Pick and Place Machine TVM802A / TVM802B

[sergeil]

someone tried 0402 parts already ?

I've populated over 10 PCB's now, each with >100 0402 parts. The first 1-2 were a pain due to alignment tuning and protective film collector tuning, the rest have been without problems, to the point that I can let it populate all the 100nf caps and just come back a few mins later. You must make sure to use the right nozzles: my exp has been smallest nozzle for 0402 and 0603.

My biggest problem now is 1) not enough reel spaces 2) crappy tray setups 3) setting up orientations for new parts, especially expensive IC's that I don't like wasting. Due to the need to still place some parts by hands it's not a fully automated solution in our case, but did make placement a whole lot easier.

For best results
1) make sure both fiducials align
2) run step by step the first time you run it and  take the time to tune out each problem
3) don't use the 1/2 nozzle feature. Nozzle 1 *or* 2. I've had weird occasional 90deg rotations with the 1/2 feature and I'm at least 25% sure it's not my fault.

-s

[harry4516]

thanks for your experience, Sergeil.
That sounds good.

I have a little trouble with transistors in SOT-23.
I tried the smallest nozzle, maybe its too small.
When I move from the tape to the camera, then just before it enters the camera
the part sometimes  turns around, one was even flipping top/bottom.
It looks like that there is something magnetic which pulls on the part, but I'm not sure.

[ar__systems]

They just turn from vibration and vacuum being not strong enough. Also, move the pick up location as close to the peeling edge as possible. Parts can make a flip from the pick transport needle hitting on the tape.

I connected external vac source to mine. There is not much to do about vibration, other than not use speeds below 70%. Somehow the vibration is worse at lower speeds.

[sergeil]

Exactly what ar_systems said: make sure the pickup location is as close to the peel-off as possible. I had flipped sot23 and similar as well before that tuning.

Too small nozzles seem to have too little rotary friction with the part, meaning when it rotates the part doesn't quite rotate with the nozzle (e.g. picking up a LQFP64 with a small nozzle). Use the bigger rubber nozzles and those parts get placed much better.

[harry4516]

thanks for the explanation.
I can do that by reducing the X coordiate of the left stack a little bit, right ?
Currently I use the values from the CDrom (from the pictures).
I will test that with the single step fuction.

Should I replace the original vacuum pump by an external one ?
If yes, how to choose the right pump ?

[ar__systems]

Today I tried two things: first, I replaced the rubber crap in the peel reels with brass pads. Works much better now.

Second, I tried a new concept for the peeling mechanism. Please don't laugh, it is a prototype.   And it works like a charm.
So basically we have a motor with a spool on a arm that rotate on a pivot and thus can bring the spool closer or further from the feeder. As it moves further, the steel sheet enters the slot of the photo-interrupter which in turn activates the motor (not seen on the photo, but the black spool is sitting on it). The force with which the device pulls the cover tape is constant, and is set by the spring (rubber band in this prototype). White plastic spring prevents the black spool from turning back when the motor is not powered.

I wonder why this simple method is not used in any of the cheap machines. Once the proper force spring is selected, it basically never needs another adjustment, unlike the shitty friction based peelers. The tiny DC gear-motors can be had for like 4 bucks or less, so for the total cost of like 300 a perfectly consistent peeler can be made.

[uncle_bob]

Hi

Ok, so here's an inherently un-answerable question:

How long will it be before something better comes along at a lower price?

Better at a higher price does not count in this case. You can always get better for more money.

Better in this case probably means a significant improvement in terms of:

1) Vision, machines without vision need not be considered. This also includes the whole part recognition / placement process.

2) Feeders, machines with fewer feeders or less functional feeders .. not so much. Quick change feeders would score in this category.

3) Nozzles and basic pick function. Auto change would be a plus in this category, so would handling a wider range of parts. 

4) Something really significant in software. Full open source would be a plus here. It would need to be more than incremental improvement or bug fixes.

I am by no means saying it must be an improvement in all three. Simply that it needs to be notably better in one of these categories. I've left out placement speed and number of heads. Both of those are very application specific in terms of their real performance. Coming up with a real speed number means running *your* typical job on the machine. I've also left out board feeders since that sort of gets you out of the "desktop" world.

Yes, it's a very arbitrary list of what's in and what's out. There could be a *lot* of very valid criticism of the picks on this list. Nit picking the list is fine and feel free to do so. My larger focus is on the "when" not on the quality of the list. The why of asking this is fairly simple: Nobody wants to buy a shiny new laptop and find out the new one comes out the next day. I'm saying that the machine becomes scrap when the new one comes out. The "old" laptop still works fine the next day.

So - any guesses? Is there some, known only to a few, "model year" cycle to this stuff? Are we two years away or five years? I will admit I only got to thinking about this after the A->C and B->D model stuff popped up.

