repiar parts when I broke the ends off the air pistons that drive the nozzles up and down.
I ran that machine sparsely over the 13 years I had it, but did make over 2000 boards on it. In that time, I had a sensor on the conveyor fail and the commutator of the rotation motor filled up with copper dust and started giving errors. That is the entire number of issues I had with the machine. Like I said, built like a tank!
The Quad QSA30 is also quite robust. I don't have long experience with it, it is WAY more complicated (8 motors, 4 cameras, 14 computers, etc.) but seems to have a lot of sophistication and self-checking built in. Now that I've experienced the Quad, I can see a lot
of things that the Philips kept hidden. For instance, when the Philips located fiducials, there was no way to know the quality of the location or what offset it was applying to the board origin. There is a test fiducial function on the Quad where it shows you the video with computer-drawn crosshairs so you can see that it has correctly registered the centroid and then gives the offset it will apply.
You can also do a "test pick" and it will pick a part from a feeder and report the X and Y size and X, Y and rotation alignment of the part. This tells you if it is picking up the part from the center of the tape pocket. Since centering on the Philips was with mechanical jaws, no such readout was possible. Recent experience shows the Quad feeders are much better at peeling cover tape than the Yamaha feeders, that was one constant issue that the feeders needed babysitting. But, the real reason I moved up from the Philips to the Quad was the QuadAlign camera system that uses line-scan cameras to center the components. The Philips centering jaws have a certain amount of unrepeatability, and seem to only handle parts with a certain aspect ratio. So, oblong parts like resistors and caps and SOIC chips were fine, but a square chip like a QFP or MSOP would not be aligned in both axes. the last board I did on the Philips had 13 parts I had to nudge into proper alignment under a microscope. This pushed me over the edge.
Jon
-
Well, the saga continues. There is something fouled up in the configuration files on this machine. There is a menu to set up the Automatic Nozzle Changer, you can set the location and what is in the nozzle pockets, and pick and put nozzles in and out of the
pockets just fine. But, when the machine goes to assemble a board, it goes to some preprogrammed position that I have no idea
where it comes from. VERY frustrating. Also, the machine doesn't seem to like to have the nozzles set up on the head by hand, it wants to pick them up itself. I have not found the config file that holds the ANC pocket positions. It took me TWO WHOLE DAYS fooling around to get it to where it would accept preloaded nozzles and just use them as is.
Jon
-
And, I may have cracked this issue. I had been copying my old pcb files to make new ones, and that apparently copies the ANC settings.
I created a new, empty pcb file and set up everything, ANC, feeders, parts, fiducials and placements, and it tried to grab the nozzles from
the right place. But, maybe the alignment wasn't perfect, and the nozzles jammed. But, I think if I just adjust the coordinates of the
ANC pockets a little, it is going to work. Having automatic nozzle change is going to expand what I can do. A board I need to run
in the near future has everything from 0603 to 20mm FPGAs on it. That is going to require the ANC.
So, I was able to go from no setup at all to running 17 panels of boards in about 6 hours. That was a big improvement from running into
all sorts of issues getting the setup to be accepted on the last board.
Jon
-
So, I was able to go from no setup at all to running 17 panels of boards in about 6 hours. That was a big improvement from running into
all sorts of issues getting the setup to be accepted on the last board.
Wow, well done! Great to read stories like this.
I've fixed up a few CNC machines, but I'm unlikely to take on a PnP requiring so much effort.
-
Wow, well done! Great to read stories like this.
I've fixed up a few CNC machines, but I'm unlikely to take on a PnP requiring so much effort.
Well, yes! Had I known that the machine had not been turned on for about 6 years before it came up for bid, or that it might have been stored in unconditioned space in Austin, TX for that time, I might not have placed my bid. I DID place a pretty rock-bottom bid, and got it for the opening bid price, ie. NOBODY else bid on it! That left me with a lot of budgeted $ to spend on repairs and accessories.
