chmt drag feeder issues (using reference drag feeder)

[vespaman]

Hi Guys,

I have been running some batches this week, and are now doing my usual analysis of my notes from the days. (I take notes of abnormal behavior, and try to improve for next run).

This time, I have gone through, what seems like a million "after feed pictures" to try to get a grip on what *really* causes pick failures.

The most common issues I have, is with 16mm tapes, where the cover tape friction to a metal bar that guides the used tape to the clutch spool. This I have seen before on various components. The problem here is that this friction is not constant, but varies over time. I think this is related to residues of the glue on the tape cover.

When this happens, the clutch is unable to pull as much cover tape as needed, and part of the components is then still covered when the nozzle is trying to pick it up. -> Failure.

The solution for this, is to 3d print 'bearing' that decouples the tape cover from the metal bar. Should be simple. In October, I had a catastrophic failure because of this; a tact switch was dropped by the nozzle just after picking it up, it got caught in the cover tape, and it dropped from nozzletip, just beside the tape, so when the machine did a re-feed, the drag pin collided with this dropped tact switch which in turned collided with the feeder base -> which made the whole drag pin with coil getting ripped out of its metal 'cage'.  The drag pin itself was of course useless after this event, and I had to re-calibrate etc etc. -> Serious down time.

But:- Then we also have a few mispicks on 8mm tapes. This time, I am only looking at paper tapes (there's a different story with black plastic tapes, but I'll deal with them once I have fixed the paper tapes) I have categorized them into 'Jumps' and 'Creeps'.

Jumps are typically 0402 resistors that occasionally jumps out or their pockets. I thought that out of my failures, this would be the most common failures, but I could only find one or maybe two such instances out the the million pictures I have eyeballed.

Here's the example I found;

First two ok;

Then the problematic;

And the next ones are ok again;

As can be seen, there's no trace of the failing ones, so I guess they where both picked up by the nozzle, and failed on vacuum or visual. Does not matter, but it is good to see that this can happen, but evidently not very often.

However, in order to try to create a solution for this, I think it would be useful to do some kind of script that just do repeated feeds (eg 1000 each run), and take the image debug pictures. However, even better would be if it would be possible to have opencv check that both components are still in their pockets and fail if not (or count the failures). Would this be possible within the sprocket hole pipepline? I guess it should be possible to call opencv from the script, or if that fails, eye ball the results. But eyeballing is not very enjoyable.. :-) And I suspect that I am not going to find a solution that works as wanted withing a few iterations..

Any suggestions on how to solve this?

Creeps

This is kind of similar to the 16mm tape issues, but not as prominent on the 8mm, so many times self-healing and missed by me.

Here is a self healing clip (that did not stop the machine)

It could be that the reason behind this is the same as the 16mm tape, but it could also be that the clutch is not 100 even in force (or a combination of the two). I think a solution here, is to make a stronger push-up spring under the tape, so a higher clutch force can be applied. (For those not familiar to these feeders, there a battle between to much clutch friction against the tape hold friction by the under tape spring; if the clutch is too strong, it will pull the cover tape so hard that the tape 'feeds' out a little bit, which is unwanted.

Anyway, suggestions/comments on especially how to automate the Jump detection problem are most welcome!

 - Micael

[vespaman]

.. Title should be 'using Reference Push Pull Feeder'.

I knew eyeballing all those pictures would make my brain hurt... :-)

 - Micael

[Wayne Black]

Hi Micael,

First off, awesome animation, very telling :) When I played around with these feeders and removed them, I found the tape opening edges to be very ruff and irregular. I also found the steel leaf springs to be very firm with minimal contact to the tape underside. I figured I could print smoother edged tape guide/openings and springs that made full tape contact along the bottom. 

I was wrong, the end result was the tape ran so smoothly in the guide that the tape collector clutch has to be set so loosely to prevent advancing the tape that it sometimes doesn't have enough friction to peel the tape.In this arrangement the drag pin needs to stay down as the peeler peels extra tape while X is stopped. I know you dont want this as it takes extra time :)

All that aside if you're doing large volumes, which it sounds like you are, the more tape that is peeled the larger the diameter of the tape collector and the resulting lever arm against the clutch. Even if perfectly tuned, I dont think these feeder are meant for large volumes without a lot of user intervention of adjusting the clutch.

Very cool Animation!

--
You received this message because you are subscribed to the Google Groups "OpenPnP" group.
To unsubscribe from this group and stop receiving emails from it, send an email to openpnp+u...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/openpnp/7545e04b-b089-4bf8-afca-a28843b22bd0n%40googlegroups.com.

--

Wayne Black

Owner

Black Box Embedded, LLC

black...@blackboxembedded.com

1.831.682.4964

[vespaman]

Hi Wayne,

thanks!

What I believe is the solution, is to press the tape so hard up, that the clutch force cannot possibly do the involuntary 'feed'. The drag pin+X stepper must of course be able to move it easily, but they are pretty strong (the force to pull on my largest components/deepest pockets are substantial, so there's a lot of headroom here on the 8mm) . If it is too weak, like you say it is almost impossible to trim the clutch. This way, it is less important to trim the clutch to the "clutch nirvana". But I also want to increase the surface area underside, like you did (if I understand you correctly) to make sure both pick places on 0402 are equally well supported, also, I think I want to guide the tape better after the pick position, so that any springiness from the drag pin punch/pull is not causing too much ripples.

Another simple solution, that I am also going to try, is to, instead of having a fixed starting position (first hole after the pick guide), I have a fixed end position, last hole closest to the screw. The rational is that I want a) to get away as far as possible from the picking position b) the screw+base below it, will give some support to the tape on drag pin leaving, and c) The start of drag here, is not in anyway possible to interfere with the leaf spring. Close to the pick area, the spring is *very* close to the drag pin, and I have seen the drag pin collide with the spring at times, now my springs are horrible, since the guy that owned the machine before me, had tweaked them, which may be the reason that I have seen collisions. But still, very close.

