Prometheus for Rapid Prototyping - Forget Everything You Know About PCB Milling

[rocco]

I just released a demo video of my Prometheus prototype making a double-sided PCB with through holes and edge cuts:

https://youtu.be/Tp81Aneil24. The end mill (bit) used in that video was only 7 mils in diameter.

Prometheus is a professional machine that can do 4-mil trace/5-mil space, runs at 50,000 RPM, and has a max feed rate of 150 IPM (3,810 mm/min). I engineered it to compete on specs with other pro machines in the $8,000 USD ballpark, but Prometheus will retail for $2,299 (currently in pre-order for $1,799). More pics are on the site, www.zippyrobotics.com.

[fonograph]

that looks very nice,I have noob question.If I have your Prometheus machine,does that mean I can make my own PCB in my home? Like,if I buy the PCB material,I think its called prepreg or something like that from Rogers,I can make PCB myself?

[rocco]

Thank you! Yes, you could use Prometheus to make PCBs at home, although it's more applicable to businesses and universities that want to prototype something same-day, or maybe multiple iterations in one day, as opposed to sending it off to a board house and waiting. That said, there are a few home hobbyists that have pre-ordered a machine. It just depends on how much you value the time savings.

As for materials, you would start with what's called "copper-clad". Copper-clad board comes in a variety of substrates - you could get FR-4, FR-1, or one of many Rogers materials. If the copper-clad you use has copper on both sides, you could make a double-sided circuit board.

[TJ232]

Looks like a very nice and clean implementation 

the youtube link above is not working. remove the end dot (.) from it:

https://youtu.be/Tp81Aneil24

[kripton2035]

idea for v2 : add a small dust extraction hose close to the drilling bit...
very nice machine !

[rocco]

Looks like a very nice and clean implementation 

the youtube link above is not working. remove the end dot (.) from it:

https://youtu.be/Tp81Aneil24

Thank you! Between the machine, software, and electronics, it's been 3 years in the making.

Thanks for letting me know about the link too; I fixed it.

[rocco]

idea for v2 : add a small dust extraction hose close to the drilling bit...
very nice machine !

Thanks!
I agree about the dust extraction - there's a small factory near me that does very affordable 3D printing in volume so once I design the mount that accepts the vacuum hose, I'll be able to add that.

[mikeselectricstuff]

Doesn't look like it's optimising the drill sequence - very noticeable on the DIPs

[mikeselectricstuff]

Looks like a nice implementation, but with the availability of many different PCB services, and trend for everything to be smaller, I wonder how much of a market there is for  in-house boards with no plated through-hole, only 2 layers and no resist ( and fairly limited size by the looks of it).

[rocco]

Looks like a nice implementation, but with the availability of many different PCB services, and trend for everything to be smaller, I wonder how much of a market there is for  in-house boards with no plated through-hole, only 2 layers and no resist ( and fairly limited size by the looks of it).

Thank you. The trace/space capability can support pretty much every SMD. BGAs may be the exception. You could probably even use very low ball-count BGAs that don't require breakout into 4+ layers. That's something I wanted to try to demo. Or an ATtiny in a WLCSP 

It's true that the services are a great solution if you don't mind waiting (most hobbyists, for example, would fall under that category) but for some engineers/businesses/schools they value the turn-around time. I don't view in-house prototyping and board houses as mutually exclusive. Just like you can prototype flyers or business cards on your home inkjet and then go online to get thousands made, there is a happy place for both methods and each have their pros and cons.

I do have a plan for a solder mask

Size is 6" x 4" which I think supports what most people would want to do with it. I mean, if you browse Adafruit, Sparkfun, or sites that make modules for Arduino or Raspberry PI, or anything really, how often is something bigger than that nowadays, with miniaturization? That's my estimate anyway, but maybe I'm wrong about that. If so I'd like to hear people chime in with the most common board sizes they design.

There's a lot you can do with two layers, but of course you are right that it's a limitation. At present day there's no real rapid prototype solution around that. Again, I plan to work on that 

Oh - and you're right about the drill sequence. Currently ProCAM drills them in the order that your Gerber exporter wirtes them into the file. It can easily be improved. Right now I have to prioritize the basic required feature set and software stability before I begin optimizing code.

