Mutual coupling/kickback/IC protection question

Thread Starter

dregan

Joined Apr 3, 2008
6
I have a friend that is working on a project and has run into some weird behavior. He has a PCB that has the control logic for an external relay as well as an AD converter that is taking some external inputs. The entire thing is powered by a battery. The entire thing worked great until a recent redesign. Now, the AD converter stops working, usually right after the external relay is fired. There are a couple of changes that I'm focusing on as being the cause of the problem.

First, the AD input and Analog ground lines have been moved to exit the same port as the control lines of the solenoid. Also, the ad converter is now monitoring the battery voltage as well.

I'm wondering if, since the ADC input lines are now in close proximity to the relay control lines, that when the solenoid is powered, there is kickback current inducted on th ADC input lines due to mutual coupling.

Also, some of the measurements are very sensitive and he found that tying the battery ground to the ADC analog ground caused the measurements to be inaccurate after the layout change, but not before. The analog ground for the inputs comes from the measured devices through the port that runs close to the relay control lines so I'm wondering if the effects of mutual inductance are causing the measurements to be inaccurate when the battery ground is tied in but not when only using analog ground (the mutual inductance is affecting the high and ground lines going to the ADC equally so it cancels them out)? Can anyone explain this to me a little better?

I suggested he put TVS diodes on all of the inputs to the ADC that tied them to analog ground (not battery ground) and set the clamp voltage to 5V. Any other ideas on what to do or troubleshooting experiments to try?

Thanks for your input!
 

mik3

Joined Feb 4, 2008
4,843
It would be helpful if you could post the circuit diagram and the PCB layout. He shall use protection diodes on the analog inputs.
 

John P

Joined Oct 14, 2008
2,025
Has this PCB been properly assembled? All power and gnd connections carrying what they should, no shorts or opens?

I'm skeptical about inductive effects. You might see a glitch when the solenoids change state, but once things have settled, it should be OK.

In any mixed analog and digital or power circuit, you need to be careful that the grounds are joined just once, and that no current from the digital/power side runs through the analog ground. If there are sensors delivering a voltage level versus a nominal ground, then you have to treat that ground as a signal too.

Try running the analog parts with no power applied to the solenoids. Does the problem exist then or not? If it does, then it obviously has nothing to do with the solenoid current.

That A/D monitoring the battery voltage: How is the A/D powered, via a step-up converter running off the battery? I hope there isn't an attempt to measure the A/D's own power supply, unless it's some really exotic thing with an internal Vref.
 

Thread Starter

dregan

Joined Apr 3, 2008
6
Thanks for the help all, I don't have access to the PCB layout so I won't be able to post it.

Try running the analog parts with no power applied to the solenoids. Does the problem exist then or not? If it does, then it obviously has nothing to do with the solenoid current.
Yes, it is still inaccurate when the solenoid is not operating so perhaps the solenoid is not to blame... This is what I'm trying to wrap my head around though: The solenoid doesn't stop being an inductor when it is not operating, yes there is no current being applied on the control lines, but the AD input lines are next to the control lines. If there is a current in the AD input lines, then a current is inducted on the solenoid control lines through mutual coupling. Since the solenoid is still an inductor even though it is not being "operated" the inducted current on the solenoid control lines is not allowed to change instantaneously and since the analog input lines are mutually coupled to the solenoid control lines, the current is not allowed to change instantaneously on them either effecting the measurement. Is this not how it works? Or perhaps the effect would be so small it would be negligible (the measured voltage on the analog inputs is around 40mV)?

That A/D monitoring the battery voltage: How is the A/D powered, via a step-up converter running off the battery? I hope there isn't an attempt to measure the A/D's own power supply, unless it's some really exotic thing with an internal Vref.
The converter is measuring a battery but I'm not sure if its the same one it's running off of (there were two power sources, I'll have to check that). It is powered by a battery through a voltage regulator. It is the same AD converter that is used to measure the analog inputs which is why I think he had the analog ground tied to the battery ground before.

I can see that measuring ADC's own power supply would not result in an accurate measurement (or any sort of meaningful measurement at all for that matter). But will this cause the ADC to fail?
 

mik3

Joined Feb 4, 2008
4,843
If there are fast rising/falling signals, a bad PCB layout will affect the circuit operation significantly. Try to inspect the PCB to find loops with with high inductance where fast signals are switched.
 

thatoneguy

Joined Feb 19, 2009
6,359
If this problem only showed up after the relays were added, I guess I need to ask if you added a snubber diode across the relay coils?
 

wayneh

Joined Sep 9, 2010
17,496
I'm skeptical about inductive effects.
Ditto. This sounds to me like a problem with "ground" not really being ground, or some similar noise issue. Most folks would be amazed at the havoc even a few millivolts can cause, and how easy it is to have that develop on "ground" traces. Take pains to separate the current paths of any loads greater than, say 30mA, from low current "information" signals. And bypass, bypass, bypass with caps everywhere, until you've proven to yourself that you don't need them.
 

John P

Joined Oct 14, 2008
2,025
It seems as if you really want this problem to be caused by inductance one way or another! Sorry, but you've pretty much proved that it's not.

Along the lines of how the A/D measures the battery voltage--if that component or any other is in a situation where one of its input lines is higher than its power supply voltage, you're likely to see things fail to work. It's something to check on the circuit diagram, or on the board with a meter if there's any doubt.
 

Thread Starter

dregan

Joined Apr 3, 2008
6
Thanks for the help, so I originally suggested tying the over-voltage tvs diodes from the AD inputs to the analog ground input from the measured device but if I want to keep the analog lines below the supply voltage, it sound like I'd want to tie the TVS diodes from the analog inputs to battery ground.

Sorry if it seems like I'm getting stuck on inductance from the solenoid, it's just that the AD converter fails only when the solenoid is fired and they are fairly isolated from each other (the control lines of the solenoid go to a micro-controller and so does the output of the ADC but the control lines going to the microcontroller are diode protected). So it seems like the solenoid must have something to do with the cause of failure though perhaps not through induction.
 
Top