Thirdly (I wasn't going to mention this, because it is not really related to this, but in my "todo-book", is also to print a guide that guides the used tape down into the waste slot, since a) this is a problem on cut tape, where you don't have enough of tape to guide it down from the beginning, and b) some tape struggles to be pushed down and causes a variable disharmony to the drag pin back-pressure. Especially seen on the thickest paper tapes, and also on plastic 16mm tapes with deep pockets, that is almost on the verge to skip the slot altogether on a 100% speed feed.

Regarding the larger diameter on peeled tape cover;

Yes, I think the fact that I have a couple of thousands(?) components already peeled, is the reason that it self heals fast enough, not to stop the machine. So a little extra peeling is actually good.

Sure, I know these feeders, and indeed this machine is not meant for larger batches, but do know what? It is kind of like a sport, and since they are so simple, they are also simple to modify/make better. I actually have a pretty good flow now, first and foremost since I added the drag pin anti stick stuff into the firmware, it takes care about stuff that is inevitable.

I am considering cutting away the deflector under the waste slot, so the used tape can just drop straight down, since the entangled and springy tape waste is huge potential for issues in themselves - you gotta watch them at least until they has passed the tape reels, otherwise they can easily get the way of the fresh component feeds. I have learned not to cut them too much after each day - a decimeter or two over the floor is perfect. Looks like shit though. :-)

 - Micael

[Jan]

Hi Micael!
It's cool that you collect all this information and try to draw
conclusions from them and let us participate! That's very much appreciated.
Concerning the pick problems: the problematic picture shows, that
pocket 1 is empty and the part in pocket two has been kicked out. I've
seen that in one of my feeders a few times as well. After raising pick-Z
slightly I've not seen it again. So my pick-Z was probably a little bit
to low pressing part and tape down when picking. Then going up the
second part gets kicked out.
Concerning the spring: on my stock setup I don't have much trouble with
them. It's more that tall parts like HC-49SM crystals do not fit very
well because they are quite strong. Also the take-up wheel is quite
forgiving but still requires attentions especially after some down time.
Concerning the metal cover to divert the cover tape backwards: it's of
poor machining quality on my machine too. It's likely a good idea to
sand the edges when you're willing to unloaded all feeders.
Concerning the imperfect pealing: with stronger springs you can pull a
little harder. That shall solve the problem. An other suggestion would
be to add some sort of knife to separate the tape from the cover tape.
That's how feeder finger peels the cover off even without pulling it.
On modern versions of the CHM-T they added some sort of deflection made
out of sheet metal to direct the spend tape into the slot. That's likely
a good idea to avoid trouble by spend tape not leaving the feeding area.
If you're considering to take out your saw anyhow, why not cut away the
entire feeder area at the slot for spend tape and add Yamaha feeders
there? They are fast and reliable and easy to swap out. IIRC there are a
few new commercial designs of diy PnPs out there that ship with them as
default. Maybe they sell the required mounting bay and controller
separately...

Jan

> tape-utils1728371303640340419.png
>
> tape-utils1728371306873340624.png
>
> Then the problematic;

> tape-utils1728371310277427423.png
> And the next ones are ok again;

> tape-utils1728371351587494324.png
> As can be seen, there's no trace of the failing ones, so I guess they
> where both picked up by the nozzle, and failed on vacuum or visual. Does
> not matter, but it is good to see that this can happen, but evidently
> not very often.
>
> However, in order to try to create a solution for this, I think it would
> be useful to do some kind of script that just do repeated feeds (eg 1000
> each run), and take the image debug pictures. However, even better would
> be if it would be possible to have opencv check that both components are
> still in their pockets and fail if not (or count the failures). Would
> this be possible within the sprocket hole pipepline? I guess it should
> be possible to call opencv from the script, or if that fails, eye ball
> the results. But eyeballing is not very enjoyable.. :-) And I suspect
> that I am not going to find a solution that works as wanted withing a
> few iterations..
> Any suggestions on how to solve this?
>
>
> Creeps
> This is kind of similar to the 16mm tape issues, but not as prominent on
> the 8mm, so many times self-healing and missed by me.
> Here is a self healing clip (that did not stop the machine)

> animation.gif
>
> It could be that the reason behind this is the same as the 16mm tape,
> but it could also be that the clutch is not 100 even in force (or a
> combination of the two). I think a solution here, is to make a stronger
> push-up spring under the tape, so a higher clutch force can be applied.
> (For those not familiar to these feeders, there a battle between to much
> clutch friction against the tape hold friction by the under tape spring;
> if the clutch is too strong, it will pull the cover tape so hard that
> the tape 'feeds' out a little bit, which is unwanted.
>
>
> Anyway, suggestions/comments on especially how to automate the Jump
> detection problem are most welcome!
>
>
>  - Micael
>

> --
> You received this message because you are subscribed to the Google
> Groups "OpenPnP" group.
> To unsubscribe from this group and stop receiving emails from it, send
> an email to openpnp+u...@googlegroups.com

> <mailto:openpnp+u...@googlegroups.com>.

> To view this discussion visit

> https://groups.google.com/d/msgid/openpnp/a79770c9-6c00-4fae-8ef7-a8935731da81n%40googlegroups.com <https://groups.google.com/d/msgid/openpnp/a79770c9-6c00-4fae-8ef7-a8935731da81n%40googlegroups.com?utm_medium=email&utm_source=footer>.

[vespaman]

Hi Jan!

Concerning the pick problems: the problematic picture shows, that
pocket 1 is empty and the part in pocket two has been kicked out. I've
seen that in one of my feeders a few times as well. After raising pick-Z
slightly I've not seen it again. So my pick-Z was probably a little bit
to low pressing part and tape down when picking. Then going up the
second part gets kicked out.

No, this is not the case here. This is _before_ picking. So this is a feed problem of some sort. If I did not have the vision, I would have thought, like you, that this would be a pick problem.

This is why I want to run a feed-test, instead of feed/pick/place, since it is easy to draw wrong conclusion if all steps are "bundled into one" :-)

Of course, I could create a test board, with e.g. 1000 resistors, panelize into say 10 boards. This, and feed/pick/place as normal (and maybe it will end up with this), but it would take time to  create such a board, and also to run tests.  This board could be useful for later placement tests as well.

 

Concerning the spring: on my stock setup I don't have much trouble with
them. It's more that tall parts like HC-49SM crystals do not fit very
well because they are quite strong. Also the take-up wheel is quite
forgiving but still requires attentions especially after some down time.

Yes, and also when you switch to a new reel, the same thing has to be done. And I also think something else comes to play, maybe temp in room, humidity or wear or something, since from time to time, I have to adjust the clutch. But most likely the variable friction of the tape cover is the bad boy here. 

 

Concerning the imperfect pealing: with stronger springs you can pull a
little harder. That shall solve the problem.

 

This is not possible on 16mm tapes that are getting into their "obstructive mode", trust me! I have been dealing with this too many times now. It seems that the increased friction is spiraling into crazy - the more clutch you bring, the harder the force -> more friction. Super unreliably. But not on every reels, and not always on the same reel. It seems to come and go.

But I have fixed this now, yesterday I installed these tape cover idlers on one bank;

- what  a difference! Now there's very no friction to talk about. One problem out of the way.

 

On modern versions of the CHM-T they added some sort of deflection made
out of sheet metal to direct the spend tape into the slot.

Aha, this I did not know!

I have done these now, they looks to be working well, but next week I'll test it when I run a new batch;

In the machine;

I'm debating if I need to print them in ESD material. For black plastic tapes, I think it is not a problem, since it is grounded just before entering the 'drain pipe' and it is only a short path. But maybe esp the clear plastic tapes could generate some static, paper tapes, should also be OK, I think.

I suspect I'll populate all feeders with these, once I have tested them thoroughly. 

I am considering having them continue further towards the pick position, as part of the stabilizing guide, but I don't want to do this until I have some statistics on feed failure, so I know I am actually improving things.

 

If you're considering to take out your saw anyhow, why not cut away the
entire feeder area at the slot for spend tape and add Yamaha feeders
there? They are fast and reliable and easy to swap out. IIRC there are a
few new commercial designs of diy PnPs out there that ship with them as
default. Maybe they sell the required mounting bay and controller
separately...

In the beginning I had those kinds of thoughts, but now, I have a fairly reliable machine, and the drag feeders are not too bad. If I would do other peoples boards, it would be a  different story, but then I'd change the machine to something else instead.
In my view, the main benefit of the drag feeders is that they are taking very little space, and doing my own designs, I can select components that I have used before (the chmt 48 has 44 8mm, 8 12mm and 4 16mm and 2 24 feeders, as you know) so it is not so often I have to change reels for different jobs.

But - I have a lot of improvements for the feeders in the pipeline, and I am determined to see it through. If my ideas work in practice, these feeders will be perfect to my setup. I'll post here, of course. :-)

Right now, I am thinking of how I can best create a better spring solution.

Thanks for sharing your thoughts,

 - Micael

 

[Jan]

Hi Micael!
This is a resend. The original message got dropped for unkonwn reason.

On 24.11.2024 12:02, vespaman wrote:
> Hi Jan!
>
> Concerning the pick problems: the problematic picture shows, that
> pocket 1 is empty and the part in pocket two has been kicked out. I've
> seen that in one of my feeders a few times as well. After raising
> pick-Z
> slightly I've not seen it again. So my pick-Z was probably a little bit
> to low pressing part and tape down when picking. Then going up the
> second part gets kicked out.
>
>
> No, this is not the case here. This is _before_ picking. So this is a
> feed problem of some sort. If I did not have the vision, I would have
> thought, like you, that this would be a pick problem.
> This is why I want to run a feed-test, instead of feed/pick/place, since
> it is easy to draw wrong conclusion if all steps are "bundled into one" :-)
> Of course, I could create a test board, with e.g. 1000 resistors,
> panelize into say 10 boards. This, and feed/pick/place as normal (and
> maybe it will end up with this), but it would take time to  create such
> a board, and also to run tests.  This board could be useful for later
> placement tests as well.

While running a job you could use either "Job.Placement.Starting",
"Feeder.BeforeFeed" or "Feeder.AfterFeed" scripts whilch all contain the
selected "feeder" as a global. This global provides the method
"feed(nozzle)" to perform the feed operation ("nozzle" is available as
global too). If you can configure the feeder to perform vision before
each feed, you shall be done.

Jan

[Wayne Black]

So using your pic animation for debugging the drag feeder I discovered cross corruption of feeders using my printed fdm springs. You can see in the vids the adjacent feeder is wiggling about. Im going to retest using my fdm top plate with the original CHMT leaf springs which are much stiffer.

[vespaman]

LOL! So there's always something, isn't it! Good that you found it though.

 

I think the original springs are not stiff enough, at least not for 8mm tapes. I am considering that I may have to do different "spring plates" for plastic an paper tapes, where the spring plates for plastic carriers has a groove for the pockets. Right now I think about using coiled springs, since I don't like the fact that the chmt leaf springs are so close to the drag pin - I suspect this might be one reason for failures. But maybe if the spring plate is long enough, it will have a proper friction even with standard chmt leaf springs. Is it problematic to print stiffer springs?

Anyway, I think you might be onto something by replacing the complete top metal, it is way to small in my opinion. I need to do one more batch next week, before I can pull my machine apart.

