Best Low-Cost Home PCB Prototyping?

[Johnyradio]

Hi

Looking for best low-cost home PCB method.
(not mail-order service)

What's your recommendation?

My Requirements:
(not saying this must apply to you or anyone else)

• Cost: About $100 or $200 ish ($300 is pushing it, but would consider. $500 is too high).
• Gear Size: Desktop, small footprint.
• PCB size: at least 5" per side.
• Volume: Low (approx 10 PCB's per day, not 100's or 1,000's).
• Activity: Just for PCB's (eg, if you say "you can get laser-cutting, 3D printing, vinyl-cutting, milling, etc etc for only $1,000!", thx but too expensive.).
• Fiddliness: Minimal. Should not require a lot of annoying extra steps.
• Finickiness: Minimal. Should not depend too much on "getting it right".
  
• Turnaround: an hour max per board.
• Proven: not "maybe you could try...".
• Precision: can easily make SMD pads and traces.
• Chemical etching: acceptable (and you don't have to include etching instructions-- just method to prep the copper. That's the challenging part).
• Trace style: If it matters, my style is to remove very little copper.
• Quality: Reliable precision, accuracy. Clean traces.
• 2-sided: Nice to have, but not required.
• Availability: Current. (Kickstarter products are ok if they are actually shipping now).
• Links to actual parts, specific brands, instructables, etc.
• If DIY (eg modding an inkjet printer):
  • 10 hours max build-time (a week or two, in evenings/weekends. A month is too long).
  • No hard-to-find parts.
  • No fancy shop equipment needed to build it.
  • Well-validated, documented, kinks worked-out, no "experimenting".

Possible Options:
(does not have to be any of these)

• Mill: remove copper (but cheapest mill i can find exceeds my budget).
• Laser cutter: remove resist (there are $100 laser cutters on ebay, but do they suck?)
• 3D printer: print copper? (not sure this exists).
• Inkjet: print resist (but, must be strong method-- most inkjet methods i've seen are too fiddly, finicky, and/or sloppy).

Thx!

 

[kellys_eye]

All you'd end up with is a routed (or etched) circuit board, un-drilled and with no solder mask and no printing and for multiple boards, presumably for sale? these are not options you can forgo.

With the already-available low cost options for 'professional' pcb manufacture (often very little more than the cost of the bare board material) you won't get a home system for those numbers of boards with a TCO (total cost of ownership) much less than a bought-in one.

Then there is the double-sided issue (or multi-layer), trace width management and thru-hole plating??????? Multi-sized drill holes?

You need to re-assess what you get for your money.

There's even a price comparison website for pcb's !!!!

https://pcbshopper.com/

 

[Johnyradio]

All you'd end up with is a routed (or etched) circuit board, un-drilled /

Hi, thx for reply.

These are for quick home prototyping, not for sale. Don't need solder mask, hole plating, text printing.

Thx.

 

[kellys_eye]

You say....

Volume: Low (approx 10 PCB's per day

10 pcb's per day is stretching 'home manufacture' somewhat. In fact I'd consider that a mini-manufacturing outfit!

The standard UV exposure box or Laser Transfer method is what I use - maybe, on average, 3/month - and even then I find it a PITA...... if I was more patient I'd get them from a cheap Chinese supplier - MUCH better than anything I can make, not much more expensive and, if I really thought about it, no inconvenience if I spent the waiting time on the 'next' project.

 

[Johnyradio]

I'd get them from a cheap Chinese supplier

I understood the first time you suggested that. No need to repeat yourself. Yes, i know about mail order. But as i said, not interested in mail order. Definitely great option for a different use-case.

This is for fast prototyping (hence the word "prototyping" in the title of this thread). Again, this is not for boards to be sold, so i don't need commercial bells and whistles like solder mask, printed text, etc etc.

I'm not likely to do 10 boards per day, prolly much less. But an acceptable solution should be able to handle that much without difficulty. Just like, when you pick components, you pick components that can handle more than you throw at them, for a comfortable margin of performance.

UV exposure box

Thx for mentioning UV box. That's another possible option to consider. But, a 1-step process (eg milling) would be more convenient.

cheers

 

[kellys_eye]

I suggested that I'd get them from a Chinese supplier - what YOU do I couldn't give a rat's arris.

