I've got a small hobby project I'd like to do. I've been working with evaluation kits from vendors, but now I'm ready to move onto a real board of my own.
Can anybody recommend design services for the hobbyist? I'm looking for someone to take my parts selection, and a high-level design, and turn it into a schematic and PCB layout. I'd like to then take the PCB layout to my PCB house of choice (say somebody like ExpressPCB or PCB123) to have the boards made. I can solder myself, so that isn't an issue.
Why not do it yourself? It's really not that difficult in many cases. For layouts that may be difficult (very low level, or high speed, or RF) you would have to be prepared to be charged accordingly.
At least give it a try. Eagle has a free version that many have used successfully. In the FOSS world there are gEDA and Kicad, both of which are more than capable of doing schematic capture and board layouts that can be sent off to a board house for single-, double-, or multi-layered designs.
Part of the problem is I don't really know what I am doing electronically. At least doing it right. I don't know what size bypass capacitors are required and what (if any) termination resistors are required. I don't even know why one would use an electrolytic cap vs a tantalum vs a ceramic, etc. These are all choices that an experience designer would know.
My experience is in RTL and firmware design. Not board design.
The other problem is that many of these software kits are lacking in libraries. The parts I want to use are not available in Eagle (or ExpressPCB or PCB123's software). And while I can create my own parts, I'm not even sure the footprints I select/create will be correct. And when I drop my $200-$300 to have the PCB's made, and I can't solder the parts on....
Honestly, I have no idea how much it would cost to have a board made for me. If what we budget for engineering time at our company is a rule (about $65/hr), I imagine having and experienced person do the schematic and layout couldn't take more than a couple of days. So I'm looking at about $1k to have a board done for me (excluding the proto build cost). And paying that extra amount seems a whole lot safer for me.
Don't get me wrong, I'd love to be able to do it myself. But lacking the electrical experience, I'm afraid I'll end up wasting more than $1k just trying to get it right.
Maybe the better question is how to get a good review of the electrical issues before sending it off to be built. And a review of footprints before sending it off.
If you post your schematic to ABSE you'll get plenty of people telling you what they don't like about it.
"And a review of footprints before sending it off."
Honestly, for low production quantities that you're just going to hand-solder, getting all the parts beforehand and physically verifying that they'll fit on (a paper version of) the board is probably the best way to go. For high-volume manufacturing one can spend lots of time trying to come up with footprints that optimize yield based on the manufacturing method (wave vs. reflow soldering, mainly) -- or just purchase the IPC libraries that contain their idea of "optimized" footprints -- but for hand-assembly this isn't worth the effort.
You also have to double check that the pin names match the schematic symbol pin names, and that they match the actual part. It's quite possible to design a board that passes DRC but is rather flawed if there are problems with that layer of abstraction.
Best regards, Spehro Pefhany
"it\'s the network..." "The Journey is the reward"
email@example.com Info for manufacturers: http://www.trexon.com
"Suudy" schreef in bericht news: firstname.lastname@example.org...
From what you tell I only can say: "It can be done and I can do it." Though I agree with others that it may be best to do it your own, I agree with you that this can become a time and money consuming option. If you don't want make your design public, you can send it to me privately. My address can be decoded easily once you know that "laatditweg" translates to "leavethisout" and "enditook" means "andthistoo".
If you get an experienced designer in at $100 to $150/hr, you'll be into the couple thousand $ range, unless your circuit is trivial. You would learn a lot if you did this yourself. A 1 month project will give you an equivalent of 2 years conventional schooling plus. As someone said, post schematics and board layouts to alt.binaries.schematics.electronics (ABSE). People will give lots of comments. One chap did this on his switching power supply design recently. He had lots of input. When you do this thru the use net, you learn, we all learn.
You also become much more valuable at work with this knowledge. It's really hard to deal with firmware folks if they don't understand electronics!
Just one piece of advice. You mentioned "take my parts selection and then ...". That's not really how it works. If you can't design a circuit on your own that's perfectly fine but it is then better to tell a consultant "Hey, I want the circuit to do this, that and the other thing. It should cost no more than $xx/unit and we want to build xxxxx of them per year". Or it's a one-off where cost doesn't really matter. Let the designer pick the parts.
In the US? Wow. That can't possibly include overhead, equipment amortization, HR, accounting and so on. EE services cost a lot more. Even our roofers and plumbers charge more and they don't need expensive stuff like spectrum analyzers. And my CPA costs $200/hr.
I imagine having and experienced person do the
Good approach. Just the $1k may not work.
A design review is a very good thing. As for footprints and such, good layouters should be able to handle that on their own. Mine certainly does, never had any problems.
What I meant by "pick the parts" was the major components. For example, the microcontroller, type of ram, USB controller, etc. Just those. I'm not talking about picking resistors, caps, etc.
That $65/hr is for engineering time, not overhead, materials etc. Those are budgeted separately. I realize that a consultant will likely cost more.
But given the types of responses, it appears that it does make sense to do it on my own. The choice of parts and the risk of getting a board that doesn't work can be mitigated. I didn't even think of placing parts on a printout to see how things align. Tips like that are great.
I think the circuit is trivial. Low-speed, single supply rail. No complex power sequencing issues. The hard part for me is determining things like size and number of bypass caps, termination resistors, etc.
That's a good point. Learning from others and sharing what I've done would be a good step. I'll surf over to ABSE and lurk for a bit to see what I can learn.
In defense, I'm not clueless on the electronics side. I'm perfectly comfortable reading a schematic. I understand pull-ups/pull-downs, why open collector/open drain outputs are used, the right way to do LEDs, why bypass caps and termination resistors are needed, etc. I just don't know how to determine the appropriate types of LEDs, caps, resistors, etc. I certainly understand the interconnect, but on a logical level. The "analog" side of things is where I don't have experience or knowledge. I know enough to work with the electrical guys.
But you are correct, learning how to design a PCB is a valuable skill. I'll surf over to ABSE and take a look there.
I'd leave the chips to the consultant. If he isn't familiar with your choice uC it'll get expensive or worst case hit a dead-end.
However, then you'd have to budget that in, else the company would not be able to have an overview of what projects really cost. Like when the marketing guys want another gizmo. Then we'd always tell them what the real costs are, for how much they'd have to sell it at how many per year, and then sometimes they didn't need it that badly anymore ;-)
Yep, don't be afraid to kick the tires. Start out with a free test version of Cadsoft Eagle which has the layout package included. It easily transfers between schematic and layout:
Later if you like it and want to go to more layers and so on you can buy a version that's big enough. It's like upgrading from a moped to a serious motorcycle and it is not expensive.
It's perfectly ok to ask those questions on a.b.s.e. You can also use a web site to post schematics, then you can ask here or on s.e.basics. Some folks think that simple questions ought to be asked there but IMHO it doesn't really matter. Lots of helpful people in either group. And heck, don't think we know it all. Almost all of us ask a question once in a while because some new stuff is unfamiliar to us.
Another option is to scope out the web or app notes from TI, Microchip, Atmel and other uC manufacturers. Tons of real-life projects there, with schematics, code and all. Sometimes even with example layouts.
Ah, but that's where the difficulty comes in, to solve all the other problems where the answer isn't one. If the answer was one, they'd just give the job to one of you pocket-protector-wearing analog dweebs. ;-)
LOL. I dont wear a pocket protector; I work from home, so clothes are optional ;)
a long time ago, in a galaxy far, far away, I started out as a digital guy, but then saw the light. so nowadays I do all the analogue stuff, and drool at the awesome FPGAs the digital guys get to play with (but dislike the PSU ramp requirements of the xilinx stuff).
the last xilinx fpga I put on a board had minimum and maximum ramp up and ramp down rates for the supplies, along with stringent power sequencing requirements. Its easy to do the sequencing, and ramp-up rates are not too hard, but sequenced ramp-down rates get a lot trickier. that particular design was a cost-down. I halved the price of the original smps, and we put in an FPGA that was cheaper than the CPLD it replaced. Then added 4 linear regulators, a pair of LM339s and an LM324 along with some discretes to do the ramp-rate control & power sequencing. the extra regulators ate the FPGA-vs-CPLD cost savings and then some, and I had to add so many bits to get ramp rate control I might as well have done the regulators discretely. all up the cost saving was negligible. a better choice of FPGA (eg cyclone III) would have made a big difference. Alas I didnt get to pick the FPGA, some digital guy did, based on the impeccable logic "I like xilinx"