Another useful thing, is to use your phone recording video in slow motion - but it really only works on the first feeders, since it is hard to position the phone in any relevant position on the feeders.

I had a thought of introducing a dynamic friction brake, so each feeder could not move if it was not allowed to. But then I extended this thought into a drag pin feeder that sits before the chmt feeder. I.e. a fixed drag pin feeder that only can do 4mm feeds, and has the drag pin stay in the sprocket hole all the time, so locking the carrier from peeler. This way there's no way the peeler can do harm.

Initially, I had a thought on having this fixed feeder being operated solely by two solenoids - one for doing the up/down of the drag pin, and one for the 4mm feed. But then I read some old conversation (maybe 10years ago?) that there's no way to have a solenoid pull a full reel, which I was kind of worried about. So if I am to pursue this solution, I most likely need to find another way to pull the tape fast. I would prefer to have the 4mm being fixed by mechanical stop, to keep things fast and simple. So some kind of motor or pneumatic. Motors are hard to use, because if they have a gearbox, its not good to slam them into fixed stop..
There's also a lot of time to gain if the drag pin would not have to be used.. :-) I have ordered some solenoids, maybe it is possible to put two in series or something, to make them strong enough. But as you know, there's only 13mm in width and about 20 in height so this might be a dead end.

  - Micael

[Wayne Black]

I'm rethinking the 'plate' vs 'point' contact of the spring. I think minimizing surface contact especially with the smooth steel spring and fdm springs makes for better pinching/retaining of the tape in place. If the spring to tape mating surfaces had more friction it may make more sense.

It's also hard for me to think I could make an fdm spring any stiffer than the original steel CHMT springs. I am away from the machine for the rest of the day, but will do a comparison tomorrow with the origins steel springs using the fdm top plate.

--

You received this message because you are subscribed to the Google Groups "OpenPnP" group.

To unsubscribe from this group and stop receiving emails from it, send an email to openpnp+u...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/openpnp/158f463b-4307-4822-aa7e-70ce6314bdacn%40googlegroups.com.

[vespaman]

I see. I guess the good thing with a point, is that you know that you have a good/best contact pressure exactly where the components are most fragile. One way might be to have both plate and point, i.e. keeping the point and adding a plate surrounding it, with its own decoupled spring solution.

[Wayne Black]

Hey Miceal,

I refitted the original steel CHMT springs and the end result is no more unintended tape movement. I loaded 16 feeders with the thinnest paper tape I have 0603 thin film Rs. I marked each tape within the pick up window and then picked each station 5x's. Afterwards I checked all the stations markings, and no movement. I made no adjustments to the cover collector clutch. I just left it as it was from the FDM springs, relatively loose to minimize the feeder cross corruption seen with the FDM springs 

In doing this I noticed the CHMT springs radius was not consistent. You can kinda see this in the picks. I kept extra attention on the longest spring vs the shortest in the above test. It made absolutely no difference. It's interesting how well the original CHMT setup works given the inconsistent spring rate and poorly machined feeder top plate.

To view this discussion visit https://groups.google.com/d/msgid/openpnp/f6335bc8-94ba-4de2-a1f8-52228a66314cn%40googlegroups.com.

[Ateeq Ishtiaq]

Thanks for sharing,I was looking for these spring mechanism pictures , Can you  Please share what is the Spring steel sheet thickness?.

[vespaman]

Hi Wayne,

Thanks for the pictures!

So have you come to a conclusion, or is this just mid-term results?

If it is of any comfort to you - my springs are _way_ more different between them. And I agree, it works surprisingly well. Also given that the friction wheels are not giving the same friction around the whole turn (at least some of mine does not).

But I am not happy anyway, while perhaps not often, the issues I have been bothered with

a) The constant tweaking of the friction wheels is annoying, and some times leads to problems. A couple of weeks ago, I changed into a new reel of TVS diodes, only to find at day end, that a couple of hundreds of them had fed through and ended on the table below the machine. (Always look at it from the bright side: Gives me reason to test the loose part feeder :-) ).

b) The sometimes jumping parts on a feed (be it related to drag pin touching spring, or vibrations or whatever).

c) The rough top plate does, on some feeders with plastic tape, carves a hairline of the tape, that sometimes ends up in the sprocket holes, and disturbs the vision. This is not a big problem for me, but still something that can be improved.

I suspect your 5x is way to quick test for any conclusive result (but I guess this was also not the intention).

Regarding a)

I think what come into play here, are the many variables - the tape type (material/thickness), how well the cover tapes is attached to the tape, the unevenness of the friction wheel, and the constant change of the reel - a full reel is heavier to pull. Also, the cover tape friction/stickiness to the metal bar (which I don't have to worry about anymore, but still a factor).

This is why I would like to add  more friction into the pickup area, so that friction is overcoming any of the variable friction, leaving a simpler adjustment of the friction wheel.

Regarding b)

This, is most likely improved by a more dead/dampened area around pickup/feed.

Therefore, maybe a and b should be handled separately, e.g. 'a' is dealt with in the unused area before the pickup area, so 'b' can be fully dealt with in, and after the pickup area?

Regarding your top plate - I see you have moved your pickup point to where the normal drag starts. How come? Now it looks like you have nothing securing the pickup point from vibrations originating from the drag/peel.

And, do you have something under the top plate, that guides the tape, or how else are you loading a tape through the area?

And lastly, what is that slot on the left side, something to guide the cover tape through?

 - Micael

[vespaman]

Thickness is 0.25mm on all springs (regardless of width).

 - Micael

[Wayne Black]

Ha, no, no conclusions at the moment other than its an amazingly simple/effective system with some maddening backs.