Bob

[ttsthermaltech]

I would honestly think that better for less is going to be hard to come by. You don't get much more simple than the low end Neoden TM-240 / Qihe TVM802 units. They are  pretty much stripped down to the bone in terms of features vs price vs build quality. Both have pretty good feedback regarding build quality. Both have similar price. If the Chinese are selling at a given price, and the price is relatively even across the market leaders, you really can't expect to get much more for less, unless the cost of the components drop due to certain components going mass market or alternate suppliers selling cheaper parts. Pretty much everything on these units is mass market items such as steppers, drivers, belts, pulleys, usb cameras, etc.

The only proprietary parts are the physical machine base and control board / software. The workmanship / precision put into the machine base is directly related to machine cost. The flatter, or more rigid it becomes, the more it weighs, and the more costly it becomes. They are made in relatively small quantities (ie compared to the mass market parts), so not much is going to change there.

The software / control boards are the only parts that really separate the different vendors, and everyone here is pretty focused on the software and how well it works / flexibility, so one could only assume if we want better software, the cost will only go up in this area. Better software requires more development time / resources.

As we already have seen with the Neoden 4 which costs significantly more than the TM240, and also with the Qihe TVM920, Charmhigh CHMT530P, adding more features drives the cost up. Once you get past a  certain point in mechanical design, you have to forgo the hobby parts, and start using lots of custom machined parts. This adds cost really quickly

I am sure that there are ways to build cheaper PNP's than china offers, but in reality would you really want one? You gotta sacrifice somewhere to drive the cost down. Sometimes the sacrificial lamb is staring right at you in terms of features, other times it is in the form of sub-par components. Both will drive you crazy when it comes to PNP's.

If you want to save money, take a look at the OpenPNP software project and the accompanying OpenBuilds PNP hardware. You can build it up in basic form for about $1500. The software is great and moving further forward every month, and the hardware is really dependent on your mechanical skill and budget. You forgo the China labor in favor of your own blood, sweat and tears. You will get more bang for your buck this way (theoretically), but you also have to consider whether you have the time.

Alternative, consider a used piece of equipment. It may not be new, but sometimes you can find a well maintained commercial PNP at a really reasonable price, which can satisfy your requirements. Often the software will be antiquated, but you can always consider using / porting to openPNP to run the machine, and exploit its commercial grade hardware.

As with all things in life, you get what you pay for... Hopefully, what you can afford, can also meet your needs / requirements.

[uncle_bob]

I would honestly think that better for less is going to be hard to come by. You don't get much more simple than the low end Neoden TM-240 / Qihe TVM802 units. They are  pretty much stripped down to the bone in terms of features vs price vs build quality. Both have pretty good feedback regarding build quality. Both have similar price. If the Chinese are selling at a given price, and the price is relatively even across the market leaders, you really can't expect to get much more for less, unless the cost of the components drop due to certain components going mass market or alternate suppliers selling cheaper parts. Pretty much everything on these units is mass market items such as steppers, drivers, belts, pulleys, usb cameras, etc.

........

Hi

Ok, there's the first "vote". Thanks for the input.

I would score that as "maybe never". The price of chunks of metal does not follow an ever declining curve the way silicon IC's do. There is a lot of metal chunks in a pick and place. There is not a lot of costly silicon in a pick and place. About the only part of the machine that does not apply to is the vision stuff. I can easily see better / cheaper cameras coming along. Do they improve the machine? I suppose they *might*.

Bob

[mikeselectricstuff]

How long will it be before something better comes along at a lower price?

Better at the same or slightly higher price would be a start.
The Chinese are too focused on lowering the price and thiking that placement speed is the only important benchmark, so corners get cut, especially with software design & implementation, and I can't see that changing any time soon.
There is such a huge gap between the more established makers and the newer Chinese machines that there has to be a market for something more polished and better supported at a slightly higher price.
There are plenty of areas for things to be improved, and none of them should be that hard to do given a bit of effort:
Auto nozzle change, taller parts, better feeders, better software, flying upward-looking camera.

[l0wside]

The cost of the cameras is not really relevant, they should not contribute much more than 100$ to the price. Some machines use a PC as a controller, so even from electronics perspective, there is not much room left (as you stated correctly).
More or less all components that go into such a unit have been on the market for a long time, be it motors, bearings or nozzles. A further decline in their cost is not really to be expected.
The Chinese also want to recoup their development cost (even Chinese engineers do not work for free). Given the modest sales figures of these machines (selling a few hundred might already be a major coup), there is simply not much room for further savings.

For the extreme low end, check out the Liteplacer (which comes as a kit). It might give you an idea about the lowest possible cost for the most rudimentary P&P with vision.

Max

[uncle_bob]

How long will it be before something better comes along at a lower price?

