state-of-the-art in rapid PCB production machines

[jmole]

Is anyone aware of anything new, other than the LPKF stuff, which seems like it hasn't changed much over the past 5 years?

I've got a good budget, but everything on the market looks like crap. Who knows different?

[Muxr]

You can look into CNC machines, the Othermill is kind of popular. But I dunno.

For me personally, because you can't do vias easily it means you have to design boards specifically for it, and that feels like a waste of work. Also lack of soldermask makes it difficult to solder fine pitch SMD parts.

I find I'd rather work on a few designs at the same time, and that way have something else to work on while I await on the board from the PCB house.

[edavid]

Isn't the LPKF ProtoLaser U3 fairly new?  It doesn't seem like crap.

At the other end of the cost spectrum, another new thing is the Voltera conductive ink printer.

[jmole]

Isn't the LPKF ProtoLaser U3 fairly new?  It doesn't seem like crap.

At the other end of the cost spectrum, another new thing is the Voltera conductive ink printer.

Fair enough. U3 is $145k according to online rumors. Not sure if it's still the case. I've got budget, but 6 figures is a huge stretch.

[jmole]

You can look into CNC machines, the Othermill is kind of popular. But I dunno.

For me personally, because you can't do vias easily it means you have to design boards specifically for it, and that feels like a waste of work. Also lack of soldermask makes it difficult to solder fine pitch SMD parts.

I find I'd rather work on a few designs at the same time, and that way have something else to work on while I await on the board from the PCB house.

I haven't used the othermill in about 6 months, so take it for what it is, but yeah. to use the othermill, you had to design for the 1 single tool you were using, or you'd just waste time. sticky tape + FR4 is not what I want to deal with at this point. othermill seems geared toward the EAGLE crowd, but the mid-market stuff just doesn't seem to be there...

[ataradov]

Take your budget and estimate how many boards you can have professionally made using 1 day service. Then it will probably be clear why there are no mid-range machines like that. They don't make much sense.

[jmole]

Take your budget and estimate how many boards you can have professionally made using 1 day service. Then it will probably be clear why there are no mid-range machines like that. They don't make much sense.

call it 2k for a 1 day run, pretend you need at least one prototype a week, and you're talking 100k over a year.

every 2 weeks, you're looking at 50k, or every month, 25k.

big numbers, no matter how you crunch it. it's less to do with that though, and more to do with availability. there's no "oh shit, get the design exactly right before we send it out" when you have your own machine.

boards come in all shapes and sizes. be honest, if getting a single layer pcb in your hands was as easy as using a laser printer (without the etching chemicals, manual drilling, etc.) , we'd all be building much better prototypes, much more quickly.

[ataradov]

I'm not sure about that $2k number.

But even then, service like this (no affiliation) http://www.4pcb.com/bare-bones-pcbs/index.html gets you a board with basically the same quality you would get on a really good milling machine, but those boards will also have plated holes. Sure, it takes a bit longer, but it is also way cheaper.

And not everyone makes a board per week. And it makes no sense to make machines for those three people who do .

[HAL-42b]

Hitachi had this flip-chip wiring process that worked on board scale. Not sure what the name was. Something like modern day wire wrapping I suppose.

[rx8pilot]

$2k? What kind of board are you designing? Solid gold traces?

For that price you can get a pile of boards with exotic features that are far beyond any mill or laser.

Sent from my SAMSUNG-SM-G890A using Tapatalk

[mikeselectricstuff]

Take your budget and estimate how many boards you can have professionally made using 1 day service. Then it will probably be clear why there are no mid-range machines like that. They don't make much sense.

call it 2k for a 1 day run,

If that's what you're paying for 1-day 2-layer PCBs then someone is taking the piss.
I pay about a tenth of that from my local supplier, though I have been quoted stupid prices by other suppliers that thought they could get away with it.

[mikeselectricstuff]

Is anyone aware of anything new, other than the LPKF stuff, which seems like it hasn't changed much over the past 5 years?

I've got a good budget, but everything on the market looks like crap. Who knows different?

And does your budget include all the staff time to set up and run an in-house process ?
Unless you're doing stuff like RF which is very layout-sensitive, or have reasons why very fast turnround is essential it isn't going to make sense to do it in-house. Leave it to people who do it all day and have all the equipment and expertise.

[AndyC_772]

be honest, if getting a single layer pcb in your hands was as easy as using a laser printer (without the etching chemicals, manual drilling, etc.) , we'd all be building much better prototypes, much more quickly.

I'm not entirely sure that's true. Every single board I ever design has a solid ground plane, and most have components on both sides. A four layer PCB is an absolute minimum, and a machine that could only produce a single layer would be completely useless to me.

What's the application, for that many PCBs, none of which is multi-layered?

[Kilrah]

big numbers, no matter how you crunch it. it's less to do with that though, and more to do with availability. there's no "oh shit, get the design exactly right before we send it out" when you have your own machine.

Huh?
Not taking the time to get the design exactly right before sending it out has the exact same result with a commercial service or your own machine, if you left a mistake then the time/money needed to fab it is lost... 
With an external provider it's time any money, with your own machine it's time, consumables and some of the money invested in the machine...

Sounds like you're about to fall into the "but I can save time on design because it's easy to just do another run!" that usually leads to the exact opposite - doing 6 runs because everytime you correct a mistake and rerun instead of also looking for other things you may have missed takes much more time and cost than doing one, taking the time to find all mistakes and correct them, then do a 2nd run that's correct.

The easier the "manufacturing" process is the more you're tempted to take a lax approach and ultimately lose time. Works with 3D printing, and even software development. When compiling/loading a new version takes 10 minutes you'll rerun the whole logic to properly check your fix before launching a build, when it's 15 seconds you tend to just half-randomly change a value without thinking it properly ("oh it must be that - maybe") then repeat 20 times when it isn't.

[mikeselectricstuff]

be honest, if getting a single layer pcb in your hands was as easy as using a laser printer (without the etching chemicals, manual drilling, etc.) , we'd all be building much better prototypes, much more quickly.

I'm not entirely sure that's true. Every single board I ever design has a solid ground plane, and most have components on both sides. A four layer PCB is an absolute minimum, and a machine that could only produce a single layer would be completely useless to me.

What's the application, for that many PCBs, none of which is multi-layered?

Everyone's requirements are different - I very rarely go to 4-layer, maybe 1 in 20 or 30 boards.

[trevwhite]

I have been down this road. Got the LPKF machine, done through hole plating. The amount of work it takes to make a decent two layer board is unreal. Also once made you still do not get the lovely solder resist finish.

If you find a local pcb house they will do you a whole panel of boards in three days for £200-£300. 

Thr number of times you can get away with longer lead times means you can pay a lot less than that half the time. Making boards yourself really is about speed to prove a design works.

[jmole]

I'm thinking simple stuff for the most part. one problem with prototyping today is that decent components only come in SMD now, & protoboards aren't really viable for some scenarios. breakout boards help, but they only get you halfway to a finished product. design is all about iteration, and the faster the better.

[jmole]

I have been down this road. Got the LPKF machine, done through hole plating. The amount of work it takes to make a decent two layer board is unreal. Also once made you still do not get the lovely solder resist finish.

If you find a local pcb house they will do you a whole panel of boards in three days for £200-£300. 

Thr number of times you can get away with longer lead times means you can pay a lot less than that half the time. Making boards yourself really is about speed to prove a design works.

this has been my gut feeling. none of the machines on the market look compelling. also, i clearly need to shop around for vendors a bit more.

[trevwhite]

What are you end needs? Finished product, research and development, hobby learning, work?

[jmole]

stupid prices by other suppliers that thought they could get away with it.

nailed it. although some suppliers around here just dont bother with the 1-2 day turn stuff, and focus on 10+ layer boards, text fixtures, etc. they've got bigger fish to fry.

[jmole]

What are you end needs? Finished product, research and development, hobby learning, work?

I want to hit print and get a single sided board in 30-60 minutes with 10/10 T/S. For a reasonable cost, and without a lot of hassle. This seems like such a simple problem to solve.

[mikeselectricstuff]

What are you end needs? Finished product, research and development, hobby learning, work?

I want to hit print and get a single sided board in 30-60 minutes with 10/10 T/S. For a reasonable cost, and without a lot of hassle. This seems like such a simple problem to solve.

No, it isn't a simple problem, and not enough people really need it for anyone to have produced a nice solution.

If that's what you want, then by far the quickest, cheapest and most reliable way is chemical etching with a spray etch tank. Always has been, an probably will be for the forseeable future. 
You can print, expose, develop and etch a board in 15 minutes. 10/10 is easily achivable with a 600dpi laser printer on tracing paper. Hand drilling can take a while, but something like a cheap CNC3040 engraver may be an option, and can also rout outline shapes.
The key thing is to be able to set up a dedicated area to do it and contain the mess.
You could easily set it all up for well under $5k

However you should really consider why you need it that fast. If you include the time to do the layout, ask yourself if it would be quicker to do quick & dirty prototyping with breakouts to prove any areas that need testing, then do a PCB that's going to be much closer to the final thing.
Also can you arrange things so you're doing the software (or documentation, or another part of the project) while you're waiting for the PCB to arrive.

[Kilrah]

design is all about iteration, and the faster the better.

I really can't see how you could need so many iterations to make a difference on something that fits on a 2-sided board when using proper design methodologies. Your PCB should be OK in 2 or 3, at most 4 iterations.
Any project requiring this has so many other aspects to take care of that it's not a couple of days waiting on a PCB that would leave you unable to do any progress. Putting documentation, manuals, software aside in the case someone else was taking care of those even a PCB with rather gross errors can be hacked with a couple of cuts and bridges to be able to continue working on a section that's OK while you wait for the next rev to come.

[nctnico]

Take your budget and estimate how many boards you can have professionally made using 1 day service. Then it will probably be clear why there are no mid-range machines like that. They don't make much sense.

call it 2k for a 1 day run, pretend you need at least one prototype a week, and you're talking 100k over a year.

Do a better planning where you don't need the PCBs next day. That is much more cost effective! A PCB isn't designed in one day and it isn't fully tested in one day so why does it need to be made in one day? There is no logic in that other than putting money in the pockets of PCB services.

Also be sure to read the other threads about PCB milling. The general consensus is that such a machine only makes sense for prototyping single sided RF boards because doing multilayer boards and vias is prone to errors and takes a lot of time, effort and toxic chemicals to get it right. You basically have to get an extra employee who deals with the milling machine and chemicals exclusively, build a lab for working with toxic chemicals safely and pay for disposal of those chemicals as well. If you add all that to the equation as well the picture doesn't look that good.

[djsb]

I use an LPKF S62 and an S103 router at work (UK university). We (well, a small team of 3 at the moment) make batches of around 100 pcb's for several of our classes and prototype pcb's for our final year EE and Design students. Boards can be made in under 30 minutes. This is only because I've been using the S62 for around 4 years and the S103 for around 2 years. As I'm the most experienced user in the University I also train other tech's and post graduate students on using the machine. I much prefer these machines to the chemical etching system we used before and it's kinder on the environment. We only do single and double layer boards though. The S62 cost £15k and the S103 was around £32k. The S62 is very reliable once user error is minimised and it can use non standard LPKF router bits from a US supplier which are half the cost. LPKF insists that we use the OEM bits on the S103 because of the clever auto depth adjustment features (using camera measurement and an adjustment ramp) on this machine. I could wax lyrical on these machines but I get enough of this stuff during the week.