I was VERY comfortable with my old Philips CSM84, knew it inside and out, but it was a very simple machine without ANE, Z axis or vision. I had very few issues with it over 13 years. Well, little did I know how messy things could get, especially with a much more complex machine. The Philips had ONE CPU to run the whole thing, and no OS. The Quad/Samsung has, I think 14 CPUs of various types, and tons of software.
I think this machine was the best one of the low-end machines in that auction, careful examination of the photos indicated that a number of the other machines had been partially disassembled and had parts missing.
So, I have now done 4 different boards on it, 3 of those were double-sided. It is going to be a while before I am as comfortable with it as I was with the old Philips, but I will get there. This machine is so much more sophisticated than the Philips. I have yet to actually use the automatic nozzle changing. The vision alignment and programmable Z is a big plus. And, the ability to make it show me the results of the fiducial pickup and test pick alignment is very nice. Also, I now know how to do array boards (panels). The Philips could do it, but I never learned how.
Jon
-
I got this machine in August, 2020. I never got the ANC to work. One of the issues is you can set up the ANC and teach the hole locations in a GUI, and it all works perfectly. But, when you go to place parts on a board, it uses some OTHER data, hidden SOMEWHERE, for the ANC coordinates. I finally figured out a sequence to get that to work, but over several months, I have NEVER figured out how to change which air and sensor port the ANC is connected to!
So, I was forced to put the ANC back on the front feeder rail and hook it back to port #1. Now, the ANC acutally tries to work. But, head 3 jams instead on nicely putting or picking the nozzles. More investigation shows the "Z-rod" on head 3 is bent. Initially, I measured the TIR runout at .015", with some struggling I have got it down to .006". I think I'll try to work on it some more. Head 1 has less than .003" TIR. That would explain ANC issues as well as being unable to pick up an 0603 part. It would come out on the side of the nozzle.
Fixing that runout and recalibrating the head offset should help things a lot.
Well, it was really good to see the ANC work like it is supposed to. I made up a test program mounting 3 parts that needed a different nozzle for each. It didn't work on head 3 due to the runout, but when I reset it to use head 2, it worked perfectly! The top side of the board I'm working on now will have everything from 0603 to a 20 mm FPGA chip, so getting the ANC working was necessary.
Jon
-
Well, I had to use shade tree mechanic skills on a $100K machine. Last year I replaced the #3 head with an eBay head, due to the trashed ball spline on the head on the machine. It all seemed to be working OK with 0805 parts, but i was
having issues with 0603's. So, I had to apply massive force to bend the Z rod closer to straight. I managed to do it without breaking anything. I got the TIR runout down to .003", which was only a little worse than the other heads.
Then, it wouldn't pick up sonme 0603 caps. I did notice those caps were a bit thinner than the tape, so they sat BELOW the surface of the tape by a few thousandths. Also, when it DID manage to pick a part, it would blow it all over the board, meaning it wasn't planting the part. So, I tried changing the Z offset on that head, and now it picks and places the 0603's fine!
Also, while dealing with this yesterday, I got some dirt on the alignment camera, and it was giving crazy dimensions of the parts on that head. I had to pull the camera to clean it properly. Once I could SEE the speck of dirt, I could clean it off. (The software has a diagnostic page where it gives you what looks like a scope trace with X as linear dimension and Y as brightness of the linescan image. So, you can easily see when there's something wrong with the camera.)
Well, I hope this is the last of these issues for a WHILE!
Jon
-
Awesome work Jon, I really enjoy reading about this process and understanding the differences between these highly engineered "classical" machines and our modern Chinese "economy" machines :)
-
Awesome work Jon, I really enjoy reading about this process and understanding the differences between these highly engineered "classical" machines and our modern Chinese "economy" machines :)
Yeah, this has been quite a saga! The machine sold at auction for $500, but I've had to throw in an addtional $7000 or so for repair parts and then feeders, a replacement head and a few other things. I built my own waffle tray holder.