[sprok]

.

[rocco]

What's the minimum track width/spacing this mill can do?


PCB services are stupidly expensive and slow for prototyping and hobby use.

I have demonstrated 4-mil trace widths, 5-mil spacing. Here is a pic of a test board that shows this. All gaps throughout the board are 5 mils wide: http://www.zippyrobotics.com/wp-content/uploads/2017/03/IMG_0944.jpg

It may be able to go even smaller, but I have not tried that yet.

[Mechatrommer]

Looks like a nice implementation, but with the availability of many different PCB services, and trend for everything to be smaller, I wonder how much of a market there is for  in-house boards with no plated through-hole, only 2 layers and no resist ( and fairly limited size by the looks of it).

i believe its the same market as how much people buying copper etchant in hobby quantity this days. not every place can order pcb services at cheap cost (time and money) esp for one prototype board and just a bit slower than "less than a day" completion time. we usually trade time with effort to stitch vias by hand (or just double side solder if we have components through that holes), and we cant bare to think of receiving 10 boards from the cheapest pcb house in China in weeks time just to realize we have design flaw in them yeah i experienced this, so much wasted boards with nice silk screen here. otoh we have processes to plate vias available in youtube for anyone interested in playing chemicals, or some "punched in copper tube" thats available in ebay. resist or solder mask or silk screening can be done in different process from this machine. i just watched Dave's uCurrent board production youtube, now i have some idea

btw i built my Bethan (similar functionality as Prometheus) at somewhere 1/10X the cost or less i guesstimate. the difference is it doesnt mill copper out, it only mill out etch resist ink applied earlier on the pcb board for later process of the usual board immmersed in etchant process. that method where i can reduce the cost of the machine the most by ruling out the necessity of very rigid machine. and i'm not a fan of pc connected machine so control is made from sd card and a number of GCode files for alignment etc. and i dont have to waste too much excess copper board to do the alignment housekeeping job, i can cut to the board size, 0 border excess, make the alignment manually, and etch i go with different GCode file. https://www.eevblog.com/forum/projects/hackable-3d-printer-for-research/msg1188994/#msg1188994

about optimising the drill sequence as you mentioned, i'm not sure what you meant and why its necessary, maybe you can give us some idea so we can implement it in our GCode generator PC SW. fwiw...

[Pitrsek]

We have one of the more expensive LPKF PCB machines(vacuum table,automatic tool changer,camera) in my day job. I can not imagine going back to days when I would wait days for pcb for experiments and proof of concepts  ...
In house proto manufacturing IMHO absolutely worth it .
Every time the guy who is in charge of the machine is off, or the mill has some problems(wit our heavy use, it breaks time to time...), there this sense of impeding doom among electronics designers
We have team of cca 25 electronics designers, and most of the day the PCB mill is working all day long. Also our colleagues from mechanical department discovered that you can make nice stuff from plexi and POM, so the CNC is used for this as well. We have used it for qfn, and even some BGA no problems.
For vias we use rivets and wire stitching
 
Regarding the working area - I would guess that 70% of the stuff(PCBs, no the mechanical stuff, that one is usually bigger) we do would fit.
For fine pitch stuff, I would argue that solder mask is more important than really fine trace/separation.
There absolutely needs to be a vacuum tube holder/ some other means of get rid of chips. 

Have you considered adding laser for exposing photo resist/solder mask?
Nice hack with the oil, what happens if I put more then optimum amount? Will it flow all over the machine, or is there defined way into some container?
How long does it take to change tool? Auto-changer would be a great upgrade.

[mikeselectricstuff]

What's the minimum track width/spacing this mill can do?


PCB services are stupidly expensive and slow for prototyping and hobby use.

Seriously ? What planet have you been on the last few years?
Have you not heard of OSHPark, Seeedstudio, PCBway etc. etc. ?

[mrpackethead]

We have one of the more expensive LPKF PCB machines(vacuum table,automatic tool changer,camera) in my day job. I can not imagine going back to days when I would wait days for pcb for experiments and proof of concepts  ...

