PCB design review

[PL]

I'd like to have the design reviewed of schematic and PCB board layout on a couple of interconnected boards. I'm personally developing several prototypes, but nearly nil electronics knowledge. I have contracted out the design work, but I think its good to double-check everything. I can't afford to have them made and assembled to find out that there are critical errors.

The boards are small and not exceptionally complicated. Well known microcontroller with WiFi, relays, ADC, IO expander, etc. Have schematics and Gerber files.

Any suggestions on finding a well experienced person to provide some reviews? I don't have the budget for the fees of an EE engineering firm, and an hoping find someone who freelances.  Preferably someone in Australia, as possible future work would go better if they're here.

[nigelwright7557]

What I do is for first batch make a set of 5 pcb's to prove the concept.
If that works well only then go on to larger batches.

I remember many years ago a manufacturer doing a hundred pcb's for a job I worked on.
They came back and none worked.
No one had told them the PIC micro had to be programmed.
Muggins here got to desolder all the PIC's, program them and solder them back in !

PCB design can be a nightmare so best left to an expert.

In my early years I designed a USB mixer.
PCB came back and I built one up.
With inputs shorted I was getting 1VAC on output ! should be zero volts.
I looked into it and I hadnt separated out power supply and smoothing caps charging impulses were modulating teh ground line whic was amplified through the op amps.
I reworked pcb separating out pwoer supply, used star grounding and that fixed it.

I also designed a USB scope. It worked but had 10KHz signal superimposed on sine wave.
I had run the scope signal beneath a ICL7660 negative voltage generator and it was picking up the 10KHz oscillator signal.
A reroute of the scope signal fixed it.

[Abdurrahman YAMAN]

If it is not confidential you can share it in here. There are kind people check it for you.

[AndyC_772]

This is exactly the kind of thing I do for a living - but let me offer a little advice for free.

If you've contracted out the design work, it's their responsibility to produce a design that works. A competent engineer should have performed checks, simulations and reviews, but there's no substitute for testing a real, physical prototype. Anyone who claims otherwise has never designed something non-trivial.

Often problems can be solved in a multitude of ways. Something that one engineer is confident will work, might be a source of concern to another. If you ask 'n' engineers how to solve a problem, you'll get at least 'n+1' different answers.

I recall one design I was asked to review; I pointed out that although it might function OK, it lacked protection components that I deemed necessary. The response from the original designer was that they weren't needed, it would be OK as-is. We may never know who was "right", if indeed it was that simple.

With that in mind, it might be worth considering what you'll do if your reviewer raises concerns. Without the ability to evaluate their responses yourself, do you go back to the original designer and insist they make changes? Or trust that they're right and your reviewer is wrong?

Might you just end up paying two people to do the same design work?

Who, ultimately, is taking responsibility for the design being right?

[PL]

it might be worth considering what you'll do if your reviewer raises concerns. Without the ability to evaluate their responses yourself, do you go back to the original designer and insist they make changes? Or trust that they're right and your reviewer is wrong?

I'm able to evaluate some fundamental aspects of design (I have experience with commercial building controls BAS/BMCS). For example - checking for proper polarity, N.O, N.C. & Com. relay circuiting, etc. Part of my current review, revealed that polarity had been reversed on a component. I advised the EE and he quickly corrected.  I have scant comprehension of the many other aspects of electronics; but with sufficient documentation and the basis for a reviewer's opinion, I have been able to draw some useful conclusions.

The circuit attached illustrates another similar concern. It appears to be laid out with reversed polarities - nominal 3vdc landed on the negative ("-") pin, and via an NPN transistor, the ground is landed on the positive ("+") terminal. It is possible that I don't fully understand the ways in which it can be successfully circuited. Also attached is a page from the relay datasheet.

 Page from Relay-V23079G2008B301.pdf (440.68 kB - downloaded 95 times.)
If in fact this is an error, it begs the question - are there additional critical issues. If its OK, then I'd glad to hear it.

With that in mind, it might be worth considering what you'll do if your reviewer raises concerns. Without the ability to evaluate their responses yourself, do you go back to the original designer and insist they make changes? Or trust that they're right and your reviewer is wrong?

It really depends. Reversed polarity is a show-stopper - it just won't work. Same applies to landing traces on the wrong pins, etc.  I'm a firm believer in QA/QC with another set of eyes. We all make mistakes and I'm glad when a reviewer catches them. However, If the design is littered with these sort of issues, then I'll have to take a hard look at how to proceed.  The extent of protection is a judgment call (BTW I did ask for additional protection and he willingly added TVS's: MOV's, Zener diodes, etc.)

So to my way of thinking, it depends on the nature and extent of errors, VS a designer's personal preferences that will work but is approached differently than another design that will also work. Example - the EE designed the board using a 12v supply, even though there was no 12v components. Only 5v & 3.3/3.0v. He wanted it just in case something was added in the future (.e.g. 0-10vdc outputs) It added cost and complexity unnecessarily, so I had it changed to 5v, as its just the first PCB prototype.

Any further thoughts are appreciated.

[exmadscientist]

First off, listen to Andy. I do this for a living (both the design side and the review side), and the question of how to handle review feedback is real and serious. If it's two coworkers who you've worked with for years and know well (maybe Alice is super conservative about everything and Bob is generally carefree, except gets really bothered about regulator feedback nodes for whatever reason), then it's not too hard to handle. If it's a hired gun versus the new guy... who knows, right? And if it's hired gun versus hired gun, well, good luck deciding who's right, who's wrong, who's panicking about a waste of time, and who's panicking about saving $1M.

Your designer's schematic is, frankly, terrible (even in such a small snippet), and that combined with your description of his behavior gives me low confidence. It is not easy to read, mixes American and European symbols, and leaves off important part numbers and component descriptions. Correctness comes first on a schematic, clarity comes second, and everything else comes later. But correctness and clarity go together, because if you can't understand it, you can't check it (or repair it). Forcing designers to match personal preferences after the fact is a waste of time, yes... but sloppy schematic capture is usually sloppy circuit design is usually sloppy final product. So pay attention!

The relay datasheet (and the one for Panasonic TX relays, a drop-in replacement for these TE Axicom guys) suggests you shouldn't drive the coil this way. I've tested the smaller versions of these relays on the bench and found that they don't appear to care too much, so this will probably work. But it's not what the datasheet says to do, it's drawn really badly, and it's not something I would do in production (though I also wouldn't stop the presses for this). So... your designer is sloppy.

[Wilksey]

Good lord, that snippet was done by a "professional" EE?  A relay circuit is very simple to draw, you usually need the transistor to provide the current as the driver (usually a micro) cannot provide enough current itself, the diode is a back emf diode when the relay is turned off the energy in the coil can still dissipate through the diode, some circuits have extra components or are driven using opto couplers etc, depends on spec and requirements.

Some people want belt and braces some want MVP, but in either case that schematic looks like a beginner has drawn it.

[PL]

that snippet was done by a "professional" EE?  A relay circuit is very simple to draw, you usually need the transistor to provide the current as the driver (usually a micro) cannot provide enough current itself, the diode is a back emf diode when the relay is turned off the energy in the coil can still dissipate through the diode,

For my education, could you give an example circuit?

[AndyC_772]

The circuit you already have is - aside from the relay coil being back to front - functionally correct, as far as we can see.

What's ringing alarm bells here is the way it's drawn. An experienced, professional engineer simply wouldn't produce a drawing that looks like that, because there are some simple conventions that are normally followed that make the schematic clearer and easier to read. These conventions are so deeply ingrained that a professional engineer wouldn't even think about them.

For example, things that leap out at me are:

• Relay coil is connected with positive to the pin marked '-' and negative to the pin marked '+'. Since this looks so obviously like an error, I'd always add a note alongside to explain why it's done here (eg. "note: reversed polarity allows easier PCB routing and is confirmed by experiment to work on this relay type"). Or just fix the error, if that's what it is!
• T4 has no part number. Why? It looks like a package with integrated resistors, why aren't their values marked either? Is there really not room? Does missing them off make it easier, or harder, to check and understand how the circuit works?
• 'T' is an unusual prefix to use for a transistor - 'Q' is more usual
• While I'm nit-picking T4... the symbol should really show circular blobs where the wires are connected, and personally I'd put a box round the entire symbol to make it clearer that the resistors are integrated. It does suggest that the designer has picked a symbol that's already in a supplied library and decided it's 'good enough'.
• Supply rails are normally drawn with the most positive toward the top of the page, so having a positive rail "Cr5_3V3" at the bottom is odd.
• The part type for D15 has just been randomly chucked somewhere near the part, with a wire cutting through which makes it harder to read. This is just sloppy.
• The name "Cr5_3V3" implies it's not a rail that's used elsewhere, so why is a rail symbol used at all? Why not just show FB4 connected directly between VCC_3V3 and the relay?
• That kink in the wire by relay pin 12 is ugly, though it's something we all have to do from time to time

Finally - and I almost know I'm going to regret saying this because someone will undoubtedly complain about my clearly unjustifiable CAD-ist prejudice here - it looks like the schematic has been drawn using Proteus, which is an inexpensive CAD tool that tends to be bought by beginners, hobbyists and the smallest of small businesses.

I've never used it myself, I've no reason to say the tool is "bad" in any objective way, but it does just hint that maybe the designer has yet to outgrow it and buy what I might call 'real' CAD software... Cadence, Altium or PADS.

OrCAD PCB Designer Standard is really not that expensive for a professional engineer, and it has an upgrade path all the way up to Allegro while still being able to read and edit existing schematics and PCBs.

[Wilksey]

I thought it was Altium

[AndyC_772]

Possibly, in which case I'm wrong. It does happen!

[Wilksey]

that snippet was done by a "professional" EE?  A relay circuit is very simple to draw, you usually need the transistor to provide the current as the driver (usually a micro) cannot provide enough current itself, the diode is a back emf diode when the relay is turned off the energy in the coil can still dissipate through the diode,

For my education, could you give an example circuit?

As Andy has said the circuit is there it is just very badly drawn, god knows what the layout is like!

[PL]

some simple conventions that are normally followed that make the schematic clearer and easier to read. These conventions are so deeply ingrained that a professional engineer wouldn't even think about them.

I've spent untold hours and days trying to understand the drawings. Had the EE followed proper conventions and drawing practice it would have removed a lot of mysteries that I had to sort out.

It really looks like back to the drawing board on this. Rather get it right at this point. Lot harder to remedy things when its on a PCB.

[PL]

I thought it was Altium

Correct

[Wilksey]

Can you share the entire schematic?  I can only imagine what other horrors lie within.
Can you say what the board is for?  Driving relays I presume, but what is the purpose?
I expect we could decode it for you or to what it is attempting to do at least.

[PL]

Can share portions. Some other parts involve IP.
Will send soon. Thanks.

[Wilksey]

Fair enough, have you already paid this EE for work?  I'd request some kind of refund if so.

[PL]

Haven't paid for most recent work.

[Wilksey]

Good, where did you find him?

[PL]

Online search for freelancing (chartered) EE's.

[Wilksey]

Ah, ok, well, I guess somebody was lying.

[PL]

I checked his EE registration. Baffling.

[Wilksey]

Is that an Australian thing?  You don't have to be registered here in the UK for Electronics, just Electrical work (house mains etc).

[PL]

Not sure if you have to here either.
Perhaps a requirement of some employers?
Certainly in commercial building services design its a standard thing for engineers.

[julian1]

For more than one relay, consider a bjt relay driver. Takes care of the resistor biasing, and incorporates flyback diodes. One part, generic, common, inexpensive etc. eg. uln2003

https://www.mouser.com/datasheet/2/308/ULN2003A_D-1814978.pdf