Author Topic: Prometheus for Rapid Prototyping - Forget Everything You Know About PCB Milling  (Read 48324 times)

0 Members and 1 Guest are viewing this topic.

Offline Kjelt

  • Super Contributor
  • ***
  • Posts: 6459
  • Country: nl
Seeing this movie I do think it is backlash that I am experiencing and the only way to correct is to upgrade to closed loop steppers or continuously calibrating with optical lineal or something similar , making it cost more than the cnc mill cost  :o
 

Online Mechatrommer

  • Super Contributor
  • ***
  • Posts: 11535
  • Country: my
  • reassessing directives...
ok i think that a backlash buildup not considered in the FW... anyone can make the compensation in the FW but... well as you realized, nothing beats recalibrating it with our own hand.
Nature: Evolution and the Illusion of Randomness (Stephen L. Talbott): Its now indisputable that... organisms “expertise” contextualizes its genome, and its nonsense to say that these powers are under the control of the genome being contextualized - Barbara McClintock
 

Online nctnico

  • Super Contributor
  • ***
  • Posts: 26752
  • Country: nl
    • NCT Developments
ok i think that a backlash buildup not considered in the FW... anyone can make the compensation in the FW but... well as you realized, nothing beats recalibrating it with our own hand.
I think it is not just backlash but also missing steps due to the mass the stepper engines have to put into motion. I don't hear any ramp-up or ramp-down. IMHO you shouldn't use stepper motors in an open loop configuration. That is asking for problems.
There are small lies, big lies and then there is what is on the screen of your oscilloscope.
 

Online Mechatrommer

  • Super Contributor
  • ***
  • Posts: 11535
  • Country: my
  • reassessing directives...
ok i think that a backlash buildup not considered in the FW... anyone can make the compensation in the FW but... well as you realized, nothing beats recalibrating it with our own hand.
I think it is not just backlash but also missing steps due to the mass the stepper engines have to put into motion. I don't hear any ramp-up or ramp-down. IMHO you shouldn't use stepper motors in an open loop configuration. That is asking for problems.
miss steps is possible. even if you manage to get a stepper motor in closed loop (with encoder), backlash is mechanical problem, the error will still build up. there is anti backlash bearing claim to solve this issue.
Nature: Evolution and the Illusion of Randomness (Stephen L. Talbott): Its now indisputable that... organisms “expertise” contextualizes its genome, and its nonsense to say that these powers are under the control of the genome being contextualized - Barbara McClintock
 

Online nctnico

  • Super Contributor
  • ***
  • Posts: 26752
  • Country: nl
    • NCT Developments
I don't see how backlash can accumulate mechanically because that would mean the size of gears, lengths of lead screws, etc would have become physically different. If there is an accumulated error then this has to do with missing steps, wrong position readback and/or rounding errors in the firmware.
There are small lies, big lies and then there is what is on the screen of your oscilloscope.
 

Online Mechatrommer

  • Super Contributor
  • ***
  • Posts: 11535
  • Country: my
  • reassessing directives...
Seeing this movie I do think it is backlash that I am experiencing and the only way to correct is to upgrade to closed loop steppers or continuously calibrating with optical lineal or something similar , making it cost more than the cnc mill cost  :o
[youtube link cut to save space]
i just realized that video showing 0 backlash/miss steps. you must have some problem with your machine, too heavy head or too small motor. here is backlash/miss steps problem with $200 mdf built machine...

« Last Edit: April 23, 2017, 06:07:28 pm by Mechatrommer »
Nature: Evolution and the Illusion of Randomness (Stephen L. Talbott): Its now indisputable that... organisms “expertise” contextualizes its genome, and its nonsense to say that these powers are under the control of the genome being contextualized - Barbara McClintock
 

Offline usagi

  • Frequent Contributor
  • **
  • Posts: 390
  • Country: us
I don't know the exact english term, we call it "speling" on the bearings probably slack or backlash.

that's not backlash. that's skipped steps. backlash does not accumulate, it is a fixed error.

your video shows a large mass being moved by steppers and it appers you have zero deceleration, you are trying to full reverse at full speed and this simply is not possible without skipping steps. in the 3d printing terms, your jerk setting is much too high.

stepper motors don't need closed loop if you use them correctly. that is kind of the point of stepper motors.  ;)
« Last Edit: April 23, 2017, 09:28:51 pm by usagi »
 