There are just two popular basic home manufacturing methods that produce acceptable results - UV exposure and laser transfer (iron on). Both methods are well documented, easy to achieve and suited to small production such as you require.

If any other method delivered consistently reliable results at a price we could afford then it would be readily identifiable by a quick search. The fact that there isn't says a whole lot.

Of all the methods I've tried over the years the UV route has always give the most consistent results and the availability of pre-coated board material makes life that much easier. The majority of the process is the etching - where a decent air-bubble or etchant-washing process gives a quick etch time and long(ish) life for the etchant PLUS reduces the potential for mess (an inevitable consequence of home manufacture).

The 'esoteric' methods are, imho, not fully debugged enough nor worth the additional effort as the end result, when compared to the simple UV method doesn't reveal a 'magnitude' of improvement - which I would only consider to be a valid reason to move to a different production method.

Your budget is the limiting factor. Spend your budget on the established production methods and get good results or spend it on edgy stuff and be prepared to fiddle with it endlessly (and frustratingly) to get the same results.

But at the $100-200 budget level you're looking at homebrewing the stuff (UVLED exposure box, DIY etch tank etc). The second hand market for pcb manufacturing equipment (box/etch) is remarkably busy and, consequently, expensive.

 

[Johnyradio]

If any other method delivered consistently reliable results at a price we could afford then it would be readily identifiable by a quick search. The fact that there isn't says a whole lot.

I've done PCB milling when i took an electronics class (with an OtherMill, now named Bantam), and it was real nice! Results were quality, and no messy chemicals. Drills the holes too. But, $3,000+

https://twitter.com/bayviewboom/status/672959165875159040

My budget can go up to about $300. There are a few sub-$400 desktop mills on the 'net, but they might be junk

https://www.dhgate.com/product/grbl...king/404463578.html#se1-1-1b;price|2836897375

http://www.ebay.com/itm/CNC1208-sup...hine-3-Axis-Pcb-Milling-Machine-/182861108458

Btw, when i say 10 boards/day, i do not mean a run of 10 identical boards.
I mean 10 different boards, part of my prototyping process.

 

[kellys_eye]

From what I've read, milling machines are getting better by the week and coming down in price too. If you're not after sub millimetre tracks and/or interpin routing then a cheap mill 'may' do the job but I reckon you'd have to ask on a CNC forum for specific details.

Given that the 'pro's' still use the UV exposure method (and you can do it for solder masks and overlays on a DIY basis too - a bonus if your prototype ever gets to the stage of buildability) there's still a lot going for it.

I've been impressed by a Youtube video I saw of a guy using a laser to burn off 'paint' to leave an etchable layout and he got the technique capable of 0.1mm tracks..... but that was a self-build rig IIRC - not exactly something you can puchase off the shelf although there are some mini CNC kits that offer laser etching add-ons. Maybe worth investigating?

Your links to CNC's are quite revealling - I've never seen them so cheap! and they do look robust (the essential part to successful milling). The DHGate device looks rather like one that another Youtuber (GreatScott? or the other Swiss electronics guy) seemed to have and is investigating its capabilities. He was impressed with the initial build quality but has yet to report on the likes of milling usefulness.... I'm even tempted myself at those low, low prices! although I suspect our local taxes may bump that up somewhat.

 

[(*steve*)]

When I do pcb's on a laser cutter, I use a device with an 80W tube in it.

It fits all your requirements except cost

Cheaper laser cutters might work. The device I use has way more power than I required and the management of the power is the largest part of the problem.

I have a colleague who has a desktop laser cutter with a 3W semiconductor laser which might work, but as I haven't tried it, it's clearly outside your requirements.

 

[Johnyradio]

I feel milling is preferable to any chemical etching process, cuz 1-step and no messy chemicals. But, as you say, etching is likely to be cheaper than anything else.


Laser Cutters

Here's a 3" square laser cutter, $100. A UV box would be cheaper.
https://www.ebay.com/itm/1000mW-USB...r-Carver-Cutter-Engraver-Machine/272937138073

This laser cutter, about 7" square working area, $135 (incl ship). A bit concerned there's no fume capture.
https://www.ebay.com/itm/500mW-Lase...-Cutter-DIY-Kit-Engraver-20x17cm/401329703929


UV Box

Here's a UV box for $65
https://www.ebay.com/itm/MEGA-KIT-B...NICO-EXPOSURE-UNIT-PCB-MASTER-UV/172979878497

but i might be able to build one for cheaper, with some high-wattage UV LED's, like this 10W LED for $4.
https://www.ebay.com/itm/3W-5W-10W-...let-High-power-LED-for-Aquarium-/142440408678

Instructable
https://www.instructables.com/id/PCB-on-a-Box/


Mills

Having read a bit more, it seems the problem with those cheapo ebay mills is they may be made of plastic or PVC, so they lack the rigidity needed for fine work. Some of those ebay mills include a waiver about "for educational purpose only, not production", which is not reassuring re build-quality.

This one looks more promising, and only $350 with switches and rigid upgrade. A bit more than i'd like to spend, but so far the cheapest decent-looking mill i've found. (not including router, router mount, work clamps, or end mills, so there's additional cost).
https://www.millrightcnc.com/product-page/millright-m3-basic-kit

Might interest you:
https://www.scan2cad.com/cnc/best-cnc-kits-beginners/


DIY Mills

Seems the cheapest way to get a quality mill is to build it myself. Haven't yet found a design in my budget that gives me confidence, but there are a ton of DIY designs yet to explore. Some designs use wood, and I've read wood is not ideal for a mill frame.

This one sure gives me some laughs. Built with plumbing supplies
https://www.instructables.com/id/Easy-to-Build-Desk-Top-3-Axis-CNC-Milling-Machine/


(Btw, just discovered my preferred routing style is called "isolation routing".)

 

[(*steve*)]

Laser cutting has the potential to be able to produce features far finer than milling.

If your surface mount components include (say) sot-323, I'd be surprised if you could mill the board.

 

[Johnyradio]

When I do pcb's on a laser cutter, I use a device with an 80W tube in it. has way more power than I required and the management of the power is the largest part of the problem.

Thx for that info! I've been ignoring the low-watt lasers.
Seems to me you could reduce power with a current limiter, constant-current power supply, or PWM?

Laser cutting has the potential to be able to produce features far finer than milling. If your surface mount components include (say) sot-323, I'd be surprised if you could mill the board.

Definitely something to investigate. A lot of people are milling PCB's.

 

[Johnyradio]

If your surface mount components include (say) sot-323, I'd be surprised if you could mill the board.

"We were a little concerned about the pads on the SOT23 parts being a little too small but it actually wasn’t a problem at all."
http://ubld.it/how-to-cnc-mill-a-solder-paste-stencil/#comment-636

 

[Johnyradio]

One of the more promising CNC DIY builds i've seen. The maker iterated 3 times on this project. Claims better than 1mm precision. Work area: 270 x 430 x 100 mm (X, Y, Z). Under $200. Lots of design-files provided.

Wow.

https://www.instructables.com/id/Low-Budget-CNC/

one question is how to fab those custom shapes without shop tools (maybe would be cheap to farm out the cut job to a local shop).

 

[(*steve*)]

better than 1mm precision

That's fine as long as your minimum feature size is 3mm. Do you mean 0.1mm? 1mm is horrible.

Seems to me you could reduce power with a current limiter, constant-current power supply, or PWM?

Yes, PWM is used to vary the average power, however, as you reduce the power (even with PWM), the time it takes the laser to turn on changes.

Semiconductor laser are far easier to control than CO2 ladders in this respect.

The greatest variation comes from mechanical factors. Surprisingly, the laser cutter I use can position the beam with an accuracy better than the kerf of the laser if you know what your doing (that's less than 0.1mm for our laser)

Because most milling machine use a V shaped but to do isolation milling, your minimum cut with is related to your vertical resolution and the flatness of your workpiece. Variations on flatness will directly vary the width of the cut. Also, milling can cause small islands of copper to be accidentally removed.

Chemical etching has its own issues, with undercutting being perhaps the worst. This can also cause the loss of small features.

Of the two, for small pitch SMD components, chemical etching is the one I purchased prefer.

I don't think you've mentioned the smallest pitch SMD components you'd use?

Note that I said SOT323, not SOT-23. If you're not using smaller than SOT-23 then you are using large components

And the other problem is the creation of a solder stencil. For very small parts it's harder to place the solder (especially where you don't have solder mask).

I'm seriously looking at using a milling machine with a scribe tool to scratch the resist off the PCB prior to chemical etching, but again, experimental

 

[Johnyradio]

Feature Size

That's fine as long as your minimum feature size is 3mm. Do you mean 0.1mm? 1mm is horrible.

Ah, i was looking at 'pitch' by mistake. Def want to use SMD parts. How small? The smallest i can get in my budget


Laser Power

Yes, PWM is used to vary the average power, however, as you reduce the power (even with PWM), the time it takes the laser to turn on changes.

Seems acceptable, no? Or, maybe you could adjust the PWM for faster turn-on time, then immediately readjust for cutting as soon as it turns on?

Semiconductor laser are far easier to control than CO2 ladders in this respect.

So you're using a CO2?

the laser cutter I use can position the beam with an accuracy better than the kerf of the laser

:O how?


Flatness

Variations on flatness will directly vary the width of the cut.

Some mills (or rather the software) scan the PCB before milling, to capture height imperfections in the workpiece, and then compensate during the cut. Prometheus does that. I'm hoping most mills conform to protocol standards, allowing me to use software of my choice.

Also, milling can cause small islands of copper to be accidentally removed.

Due to?


Etching

Chemical etching has its own issues, with undercutting being perhaps the worst.

IF i chemical etch, my plan is to use sponge etching, not bubbling. Praised by many. Said to reduce undercutting (due to quicker etch-time), and reduces amount of etchant required.

"wiping the sponge across it is much more effective. It takes much less time to remove the copper using this method, which of course helps prevent undercutting of the mask and allows for very fine traces.". https://www.rcgroups.com/forums/sho...eb19c796b9e158c46caaf4&p=12397596&postcount=4

Also planning to use salt, hydrogen peroxide and white vinegar (or other mild chemicals-- there are various recipes out there). I think this may reduce undercutting (due to less aggressive chems). Faster etch time given by sponge method may compensate for the weaker chems.

There's also electro-etching, which some metal artists are using, and PCB hobbyists.

Yet another method: sprayed etchant. But seems too bleeding edge for me.


UV
Some recommend using "glass to hold the image flat to the PCB to prevent undercutting the photo-resist" during UV exposure. These other guys say to stack a bunch of books on the glass, but so janky! Some strong clamps or clips would be simpler.


Stencils

other problem is the creation of a solder stencil.

I expect to use the same process for the stencil as i use for the PCB. There are recipes out there for milling, lasering, and etching stencils. I believe if i solve the PCB, i'll have solved the stencil.

http://ubld.it/how-to-cnc-mill-a-solder-paste-stencil/#comment-636
https://learn.adafruit.com/laser-cut-pcb-stencils?view=all
https://www.instructables.com/id/DIY-Stencil-for-PCBs/


Milling

looking at using a milling machine with a scribe tool to scratch the resist off the PCB prior to chemical etching

might work great! Tho' seems very delicate, since you don't want to mill away the copper. I mean, if your mill is precise enough to scratch off the resist without scratching the copper, at SMD sizes, then i think your mill is precise enough to simply mill the traces (and forget etching).

 

[(*steve*)]

Feature Size

Ah, i was looking at 'pitch' by mistake. Def want to use SMD parts. How small? The smallest i can get in my budget

Well... it's pretty hard to create an object less than three times the size of your error. So, if you want 0.5mm pads, you need positioning to be better than 0.2mm

Seems acceptable, no? Or, maybe you could adjust the PWM for faster turn-on time, then immediately readjust for cutting as soon as it turns on?

It depends on how much control you have over the laser. I gt the most accuracy and evenness of power per unit area by using the etch function of the laser cutter. Perhaps a lower powered laser cutter with more software control could do the isolation milling by drawing along the lines. I find that with a high power CO2 laser there is too much power on corners and ends which reduces the available resolution.

So you're using a CO2?

Yep

:O how?

The Y movement of the gantry is in 0.07mm steps. The kerf is often assumed to be 0.1mm that's a reasonable estimate when the laser is focused properly)

X movement when "etching" bidirectionally has a visible (under magnification) offset of about 0.1mm between left to right and right to left passes. When etching unidirectionally there is no visible misalignment (under the same magnification where 0.1mm sticks out like dog's balls).

