Finding Common Ground

[lnadolski]

Hello,

I am working on a PCB with relays that require a 24VDC coil voltage and op-amps that require +-12VDC supply voltage. I am using two AC/DC power supplies from Automation Direct (PSB24-060S-P (24V) and PSE12-215 (+-12V)). My original thought was to connect the negative terminal of the 24V supply and the COM terminal of the +-12V supply to the common ground plane located on my board (which is also attached to a DAQ ground). The problem is that I am unable to connect the 24V ground and +-12V COM because there is potential, measured via DMM, between these two points. If I attempt to connect the +-12V COM terminal to the 24V negative terminal the LED on the +-12V quickly goes out...I assume that I am creating a short and the +-12V is powering down.

Additionally, it appears that the +-12V supply is not sharing any common voltage with respect to the 24V supply. With respect to 24V supply negative terminal, the +-12V supply positive rail is 21.0V above, the COM terminal is 9.0V above and the negative terminal is -3.0V above (3V below).

Originally, I would have guessed that the 24V negative terminal and +-12V COM terminal would have shared a common point (guess not since the outputs are each isolated from input ) and +12V, -12V (wr to 24V negative terminal) for the other +12V terminals.

Edit: A requirement of the PCB is that the 24V supply be separate from the +-12V supply since it will be powering the PCB relays and may cause some switching noise which may cause noise if connected to +-12V power suppy (op-amp power supply terminals). If I do not intend to do any relay switching while measuring the output of the op-amps with my DAQ do the power supplies necessarily need to be isolated? If they are isolated can I ever refctify the grounding issue?
 
Has anyone ran into this issue before and found a good solution to 'force' the 24V negative terminal and +-12V COM to same potential or can recommend a system/power supply combination that would resolve the grounding issue?

Thank you so much in advance and I look forward to learning from you all.

Cheers,

L

[tautech]

Welcome to the forum.

Simple but not elegant workaround would use miniature relays in a cascade configuration to switch the 24V units.
Any spikes that might appear on the power rails will be addressed by sufficient local bulk capacitance and appropriate decoupling.

Interesting problem.... 

[madires]

Is the negative/common output of both PSUs connected to mains earth?

[lnadolski]

@tautech  Thanks for the response.

By 'Simple but not elegant workaround would use miniature relays in a cascade configuration to switch the 24V units.' are you suggesting to use the relays to somehow shift the output voltage of the 24V supply or Sorry I am slightly confused by the wording.

I have designed the +-12V supply rails to have local bulk caps (6.8uF) at PCB input terminals and decoupling caps (0.1uF) located in close proximity to the op-amp rails. I simply used the suggested values provided by the manufacturer (these were also values used by manufacturer on IC eval board).

Thanks again and look forward to more feedback.

@madires   Thank you very much for the follow up question.
No. I believe that is part of my issue.
The negative terminal of the 24V supply is connected to a 'ground' plane on my PCB. Also connected to this 'ground' plane is the GND terminal of a LabJack U6 DAQ which is connected to my computer, via USB, that is plugged into the wall using a 3-prong plug.

Thanks again.

[Chris Mr]

Well, the data sheets say that the isolation voltage of each is 3Kv or more.

It also only shows two input terminals for the mains on one of them (the +/-12v one).

Hmmm

If it's isolated, the one with two input terminals, then you should be able to connect a (large wattage) 1K, or whatever, resistor from any output terminal to local ground and then measure all the voltages you expect.  Check that the voltage across the resistor is low.

You can do something similar with the 24v one- try disconnecting the earth from the input side first.

Over 

[Chris Mr]

Ah, another post in-between posts.

Best to try things in isolation.  Take one PSU and try the resistor to Earth (I said ground last time meaning Earth |), then the other, then the resistor between the two and so on.  All this without them being connected to the board; just on their own.

See how you get on.

[tautech]

@tautech  Thanks for the response.

Simple but not elegant workaround would use miniature relays in a cascade configuration to switch the 24V units.
Any spikes that might appear on the power rails will be addressed by sufficient local bulk capacitance and appropriate decoupling.

I have designed the +-12V supply rails to have local bulk caps (6.8uF) at PCB input terminals and decoupling caps (0.1uF) located in close proximity to the op-amp rails. I simply used the suggested values provided by the manufacturer (these were also values used by manufacturer on IC eval board).

I wouldn't consider 6.8uF to be nearly enough bulk capacitance.
100uF is the minimum I'd place on a remotely powered PCB.

Some time spent with a scope on the rails will show you why. 

[lnadolski]

@Chris Mr   Thank you very much.
I will report back with results.

@tautech  Thanks again.
Yes, I have noticed quite a bit of noise on rails and probably some on the op-amp output given the insufficient size of the bulk caps I am using. Will update and scope. Thanks.

[CatalinaWOW]

The suggestion of a resistive coupling between the grounds is a good investigative check, but I really think something is wrong.  You should be able to connect the grounds as you originally proposed.  A defective supply is a possibility, but is unusual to get one of these defective out of the box.

Other things I would look to for clues.

Does this occur when nothing else is connected?  You may have something going on elsewhere in your setup.

What is the voltage between ac ground on the 24v supply and the 24v negative?

[lnadolski]

@CatalinaWOW   Thank you very much for the response.

I have remove the PSUs from the PCB and connected the 24V negative terminal and +/-12V COM terminal without any negative side effects. Once connecting the two previously mentioned terminals,  I measured the voltage between ground (24V negative terminal and +/- 12V COM connected)  and the PSU rails and all the measured voltages made sense...+24V, +12V and -12V. Thank you.

So I would guess that my problem is the grounding of my laptop/LabJack U6 DAQ?

Thanks again.

[CatalinaWOW]

You will have to trace possible ground paths one by one.  You at least have the advantage of a clear and obvious fault.  Often grounding issues only show up as noise and can be very difficult to run down.  I don't know that I would suspect the DAQ at first.  Everything is in the suspect list, and I would look at your PCB first.

Without knowing more about your PCB it is hard to give specific advice, but I would probably start by connecting the 24 v supply and the ground of the 12  v supply to the PCB without hooking up the 12 v power lines.  Then add the ground to the DAQ.

If you have the schematic for the PCB look for all the places that ground is hooked up and think about them.  If it is a PCB of your own design spend time making sure the PCB does not have ground connections you didn't intend.

One DAQ related thing I would check.  The terminals on this DAQ are fairly closely spaced.  If you used stranded wire to make your connections it is not that difficult to have a strand of wire bend out from the bundle and touch adjacent positions.  Use a magnifying glass or other means to closely examine your connections.

[lnadolski]

@CatalinaWOW  Thank you very much for the follow-up.

While going through all ground connections on  my PCB I found that the bulk cap on the -12V rail was installed backwards (reverse polarity). After removing this cap and installing a new one with the correct polarity I have been able to successfully connect all PSUs and my laptop\DAQ to the PCB for measurement. I believe this issue has been resolved...on to the next one! 


Thanks again everyone for you comments and help troubleshooting this issue.

Cheers,

L

[tautech]

While going through all ground connections on  my PCB I found that the bulk cap on the -12V rail was installed backwards (reverse polarity).

Really ? 
Holy hell. 
From the factory that's an inexcusable mistake. 

Automation Direct PSU's you said ? 

[lnadolski]

@tautech

That was my mistake NOT Automation Direct. The PCB is something I designed and hand assembled the first prototypes and accidentally flipped one of the bulk caps.

Holy hell is right 

[CatalinaWOW]

And my congratulations to you.  I probably wouldn't have found that fault as quickly as you did.