Online Mechatrommer

  • Super Contributor
  • ***
  • Posts: 11535
  • Country: my
  • reassessing directives...
yeah you are right but the video is not his... no trace of miss steps there...
Nature: Evolution and the Illusion of Randomness (Stephen L. Talbott): Its now indisputable that... organisms “expertise” contextualizes its genome, and its nonsense to say that these powers are under the control of the genome being contextualized - Barbara McClintock
 

Offline ebclr

  • Super Contributor
  • ***
  • Posts: 2328
  • Country: 00
Use deacceleration and acceleration routines will reduce miss steps
 

Online nctnico

  • Super Contributor
  • ***
  • Posts: 26752
  • Country: nl
    • NCT Developments
stepper motors don't need closed loop if you use them correctly. that is kind of the point of stepper motors.  ;)
In theory. Wait until the drill bit gets stuck or something else goes wrong which makes the stepper miss a step. AFAIK this is where the (older/cheaper) LPKF machines fail as well.
There are small lies, big lies and then there is what is on the screen of your oscilloscope.
 

Offline Dubbie

  • Supporter
  • ****
  • Posts: 1114
  • Country: nz
When I first got into electronics, I thought a pcb mill would be great. So I built a very nice one. It has ground ballscrews, double ball nuts, well overspecced linear rails etc etc. it is extremely accurate and fast. However I have cut the grand total of 1 board with it. The pain the neck and vastly sub par result compared to a commercial etched board put me off. Now I use the machine for general machining of housings etc. it works great for that!

Also on the topic at hand, steppers don't miss any steps if they are operated within their design constraints. If you have sized them properly to the task you can expect to never miss a step unless you have a crash.
 

Offline 691175002

  • Regular Contributor
  • *
  • Posts: 64
stepper motors don't need closed loop if you use them correctly. that is kind of the point of stepper motors.  ;)
In theory. Wait until the drill bit gets stuck or something else goes wrong which makes the stepper miss a step. AFAIK this is where the (older/cheaper) LPKF machines fail as well.

Servos won't solve that problem, only detect that something has gone wrong.  If you lose position the chances are that something has snagged, broken, or been cut in the wrong place and that can't be undone.

Fault detection is okay when continuing a job might damage the tools or machine, but does't really provide many advantages in a PCB mill.  The real advantage of servos is that a tight control loop is orders of magnitude more precise and rigid than a stepper motor, and you aren't running your motors at full current regardless of load.


I think if you could somehow do automatic vias prototyping would become very attractive, but I'm not even sure how one would approach that feature.

« Last Edit: April 23, 2017, 11:17:23 pm by 691175002 »
 

Offline usagi

  • Frequent Contributor
  • **
  • Posts: 390
  • Country: us
stepper motors don't need closed loop if you use them correctly. that is kind of the point of stepper motors.  ;)
In theory. Wait until the drill bit gets stuck or something else goes wrong which makes the stepper miss a step. AFAIK this is where the (older/cheaper) LPKF machines fail as well.

if you miss a step or get stuck you are screwed even with feedback -- you will still have mangled output. feedback doesn't prevent that. in my experience you will break a bit sooner than you will skip steps.

run your steppers properly within their constraints, no problems.

steppers without feedback are sufficient for many $50,000 industrial machines. they're certainly ok for $2000 hobbyist machines.

Online G0HZU

  • Super Contributor
  • ***
  • Posts: 3012
  • Country: gb
Quote
You can test an idea within a single day, that is priceless.
I agree and I don't understand why the 'experts' here can't grasp this. They might as well claim that building a test board in dead bug style is a waste of time too because you can go to a PCB house instead and get a PCB with silk screen and vias ;D

Maybe if they spent some time in a quiet corner they might eventually realise that there is a need (for some people) to be able to do rapid prototyping that takes the technology beyond dead bug style.

This could be for a rapid fix to a problem with a production board. eg a problem is spotted and a meeting takes place and the decision is to make a little piggy bodge board to prove the concept of the 'fix'. The chap who operates the mill often designs the piggy PCB in cases like this and this speeds things up. He may well fit components to the PCB and test it in time for a technical meeting the next morning.

How are you going to compete with that using a PCB house? It just isn't going to be as quick and if the piggy board needs a tweak to the circuit (or to make it fit in a tight corner) then you lose another two days to the PCB house.