RE a, As we both have noted this is quite problematic if doing large quantities. I'm doing maybe dozens a build at most, so not that problematic for me. Sounds like a nightmare for you though :(

RE b, Yes, this is an issue especially for me. I redesigned the head and used a pneumatic dragpin. The actuating is not an issue, but the longer pin arm (which is currently FDM mounted) seems to stutter a little when pulling thick or plastic tape. Your deadening fix I dont think will fix my issue. I need to get the mount alloy machined and or some closer positioning.

RE c, Yes, this was a huge problem for me before the FDM top plates. It would create thin curly hairs that would widen and finally just make the plastic tape crack and break. I dont foresee this happening ever again w FDM top plates. If you had the time and patience you could try filing the original steel plate. Originally that was my intent when I first took the plate off, but the machining is very irregular and modding it further may create other problems.

I'm not sure I understand;

"Therefore, maybe a and b should be handled separately, e.g. 'a' is dealt with in the unused area before the pickup area, so 'b' can be fully dealt with in, and after the pickup area?"

You're correct, I moved my pick up window so the 2nd nozzle can reach the feeder which biases the spring forward of the pick up window. It doesn't appear to cause any problems with paper tape. Ive run 2 slots of sot23-3 with them and so for no probs either.

I'm modding my head still further. The original 36V X carriage block is too wide so Ive redone the block and narrowed the Z plate. This will draw the dragpin and 2nd nozzle closer So I can go back to the original pick up location for the drag feeders. I Also designed a custom Duet 3 expansion board to replace the x2 1TLCs and x1 1XD.

--

You received this message because you are subscribed to the Google Groups "OpenPnP" group.
To unsubscribe from this group and stop receiving emails from it, send an email to openpnp+u...@googlegroups.com.

To view this discussion visit https://groups.google.com/d/msgid/openpnp/516005ed-e765-409e-993c-6697020eef7bn%40googlegroups.com.

[vespaman]

>I'm not sure I understand;

"Therefore, maybe a and b should be handled separately, e.g. 'a' is dealt with in the unused area before the pickup area, so 'b' can be fully dealt with in, and after the pickup area?"

Sorry, what I meant was that I think separating the two issues, instead of trying to achieve a solution for both of them in the pick up area, which I previously thought would be a good idea. But with your findings, I think that it is better to address each problem separately;

Where the 'Friction area' is only trying to do just that  - create a significant friction, so there's no way the peeler can have a say about pulling the tape. And, of course the yellow peel/pick area is only about trying to minimize jumping components.

Take note on my springs - they have a flat start/finish.. Something I suspect was done by the previous owner of this machine, since it looks 'hand made'  (but I could be wrong though, maybe someone was bored the day in the factory, when my machine was made.. :-) ).

For the friction area, I think that a strong spring pressing the tape against a top plate should do it, but I have also been thinking about doing some dynamic break, for instance, like this, where the drag operation needs to pull the (red) tape, first or simultaneously with the peeler (green-ish), without drag, the peeled tape will lock between the rubber wheel and the walls. (the simple drawing shows a distance between the wheel and the top, but they should be in contact, the wheel is free to move, but once it hits the left wall, it will stop, and therefore put a breaking force to the cover tape. The spring below the 'full tape' is just to make sure the full tape will move the wheel to the right, regardless of the thickness of the tape.

... but I don't think this is needed, I'll certainly try for a simpler solution first.

 - Micael

[vespaman]

Regarding you new beefy head - cool! Suddenly my head looks so tiny and underdeveloped... :-) I think this is what I like the most with these machines - their sturdiness, being able to host a head like that.

The custom expansion board interests me, even though I am using the original charmhigh mainboard - are you going to mount it on the head, for driving the steppers locally? I have been investigating to do this on my machine, since I anyway have a plan to remove the original stepper drivers.

Regarding stepper drivers, may I ask how much current you drive your Y motor with? I know you have 48V, which is where I'm going too (at least), but the original Y motion is acceleration limited, still the motor is never really hot, something that don't really make sense to me. (Why did they not increase the current a bit more?). While it would stress the belts a bit more, they are not too tiny, and they are two, so..

onsdag 4 december 2024 kl. 04:54:45 UTC+1 skrev black...@blackboxembedded.com:

[Wayne Black]

Ah yes, I understand now. I think you're right in both decoupling the problems and how to address them. Though they may have the same solution ;)

RE friction, biggest problem using figuring a constant spring rate for varying tape thickness using individual coils or leafs. Now if the CHMT drag feeder block was also pneumatic manifold you could control all the feeders spring rate/friction via a single regulator.

RE Dampening I'm not sure how important that is once proper friction and picking are accomplished. Still though, having loose plastic tape vibrating around isnt good. A simple FDM leaf would be sufficient.

Looking at your machine and I think you're right about the springs being modified by an end user. My 36 being as modded as it is doesn't have that. I just checked my 48V that is still in the crate and from what I can see the ends of those springs are not rolled either. I can understand the theory of 'rollling' the ends, but I think practically it makes your lack of friction problem even worse. 

[Wayne Black]

RE the 'big' head, yeah, I'm not proud of how big the head got and I'm trying to parse it down. The impetus of all this was to make the CHMT head capable of tool changes and accommodate taller components. I don't want to completely derail your Dragpin thread so Ill update 36V Upgrade Thread as there has been a lot of changes.

[Jan]

Hi Micael!

On 04.12.2024 15:53, vespaman wrote:
[...]

> For the friction area, I think that a strong spring pressing the tape
> against a top plate should do it, but I have also been thinking about
> doing some dynamic break, for instance, like this, where the drag
> operation needs to pull the (red) tape, first or simultaneously with the
> peeler (green-ish), without drag, the peeled tape will lock between the
> rubber wheel and the walls. (the simple drawing shows a distance between
> the wheel and the top, but they should be in contact, the wheel is free
> to move, but once it hits the left wall, it will stop, and therefore put
> a breaking force to the cover tape. The spring below the 'full tape' is
> just to make sure the full tape will move the wheel to the right,
> regardless of the thickness of the tape.

