You are absolutely right. Your process is the way to go for professional
and especially for safety critical hardware.

My description was more focused on the hobby developer or the prototype
board to do some experiments. As you mention to test stuff that is not
in the data sheet.

I hope and wish that all product I buy or use have been developed using
the process you described.

But when I do private projects for me I work like what I described. I'm
not proud on it but it is sometimes faster. It may also have something
to do with that I'm more on the Software side and have a little bit of
two left hands when I need to solder stuff. I would not do it that way
with a product that I or the company I work for would sell.

So thank you for the clarification Jack.

Best Regards

Lars

Am 19.08.2011 22:51, schrieb Jack Carroll:
> Generally, my boards do work on the first attempt, and are good enough to 
> ship to the customer.  The key to success is a rigorous top-down design 
> discipline.  It works just well in hardware as it does in software.  I have a 
> couple of papers on how I do it, but briefly, you need to start with a formal 
> specification.  You don't start design until the client signs off on the 
> specification document.  Then you develop the conceptual design, and check it 
> for correctness and completeness.  You prove that it implements the 
> requirements without error.  You fix errors at this stage, before investing 
> effort in detailed design.  Then you proceed to detailed design, and repeat 
> the checks for correctness and completeness.  To design hardware correctly, 
> you need a full understanding of the components you're designing with.  You 
> have to read all the parameters and behavioral descriptions on the data 
> sheet, and _understand_ them.  You use only the min/max parameters, not the 
> typicals, and s
omething is incomplete or not clear, you contact the application engineers 
and/or perform experiments to answer the residual questions.  You do _not_ 
attempt a prototype design if there's anything you don't know about the 
components, materials, and technology you're working with.  Once all the 
necessary information is in hand, you complete the circuit design and check it 
for correctness and for completeness, referring back to the specification and 
the conceptual design documents.  If at all possible, you get an equally 
qualified engineer to go through the complete design, checking all logic, 
calculations, etc.  You also invite the manufacturing engineer and an assembler 
to review it, so you capture their needs and advice while there's time to act 
on it.  At this point you produce a documentation package for handoff to the 
circuit board designer.  This consists of schematic, parts list, outline 
drawing, data sheets of all components, and a design note stating all layout 
requir
ements not obvious from the other documents.  This is where you state creepage 
and clearance requirements, keep-out zones, controlled impedance transmission 
lines, placement rules to meet stray inductance and capacitance requirements, 
requirements for guard rings and shielding, the standard design rules issued by 
the board fabrication vendor, and so on.  When you get the artwork back, you 
print the layers out, stack them on a light table, trace out every track on the 
board against every line on the schematic, and go through the whole submission 
package line by line to verify that every layout requirement was complied with. 
 It usually takes two full trace-out cycles and an overlay check of the final 
corrections.  If you take your time and don't skip any steps, you should get a 
board you can put into production, on a project schedule you can estimate 
within 20%.  This isn't airy theory, this is how I've done it for the last 30 
years.
> 
> Jack Carroll
> 
> ----- Original Message -----
> From: "Lars Poetter" <[email protected]>
> To: [email protected]
> Sent: Thursday, August 18, 2011 12:17:58 PM
> Subject: Re: [Open-graphics] Economics of open hardware
> 
> On the Hardware side you need not only the parts (resistors, capacitors,
> the ICs,..) but also the PCB.
> 
> For the PCB you have 2 options. 1) produce it yourself. Then you only
> need to by the copper plated FR4 and need some hours to do it. 2) let it
> produce. Then you have to pay for the PCB and they are not cheap. Just
> google for a PCB Company and they have the prices on their homepage.
> 
> But then you still need to put it all together. Therefore you need
> soldering tools. And don't think of those cheap solder irons. We want to
> do SMD and BGA right ?
> 
> If you have come so far then in reality the board will not work.
> 
> _______________________________________________
> Open-graphics mailing list
> [email protected]
> http://lists.duskglow.com/mailman/listinfo/open-graphics
> List service provided by Duskglow Consulting, LLC (www.duskglow.com)
> 

_______________________________________________
Open-graphics mailing list
[email protected]
http://lists.duskglow.com/mailman/listinfo/open-graphics
List service provided by Duskglow Consulting, LLC (www.duskglow.com)

Reply via email to