In my case I might need to have a PCB milled rapidly to test out a special RF filter. This could be a regular printed filter or something more exotic using defected grounding. ...but the benefits here are the same. With a PCB mill you can do several iterations in a single day if needed?

Quote
In theory. Wait until the drill bit gets stuck or something else goes wrong which makes the stepper miss a step. AFAIK this is where the (older/cheaper) LPKF machines fail as well.
My T-Tech machine is over 20 years old and still operates on the original mechanicals and the only way the stepper motors play up is if I have recently serviced it and the thread on the linear ways gum up with excess grease. So each time I service it I have to make sure there is not any excess grease and I also run it back and forth several times in X and Y to make sure it won't gum up.

At work our LPKF machine must have milled a lot of PCBs in the time we have had it. I would guess well into the thousands. Plus various panels and spacers and tools. I think it must be getting towards 15 years old because the CAM SW that came with it is dated 2001/2. Our first T-Tech mill was purchased by a director back in about 1992 and we bought another one a few years later. Then we upgraded to the LPKF in about 2002.
What are we doing wrong in order to get such success? Have we just been lucky for 25 years? Because according to the 'experts' here we should have thrown these machines into the skip years ago?




« Last Edit: April 24, 2017, 12:51:43 am by G0HZU »
 

Online G0HZU

  • Super Contributor
  • ***
  • Posts: 3012
  • Country: gb
Quote
You didn't miss it in the video - there is no pressure foot because Prometheus doesn't need one! Instead, ProCAM runs a ~90 second probing cycle before milling begins, where the tip of the tool moves down until it makes electrical contact with the copper-clad surface. When it does, it records the height with 1.25-micron resolution. After probing a grid of points, ProCAM forms a surface map behind the scenes and automatically tells Prometheus how to adjust the Z axis on the fly to keep a consistent milling depth. Basically, it follows the peaks and valleys of the board on its own. This has proven to produce fantastic results and users don't have to worry about pressure foot limitations or the pad wearing or snagging or anything like that. It just works.

Sounds impressive. How well will that work with very thin PCB material that tends to bow/flex upwards? eg 0.02" Rogers 4003C or thinner?
The thinnest RF PCB material I've milled on my old T-Tech machine was two sided copper with a 60 micron thick dielectric. It was like copper paper. However, that is very much a rare extreme :)

 

Offline mrpackethead

  • Super Contributor
  • ***
  • Posts: 2845
  • Country: nz
  • D Size Cell
Quote
You can test an idea within a single day, that is priceless.
What are we doing wrong in order to get such success? Have we just been lucky for 25 years? Because according to the 'experts' here we should have thrown these machines into the skip years ago?

(a) perhaps the jobs are reletively simple.
(b) I'd challenge you to think about your overall workflow.
On a quest to find increasingly complicated ways to blink things
 

Online G0HZU

  • Super Contributor
  • ***
  • Posts: 3012
  • Country: gb
Quote
(a) perhaps the jobs are reletively simple.
(b) I'd challenge you to think about your overall workflow.

It doesn't matter if the jobs are simple and in the real world you can't plan for a perfect workflow. Our production RF boards are typically a complex sandwich of Rogers material on the outer layers with a multilayer FR4 core. The one off price might be £1000 for a bare prototype PCB but some of the larger boards cost many thousands each. The mill can be used as a rapid prototype machine to try out new/alternative concepts for the various circuits that will eventually go on the finished RF PCB. Sometimes we want to experiment with alterations to the copper on these exotic boards but this can be done on a milled copy of the relevant section.

We generally go to Graphic PLC or Exception PCB for our production boards but we also use Labtech and Trackwise for some of our RF boards. Eg metal backed RF boards for RF power amplifier design. But despite this we still have a PCB mill and it is in regular demand. You don't have to just convince me that the mill is a waste of time, you would have to convince many of my colleagues over a 25 year span :)

« Last Edit: April 24, 2017, 01:47:36 am by G0HZU »
 

Online Mechatrommer

  • Super Contributor
  • ***
  • Posts: 11535
  • Country: my
  • reassessing directives...
(a) perhaps the jobs are reletively simple.
(b) I'd challenge you to think about your overall workflow.
if i am a designer of a pentium xeon extreme motherboard with length matching, differential pair traces with anual margin in at least 5 digits from the job, then i will be in the same boat as a bunch of you. but i'm not. maybe this thread is about a fight between professionals and hobbiests where professionals chime in to tell how suck the machine is. well, hobbiests should get a life. ::)
Nature: Evolution and the Illusion of Randomness (Stephen L. Talbott): Its now indisputable that... organisms “expertise” contextualizes its genome, and its nonsense to say that these powers are under the control of the genome being contextualized - Barbara McClintock
 

Offline ebclr

  • Super Contributor
  • ***
  • Posts: 2328
  • Country: 00
This machine is a total nonsense, making pcb's with UV Light is very easier,  faster, better results and cheap.
 

