I just found out that oscilloscope probes shouldn't be connected to different ground reference points. I was probing the input and outputs of an audio amplifier using separate probes. So both probes negatives were not clipped at the same place. I ended up frying my phone which was the input to the amp, it was connected at that time.

I am clearly lacking in fundamental knowledge every newbie should know, potentially endangering myself and others.

So how is this possible? Arent oscilloscopes supposed to be infinite impedance? And I guess the inputs and outputs of the amp don't same a common ground as I incorrectly assumed.

What is the proper procedure?
Thanks in advance!

 

On most mains powered scopes the vertical input commons are common to each other and also ground. There are various differential vertical inputs but for the most part the vertical inputs are BNC connectors and the outer shells are common ground. Oscilloscopes have an input impedance which is typically 1 Meg Ohm shunted by some capacitance. When using a 10X probe that input impedance goes to 10 Meg Ohm in most cases. All of this depends on the scope and any probes used. So I guess the question is what scope are you using? The input impedance is not infinite.

The procedure will depend on what you are trying to measure and the scope you are using, they are not all the same.

Ron

 

Oh ok I guess 1MOhm isn't really that high. Dumb me. Well the scope I am using is a Keysight DSO. I was probing the inputs and outputs of a Class D amp. TPA3118 amp with a 24V DC supply to be exact.

 

"Bridge tied load" (BTL) audio amplifiers have become quite common, especially when high power with low supply voltage is desired.
Instead of connecting the load between the amplifier output and ground, the load is connected between the ouputs of two amplifiers which are configured so that their outputs are 180° out of phase. This gives the same effect as doubling the supply voltage. The low cost and compactness of class D amps abets this configuration.

Check the Keysight (oooh, I detest that name!) website for applications notes on using oscilloscopes and probes.

The nominal resistance at the tip of a 10:1 high impedance attenuator probe is 10 megohms. BUT even a good quality probe with moderate bandwidth (e.g. in the range of 100-200 MHz) also has a shunt capacitance to ground of about 12 pF, give or take 2 or 3 pF. By the time you get to 20 kHz, the impedance at the probe tip, assuming 12 pF, is down to about 663 k. At 10 MHz it is down to about 1330 ohms. You can buy highish impedance probes with capacitance of around 1 pF at the tip. These are "active" probes that have an amplifier right in the probe body. They are quite delicate, easy to kill with overvoltage and one with reasonably decent bandwidth will be $1000 plus.

Zc = 1/(2π × f × C) Z in ohms, f in Hertz and C in farads
For actual impedance you should include the resistance in parallel, but as you can see from the values above, it can be ignored once you're up to a few kHz.

 

Quick perusal of the TPA3118 data sheet - it looks like it has a bridged output, neither of the output pins are at actual ground, they are both 'hot'. By connecting the scope probes the way you did, you shorted half the output of one channel.

Don't know how that would have blown out the signal source though.

 

Perhaps you accidentally connected one of the probe grounds to a hot line while the other probe was connected to ground.

 

Ouch! I hope I didn't cause any permanent damage to anything else...

The amp?
Especially the scope?

 

Quick perusal of the TPA3118 data sheet - it looks like it has a bridged output, neither of the output pins are at actual ground, they are both 'hot'. By connecting the scope probes the way you did, you shorted half the output of one channel.

Don't know how that would have blown out the signal source though.

Perhaps you accidentally connected one of the probe grounds to a hot line while the other probe was connected to ground.

Along those lines would be my guess. I would suggest reading the scope manual to see how the vertical input channels can be used.

Ron

 

Yes, it is very odd that it should take out the signal source. Perhaps if we knew the complete circuit down to all the details we might be able to figure it out. There is a certain perversity in electronic devices that somehow always manages to cause damage to the hardest to fix or most expensive part.

The scope itself is unlikely to have been damaged. You may have put quite a lot of current through the ground leads and shields of the probes, but there is usually a lot of copper there so it takes a lot of current for some time to get it very hot.

I haven't looked at a lot of them, but overcurrent protection seems to be very common on the outputs of class D amp ICs, so they'll usually survive considerable indignity.

I once has a very bright but not very experienced high school student working with me. He was repairing a video terminal. I told him to be careful, that that the neggy end of the big capacitor wasn't ground, but I didn't explain properly. He burned the insulation off the ground lead of one of the probes (worth about $300 each). No other damage that we could detect.

 

from the data sheet
Integrated Self-Protection Circuits Including
Overvoltage, Undervoltage, Overtemperature, DCDetect,
and Short Circuit With Error Reporting [emphasis mine]

which all sounds good, except it's ratted you out to ???

 

So both probes negatives were not clipped at the same place. I ended up frying my phone which was the input to the amp, it was connected at that time.

The way I read this, you are saying that the grounding clips on the probes were connected to different places in your circuit.

Bad! Bad! Bad!

No wonder you blew something!

Guess what?

You shorted your circuit through the grounding clips.

 

Oh ok I guess 1MOhm isn't really that high.

No, that is not the problem. The problem is that the two probes each connect to the same ground. The impedance between the two ground connections is basically zero.

Edit: This gets even worse if you are working on a line powered device. You CANNOT connect the probe ground to anything but the neutral / ground connection on the device under test or you will literally see sparks fly. The recommended method of doing this is to use an isolation transformer for the scope so that there is not connection between it and the device under test.

Bob

 

Watch this:

 

No, that is not the problem. The problem is that the two probes each connect to the same ground. The impedance between the two ground connections is basically zero.

Edit: This gets even worse if you are working on a line powered device. You CANNOT connect the probe ground to anything but the neutral / ground connection on the device under test or you will literally see sparks fly. The recommended method of doing this is to use an isolation transformer for the scope so that there is not connection between it and the device under test.

Bob

No! Floating the scope and connecting a ground clip to a live point will put all metal on the scope at a high voltage (unless you are using a battery-powered scope that is designed with the proper insulation). Connecting the scope ground to neutral is not legal per NEC since it will create a parallel neutral path and all neutral current on that circuit can also flow through the scope probe and PCB...

The only safe way to do this is to use the isolation transformer on the unit under test (but different scope channels will still have a common ground connection) or better yet, use differential probes.

I once had to use a crowbar to remove a scope from a grounded metal bench when someone floated the scope and the rubber foot on the scope failed.

 

"I once had to use a crowbar to remove a scope from a grounded metal bench when someone floated the scope and the rubber foot on the scope failed."

I know I'm not supposed to laugh at such things, but I did. Someone could have wound up seriously injured or dead.
Floating the scope ground is futile for high frequencies, which of course includes the fast edges of low frequency signals, because of the capacitance between the scope chassis and - well, everything.

Has anyone here used any of the relatively low-cost differential probes - the sort with input leads much like meter leads? They seem to me like an excellent solution for a good deal of industrial work on things like motor controls. Some claim surprising high bandwidth but I never looked carefully at the common mode rejection at higher frequencies.

 

Iamroot:

Simply sounds like you have a floating power amplifier. It is not ground referenced. And yes you could burn out your scope PCB ground trace or pop the probe open. For your case I suggest using two probes and set the scope in differential mode. The scope gets grounded to the chassis separately. Probe one is GND reference on the amps - output, and probe two is hot +. This is now an isolated input. See scope manual for set up. Second choice, get a differential probe, and directly connect to amp output terminals. Hope this helps.

 

http://download.tek.com/document/51W_10640_1.pdf
https://www.google.com/url?sa=t&rct...tOscopes.pdf&usg=AOvVaw05dghFdGtskROhJLeOcPmf

https://www.google.com/url?sa=t&rct=j&q=&esrc=s&source=web&cd=6&ved=0ahUKEwjpsumL9MrZAhVIdt8KHQFrBe8QFgh_MAU&url=http://www.newark.com/pdfs/techarticles/tektronix/FFM.pdf&usg=AOvVaw2udBMlAswXXiYwjquqYBRQ


Regards, Dana.

 

No, that is not the problem.

....

Bob

Part of the original post:

"So both probes negatives were not clipped at the same place. I ended up frying my phone"

MrChips, is correct. The OP connected two different points, in his phone, together, by way of the ground clips. ... I see smoke escape !