> SmartSelect_20241204_152504_Gallery.jpg

After you started this thread, I revisited my machine and inspected the
cover plate of my feeder bank. It suffers from the same manufacturing
"features" as yours. So I thought what could be do and the first idea
was to remove it and sand or rasp the edges or bend some thin sheet
metal over it or replace it with a plate with roles in it all for extra
smooth transition (as you reported for cover tape guide). Then, this
night, I had to notice, that this is likely counterproductive because,
as you pointed out, you don't wont the peeler to pull the tape. So the
180° turn back is likely a good starting point for that. If that's to
rough though, peeling and dragging could be separated again as in former
times. With dragging first and peeling second, you might also fix some
of your components-are-jumping problems. For the ReferencePushPullFeeder
we could split the peeling into two arbitrary steps, like 3mm while
dragging (leaving the cover tape a little loose) plus 1mm while
releasing the drag pin which would then fasten the cover tape again. One
could likely extend this using some kind of rubber tube acting as break
that gets pulled by the peeler and loosened when the tape is dragged.

Jan

[...]

> onsdag 4 december 2024 kl. 04:54:45 UTC+1 skrev
> black...@blackboxembedded.com:
>
> Ha, no, no conclusions at the moment other than its an amazingly
> simple/effective system with some maddening backs.
>
> RE a, As we both have noted this is quite problematic if doing large
> quantities. I'm doing maybe dozens a build at most, so not that
> problematic for me. Sounds like a nightmare for you though :(
>
> RE b, Yes, this is an issue especially for me. I redesigned the head
> and used a pneumatic dragpin. The actuating is not an issue, but the
> longer pin arm (which is currently FDM mounted) seems to stutter a
> little when pulling thick or plastic tape. Your deadening fix I dont
> think will fix my issue. I need to get the mount alloy machined and
> or some closer positioning.
>
> RE c, Yes, this was a huge problem for me before the FDM top plates.
> It would create thin curly hairs that would widen and finally just
> make the plastic tape crack and break. I dont foresee this happening
> ever again w FDM top plates. If you had the time and patience you
> could try filing the original steel plate. Originally that was my
> intent when I first took the plate off, but the machining is very
> irregular and modding it further may create other problems.
>
> I'm not sure I understand;

> /"Therefore, maybe a and b should be handled separately, e.g. 'a' is

> dealt with in the unused area before the pickup area, so 'b' can be

> fully dealt with in, and after the pickup area?"/

>
> You're correct, I moved my pick up window so the 2nd nozzle can
> reach the feeder which biases the spring forward of the pick up
> window. It doesn't appear to cause any problems with paper tape. Ive
> run 2 slots of sot23-3 with them and so for no probs either.
>
> I'm modding my head still further. The original 36V X carriage block
> is too wide so Ive redone the block and narrowed the Z plate. This
> will draw the dragpin and 2nd nozzle closer So I can go back to the
> original pick up location for the drag feeders. I Also designed a
> custom Duet 3 expansion board to replace the x2 1TLCs and x1 1XD.
>
>
>
>
> On Tue, Dec 3, 2024 at 1:13 AM vespaman <micael....@gmail.com> wrote:
>
>
> Thickness is 0.25mm on all springs (regardless of width).
>
>  - Micael
>
> tisdag 3 december 2024 kl. 06:31:04 UTC+1 skrev ateeq....@gmail.com:
>
> Thanks for sharing,I was looking for these spring
> mechanism pictures , Can you  Please share what is the
> Spring steel sheet thickness?.
>
> --
>
> You received this message because you are subscribed to the
> Google Groups "OpenPnP" group.
> To unsubscribe from this group and stop receiving emails from
> it, send an email to openpnp+u...@googlegroups.com.
>
> To view this discussion visit

> https://groups.google.com/d/msgid/openpnp/516005ed-e765-409e-993c-6697020eef7bn%40googlegroups.com <https://groups.google.com/d/msgid/openpnp/516005ed-e765-409e-993c-6697020eef7bn%40googlegroups.com?utm_medium=email&utm_source=footer>.

>
>
>
> --
> Wayne Black
> Owner
> Black Box Embedded, LLC
> black...@blackboxembedded.com

> 1.831.682.4964 <tel:(831)%20682-4964>

>
> --
> You received this message because you are subscribed to the Google
> Groups "OpenPnP" group.
> To unsubscribe from this group and stop receiving emails from it, send
> an email to openpnp+u...@googlegroups.com

> <mailto:openpnp+u...@googlegroups.com>.

> To view this discussion visit

> https://groups.google.com/d/msgid/openpnp/652354ac-f695-42a4-a344-e1bac8c5d004n%40googlegroups.com <https://groups.google.com/d/msgid/openpnp/652354ac-f695-42a4-a344-e1bac8c5d004n%40googlegroups.com?utm_medium=email&utm_source=footer>.

[vespaman]

Hi Jan,

peeling and dragging could be separated again as in former
times. With dragging first and peeling second,

Yes, I thought about that, but now my gut tells me higher friction to pull, less everywhere else.

releasing the drag pin which would then fasten the cover tape again. One
could likely extend this using some kind of rubber tube acting as break
that gets pulled by the peeler and loosened when the tape is dragged.

You mean like the solution I drew? Yes, this is likely the best "stop", since it is rather definite.

However, I'm not too keen to go there just yet, since it might open another can of worms.

By accident, I found two coil springs (well, not really accident, I'm always collecting small springs :-) ) that I decided to test, and the new friction seems *perfect*. First try :-o

I know this is not what I was supposed to do - I had decided to put friction outside pick area, well...

 

The springs are off center, I wanted to have most pressure under the sprocket holes, so I can make a groove for the plastic tape pockets, having most pressure going where the sprocket holes are, in order not to be so sensitive to the different pocket depth. However, I just wanted to test the springs first, which is why I just whipped the flat version together first.

