Author Topic: Philips CSM-84 performance  (Read 5974 times)

0 Members and 1 Guest are viewing this topic.

Offline jmelsonTopic starter

  • Super Contributor
  • ***
  • Posts: 2758
  • Country: us
Philips CSM-84 performance
« on: February 24, 2019, 03:46:02 am »
Some months ago there was some discussion about performance of my CSM-84 P&P machine.  I just did a board that was very suited for this machine.  it had 230 parts, mostly in 8 mm tape, but 3 different ICs in tubes, on the vibratory feeder.  The machine recorded a mounting time of 239 seconds, and I'd guess about 10 seconds for conveyor transport and fiducial detection.  So, that is 230 parts /249 seconds, or 0.92 parts/second or 3325 components/hour.

Now, some of the things that slow the machine down are requiring more than one push to advance the feeder and using the mechanical alignment station for the larger parts.  This particular board did not need any of those, and it was able to pick up one passive on nozzle 1 every time it picked up a chip on nozzle 2, saving just a little motion.  This is admittedly about the best this machine can do.

Jon
 

Offline Reckless

  • Frequent Contributor
  • **
  • Posts: 367
  • Country: us
Re: Philips CSM-84 performance
« Reply #1 on: February 25, 2019, 05:44:50 am »
Thanks for posting this up.  It looks like CSM-84 was rated at 4800cph.  I have a sapphire which is rated for 25k cph but I haven't used it yet (need to get feeders).  I'm running Universal GSMs primarily and getting terrible CPH as I need to get more nozzles.
 

Offline jmelsonTopic starter

  • Super Contributor
  • ***
  • Posts: 2758
  • Country: us
Re: Philips CSM-84 performance
« Reply #2 on: February 25, 2019, 09:40:41 pm »
The CSM can use all three nozzles at once, and save some motion time between the feeders and the board.  I have the chuck jaws off nozzle 3 to accommodate FPGAs and other large parts, so those have to be aligned by the mechanical aligner on the back rail, and that adds about 7 seconds/part.  So, that doesn't actually speed anything up.  If you were just doing all passives, then you could put 3 sets of jaws on it and it could grab 3 parts every time it goes by the feeders.  That is probably needed to hit that 4800 CPH figure.  Also, grouping the components in nearby feeders would help.

The Sapphire can grab a bunch of parts at one time, but I don't have 3-phase power here.  All the Gem series machines need 3-phase power.  I don't know how much hacking it might take to get one of those to run off single-phase power.

I'm pretty happy with the speed of my CSM-84.

Jon
 

Offline Reckless

  • Frequent Contributor
  • **
  • Posts: 367
  • Country: us
Re: Philips CSM-84 performance
« Reply #3 on: February 26, 2019, 07:46:23 am »
What feeders are you using?
 

Offline jmelsonTopic starter

  • Super Contributor
  • ***
  • Posts: 2758
  • Country: us
Re: Philips CSM-84 performance
« Reply #4 on: February 26, 2019, 10:02:26 pm »
What feeders are you using?
I don't know.  They are Philips, quite possibly made by Yamaha, like the machine is.  The older machines had mechanical advance for the 8 and 12 mm feeders, the air piston that drives the nozzle down also pokes the advance lever on the feeder.  Pressing the advance lever retracts the cover blade, exposing the part.  Releasing the lever advances the tape and cover blade simultaneously.

The 16mm and wider feeders have an air cylinder in them, which is activated by an air valve operated by the same trip lever.  The air is fed through a hose that is inserted into a quick-connect valve fitting under the feeder rail.  It is a MASSIVE pain to disconnect these hoses every time a feeder needs to be pulled off to re-thread or otherwise fiddle with the tape.  So, I see why the later machines CSM /// and Gem series) went to air through the feeder rail alignment pins.

Other than the air activation thing, these feeders look VERY much like the later Yamaha/Philips feeders.  Almost all my feeders came with the machine, I did get a few more 12 and 16mm ones in a swap for spares.

