[Diy_efi] PCB layout/circuit design

[Bill Washington]

James,
    You said "all ICs should have at least a100uF cap as close as 
possible to the supply " do you mean 100nF?
Steven,
    In PCB layout it is good practise to have a large reservoir Cap 
(your 1 Farad) close by plus smaller (ceramic 100nF and sometimes 10nF 
as well) caps right on the IC supply pins (as close as you can get 
them). The reason for this is that the large reservoir cap has some 
inductance and resistance in its construction which slows down its 
response time - the ceramic Caps (which very low ESR) on the IC supply 
pins handle the sharp rising and falling edge surges and minimise ground 
bounce which can cause all sorts of problems ....
Regards
Bill
>
> Subject:
> [Diy_efi] RE: More ECU progress (Steven P. Donegan)
> From:
> "James Holland" <j_holland at btopenworld.com>
> Date:
> Sat, 24 Feb 2007 22:41:30 -0000
> To:
> <diy_efi at diy-efi.org>
>
> To:
> <diy_efi at diy-efi.org>
>
>
> I have been following this project with some interest. I had a look on the
> website but you seem to have the same schematic for the datalogger and I/O
> board. I presume the I/O board schematic hasn't been posted. I've designed
> electronics for military automotive issues so I'm aware of the problems of
> designing for this kind of environment.
> I like the idea of this but I think that you will have some noise issues
> with the design as it stands. The PCB design lacks a good ground plane and
> the decoupling could do with being improved, all ICs should have at least a
> 100uF cap as close as possible to the supply. The PIC should have two, one
> on each supply pin. I'm a bit surprised that the PIC doesn't have a separate
> Analogue supply for the A/D as it stands the analogue and digital returns
> are mixed which isn't good. The A/D inputs should really have anti-aliasing
> filters close to the inputs, a simple RC would do and only the C really
> needs to be up by the input. The R would also provide some input protection
> to the PIC. Input voltage clamping is also a good idea. The PIC is pretty
> well protected but voltages outside of its supply rails throw the A/D
> readings all over the place. 
> It looks like you are using the 5V supply as the reference for the A/D, what
> supply are the I/O lines referenced to?
> The digital I/O could do with some resistance in line as well, a couple of
> hundred ohms would provide output short circuit protection and give some
> filtering on the inputs. If your digital inputs are coming off another board
> then you could see some ringing on them, that could be more of a problem on
> PortB which doesn't have Schmitt inputs. 
> The 5V regulator has plenty of protection, that's good. Automotive supplies
> are horribly noisy. The 28V military stuff I design has to be protected
> against +280V/-130V spikes. The EMC test involves 100V for 100mS. You could
> reduce the protection needed by using a regulator specifically designed for
> automotive use, there are a few of them out there. I have used the LM2940S
> in the past and that works well but I think the best spec only comes in an
> SMD package.
> I hope this doesn't sound too harsh, I don't see any reason why it won't
> work but I think you will lose some A/D resolution by having to digitally
> filter the inputs. I learnt the hard way with a few board redesigns along
> the way. The Microchip website has a very good on line seminar about PCB
> layout and design for A/Ds, its well worth checking out.
> HTH
> James
>
>
>   
>
> ------------------------------------------------------------------------
>