Unfortunately, either the colour is not to the likes of the vision (was what I had in my printer), or perhaps the top plate is too close to the holes (this is a feeder that I have had issues with earlier, and don't use bc of this), so the down vision give up on sprocket hole detection, and when I was editing the pipeline, OpenPnP hung, and I had to ctrl-c out of it. Will revisit this later, since it felt very promising.

For paper, feed looked 101%, have not tested real feed on flimsy plastic tape yet. The friction might be too much with deeper pockets, I fear drag operation may tear the tape.

Anyway, this was a good start. Hopefully it is possible to purchase similar springs if they indeed turns out to be perfect.

 - Micael

[vespaman]

Hi Guys,

So, I thought it is time a little update on my road to feeder nirvana..

Since last time, I have evolved my setup a little, and I have run a few boards with this setup now. Only about 60 boards though, so this is just a early heads up.

These where the points that I wanted to address with regards to the original feeders;

1. Better springs and controlled pressure at drag/pick area.
Especially on my machine, the springs where not consistent, and generally too weak. Also, the original springs are pushing on the component pocket/compartment, so the depth of it defines the pressure. (And almost pushes the components out of the pockets).

2. Better overall friction control and reduction.
The main reason for this, is to get rid of especially random/fluctuating friction, in order to getting control over how much the friction wheel needs to pull, and on the other hand, being able to increase the friction at the pick/peel point.

3. Remove intermittent bad behavior of the used carrier tape.
Some carrier tape (esp flimsy plastic carriers with max depth pocket) sometimes flee the disposal slot, and instead where fed out onto the main surface, ending being caught by the gantry if left unattended.

4. Be able to start with a cut tape and start feeding only with one sprocket hole for drag.
In the original setup, this would not work, since the starting edge of the tape would in 95% not arrive to the screws properly. Most likely it will go below, and crash in to the alu blocks, creating a mess.


But in doing this, I also realized that I wanted to have a go at other irritating areas around the feeders.

5. The reels are normally mounted on the alu-rod as you know. This is a complete pain if one need to change during board run, since it has to be lifted up, before pulled out, making the other reels skip their position into their neighbors etc.
It is sometimes possible to run without the alu-rod, but this is not safe esp if the reels are full, and there's space between them.

6. Also, there are some reels that are full to the edge of the reel. This means that the actual components will ride on the blue positional guides, lifting the hole rod, until the first couple of mm has been unspooled from the reel.

7. There's a lot of messing making sure that the tape is pulled over/under bars etc. Esp. on a loaded machine, this means time lost, and possibility to accidentally mess up some other neighboring reels/position in pick area. And it takes valuable time during production days.

So in order to mitigate the above, here what I did..

1. As I had good luck with my first test before Christmas, I decided to order similar springs, before I tweaked too much.
The ones I started with was 0.5mm x 5mm (d) x 7.5mm (h). They where about right, maybe a little long (h). So after spending some time on aliexpress, I found a seller having 0.5x5x5, and they turned out to be good quality, once the arrived early January.
So I tweaked the "back plate", added a slot for the pockets so that the pocket depth would not be the defining factor, and stop flanges so the spring plate would not come out if pulling the tape either way.

The pressure is quite high with this. But there's absolutely no way the cover tape can do unwanted "feed" now. So this means it is much easier to adjust the friction wheel.
I chickened out on the 1.3mm thick paper tape I had on two rolls (large capacitors), by doing special back plates with 1mm less spring preload. I did this, since I did not want to mess with this during the run of my batch.
When I did some 50-100 test feeds, it worked on the 'normal' spring-plates, but still, it felt like too much. So I might continue having two sorts of back plates, or maybe I dare to use the 'normal' as confidence grows.
Super happy with the result, and endlessly better than my crappy original springs.

2a. The main issues here, without a doubt, is the friction on the cover tape glue residues and, I think, some materials of cover tape. Especially on 16/24mm tapes.
With the introduced idlers/rollers described earlier in this thread, the difference is substantial. From the last time, I have added "stop sections" between the rollers, since they tended not to stay in their intended places when unloaded, so it became a abacus.

2b. Another potential friction issue was the reel mounting system mentioned in #6. I'll describe this down below.

3. This is where the drain pipes comes in. Always making sure the tape goes down. This is also part solving paragraph 4.

4. From the beginning I thought I should have some spring loaded solution in the drag area, in order to stabilize the drag. But I gave up on that, reason being that the reference of the tape is on top, which means that any stabilization needs to cover the drag area, which I did not want - I want to see the tape here. And also it got very complicated trying to solve this in general.
 So instead, I have done a small guide, that is on the same level as the metal shelves below the screws. This guide makes sure a tape start cannot deviate from its destiny - it will arrive into the screws and down the drain pipes.
 This guide, I only push down when I have/use cut tapes, to keep it easier to clean, and to avoid potential issues where a larger component 'on the run' is blocking the drag slot in the guide. (Small risk, but still possible).

The above pic is how I started feeder 21, something that would not have been possible before. I think it would be great create an updated pushpullfeeder that could begin with picking a selected amount of open pockets. (Set to 4 in this case). If I ever learn Java, this will be the first thing to try.  :)

Perfect for prototype builds.

 - Micael

[vespaman]

..continuing..

5 & 6:

I removed the rod, the baby blue plastic guides on all three places. On the lower two rods I instead mounted rollers. So now the reels stands on these rollers. No more rod, no more issues with super full reels!
The rollers has the cheapest ball-bearing I could find, one for each slot/axle (this might have been unnecessary, but the cost was so little, so it does not matter, and it means the the fit against the axle is good and stable).