I've now completed 4 production runs with it, each time going to a more challenging board. Now that I have the ANC working, I can move up to more challenging stuff. I'm doing the back side of another board now, then I will do the front side, which is about as difficult as anything I expect to do. It has everything from 0603 passives up to 144-pin 20mm TQFP FPGAs with 0.5mm lead pitch, and MSOP10 chips. These are the ones that started the whole process of upgrading. My former Philips CSM84 could do a fair job with the big FPGA, but the centering jaws just did not work AT ALL on the MSOP10! It would put them randomly a whole lead pitch off, and there are 8 of them per board. A total of 14 ICs that had to be manually poked into alignment. And, of course, if you fail to properly realign one, it is a pretty big hassle to fix it after reflow.
Although my Philips could do panels, I never tried it, as there was no way to know if it was set up right.
The Quad has a test feature for nearly everything. Once you have entered the array X and Y spacing, number of repeats and the fiducials, You can load a board, and walk it through each board of the panel and view the (0,0) coordinate on the camera.
The QSA30 has actually quite nice software, once you understand it. The manuals are about the worst I've ever seen.
So, after 4 failed attempt so pick a part, it marks that feeder as skipped with a message, and then goes on to complete the rest of the board or panel. Then, at the end, you go in and fix whatever was wrong, like cover tape not peeling, and press start. It then does only the missing parts -- pure MAGIC!
It also has LOTS of online diagnostics, which has been a huge help. So, you can select a feeder and do a "test pick".
It picks the part, measures it with the alignment camera and gives a dialog box with the X and Y correction required, as well as the X and Y size of the part. So, you can verify the alignment camera is properly analyzing the part.
You can also view what the flying alignment sees at that point, so it gives you a "live image" of the part on the nozzle.
This flying vision is a line-scan camera at each nozzle, mounted on the head, so it gauges the XY size and alignment correction while the head travels to the board. You can also pick up and put away nozzles in the ANC, or tell the machine that you have manually placed a nozzle on a head. You can teach the position of the ANC pockets with the down-looking camera. You can also program the location to pick up a part from a feeder if it is not at the normal position, using the camera. As for fiducial recognition, you program the location and size/shape of the fiducials, load a board and you can then tell it to scan the fiducials. It shows the camera image on the screen and a black crosshair where it thinks the centroid is. So, you can tell if it is properly doing all this. This was invaluable when I was diagnosing missing solder joints in the DSP board. And, when I fixed that, it was immediately CLEAR that I had, in fact, fixed the problem.
It took some time to get it to work, but I found a way to download a "CAD import" file and then run the "optimizer" which figures out the best place for feeders and the best order to place parts, optimizing for head motion and nozzle changes. I wrote a C program that takes in the P&P file from my CAD package and outputs what the QSA needs.
So, this machine has X and Y servos, 3 steppers for head rotation and 3 servos for Z position. It has 4 cameras, 3 of the line-scan alignment type and one down-looking for teach and fiducial pickup. It has a PC for the GUI and program loading, a Motorola 68040 VME computer to run the motion system, and a total of about 14 computers, all the rest are single-chip micros or TI DSP processors.
This machine is VASTLY more complicated than my old Philips, or the Chinese machines. In 2000, they needed a distributed system to handle the workload, all the vision stuff going on, etc. But, I'm sure glad they did put in all that diagnostic and test functionality, it makes it a lot easier to be sure you are setting up the program right.
When I first got it, I thought the Quad electronic feeders were over the top complicated, and would be a lot of hassle to use, compared to the totally mechanical feeders on my old Philips. But, after running a couple boards, I see how good they are. The Philips/Yamaha feeders had LOTS of issues, mostly peeling cover tape, and the tape binding under the hold-down plate. Those didn't have a latch to keep the holdd-wn plate from rising up, and the advance was made by a spring after the head released the trip lever. So, I had to hover over the machine waiting for the inevitable feeder jam every couple minutes.
The Quad feeders are not quite so easy to thread, but since I replaced the rubbery peel rollers, they are VASTLY more reliable. I did a batch of boards in 2 runs, placing about 2600 parts with ZERO feeder errors in each run!
They also have electronically programmable pick-up position and parts pitch, that you set by pushing buttons.
Jon
-
Luckily 4-6k is still a really good speed for a lot of product types and certainly radically faster than conventional assembly,
In my world 4-6k is terrible. I went through alot of pick place machines before getting my Fuji Turret head. Everyone gave me a lot of jive about cph and unless its a turret head it wont come close to rated figure. Turret heads are generally within 10% of rated figures ~36k cph. Since then to now I have been very anti piano head as I like to call them. Supposedly the old phillips multiflex/assembleon ax series can do high rated cph figures. I was trying to find an old phillips to play with. My Fujis CP6's have been great. I wish I hadn't listened to everybody who pushed gsms/jukis/etc. Wasted alot of my time and energy trying to get near IPC cph figures. I do like Samsung machines as they seem pretty straight forward.
-
Luckily 4-6k is still a really good speed for a lot of product types and certainly radically faster than conventional assembly,
In my world 4-6k is terrible.
Well, I'm not in that world. I make batches of 20 - 60 boards at a time, usually. I have a converted toaster oven that can take 4 - 8 boards at a time. So, really, 4K CPH is pretty good, and can keep ahead of manual stencil printing and the batch oven.
Jon
-
Sorry I didn't mean to come off patronizing, just trying to contribute to discussion. I love pick place machines but have a hard time finding information about each brand/model without buying them and playing with them. I was trying to complain how CPH figures are seriously misleading especially with piano style head.
Very impressed with your Quad story. More because the model you got was a samsung. Somehow Samsung made units easier to work with 20 years later and easy to keep alive.
-
Sorry I didn't mean to come off patronizing, just trying to contribute to discussion. I love pick place machines but have a hard time finding information about each brand/model without buying them and playing with them. I was trying to complain how CPH figures are seriously misleading especially with piano style head.
Yes, this machine has a 17,000 CPH rating. Clearly a bunch of baloney on a job specially designed only to give the highest number possible. They do have synchro-pick, where the spacing of the heads is a multiple of the feeder rail spacing, so if the parts are put in the right feeder slots, it can pick 3 parts at the same time. I've set up a job to use that feature. It really doesn't speed thing up by a large margin. Oh, and for now, I'm running the machine at the defauly 60% rate. Sometime, I will have to see what it does at 100%.
Very impressed with your Quad story. More because the model you got was a samsung. Somehow Samsung made units easier to work with 20 years later and easy to keep alive.
Well, I did play with a Samsung CP45 with the mirror-flip bottom camera scheme at an auction. It looked quite nice, but was way more machine than I needed. Since the CP30/QSA30 is pretty old, now, the vultures will let parts go for cheap!
The manuals are absolute CRAP. Clearly, Quad just took the Samsung manual and edited in the pages relative to their feeders and other details. So, that's mostly on Samsung. The operations manual is basically a guide through their GUI, teling you what each on-screen button does, without telling you ANYTHING about how data is stored or interacts at all in the control software. There is a HUGE amount of interaction as data is entered, you MUST do the setup in the right sequence, or you get a total mess! They don't tell you ANY of this in the manuals. Believe it or not, they actually DON'T tell you how the XY coordinate system runs in the machine! In fact, the X+ X- labels in the machine are BACKWARDS, as represented in the board. So, the machine coordinate system has +X to the right, as one might expect. BUT, in the placement scheme, +X is to the LEFT!
But, with a HUGE amount of help from the guy at Quad who did the factory training, I've figured this stuff out and now know how to do a setup. ALL of the users of QSA30 and CP30 machines say they manually teach all placements.
I figured out in less than a week how to create a file for my old Philips machine from the CAD data, but it took me more like 3 weeks to nail down all these details. If you don't have component profiles entered first, it screws up the import of data. Due to the possibility of errors, I do NOT want to teach placement and manually enter all the data (part type, feeder, nozzle, etc.) when the machine CAN do it. And, now that I have it figured out, it does do it.
My Philips machine was delivered to me out of a working shop, and having never even seen a P&P machine, I had it building boards in a week. (The first-generation Yamaha/Phiips manuals were not great, the 3rd revision was really GOOD!) With all the breakdowns and crappy repairs on this machine, it took me over 2 months to do the first trial board. (It seems like it was a LOT longer that that, with all the ups and downs! I was close to scrapping the machine a few times, mostly over software/programming issues!)