I also had an expensive LPKF system and it was the biggest pile of junk and waste of money that i could ever imagine.   Terrible support from teh OZ agent,  who when it got returned to them to fix it, informed me that they dropped it off the forklift..    But seriously, with two or three day turn around on pcbs, and the need for 4 layers, milling is'nt an option any longer.    For any serious design work, i can't see it being an option any longer.

[Mechatrommer]

those fancy pcb houses have their own fixed format pcb in order to get cheap, once we need slightly bigger board, the price go up exponentially. for non local customer, shipping cost can just the same price as the item price, double or tripple if using express service. not justificable in design/prototyping process, mass production of proven design maybe.

[mikeselectricstuff]

about optimising the drill sequence as you mentioned, i'm not sure what you meant and why its necessary, maybe you can give us some idea so we can implement it in our GCode generator PC SW. fwiw...

It's moving more than it needs with the drill up.
e.g. for a DIP8 :
It appears to be drilling pin 1,8,2,7,3,6,4,5, so moving 7 x 0.3" plus 3 x 0.1" = 2.4"
if it did 1,2,3,4,5,6,7,8 , it would move only 6x0.1" + 1 x 0.3" = 0.9"

Although getting a completely optimal solution is hard ( travelling salesman problem), getting a huge improvement is very easy.
Simply :
Drill first hole, mark as drilled.
Move to nearest undrilled hole.
Repeat.

if you don't have enough memory to store the whole list, or the list is so long that searching for next undrilled hole is slow, split in to blocks and optimise within each block.

 
 

[nctnico]

We have one of the more expensive LPKF PCB machines(vacuum table,automatic tool changer,camera) in my day job. I can not imagine going back to days when I would wait days for pcb for experiments and proof of concepts  ...
In house proto manufacturing IMHO absolutely worth it .
Every time the guy who is in charge of the machine is off, or the mill has some problems(wit our heavy use, it breaks time to time...), there this sense of impeding doom among electronics designers

So your company is spending a whole year's salary + maintenance costs on having PCBs made quickly. How many next day PCBs with metalisation, solder mask, silkscreen, etc can you have made for that amount of money?

[rocco]

We have one of the more expensive LPKF PCB machines(vacuum table,automatic tool changer,camera) in my day job. I can not imagine going back to days when I would wait days for pcb for experiments and proof of concepts  ...
In house proto manufacturing IMHO absolutely worth it .
Every time the guy who is in charge of the machine is off, or the mill has some problems(wit our heavy use, it breaks time to time...), there this sense of impeding doom among electronics designers
We have team of cca 25 electronics designers, and most of the day the PCB mill is working all day long. Also our colleagues from mechanical department discovered that you can make nice stuff from plexi and POM, so the CNC is used for this as well. We have used it for qfn, and even some BGA no problems.
For vias we use rivets and wire stitching
 
Regarding the working area - I would guess that 70% of the stuff(PCBs, no the mechanical stuff, that one is usually bigger) we do would fit.
For fine pitch stuff, I would argue that solder mask is more important than really fine trace/separation.
There absolutely needs to be a vacuum tube holder/ some other means of get rid of chips. 

Have you considered adding laser for exposing photo resist/solder mask?
Nice hack with the oil, what happens if I put more then optimum amount? Will it flow all over the machine, or is there defined way into some container?
How long does it take to change tool? Auto-changer would be a great upgrade.

Nice! I'm curious - what kind of work does your company do?

I'm considering a few methods to give Prometheus the ability to make solder masks, laser being one of them but I first want to see if I can do it without adding more components to the machine. We'll see
I did solder the .5 mm-pitch QFP for the board in the video by hand without a solder mask without a problem. I just had to check it with a microscope and correct some bridging, but it wasn't bad. Of course a solder mask would be better, and I really want the capability for myself so I'm going to try to make that happen.

The oil will just flow onto the table if you put too much but it's not a big deal to clean up. The oil is great for milling operations. Prometheus can run at higher feed rates and it keeps the chips/dust in the oil and thus out of the air.