Jon
 

Offline Reckless

  • Frequent Contributor
  • **
  • Posts: 367
  • Country: us
Re: Philips CSM-84 performance
« Reply #5 on: December 03, 2020, 03:24:40 pm »
I have some for universal for sale
 

Offline jmelsonTopic starter

  • Super Contributor
  • ***
  • Posts: 2758
  • Country: us
Re: Philips CSM-84 performance
« Reply #6 on: December 03, 2020, 04:31:31 pm »
I am also looking for vibratory feeders for my machine, can anyone suggest a good place to get them?
For a Philips/Yamaha machine, or some other?  There are people selling used vib feeders on eBay for various machines.  As long as the height of the unit is not too high for your machine, most of them could be used on other models.  Note that Quad feeders run off 24 V AC.  Most of the others run off regular AC power.  I have a Contact Systems feeder on my CSM84.  I got a Quad vib feeder for my (new to me) Quad QSA30, and it uses the same top plate as the Contact System.  I had made a bunch of custom feeder lanes for odd-sized parts that bolt to the side of the vib feeder top plate, so i could continue to use these as needed.

But, I'm actually moving away from vibratory feeders when I can get the parts on tape.  The vib feeders need a lot of fiddling, parts leapfrog other parts and end up upside down, etc.

Jon
 

Offline Mangozac

  • Frequent Contributor
  • **
  • Posts: 470
  • Country: au
Re: Philips CSM-84 performance
« Reply #7 on: December 04, 2020, 10:05:32 pm »
But, I'm actually moving away from vibratory feeders when I can get the parts on tape.  The vib feeders need a lot of fiddling, parts leapfrog other parts and end up upside down, etc.
Vibration feeders are for masochists! When we first bought our PnP I intended to do a few parts in tubes but now we avoid is as much as possible. About the only things we do now are SOIC microcontrollers and eeproms being used in very small quantities.
 

Offline SMTech

  • Frequent Contributor
  • **
  • Posts: 845
  • Country: gb
Re: Philips CSM-84 performance
« Reply #8 on: December 05, 2020, 12:06:31 am »
There are attempts at better ways of feeding tubes but they end up more expensive as you often need a custom machined bit for every package. Mydata shuffles (or did) instead of vibrates, some blow air down the tube, and Europlacer tips the parts down a shute with a belt at the bottom of it.
I hate tubes but very often, with lower volumes or price sensitivity a tube is all you can get readily. Common examples seem to be eeproms, PICs, maxim tranceivers (who also don't mark or colour code their tube ends or use very legible case markings HATE them). I find that SO14/16 seems to be the sweet spot with vibratory feeders, they seem to have just the right weight to feed first to last in a tube without being stubborn. TSSOPs are the pretty horrible, much fonder of flipping (especially if you've made a feeder slot by cutting the top of the feed end of the tube) because they are so thin and light, larger SOICs are just slow to move down tubes.
 

Offline Mangozac

  • Frequent Contributor
  • **
  • Posts: 470
  • Country: au
Re: Philips CSM-84 performance
« Reply #9 on: December 05, 2020, 12:05:22 pm »
TSSOPs are the pretty horrible, much fonder of flipping (especially if you've made a feeder slot by cutting the top of the feed end of the tube) because they are so thin and light, larger SOICs are just slow to move down tubes.
Don't tell me that - we're trying some TSSOPs in a tube for the first time on Monday 😬

I have been considering machining a "receiver" for each of the parts we do in tube but haven't had time to make it yet...
 

Offline 48X24X48X

  • Frequent Contributor
  • **
  • Posts: 509
  • Country: my
    • Rocket Scream
Re: Philips CSM-84 performance
« Reply #10 on: December 05, 2020, 03:59:21 pm »
I think vibration tubes are the easiest to work with among all and with minimal setup. Just tell the machine where to pick and it's height. I usually just cut out the upper half of the tube at one end with the length of slightly more than the IC length. Tried with TSSOP, SOIC and even 3x3mm QFN, they work just fine  The only thing I hate about vibration feeder is the amount of feeder slots it takes which is 6 on my HWGC machine. Other than that, if you can't afford some expensive IC in reels, vibration feeder is the way to go.

Offline Kinguru

  • Newbie
  • Posts: 1
  • Country: us
Re: Philips CSM-84 performance
« Reply #11 on: December 07, 2020, 03:57:47 pm »
Hi, you mentioned that there is a mechanical aligner in the CSM 84, can you please explain how to program a part for mechanical alignment? thanks
 

Offline jmelsonTopic starter

  • Super Contributor
  • ***
  • Posts: 2758
  • Country: us
Re: Philips CSM-84 performance
« Reply #12 on: December 08, 2020, 04:15:35 am »
Hi, you mentioned that there is a mechanical aligner in the CSM 84, can you please explain how to program a part for mechanical alignment? thanks
OK, you have a CSM84 or the equivalent yamaha model?  First, the head must be set to an IC head, I think they call it.  That means it has no alignment chuck on that head (nozzle).  The software will declare an error if not set that way.  So, if you will use the mechanical aligner, then you have to dedicate one whole nozzle to that for the entire board.

Then, in the feeder setup, you have a selection (use mech align) and 1 or 2.  The 1 or 2 selection determines if the mechanical aligner lifts up to align low profile parts or stays down to align thick parts.  The coordinates of the mech. aligner have to be set accurately for it to properly align larger parts.
The offsets for this device are in the MCH setup menu.  It gets quite complicated, as the part will be rotated after alignment, so you need to compensate for that to know which axis to adjust.

The mechanical aligner is a block at the right rear of the machine that has pairs of jaws that close on the component leads to center it on the nozzle.
I don't know how accurate and repeatable it can be, I know that mine is not all that great.  I get repeatability of maybe 0.25mm at best.

Did I actually answer the question you were asking?  This device is really "old school" and various centering cameras have to be a LOT better.

Jon
 

Offline Marina CSM

  • Newbie
  • Posts: 8
  • Country: it
Re: Philips CSM-84 performance
« Reply #13 on: January 01, 2022, 02:23:04 am »
Around 15 years ago I learned how to work on an Emerald, right after that I worked on a Philips CSM 66 for a few years, then I quit. Remembering how efficient and sturdy they were, I bought a Philips CSM 84 V III, but now I don't remember how it works. I'm very scared, it came with 6 original Philips manuals, and I hardly ever find what I'm looking for in those. When I learned how the Emerald works, I also had a very simplified user manual. really friendly. Could someone help me find it? Thank you all.
 

Offline jmelsonTopic starter

  • Super Contributor
  • ***
  • Posts: 2758
  • Country: us
Re: Philips CSM-84 performance
« Reply #14 on: January 01, 2022, 04:57:02 pm »
Around 15 years ago I learned how to work on an Emerald, right after that I worked on a Philips CSM 66 for a few years, then I quit. Remembering how efficient and sturdy they were, I bought a Philips CSM 84 V III, but now I don't remember how it works. I'm very scared, it came with 6 original Philips manuals, and I hardly ever find what I'm looking for in those. When I learned how the Emerald works, I also had a very simplified user manual. really friendly. Could someone help me find it? Thank you all.
Scared?  Well, try to calm down.  I have to assume the manuals you have are newer than the ones I had with my machine, which was the original CSM84, no vision, no /// (AC servos instead of brush DC).  The vison programming makes everything a lot more complicated.  On the CSM84 V ///, I think you only have vision on nozzle 3, and just the old-school chuck jaws on nozzles 1 and 2.  Since my machine didn't have vision, I just glanced at that part of the manual, but did observe there was a LOT to the vision programming.  You have to enter the dimensions of all the parts that will be vision-aligned into FIOS.  One thing on these old machines it seems is that the vision system is quite far removed from the rest of the programming system.
Jon
 

Offline Marina CSM

  • Newbie
  • Posts: 8
  • Country: it
Re: Philips CSM-84 performance
« Reply #15 on: January 02, 2022, 05:33:56 pm »
 Yes, we have the same machine. My CSM has only one monitor, I don't know if it can work with only one, or if I have to add another one, and what is the limitation of having only one. I started reading the manual, I got halfway there, and I still don't understand how to make a new PCB !!. I didn't understand if the camera works, maybe the machine is set up for 2 monitors .. Despite this, I remember that it was so easy to program it.  |O
 

Offline jmelsonTopic starter

  • Super Contributor
  • ***
  • Posts: 2758
  • Country: us
Re: Philips CSM-84 performance
« Reply #16 on: January 02, 2022, 06:09:56 pm »
Yes, we have the same machine. My CSM has only one monitor, I don't know if it can work with only one, or if I have to add another one, and what is the limitation of having only one. I started reading the manual, I got halfway there, and I still don't understand how to make a new PCB !!. I didn't understand if the camera works, maybe the machine is set up for 2 monitors .. Despite this, I remember that it was so easy to program it.  |O
The /// versions only have one monitor, the computer switches what is displayed as needed.  Yes, the old CSM84 with no vision was very easy to program.
When you add component vision, it gets a lot more complicated.  The placement file B<filename> is pretty similar, but for head 3 you have to provide the vision template name.  I don't quite know what that looks like, as I didn't have vision.
I wrote a small C program to take in the P&P file from my CAD system and convert it to the placement file that my CSM84 needed.  Does your machine have any files from the previous owner left in it?  You could read those to see the general format.  In the back of the operation manual it has a one page listing of the placement file with annotation of what the fields mean.  The file is human-readable ASCII text.  You can create a scheme (spreadsheet or computer program)
to create such a file automatically from your CAD system, or just hand-edit it.  You can edit the placement file at the machine, as well as the machine setup file <MCH> and feeder file <FDR>.  Be sure to back up the <FDR> file.
Jon
 

Offline Marina CSM

  • Newbie
  • Posts: 8
  • Country: it
Re: Philips CSM-84 performance
« Reply #17 on: January 02, 2022, 06:32:44 pm »
Thanks for your clarification. When I get back to work from my covid I will also look at what you suggested.
 

Offline Marina CSM

  • Newbie
  • Posts: 8
  • Country: it
Re: Philips CSM-84 performance
« Reply #18 on: January 03, 2022, 06:41:36 pm »
I couldn't wait and went to the P&P. We discovered why we didn't see the PCB images on the monitor, because this machine was equipped with 2 monitors! It came with a separate PC, which I have not unpacked yet, perhaps there is some program in it to convert the cad file.
 

Offline jmelsonTopic starter

  • Super Contributor
  • ***
  • Posts: 2758
  • Country: us
Re: Philips CSM-84 performance
« Reply #19 on: January 04, 2022, 05:12:47 pm »
OK, I thought the /// machines only had one monitor, but I never used one of that model.  My machine was an earlier version.

Jon
 
The following users thanked this post: Marina CSM

Offline Marina CSM

  • Newbie
  • Posts: 8
  • Country: it
Re: Philips CSM-84 performance
« Reply #20 on: January 10, 2022, 01:42:37 pm »
I think I understand how to make the machine run, but i didn't find where to set the speed, it moves realy slow, could you tell me how to change it?
 

Offline jmelsonTopic starter

  • Super Contributor
  • ***
  • Posts: 2758
  • Country: us
Re: Philips CSM-84 performance
« Reply #21 on: January 10, 2022, 04:21:03 pm »
I think I understand how to make the machine run, but i didn't find where to set the speed, it moves realy slow, could you tell me how to change it?
There is a key in the bottom right "foot" of the machine, near the DB25 connectors and run time meters.  This key allows the machine to be run with the guard doors up, but VERY slowly.  Turn the key counterclockwise to go to normal mode.
Then, on the screen, there is a speed number that can be set from 10-100% in 10% steps.  I now forget how you set this, but it is in the main page for the placement job.
Jon
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf