I keep ruining circuits with oscilloscope

[cabalist]

Hey there,

I've fried a couple of Servo control boards and I'm hoping for some guidance to what I'm doing wrong.   

The equipment in question:

Mg996r servos
Hiwonder Digital Servo Tester Servo Controller with Voltage Display (https://www.amazon.com/gp/product/B07485M6PH)
Hiwonder 6 Channel Digital Servo Tester with Over-Current Protection (https://www.amazon.com/LewanSoul-Channel-Digital-Over-Current-Protection/dp/B073XZH264)
Rigol 1054Z
Tenma 72-2660 Portable Power supply (https://www.newark.com/tenma/72-2660/power-supply-handheld-3-75a-12v/dp/56Y1289)

The power supply is connected to the Servo Controller @ 6V limited to 1A.  The servo is plugged in with Signal, Power, Ground in what I believe to be the correct pins. When power is turned on the servo twitches but turning the knob and/or pushing the button seems to do nothing.  I wanted to check the signal out and so I connected the oscilloscope to the Signal pin and the ground pin.  I smelled what I now know was the MCU getting very very hot.   I removed power and the probes and now the controller doesn't work.   I foolishly did this AGAIN and fried another board.   :/

Any hints on what I could be doing wrong?   The second board I fried was using a different servo...  The Power supply doesn't have a ground pin and so I believe I am not creating a ground loop.  Should I suspect the oscilloscope?  The servos? 

The oscilloscope reads its compensation terminal just fine.   I'm new to this and I'd really like to get past the point of mysteriously burning out boards.  Any help would be really appreciated.  Thanks!

[StillTrying]

"The Power supply doesn't have a ground pin and so I believe I am not creating a ground loop."

Check it's really floating, just in case. I don't think you're doing anything wrong, but on boards like that never trust the "+" and "-" markings, they're sometimes the wrong way around, check all the "-" are connected to GND. And of course keep the board on something no-conducting.

[bdunham7]

The scope probe input has 1M or 10M (you should start in the 10X setting) input resistance, so 6V will get you 6uA at the worst.  That won't heat anything up.  So you only have two connections to the scope--the probe tip and ground, right?  So the only way this happens is if you have current going thru the ground connection. 

First, watch Dave's video:



In fact watch it twice, since you've blown up two boards! 

Then use my own oscilloscope safety technique--remove your ground clips!  Just use the probe tip and no ground to measure and observe all of the outputs and the ground connection.  You'll see that your "ground" on the board in fact is neither at ground potential nor isolated, and has measurable potential to ground.  That or the scope has nothing to do with the problem.

[Psi]

It does sound like you are shorting something with the scope ground clip.
But from what you describe i cant see why its happening.

Things i would check
- Silkscreen markings on china PCBs are not always correct. Maybe the polarity is backwards somewhere.

- Before connecting the scope ground clip to anything put a small 5-12V lightbulb in series between the ground clip and what you're about to connect it to.
The light bulb should NOT turn on. If it does there is voltage between scope ground and that point. You will short it out if you connect the probe there.

- Do any of those boards have 79xx written on the voltage reg ICs?  It's pretty unlikely, but if those are negative regulators it might make sense why things blow up when you connect to ground.

[David Hess]

Something might be using ground side switching or current sensing in which case connecting the oscilloscope ground will short that function out.

Combine two oscilloscope channels using add and inverter mode (or subtraction) and use them as a differential probe with the ground leads connected together and not to anything else.

Also to verify what is going on, try connecting an oscilloscope probe tip to the ground point with the oscilloscope probe ground floating and see what you find.  If the voltage is not zero, then something is going on.

[cabalist]

Check it's really floating, just in case.

@StillTrying  In checking continuity between the negative terminal of the power supply and earth I am getting an intermittent continuity beep while the multimeter reads OL. I don't know how to interpret that. . It seems to be very briefly ground referenced?  The plug on the power supply is a C7 connector. 

@bdunham7 . First thing I watched after the first board died.    Thought I was safe since the PS only had a two wire plug.

I am hearing a lot of distrust for the silkscreen markings.   I would not have been that suspicious.  I'll be checking it on the next board that should arrive tomorrow.  Ground clips aren't touching that one.

The voltage regulators on these boards are marked to be: http://www.advanced-monolithic.com/pdf/ds1117.pdf  3.3v for the STC15L2K08S2 and STC15W404AS. 

 

Combine two oscilloscope channels using add and inverter mode (or subtraction) and use them as a differential probe with the ground leads connected together and not to anything else.

@David Hess
 I never would have though of that.   Very interesting.  Thank you.


Really appreciate all the insight!

[Psi]

I am hearing a lot of distrust for the silkscreen markings.   I would not have been that suspicious.

When a good manufacture makes a mistake and ends up with PCBs that have incorrect silkscreen markings they throw them away and make a new batch.
But in china someone takes them out of the trash and thinks.. "hmm... i can sell these"

Very common on aliexpress to get factory seconds that have incorrect silk markings.

[tggzzz]

There are many ways improper probing techniques can destroy scopes, equipment, and people. You need to understand probes and probing, since the probe and scope become part of the circuit being tested.

Before you think of it, never ever disconnect the earth in the scope (i.e. float the scope).

To understand the different classes of probe, when and how to use them safely, have  a look at the references in https://entertaininghacks.wordpress.com/library-2/scope-probe-reference-material/

[mikerj]

I'm a little suspicious of the power supply.  It's clearly a switcher and I expect will have a class y capacitor across the transformer which will give some leakage current.  Measure the output voltage of the supply w.r.t. ground (e.g. scope probe ground) and you'll probably find it's floating at a fairly high voltage (maybe 100v or so).  This should be a high impedance, try putting your multimeter in mA mode and measuring current between PSU outputs and ground, this should give a very small current in the uA range.

You could rule this out by powering the board from a battery so it's completely isolated.

[StillTrying]

Thought I was safe since the PS only had a two wire plug.

Yes, I thought the cable in the pic was only 2 core, the PSU must be floating.

Nothing else at all connected to the PSU?
Nothing else at all connected to the servos?
Using just 1 scope CH and GND? It's easy to create a short by connecting one probe's GND to GND, and the other probe's GND to something you thought was, or should be GND.

You / we will have to find the problem with the blown board(s) - before you connect the 3rd.

What mikerj just wrote, check for VAC leakage first, then current.

[DDunfield]

This should be a high impedance, try putting your multimeter in mA mode and measuring current between PSU outputs and ground, this should give a very small current in the uA range.

Only problem with using a meter in current mode to verify that no significant current is flowing is -- what if you are wrong (isn't that why you're testing)?
If this happens, at best you'll pop a fuse in the Multimeter (proper replacements can be expensive) or if a cheap unfused meter you could smoke the meter and/or the DUT.

Leave your meter in voltage mode and measure voltage across a couple-K resistor between the PSU and ground. If it is a high impedance leakage, you should read a very low value, (you can work out the actual current with Ohms law if you want). If you are wrong, the worst you'll do is smoke a cheap resistor (be aware that it could get very hot in this case).

Dave

[Shock]

Someone brought up a good point a few weeks back about charging the cap in an oscilloscopes input and then on the next measurement unknowingly discharging into a sensitive circuit/component.

[dnwheeler]

Are you sure it's your o-scope probing causing the problem? Perhaps the timing is just coincidence - you're doing the same steps in approximately the same order, so the same failure could just be happening at the same time after power-up.

[djacobow]

Why do you believe that PSU has isolated outputs? Looking at the datasheet I see no mention of isolation at all.

http://www.farnell.com/datasheets/2718447.pdf

Assuming it is not isolated, depending on the orientation of the power plug coming in, I'd expect the negative terminal to be shorted to neural or hot. If the latter, then hooking up the scope "ground" to that should create excitement, no?

[Audioguru]

It is illegal (and deadly) to have a power supply to have one of its output wires connected directly to an AC wire. Maybe this one is shorted like that.
Since the servos did not work properly BEFORE the 'scope was connected then the 'scope is not the problem.
I agree that the cheap Chinese products might have their power polarity marked backwards.

[JxR]

Now, I'm not saying this is your problem, but:

You need to be careful with this power supply.  It has (or at-least used to have) a design flaw, where it would supply a negative voltage when the output is turned off.  I killed a couple microcontrollers with it until I fixed the problem.  You can test this with a multi-meter to be sure.  The repair requires removing one of the transistors and adding some extra resistors.  It is discussed in this thread:

https://www.eevblog.com/forum/testgear/tekpower-tp3016m-power-supply-output-funkiness/  (same power supply, different brand)

Honestly, I probably haven't used mine in over a year, but the outputs are certainly isolated from mains, and after the repair it is a decent little unit for the portability factor.

Also, if you discover that you have this same problem I would consider adding a switch to physically turn the powersupply off while you already have it open.  It always bothered me that the powersupply remains on whenever its plugged in.

[Jeroen3]

Above power supply is isolated, see the double square symbol on the back, but it is not linear. (Maximum float voltage is on the back I believe.)
This means it is a switching power supply, and it has a high leakage current.

It also has a two capacitances:
- Capacitance from the Y capacitor, because it is switching.
- System capacitance that must discharge when you attach ground to it. This is what kills your board, as it is similar to a severe ESD burst.
The second one will also happen to any people who have "floated" their scope and attach the ground while power is already on.

A linear power supply with a foiled isolation transformer does not have the first problem, and the second one is negligible.

[David Hess]

Combine two oscilloscope channels using add and inverter mode (or subtraction) and use them as a differential probe with the ground leads connected together and not to anything else.

@David Hess

I never would have though of that.   Very interesting.  Thank you.

I have occasionally done this is the field when I do not know if "ground" on the device under test is the same as the ground my oscilloscope is plugged into.  Watching an oscilloscope ground lead turn white hot and melt once was enough.

[Psi]

Your mileage may vary using the two probe differential mode trick.

It works better on some scopes than others.
With some scopes it turns into a noisy mess very quickly.