Better at the same or slightly higher price would be a start.
The Chinese are too focused on lowering the price and thiking that placement speed is the only important benchmark, so corners get cut, especially with software design & implementation, and I can't see that changing any time soon.
There is such a huge gap between the more established makers and the newer Chinese machines that there has to be a market for something more polished and better supported at a slightly higher price.
There are plenty of areas for things to be improved, and none of them should be that hard to do given a bit of effort:
Auto nozzle change, taller parts, better feeders, better software, flying upward-looking camera.

Hi

I deliberately left out the "better at same to slightly higher price" simply because it opens the door to "higher price". You then get into all sorts of debates about the value of each item added. Exchange rates / prices where also was omitted since that gets into other debates.  The omission of placement speed also was deliberate. The manufacturers are to focused on being able to advertise some sort of nutty placement speed number. A lot of things get compromised as a result.

I think I would score this one as "it could happen, but not at a noticeably lower price".

Thanks!

Bob

[uncle_bob]

The cost of the cameras is not really relevant, they should not contribute much more than 100$ to the price. Some machines use a PC as a controller, so even from electronics perspective, there is not much room left (as you stated correctly).
More or less all components that go into such a unit have been on the market for a long time, be it motors, bearings or nozzles. A further decline in their cost is not really to be expected.
The Chinese also want to recoup their development cost (even Chinese engineers do not work for free). Given the modest sales figures of these machines (selling a few hundred might already be a major coup), there is simply not much room for further savings.

For the extreme low end, check out the Liteplacer (which comes as a kit). It might give you an idea about the lowest possible cost for the most rudimentary P&P with vision.

Max

Hi

I have looked at a number of other machines and pretty much decided that anything less capable then the TVM802 is not going to do a reasonable job. That is a *very* subjective evaluation and in no way a knock on the other gear on the market. Truthfully the TVM802 *barely* does what I think it needs to do. Barely doing the job is still OK.

Again, I think I would score this one as "not going to happen".

So far, nobody seems to believe that the $1500 replacement for the TVM802 is coming out next week I certainly do not disagree with anything anybody has come up with. It is good to see that most people have a fairly similar view of where this all is going.

Thanks!

Bob

[mikeselectricstuff]

I deliberately left out the "better at same to slightly higher price" simply because it opens the door to "higher price".

I'm not convinced that there is much benefit to lowering prices further - performance and usefulness will inevitably suffer.
If you can't afford a P&P now, chances are you don't really need one.


 

[uncle_bob]

I deliberately left out the "better at same to slightly higher price" simply because it opens the door to "higher price".

I'm not convinced that there is much benefit to lowering prices further - performance and usefulness will inevitably suffer.
If you can't afford a P&P now, chances are you don't really need one.

Hi

Ok, one half of that one I would not agree with. (lowering price does not get more customers). I have never seen any market study on a piece of gear where volume did not go up as price goes down. Yes there may be some inelastic demand gizmo somewhere, but I don't think pick and place machines are it. Drop the price by half and you more than double the number of units sold (provided it's the same quality /type / capability machine). There is always a "if we add another machine will it pay for it's self" decision. Lower price gets more of those decisions turned into a purchase.

I'll score this one as "not going to happen". The real core argument is still that the quality requirement can't be met at a lower price.

Bob

[ttsthermaltech]

Lets remember that the Chinese manufacturers are typically working on very slim margins. Either materials must go down in price, or their labor must go down in price, as the hours to put one together will likely not go down significantly to impact price. I don't see materials dropping, and wages in China are on the rise. Your best bet is to hope for currency swings to save you some money.

So yeah, as everyone is pretty much saying, the cost isn't going down anytime soon.

But from another perspective, consider the benefits and time / cost savings of having a reliable, well built PNP in your office. You should be able to recoup the cost of almost any small, well built machine rather quickly. At that point, any minor savings in purchase price are soon negated.

[uncle_bob]

Lets remember that the Chinese manufacturers are typically working on very slim margins. Either materials must go down in price, or their labor must go down in price, as the hours to put one together will likely not go down significantly to impact price. I don't see materials dropping, and wages in China are on the rise. Your best bet is to hope for currency swings to save you some money.

So yeah, as everyone is pretty much saying, the cost isn't going down anytime soon.

But from another perspective, consider the benefits and time / cost savings of having a reliable, well built PNP in your office. You should be able to recoup the cost of almost any small, well built machine rather quickly. At that point, any minor savings in purchase price are soon negated.

Hi

Rather than get into all sorts of messy details, let's just say that not every machine gets used the same way in the same setting. Put another way, you can justify a $250,000 or a $1,000,000 machine with "it will pay for it's self". Been there done that.

"Well built" is pretty elastic and it will move you up from a TVM802 pretty fast. That's where the original "at a lower price" came in. Without that, you are on a very slippery slope. In some cases budgets do matter

I'll log this one as another "not getting cheaper any time soon, far more likely to go up in price".