Offline mrpackethead

  • Super Contributor
  • ***
  • Posts: 2845
  • Country: nz
  • D Size Cell
It doesn't matter if the jobs are simple and in the real world you can't plan for a perfect workflow.

Oh yes, you can.   You've absolytely confirmed my suspsisons.

Do we get it wrong.  Yes of course we do.  But  we get it wrong a lot less because we take the time to catch as many problems as pssible before we commit to somethign physical.

Anyway, glad that milled pcb's work for you.  My suspision is that for most folks. its not viable.
On a quest to find increasingly complicated ways to blink things
 

Offline janoc

  • Super Contributor
  • ***
  • Posts: 3781
  • Country: de
your average hobbyist doesn't need 1hr turnaround, and wants something like oshpark silkscreened PCBs with vias anyway.

That's not really true, I would love to have 1h turnaround time as a hobbyist. OTOH, I am certainly not going to spend so much money for a small dedicated machine doing only PCB milling.
 

Online mikeselectricstuff

  • Super Contributor
  • ***
  • Posts: 13695
  • Country: gb
    • Mike's Electric Stuff
Quote
You can test an idea within a single day, that is priceless.
I agree and I don't understand why the 'experts' here can't grasp this. They might as well claim that building a test board in dead bug style is a waste of time too because you can go to a PCB house instead and get a PCB with silk screen and vias ;D

If you want to test an idea, you rarely NEED a PCB to do it.
Youtube channel:Taking wierd stuff apart. Very apart.
Mike's Electric Stuff: High voltage, vintage electronics etc.
Day Job: Mostly LEDs
 

Online mikeselectricstuff

  • Super Contributor
  • ***
  • Posts: 13695
  • Country: gb
    • Mike's Electric Stuff
This machine is a total nonsense, making pcb's with UV Light is very easier,  faster, better results and cheap.
Totally agree. The only downside of this is the drilling, but it's way cheaper in terms of initial investment, and by the time you factor in tools & wastage is probably not much more expensive in materials.

I don't think anyone would argue that there is no demand for in-house PCB making. There will always be some niches where it makes total sense.
But it will always be a niche market, for users that need it quickly and can live with the substantial limitations.
A lower cost and/or better performing solution is always welcome, and will increase the size of the market for whom it makes sense, but it's still a small market and always will be.

Youtube channel:Taking wierd stuff apart. Very apart.
Mike's Electric Stuff: High voltage, vintage electronics etc.
Day Job: Mostly LEDs
 

Online nctnico

  • Super Contributor
  • ***
  • Posts: 26752
  • Country: nl
    • NCT Developments
Quote
You can test an idea within a single day, that is priceless.
I agree and I don't understand why the 'experts' here can't grasp this. They might as well claim that building a test board in dead bug style is a waste of time too because you can go to a PCB house instead and get a PCB with silk screen and vias ;D
If you want to test an idea, you rarely NEED a PCB to do it.
Unless it is for RF where the PCB itself is a component. But then again you can do a lot using RF simulation software these days and get results within minutes.
There are small lies, big lies and then there is what is on the screen of your oscilloscope.
 

Offline fcb

  • Super Contributor
  • ***
  • Posts: 2117
  • Country: gb
  • Test instrument designer/G1YWC
    • Electron Plus
It's really simple. If there is a demand for the product at that price point then the OP will make money.

If the demand is high, then they will hopefully make tons of money as we flock to buy something that we have largely survived without all these years.  If the demand is low, then so be it.

I have a three axis CNC here, which gets used once or twice per year for CNC things. Only once did I try to mill a board on it - it was painful. Also I have several kg's of ferric chloride and a large UV light box, used it once perhaps 15 years ago on an RF project.  These days I do little prototyping (dead bug or other methods), lots of simulation (LT Spice) - mainly I spin boards and wait a week, much more efficient use of my time.
https://electron.plus Power Analysers, VI Signature Testers, Voltage References, Picoammeters, Curve Tracers.
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf