Anyone make PCBs with Othermill?

Vias are pretty easy to do:

You can even do multilayer:

Cheers

Klaus

We have zero prototypes. Ten protos sounds like design by trial and error to me.

To speed up the time to production, eliminate all of the prototypes.

A modern design, with BGAs and fine-pitch parts, couldn't be made on a milled board. And we can buy a lot of 2-day or 1-day turn PCBs for the cost of a PCB milling machine.

I'd expect that a 3-day turn of a few pieces of proper boards (multilayer, 5 mil features, PTH, blue solder mask, yellow silk, routed to the right size, tested) would cost less than the labor needed to make it on a milling machine. We usually order six or eight, assuming that they will work and that several can be sent to customers. Nearly all the time, they do work.

Quick-turn boards cost us a lot more than that! But people are expensive here, and somebody has to operate that milling machine.

$100, or $800, is in the noise. What matters is time to market. Engineers are expensive, so we want to finish a product and move on. I think that multiple prototypes, or even one prototype, slow down the process.

Or too often, specification by trial and error.

"No, that's not what I want." "No, that's not quite right either." "How about..."

Eliminate customers and management and it may be possible.

Not to mention TPH connectors and perhaps capacitors.

That's what we found. The milling machine sounds like a great idea but it just sits there taking up space. It is a good source of copper clad, though. My standard order is 25 boards. I should probably double that for the board I'm pushing out now but if I run out of them, I'll get to do another pass. ;-)

$100?!!! You'd spend a *lot* more than that just babysitting the milling machine and it takes a *lot* of $100s to buy one of those things (the copper clad isn't free, either).

use it

to use

ve

t a

least 4-5 times per week and it was great for producing fast prototyping. W e could come up with an idea in the morning and have a working PCB in the a fternoon. With PCB prototyping that takes a week

full scale production. In the place where we used the milling machine, we could have 10 prototypes before the final solution

Not exactly. Sometimes topologies can be proved by simulations, but I do no t trust the simulation without backup real life measurements. For switch mo de converters, often the simulations can give you the incorrect result.

So ten prototypes is not the same incremental design, but includes very dif ferent topologies. Good luck ordering PCBs for that, and using all your pro ject time waiting for the PCBs (exaburation intended)

to speed up the process (no waiting for PCBs)

Ever heard of field tests and real life problems? (your statement only fits for extremely simple interfaces and fields of scope)

Cheers

Klaus

e to use it

one to use

r this

es I've

. By

st get a

t at least 4-5 times per week and it was great for producing fast prototypi ng. We could come up with an idea in the morning and have a working PCB in the afternoon. With PCB prototyping that takes a week

do the job but for hobby use it doesn't make much sense, you'll spend more in tool bits than it cost to get a prober pcb made and delivered in a few w eeks

ress in 8 days ~$30

th

uring the week or 2 for the PCB to arrive, the milling option will have a f inished prototype, probably allready tested etc. (we always have most parts in stock, so we do not lose time on that)

The labor for the milling is very low. A technician runs the machine. He se ts it up, does other work while it runs, and shuts it down when finished. P robably 30 minutes of work. Even with insane Danish hourly wages that's les s than 10USD.

Cheers

Klaus

Are you imitating my customers?

But if rapid protyping is not available, you can just tell them that each of their new ideas will take four months and another $70K to do.

o use it

e to use

his

I've

By

get a

t least 4-5 times per week and it was great for producing fast prototyping. We could come up with an idea in the morning and have a working PCB in the afternoon. With PCB prototyping that takes a week

e,

t.

re full scale production. In the place where we used the milling machine, w e could have 10 prototypes before the final solution

se

ay to speed up the process (no waiting for PCBs)

and/or practice "defensive design" footprints and zero ohm resistors are practically free, you can add a lot of flexibility for the price of a bit of thinking

-Lasse

It's reasonable to breadboard a non-standard switcher. I wouldn't breadboard a Simple Switcher or some other vanilla design.

This is a constant-current switcher that didn't work:

I proudly note that it only took me ONE iteration to prove that it wouldn't work.

One Dremel is worth a hundred PCBs.

Most of our products can be sold as production-built rev A, with no prototypes. They might have a kluge wire or two, and we might make improvements later on, but we usually get paid for the rev A's. Few of our products are simple. First-time-right is a culture. Multiple prototypes is a culture, too.

Copperclad FR4 is cheap on ebay. Gold plated FR4 isn't cheap, but it looks great.

We don't have any engineering technicians. I've always found it difficult to keep them busy.

You've still got to do the board layout and stuff. I just grab my Dremel.

You Californians are so all about bling. Who cares if it works? ;)

Cheers

Phil Hobbs

Nah, they're all the same.

Management has lots of money. $70K might pay for the hardware, certainly not my time.

You don't have a government? Management? Benefits? They're not free. Our L&B rate is *way* over $100/hr.

Sure, we do that in places. FPGAs and software add a lot of flexibility, too. But careful design and design reviews can prevent most hardware design errors. Building a prototype is a terrible way to find dumb mistakes.