The use of the etch function (which essentially does a raster scan of the job) operates in a way that maximises the repeatability by limiting all motion to (essentially) a single direction.

Due to?

The sharper the bit, and the faster the blade, , the smaller the feature you can cut. However a V shaped bit has a zero velocity at the tip. The problem is that a slow blunt edge can rip the material. Copper is only glued to the board and you can easily impose a shear stress that will tear it away.

IF i chemical etch, my plan is to use sponge etching, not bubbling. Praised by many. Said to reduce undercutting (due to quicker etch-time), and reduces amount of etchant required.

"wiping the sponge across it is much more effective. It takes much less time to remove the copper using this method, which of course helps prevent undercutting of the mask and allows for very fine traces.". https://www.rcgroups.com/forums/sho...eb19c796b9e158c46caaf4&p=12397596&postcount=4

Sponge etching only causes less undercutting because you are observing the etch as it takes place, and because you have some control to prevent overetching by not sponging some areas. Other than that, there's no magic.

Also planning to use salt, hydrogen peroxide and white vinegar (or other mild chemicals-- there are various recipes out there). I think this may reduce undercutting (due to less aggressive chems). Faster etch time given by sponge method may compensate for the weaker chems.

unlikely (same with the other methods too).

Some recommend using "glass to hold the image flat to the PCB to prevent undercutting the photo-resist" during UV exposure. These other guys say to stack a bunch of books on the glass, but so janky! Some strong clamps or clips would be simpler.

Different issue. You're talking about exposure of negative material. minimising the gap between the photosensitive layer and the pattern (and minimising the thickness of the photosensitive layer) are the key. Having a light source further away can also help.

I expect to use the same process for the stencil as i use for the PCB. There are recipes out there for milling, lasering, and etching stencils. I believe if i solve the PCB, i'll have solved the stencil.

Yes and no. The stencil is a lot thinner than the PCB material, but the thickness of the material you're removing is larger. Both of these will affect the process.

might work great! Tho' seems very delicate, since you don't want to mill away the copper. I mean, if your mill is precise enough to scratch off the resist without scratching the copper, at SMD sizes, then i think your mill is precise enough to simply mill the traces (and forget etching).

The point is that with a sharp scribe I can remove paint and minimal copper without the risk of damaging the remaining copper. It still needs to be chemically etched. Anyway, I haven't tried this yet.

 

[(*steve*)]

If you're interested in what can be achieved, see here.

The whole thread is about my many trials, and you might find stuff there of interest if you want to experiment (but you don't ).

 

[Johnyradio]

I find that with a high power CO2 laser there is too much power on corners and ends which reduces the available resolution.

Then software should reduce power on corners, no?

a V shaped bit has a zero velocity at the tip. The problem is that a slow blunt edge can rip the material.

Wow, interesting. Never thought of that. Is there a different tip-style you could use? Something with tiny serrations, instead of a point?

Sponge etching only causes less undercutting because you are observing the etch as it takes place, and because you have some control to prevent overetching by not sponging some areas.

Not sponging which areas?

with a sharp scribe I can remove paint and minimal copper without the risk of damaging the remaining copper.

If you can scratch the resist, why can't you just cut the copper?

if you want to experiment

I don't own a mill on which to experiment, so i need to develop a plan that's been pretty validated by others before i start buying gear. Unfortunately, I can't afford to buy gear just to experiment

 

[(*steve*)]

Then software should reduce power on corners, no?

Yes, but the control, whilst sufficient for most normal tasks is not fine enough for ablating paint.

Wow, interesting. Never thought of that. Is there a different tip-style you could use? Something with tiny serrations, instead of a point?

I imagine you could use a tool with parallel cutting edges, but you couldn't do fine lines then (and it would be far too easy to snap)/

Not sponging which areas?

The areas already etched

If you can scratch the resist, why can't you just cut the copper?

because it requires a far more rugged tool, and because it risks damaging adjacent areas of copper.

I don't own a mill on which to experiment, so i need to develop a plan that's been pretty validated by others before i start buying gear. Unfortunately, I can't afford to buy gear just to experiment

Have you considered toner transfer?