I'm by no means saying that I disagree with any of the basic points made. For a lot of settings, you would be foolish to buy one of these over a machine costing 2 to 5 times as much. The delta in purchase price would quickly be paid back on a better machine that is kept full 80 hours a week 52 weeks a year. You would be banking the delta long before the setup turned 1 year old.

Bob

[ttsthermaltech]

Hi

Rather than get into all sorts of messy details, let's just say that not every machine gets used the same way in the same setting. Put another way, you can justify a $250,000 or a $1,000,000 machine with "it will pay for it's self". Been there done that.

"Well built" is pretty elastic and it will move you up from a TVM802 pretty fast. That's where the original "at a lower price" came in. Without that, you are on a very slippery slope. In some cases budgets do matter

I'll log this one as another "not getting cheaper any time soon, far more likely to go up in price".

I'm by no means saying that I disagree with any of the basic points made. For a lot of settings, you would be foolish to buy one of these over a machine costing 2 to 5 times as much. The delta in purchase price would quickly be paid back on a better machine that is kept full 80 hours a week 52 weeks a year. You would be banking the delta long before the setup turned 1 year old.

Bob

Yeah, I agree with you UB. Buying any piece of equipment, be it a table saw, or a PNP, you will find may different products meeting many different needs. You need to fully understand your requirements, and purchase accordingly...

[protoneer]

The TVM802A has been my first P&P. I have had it just a few weeks and it has been running an hour or two per week. I will go in full production in a week or two.

For the capital investment I need to make about 1500 units and I have already done about 10% of that up to now.

It has been a pleasant budget machine to work with. For me this machine compares to the experience I had with my first 3d printer. I have a commercial 3D printer that was about 4 times the price of an RepRap. It printed stuff with in the first 15 minutes. Even at 4 time the price it has paid it self of with in the first few months.

Getting back to P&P. I could have gone for the DIY version but I would have had more stress keeping it running and making sure production stays online.

Upgrading the software on this machine will make it much better. Not sure how far @ar__systems is with decoding the protocol. Would be interested in writing add on't that use movement of the machine but better camera's or higher components.

My question is, once I pay this machine off where do I go from there? Is there a $10K machine that has quick reel changing and software that is more forgiving. (At $10K and up unfortunately I don't think there will be Open-source options)

[ttsthermaltech]

I will let yo know once my Qihe TVM920 shows up. I will post a complete review of everything from setup to first runs. Hopefully it will fill the gap between the TVM820 (fixed drag feeders) and the Neoden 4 (proprietary fixed feeders). I am confident that the interchangeable Yamaha style CL feeders used on the 920 will make all the difference.

Fingers crossed....

The TVM802A has been my first P&P. I have had it just a few weeks and it has been running an hour or two per week. I will go in full production in a week or two.

For the capital investment I need to make about 1500 units and I have already done about 10% of that up to now.

It has been a pleasant budget machine to work with. For me this machine compares to the experience I had with my first 3d printer. I have a commercial 3D printer that was about 4 times the price of an RepRap. It printed stuff with in the first 15 minutes. Even at 4 time the price it has paid it self of with in the first few months.

Getting back to P&P. I could have gone for the DIY version but I would have had more stress keeping it running and making sure production stays online.

Upgrading the software on this machine will make it much better. Not sure how far @ar__systems is with decoding the protocol. Would be interested in writing add on't that use movement of the machine but better camera's or higher components.

My question is, once I pay this machine off where do I go from there? Is there a $10K machine that has quick reel changing and software that is more forgiving. (At $10K and up unfortunately I don't think there will be Open-source options)

[RobK_NL]

I will let yo know once my Qihe TVM920 shows up. I will post a complete review of everything from setup to first runs.

Oh yes, please do!

I'm looking for a smaller replacement for my somewhat ancient Dima Optimat and the TVM920 seems to be a strong contender. I'd love to know if it's as good as it looks.

[ServoKit]

Latest version of automatic PCB loading and unloading:



As you can see, it looks very different from the first version and runs much smoother; list of changes I made:

http://servokit.com/blog/2016/05/03/automated-pcb-handling-with-the-tvm802a-pick-and-place-machine-iv/

Regards, Axel

[ServoKit]

Close-up of PCB pickup and drop off:



Regards, Axel

[thommo]

Hi @ ServoKit,

Looks great. Can't wait to see it make its way into your oven eventually.

Out of interest - what are you using for the 'initial lift, and drop actions' on the PCB carriage?

Some detail pics, parts list, would be appreciated.

Thnx

[ServoKit]

Details are here:

http://servokit.com/blog/2016/05/04/automated-pcb-handling-with-the-tvm802a-pick-and-place-machine-v/

Regards, Axel