The connector is polarized, so that's not an issue.  A lot has to happen=20
simultaneously -- analog sampling, solenoid driving, etc. But yes I am=20
averaging and filtering the analog signals -- up to 128 samples on some=20
of the signals that don't change rapidly.  My concern is more with the=20
digital serial signals getting corrupted by the higher-voltage signals.

I do have one signal] which has a pull-up and that's a shutoff-switch=20
input (ie: tells the system to shut off completely). I'll have to look=20
into keeping this further away.

I had not thought about ferrites but I'll look into those.  I did find=20
this neat doc on EMI which I'm still going through, and it too mentions=20
ferrites...
http://www.ti.com/lit/an/szza009/szza009.pdf

None of the serial is differential -- one is RS232 (actual RS232 level=20
from a MAX232) with error checking.  Another is also RS232, with no=20
error checking, but that comes from a different system so I don't have=20
much control over that.  The third serial data-stream is a 5V level=20
custom protocol.  It does have a checksum in both directions.

I'm confused about using 2 grounds yet merging them where they enter the=20
board.  Wouldn't that mix any noise generated by the motor side onto the=20
logic-side?

Cheers,
-Neil.


On 8/26/2015 3:07 AM, Jesse Lackey wrote:
> Hummm.... neat question.
>
> One aside: if the connector isn't polarized, you could use strategically
> place the spare pin such that if the connector is inserted backward and
> power applied, the +12V goes to this unused pin, and the user gets
> nothing happening vs. smoke.
>
> Can you arrange things such that you aren't sampling the analog sensors
> while motors are running (or starting/stopping), or solenoids switching?
>    Or maybe at least take an average over time to minimize bad data.
>
> Digital I/O ... it would be best if all signals are driven, i.e. not
> have pullups, for better noise immunity.  Not sure how much it matters.
>
> You could use that spare pin for a ground for digital and/or analog, and
> have the digital and/or analog ground therefore be separate from the
> rest, and connect the grounds together where the power comes in from
> whatever is plugged into this and making the +12V.  That said, I have
> seen cases where the best performance was to use the same ground all over=
..
>
> A ferrite on the +12V @ the connector would be a polite thing to do,
> maybe on the motor as well.  They're cheap.
>
> I assume the serial is differential (rs485?).  Doing it packetized with
> a CRC and having both ends of the communication keep count of how many
> packets are damaged is a good way to monitor signal integrity.  i.e. do
> things all command/response style and be able to ask that count to
> determine % damaged.
>
> I look forward to this thread!
> J
>
>
>
> Neil wrote:
>> I'm trying to figure out how best to allocate connector pins for a
>> mixed-signal board.  The system has high-current, high-voltage,
>> repeated-switching signals (motors and solenoids), serial data signal
>> pairs, and a number of analog sensor signals coming in.  The catch is
>> that this board has only one 30-pin connector (3 rows of 10 pins each).
>>
>> Currently, I have, in order starting from one end towards the other:
>> - Motor + and - (12V, 6A max)
>> - Solenoids (up to 4A, but with reverse spikes will see 30V+)
>> - Power & ground (12V)
>> - Serial signals (3 pairs)
>> - Digital I/O
>> - Analog sensor signals
>>
>> The PCB has the motor and solenoid drivers on one side of the board,
>> power supply in the middle (but off towards the back), and 2 PICs that
>> handle all the logic near the other side.  The analog circuitry is kept
>> really short and off in one corner of the board nearest the end of the
>> connector with those signals.
>>
>> So what I'm wondering is:
>> - Is there a better way to organize the signals on the connector?
>> - I have one spare pin -- should I run a separate logic ground?
>> - If so, where would I connect these on the board (or wouldn't this
>> cause a ground loop)?
>> - If I do or don't, should I split the ground plane so the logic side is
>> separate from the motor/solenoid side?
>> - Anything else I should do?
>>
>> Cheers,
>> -Neil.
>>

--=20
http://www.piclist.com/techref/piclist PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist
.