Watch your ground loops....

[Th3_uN1Qu3]

I've been battling with oscillation issues on my +/-60v adjustable supply for days. On paper and in simulation the compensation was correct but it just didn't work in the real world. More so - the sim said that from a certain load level up, the thing would not need any compensation at all! And yet it was oscillating at all loads...

I noticed i could influence the oscillation frequency by touching the ground of the TL494 board. Hey wait a second... The 494 was wired to ground at the standby supply. As the last thing to try, i soldered a ground wire from the 494 board to the ground of the filter capacitors (which all this time have been hanging by wires outside the supply). And i'll be damned... the thing shut up as soon as i did that. Just like that.

Morale: When you're prototyping, keep wire lengths short, and always, always, always, always have your control circuit's ground connected directly to the ground of the output filter caps, with a separate wire.

PS. Provided the load remains constant, a power supply that is stable will not start oscillating by itself. If you hear something crackling, especially if it's combined with a rise in the current consumed by the PSU, that's not oscillation, that's a part just about to blow up. Can you guess how i know that one?

[momaka]

Re: Watch your ground loops....

Can you guess how i know that one?

Oh, I don't know... the gray magic smoke perhaps? Or the wonderful smell of hot, miserable electronics?

Yeah, ground loops or missing grounds can be a pain to troubleshoot sometimes. Just until recently, we used to have these old Heathkit PSU/breadboards in our labs at college. They weren't bad, but they were old and have probably withstood quite a bit of abuse over the years. Thus, a few of them had some dead spots on the breadboard. Took me a while once until I found why my circuit was producing complete nonsense.

They finally changed them this year, though.


We also had an experiment for one of our electronics classes where we were supposed to measure the signal across different components in a LRC-series circuit with an oscilloscope.
Of course, the problem with this experiment is that you can't measure the signal across each component with an oscilloscope - not unless you re-arrange the order of the components in the circuit each time. If you don't yet understand why you have to change the order, think again - the oscilloscope probe has 2 wires: (+) and ground. So, when you try to measure a component that is in the middle of the LRC-series circuit, the ground wire on the scope probe will completely mess up the ground of the circuit.

This can be avoided easily, though, if the component being measured across is moved all the way at the end of the LRC circuit where the ground is on one side. Then the scope probe can be connected to that same ground, and the grounding issue would be gone.

Of course the the lab manual for that experiment never mentioned this. Moreover, this class was an introduction to linear electronics (i.e. fairly basic level course), so you can imagine how many students had no idea what was going on and got wrong results. Even my group did, but by the time this issue caught my attention, we already had acquired the results the week before and were already working on a different experiment. Not like it mattered much, though - the teaching assistant that was teaching our lab was no good at all and she never noticed any problems in anyone's results for that experiment, including ours. She even told one group they are pretty much SOL because she couldn't help them - that's actually when I found out about this grounding issue (I went to that group and helped them fix this problem).

[Th3_uN1Qu3]

Re: Watch your ground loops....

Of course, the problem with this experiment is that you can't measure the signal across each component with an oscilloscope - not unless you re-arrange the order of the components in the circuit each time. If you don't yet understand why you have to change the order, think again - the oscilloscope probe has 2 wires: (+) and ground. So, when you try to measure a component that is in the middle of the LRC-series circuit, the ground wire on the scope probe will completely mess up the ground of the circuit.

I remember one of our labs - someone in my group was trying to measure current with the meter set on volts... The teacher didn't notice it - i did, and it saved our results.

[momaka]

Re: Watch your ground loops....

I remember one of our labs - someone in my group was trying to measure current with the meter set on volts... The teacher didn't notice it - i did, and it saved our results.

Oh yeah, I've done that too, although I noticed it right away when we weren't getting proper results.
For some reason, towards the beginning of last semester there was very few probes left for the multimeters so we had to use only 1 multimeter for an experiment that required both current and voltage measurements. Took much longer to complete the experiment since we had to switch wires back and forth all the time.

But yeah, there are some kids in my classes that really have no idea what they are doing. Most of them are doing EE just because they think they will get a good paying job very easily. Sometimes it's kind of funny to watch them wire a circuit because they just try and wire things until the circuit works. Luckily most of the experiments we do are low voltage, low current, so it's rare to see components letting out the magic smoke. There are people who have done it, though (I'm not exactly sure how, but anyways).