7. I did not like this at all, so I did some long guides that is acting like pipes for the tapes, from the reel to the pick/drag area. So now it is just to push the tape in on the reel side, and it (surprise, surprise) arrives exactly where it should without any messing around.
This guide does add some friction, truth to be told, but at least it is not much, and it is consistent.
The red/black colour coding is to make sure that the reel position matches the guide start and the guide end, so it is easy to see witch reel slot belongs to a particular feeder.
The guide is printed in ESD PETG, and is in pressure with the second rod from the reel side, so it should be properly grounded afaict, but I am thinking of adding a connection to ground on the pick side by printing all ends with ESD PETG, and instead add some red markers on top somehow.
 
With (7) in place, there's no more need to be able to access anything from below the tape feed, so my next, and final project, is to print some cover plates, so that the used tape, cannot arrive up in the reel area any longer.
 However, after doing my first batch, I realized that it does not matter as much anymore, since the guides are acting as a cover themselves, and afaict, there's no longer any risk that machine will try to eat the used tape, which could happen before, when leaving the machine unattended.

So all well now? No, that is not how the world works!
 
I realized that some reels was very flimsy, making it hard to get them into their intended slot - they would easily make their way into their neighboring slot. And it is not so easy to see this, when the machine is fully loaded.
So I designed a special tool for spool loading.

It acts also as a helping hand, since while feeding the tape/cover tape into the guide, the spool rests outside, so one can use both hands to handle/push the tape through the guide entry. Then, is is just to roll the reel into place, and pull the tool.

 
Now, changing a roll is done in a matter of seconds, and a joy. The only thing that takes time now, is to align the sprocket hole, and tie the knot.

I have so far only done the tool for 8mm reels, so I'll do 12 and 16 as well (not sure I need 16, since those reels are not flimsy, but the 'extra hand' can always be good, I suppose).

I have so far only tested all this on the west feeder bank, and are about to do the updates to the east bank as well (with some updates of the parts that I learned from the west bank).

The east bank has a different distance between the feeder pick area and the reels, so I need to change the guides accordingly.

Also, on my east bank, I have my 12" reels, which also needs some special TLC with the guides

[vespaman]

While not part of the above, part of my echo system;

Since before, I have these QR code 'flags', and I thought I should also show how I keep my reels outside of the machine;

The QR flag is fixed on a clip, which is nice, since one can see what component the spool holds, from above, without having to pull/search for each reel. Also the little clip holds the extra cover tape, so no tape is needed.

OK, end of reporting for this time.  :-)

 - Micael

[Wayne Black]

Great work Micael! 

I took a different approach to eliminate the reel binding problem, but my set up is different where the reel are lower than the machine relatively. I like the way you did this :)

Re the spring issue. Wouldn't be easier to forego the coil springs and just use grub screws to adjust pressure? It would be more tuneable I would think

--
You received this message because you are subscribed to the Google Groups "OpenPnP" group.
To unsubscribe from this group and stop receiving emails from it, send an email to openpnp+u...@googlegroups.com.

To view this discussion visit https://groups.google.com/d/msgid/openpnp/9e6f89dc-57ca-440e-8d1d-ad30d3335b3cn%40googlegroups.com.

--

Wayne Black

Owner

Black Box Embedded, LLC

black...@blackboxembedded.com

1.831.682.4964

[vespaman]

Hey Wayne,

I actually think my drag feeder tuning/tweaking days are a thing in the past now. I felt so much confidence with the coil spring solution when I had my test batch of 60 boards. Not one fault, not one tweak of the friction wheel. And both thin plastic as well as paper was good. Even a short cut 'flimsy' plastic tape was securely held. I think (have not tested) that the cover plastic might break if I pull, and there will still be no motion-advance on the tape.

The east side had one or two feeder issues at the beginning of the job, as a comparison. (with untouched springs/no idler wheel etc).

I did some changes on the coiled plates before doing the east side (removed all the earlier coil spring test sockets, tweaked cut corner so drag pin can be started as close to the metal as its diameter will allow, and also changed the side walls angle, so now it is very easy to put them in (have to be tilted 10-40° for the front/back-stoppers to come clear).

Not entirely sure about the 24mm - maybe it needs a support spring on the other side, not to sag.

On thing that I am curios about, is if paper tapes might wear the coil plates down over time, given that the pressure is only on the sprocket hole area. I guess time will tell. Easy to replace if this happens. I was playing with the thought of doing them in aluminium (aluminum for you guys), but I think it is not cost effective.

 - Micael

[Wayne Black]

Well done :)

To view this discussion visit https://groups.google.com/d/msgid/openpnp/fa85a908-c606-439c-a699-f52dec7e672fn%40googlegroups.com.

[Mike Menci]

Just a tip - why not 3D printed with spring - something like this for example;  
https://cults3d.com/en/3d-model/gadget/headset-holder-sliding-with-springs-underdesk

[Mike Menci]

Or under layer of Foam based rubber strip(s)

Magic by iPhon

Dne 20. feb. 2025 ob 21:48 je oseba Mike Menci <mike....@gmail.com> zapisala:

Just a tip - why not 3D printed with spring - something like this for example;  

To view this discussion visit https://groups.google.com/d/msgid/openpnp/652e804d-8322-407c-8c3a-24a0bf7ddcb1n%40googlegroups.com.

[vespaman]

Well, I guess my first question would be "why"?  It is not like it is huge cost, and the performance of metal springs are way better. I think I bought 100 for about 8€

And also, and this might not be very visible in the pics - the space is only a few mm, so it would not be possible to get a good strong plastic spring here, and still make it flex these mm.

(And maybe the needed pressure is higher than you might think)

Although, I must admit that I was very lucky arriving to a good start, with the first spring I chose, being almost perfect. It was easy to iterate from that point on. It can be a time consuming task to iterate springs.

In other situations, 3d printed springs can be good, since they can be made part of the structural stability. Esp. with a little bit bigger parts.

 - Micael