Changing tools takes maybe 10 seconds because you don't deal with wrenches. You can see it in the first part of the video. Yes, auto tool change would be a cool upgrade.

[Kjelt]

Great that you can build such a nice piece of equipment and software.
However I think you are a bit late, these kind of machines have seen their days in the past.
Nowadays with multilayer smt pcb's you need lots of (stitching) via's multiple inside power/ground planes etc.
Also I wonder if electrically these pcb's will perform identical to the produced pcb's esp. with higher frequency designs, because they are physically just not the same.

[sprok]

.

[rocco]

We have one of the more expensive LPKF PCB machines(vacuum table,automatic tool changer,camera) in my day job. I can not imagine going back to days when I would wait days for pcb for experiments and proof of concepts  ...
In house proto manufacturing IMHO absolutely worth it .
Every time the guy who is in charge of the machine is off, or the mill has some problems(wit our heavy use, it breaks time to time...), there this sense of impeding doom among electronics designers

So your company is spending a whole year's salary + maintenance costs on having PCBs made quickly. How many next day PCBs with metalisation, solder mask, silkscreen, etc can you have made for that amount of money?

Just to add some more info to this topic, the cheapest and fastest board houses I found via www.pcbshopper.com when searching for a 5"x4" board (which btw, takes 2 days total to receive) is $503.16, including shipping cost. There might be cheaper out there (or maybe you work across the street from a factory), but PCBShopper aggregates prices from many houses and that's the fastest/cheapest that it showed me (see attached screenshot).

You have to factor in the time value into the equation and when you do, this absolutely does make sense for some people. When the machine is $2,300 and not $17k-$20K, it makes sense for an even larger group.

[rocco]

Seriously ? What planet have you been on the last few years?
Have you not heard of OSHPark, Seeedstudio, PCBway etc. etc. ?

Some people say this but I think you're all on some hard core drugs. Please tell me any PCB service can give me an 6x4 board in 30 minutes for $2. Or a one-off 2x2 board for $0.25. It's not even possible, let alone at a price that isn't astronomically higher!

You're right sprok. I've found that disagreement always occurs if people don't assign a dollar value to the time they spend waiting, and if they don't realize that this time value varies greatly between users. For some, Prometheus or other rapid prototyping machines don't make sense at all. For others, they do.

[nctnico]

We have one of the more expensive LPKF PCB machines(vacuum table,automatic tool changer,camera) in my day job. I can not imagine going back to days when I would wait days for pcb for experiments and proof of concepts  ...
In house proto manufacturing IMHO absolutely worth it .
Every time the guy who is in charge of the machine is off, or the mill has some problems(wit our heavy use, it breaks time to time...), there this sense of impeding doom among electronics designers

So your company is spending a whole year's salary + maintenance costs on having PCBs made quickly. How many next day PCBs with metalisation, solder mask, silkscreen, etc can you have made for that amount of money?

Just to add some more info to this topic, the cheapest and fastest board houses I found via www.pcbshopper.com when searching for a 5"x4" board (which btw, takes 2 days total to receive) is $503.16, including shipping cost. There might be cheaper out there (or maybe you work across the street from a factory), but PCBShopper aggregates prices from many houses and that's the fastest/cheapest that it showed me (see attached screenshot).

You have to factor in the time value into the equation and when you do, this absolutely does make sense for some people. When the machine is $2,300 and not $17k-$20K, it makes sense for an even larger group.

I have two sort-off generic remarks about what you write:
1) The LPKF machines are notorious for needing a lot of baby sitting and learning time to get them to run right (there are various factors contributing to this but let's not get into that in this thread). This has given PCB milling a bad reputation in general which is hard to combat especially with low cost machines.

2) When you need boards quick then there usually is something wrong with the planning. When I work on a project I get the PCB design out first, work on a different (part of the) project and continue when the PCBs arrive. In many cases the PCBs are waiting for me instead of the other way around. I can't see how a board which is designed in a rush will work properly anyway. That leaves milling interesting for small prototype boards which may as well be build on veroboard and RF microstripline circuits.