Jon
-
IPC speeds were intended to cut down on the cheating of these numbers as it does apparently specify a mix of parts on a test board. However there seem to be subsets of the test for specific cases quoted by some manufacturers (they put a package size in brackets and things like that) that allow them to game the numbers. It is still a more useful number than the "optimum speed" BS you will also find, and the contrast between those two numbers also tells a tale.
Due to the way they work its much harder to game the numbers on a turret head, it picks one at time until it has a full turret and then it places them, there can never be a gang pick. Equally high speed turret heads are never flex heads and are basically exclusive to a chipshooter role placing things often 5mm high or even less. If you have an application that can exploit gang picking on a machine with 8+ heads tho' it could be virtually unbeatable but that's what... LED placement ?
Flex turrets like you find on a Europlacer or an ASM E-place (look at the difference between a CP14 and a CP12 head https://www.assemcorp.com/en/products/production/surface-mount/asm-siplace/urun/3701-e-by-siplace.html (https://www.assemcorp.com/en/products/production/surface-mount/asm-siplace/urun/3701-e-by-siplace.html)) are way slower, probably very little difference to many users between them and the inline heads.
-
OK, I definitely needed the most flexible machine I could get, I only have one. And, my old Philips was limited to 6.5 mm height. I forget, but I think the Quad QSA can go up to 12 mm height. That should cover anything I'd ever need.
But, I had to watch what I specified for the Philips, I could have easily designed in a tall electrolytic cap and been in trouble. Also, the Yamaha feeders on the Philips just couldn't handle tapes with anything taller than about 5 mm, it lifted the tape out of the sprocket. So far, I have not seen any of these issues with the Quad feeders.
Jon
-
Jon,
Send me a copy of a your original CAD file and what you think the output should look like,
I will endeavour to add it to PCBSynergy
My email is on the PCBSynergy web page.
sarason
-
Jon,
Send me a copy of a your original CAD file and what you think the output should look like,
I will endeavour to add it to PCBSynergy
My original CAD data is from Protel99 SE (with SP6). The Quad QSA30 (and apparently Samsung CP30 and CP40)
put the board origin at the lower right corner. So, I set that corner as the origin in Protel, and then all X coords are negative.
For the top side of the board, I just convert minus to plus. For the bottom, I have to offset the coords by the X dimension of the board, and mirror the rotations. But, I have a program that does all that, and it seems to work perfectly. The binary .pcb files on the QSA are an undocumented 64 - 96 K of inscrutable stuff.
Jon
-
I spent the weekend doing a bunch of calibrations and tests on the #3 head, which had been giving trouble, and got it to where it ran pretty reliably doing a test 3-piece assembly onto Scotch double-sided tape. Today, I ran 12 boards with 117 parts on the back side. I did a couple more tweaks to get optimum placement. It ran through 1404 parts with minimal issues after those last tweaks. And, the ANC is working perfectly, now!
Jon
-
Well, the saga continues. I have done a couple not very demanding boards with quite good results. Now, I'm doing the top side of the board that started this whole upgrade. It is a 4" square board with a mix of (for me) fine lead pitch devices. There a 144-lead FPGA and 8 MSOP10 chips with 0.5 mm pitch, and 5 SSOP24 chips with 0.65mm pitch, and an SSOP16 that I guess is also 0.65mm. My old Philips machine could not place any of these within half a lead pitch, so I had to put every board under a microscope and poke each chip into alignment. Thus, the need to upgrade.
So, I knew this was going to be a bit of a stretch for the Quad QSA30. Well, after just a little tuning, the FPGA is better than 90% accurate enough to require no touchhup. But, all the other fine pitch parts were getting thrown all over the board. I played around with the Z adjustments for the heads and for the individual component heights, but really wasn't fixing anything. But, while watching the assembly, I could see some inductors being forcefully tossed around.
So, I began twiddling the setting for that part, but it didn't help. Finally, I just told the machine to skip that part, and all the rest of the parts came out pretty good. not perfect, still had to poke some of them, but WAY better.
Examining the board, I could see depressions in the solder paste that seemed to indicate the parts were initially placed properly, but then got moved.
So, I'm thinking that my solder paste is kaput. It is way past the expiration date, but hasn't failed me until just now.
Those MSP10 pads don't have very much solder paste on them at all!
I'm going to try some water-soluble paste I have here that is much newer. I'll have to wash the boards after, but that will tell whether the paste has lost its stickiness.
Jon
-
Examining the board, I could see depressions in the solder paste that seemed to indicate the parts were initially placed properly, but then got moved.
So, I'm thinking that my solder paste is kaput. It is way past the expiration date, but hasn't failed me until just now.
Those MSP10 pads don't have very much solder paste on them at all!
So the part is basically sticking to the nozzle while the nozzle lifts again after placement? Possible causes:
- Blow-off isn't doing it's job. This might be an indication of a clogged nozzle, filter or sticky blow-off valve, or the regulator needs to give just a little more flow.
- The nozzle can be just dirty and sticky with solder paste.
Something else that could cause this:
- The placement height is too low and the board bounces back after placement, launching the part. This is an indication of bad machine calibration, nozzle length setting or wrong component height setting (most likely). I hear a loud *thunk* on each placement when something like this happens with our machine.
Were it not for the indentations then I would also check the camera recognition of the part location/rotation on the nozzle.
I use the slo-mo camera function of my phone to diagnose these sorts of problems.
I'm amazed that such a complex machine has difficulty with 0.5mm pitch components, let alone 0.65. Those are even placed accurately with our simple & old Dima Optimat.
-
So the part is basically sticking to the nozzle while the nozzle lifts again after placement?
No, I don't think that is the problem, as NOT placing the inductors allows the fine pitch ICs to be placed accurately.
Something else that could cause this:
- The placement height is too low and the board bounces back after placement, launching the part. This is an indication of bad machine calibration, nozzle length setting or wrong component height setting (most likely). I hear a loud *thunk* on each placement when something like this happens with our machine.
Yes, this seems to be the problem, but I don't think the calibrations and settings are very wrong. I did have some wrong settings a week ago, and I could easily see the board flex when that component is placed. Also, if the inductor ended up end-on to the nozzle, it would then plunge too deep and go flying several feet! (The alignment camera SHOULD have rejected the part in that case.) Now, I cannot see any flex when the parts are placed, and it doesn't click either.
Were it not for the indentations then I would also check the camera recognition of the part location/rotation on the nozzle.
Again, when the inductors are skipped, the parts are placed quite accurately. I think the paste is kaput, the fine pitch parts are not adhering to the board. Also, when the board leaves the machine, the conveyor slams it into the exit stop pretty hard. That's never been a problem before, but might contribute to parts moving.
I'm going to try different (newer) paste tomorrow. It is water soluble, so I'll have to wash the boards after reflow.
But, it will verify if the paste is the cause of the issue.
I use the slo-mo camera function of my phone to diagnose these sorts of problems.
I'm amazed that such a complex machine has difficulty with 0.5mm pitch components, let alone 0.65. Those are even placed accurately with our simple & old Dima Optimat.
The head is so big, it covers the whole board. The only way to see anything is to kneel down and look down the conveyor. Not the best place to see what is happening.
I don't think the pitch is the problem, just that the paste apertures for the fine pitch parts are REALLY small, to avoid solder bridges. So, that means the small, fine pitch parts have VERY LITTLE paste to hold the part in place. The MSOP10 have just 10 TINY dots of paste to adhere them. Any bounce or banging of the board will displace components. And, if the paste has separated and most of the flux has already been used up, it won't be sticky enough to hole the parts. I will find out soon.
Thanks for the comments,
Jon
-
And more saga! I was placing some boards, all was going well, then the machine started making HUGE banging noises, like a crash.
I was thinking maybe an encoder cable was going bad, so I tried to examine the issue by manually jogging around the area where it seemed to make the banging. It would NOT jerk or bang when jogging, only when under control of the placement program. Also, it would home smoothly. So, I thought "maybe it is software". So, I powered down and rebooted it, and the problem totally disappeared! I made about 8 more of those boards just fine. Kind of crazy stuff you run into on an old machine.
Jon
-
Well, more saga! I went to make another batch of boards I had made before, thought this is going to be simple, no programming needed. I powered up the machine and it would not home. I had had this problem before after not using the machine for a while, but the usual fixes didn't get it running. Then, I went in the back and saw an error code on the Y axis servo amp. I'd never checked this before when it wouldn't home, so I don't know if this was the issue in the past.
Anyway, I ordered a replacement servo amp from Korea, and put it in and it worked perfectly. I was worried that I might need to get the servo tuning parameters out of the old amp, but the new one had smooth and quiet response and all coordinates seemed to be right on.
Jon
-
Finally some good news. I had not run the machine for a LONG time, due to really slow sales. I am now running low on stuffed boards, so had to get going. After the recent fix of a servo amp, I made 25 back sides of my resolver converter board, and it all ran fine. I had to relearn a few things about running it. The last hard to get part will come next tuesday, then I'll set up and run the front side.
Jon
-
And, the saga continues! I was setting up to run a batch of my servo amp boards. When I fired the machine up, the X servo amp came up with the same symptoms as the Y amp did eariier this year, so I knew it would need to be replaced. That fixed it. The symptoms were violent banging moves rignt after startup, then some time later the amp would just not come up. Anyway, I have now almost completed 50 of these boards (both sides). And, I had an issue which had appeared before, where one specific part type would not be placed accurately, just tossed all over the board.
I could see perfectly-placed dents in the solder paste, so I knew the part was initially being properly placed, but it then must have stuck to the nozzle. Now, the pattern became clear! These parts were ones where I created a NEW part type, rather than just adding a new part to an existing part type. I figured out there must be some menu that I was not filling out when creating a type. "TYPE" refers to the most generic level, like "DIODE" or "SOIC". So, All i had to do was look, and there was a button "Edit Mount Data". This brings up a page that has all the vacuum/blow-off timings and Z motion speeds. I had never even LOOKED at that page before! I copied the settings from SOIC to the new type, and boom, it all worked perfectly!
Well, there was one more issue that I had seen before, a long time ago. My workflow is to make sure all parts are entered in the compoent library, then select all parts from the library to the parts list, then assign those parts to feeder locations. Then, import the placement file with "CAD IMPORT" that puts it in a spreadsheet-like format, with designator (like R101, Q12) and XY coords, and part type (like 2N2222 or 74HC123).
Then, you run the "optimizer" that assigns nozzles to the parts and optimizes the order of placement to minimize motion time. I had some small passives dropped on the board. I found that 2 parts had been scrambled in the placement spreadsheet. Instead of a part designator (R1 and R8) it had the same coordinate-like number 59.106 (which did not appear anywhere in the imported file) placed there. The XY coordinates were not the same as anywhere in the imported file. I had seen this sort of scrambling before, when I was first testing out the machine in late 2000. It is POSSIBLE that I did this by bobbling the mouse keys, it is awfully easy to drag and drop a cell into the spreadsheet. I will have to do more testing and carefully scan for botched designators in the future.
Oh, one other thing, I needed to place 27 different parts on this board/side, which is more that I have ever done before. All those feeders would not fit on the front rail, so I had to use the back rail. There was no feeder power on the back rail, so I had to rig a power supply for that. And, I assumed that you had to rotate the parts when picked from the back rail. But, I discovered the software automatically rotates the parts 180 degrees for you, so you program the rotation AS IF they were on the front rail! Cool!
Jon
-
And, I assumed that you had to rotate the parts when picked from the back rail. But, I discovered the software automatically rotates the parts 180 degrees for you, so you program the rotation AS IF they were on the front rail! Cool!
One of the frustrating aspects of the Chinese machines is that they don't have the ability to correct rotation for parts mounted on the rear.
-
Well, what's worse? Not having it automatically rotate the part, or not TELLING you it will do that for you? I don't THINK this feature is in the manual, but then I have written before that the Quad/Samsung manuals are really quite bad!
Jon
-
That's true, but before we invested in SMT equipment I had just assumed the machine would automatically account for it. It seems a trivial thing for the machine to do and potentially saves a lot of headache.
-
And, I assumed that you had to rotate the parts when picked from the back rail. But, I discovered the software automatically rotates the parts 180 degrees for you, so you program the rotation AS IF they were on the front rail! Cool!
One of the frustrating aspects of the Chinese machines is that they don't have the ability to correct rotation for parts mounted on the rear.
HWGC machine has this feature for the back feeder. I think the Kayo software is a bit behind in terms of feature compared to HWGC's in my opinion. But both boss did originate from the same company last time.
-
Annnd, more saga! I was assebling a batch of boards, and the first board went fine. Then, all of a sudden, it started rejecting one part. I had no idea what was going on, but a closer look at the error message indicated a bad size reading on the alignment camera. I did a "test pick" on that part, and got crazy dimensions, like 19mm on a 7mm part. I then ran the diagnostic on the alignment camera, and saw that there was a huge shadow at the right end of the image. This machine has QuadAlign line-scan cameras that look across the part with a laser line source and a linear CCD camera chip. The diagnostic give a screen that looks like a digital scope where horizontal is the line of pixels and vertical is the darkness of the shadow produced by the part. I tried cleaning the camera in place with no improvement, then pullled the camera off and cleaned it with Q-tips and alcohol. No better. Then, I took it off again and did a much more vigorous cleaning of the illuminator side of the camera assembly, and it knocked the dark spot down enough that the machine would run.
2 weeks later (after some parts came in) I set up to do the front side of the board and decided to check the camera again. The dark spot is TOTALLY gone! I NEVER expected that! Glad I didn't runs out and buy another camera from the brokers. I have NO IDEA what caused the spot to totally disappear, although there WAS some particulate dirt in both the sensor side and the illuminator side of the camera. But, I thought I got it ALL out of there in the last cleaning.
Well, keeping this machine running has been quite a saga, indeed.
Jon
-
2 weeks later (after some parts came in) I set up to do the front side of the board and decided to check the camera again. The dark spot is TOTALLY gone! I NEVER expected that! Glad I didn't runs out and buy another camera from the brokers. I have NO IDEA what caused the spot to totally disappear, although there WAS some particulate dirt in both the sensor side and the illuminator side of the camera. But, I thought I got it ALL out of there in the last cleaning.
Maybe it was some moisture?
-
2 weeks later (after some parts came in) I set up to do the front side of the board and decided to check the camera again. The dark spot is TOTALLY gone! I NEVER expected that! Glad I didn't run out and buy another camera from the brokers. I have NO IDEA what caused the spot to totally disappear, although there WAS some particulate dirt in both the sensor side and the illuminator side of the camera. But, I thought I got it ALL out of there in the last cleaning.
Maybe it was some moisture?
Yes, possibly I didn't wait long enough for the alcohol drops to evaporate. That seems lime a pretty good guess.
Jon
-
Well, I had not run the machine in a long time. Last weel I ran the back side of my resolver boards, went just great. Zero pick errors for 32 boards with 22 components each (5 different types), total of 704 placements. Then, this weekend I set up to do the front side. All setup went fine, then fired up to do the actual placement. The machine would not home, one of the rotation motors wasn't moving properly. Trying to jog it manually, I saw that it rotated the wrong way when I pressed the button. Well, obviously, the direction signal was not getting to the step driver. I tried reseating connectors there, and then it wouldn't move at ALL! I figured this had to be a connection issue, but swapped out the step driver, no change. Finally, after ohming out the cables and finding no connectivity on that drive's inputs, I spotted what I'd done. There is an interface board that feeds the differential drive to the step & dir inputs of the driver, and I had pulled on the connector enough to lift the header body up from the PC board enough so the pins didn't make contact. I pushed the header body back flush to the board, and it all started workig again! Yikes, I was thinking I might have a bad cable or board someplace that might be hard to diagnose.
So, got the batch of front sides all done yesterday.
Jon