Limits of test equipment

[Ahrenp]

I have a Rigol DS1054Z scope and a Siglent SDG2042X wavegen and I am working with signals around 25-50mVpp and 1 - 10kHz. I'm feeding the sine wave through a BNC tee with one end going directly into the scope and the other going to my circuit, connected with a BNC -> Grabber adapter (https://www.digikey.ca/product-detail/en/pomona-electronics/3788/501-1029-ND/603339). As soon as I connect the adapter the signal starts to jump all over the place on the scope and it becomes very difficult to get it to consistently trigger. It triggers fine with the circuit end of the BNC left unconnected. The leads on the adapter are fairly long so I cobbled together my own connector with a female BNC connector with some short wires soldered to it in an attempt to mitigate any problems that may be associated with the cable length, but that only offered minor improvement. Amplifying the signal and triggering of off that offers some improvement as well but does not completely solve the problem. If I do a single shot measurement the waveform generally looks pretty clean.

Are my issues due to limitations imposed by my test equipment (i.e. the ability to trigger on a low amplitude signal, ability to generate that signal) or my test set up? What can I do to improve my results? As a workaround I can generate a pulse to trigger on for testing purposes but if I can improve my set up or input circuitry that would be ideal.

As an aside - I've been using the scope to measure RMS voltage and I'm finding that the measurements are wildly different between channels for the same signal, vertical scale, etc between channels. I don't expect great accuracy but they differ by as much as a couple volts and neither agree with my multimeter. Is this typical of scope measurements?

[Andy Watson]

Are my issues due to limitations imposed by my test equipment

25mV is at the lower end of what I would consider a reasonable signal level for a normal oscilloscope. It should be possible to operate at this level but, as you have noticed, the signal can be dominated by noise. Limit the bandwidth as much as possible.
But :

As soon as I connect the adapter the signal starts to jump all over the place on the scope

Consider that you are also making a connection to the ground of your target circuit - this can be a source of noise.

Can you work around the low signal level by operating the sig-gen and oscilloscope at, say,  250mV, then use a simple resistive divider to feed your circuit? Or, since you are using audio frequencies, would it be practical to isolate your circuit with a transformer - this isolates the ground and offers the opportunity to scale the signal level - either up or down?

[Ahrenp]

Consider that you are also making a connection to the ground of your target circuit - this can be a source of noise.

I guess I should clarify that the adapter alone begins to cause problems - even before it is connected to the circuit. This is what prompted me to use to something with shorter leads. The wires on my homemade adapter are then connected via a screw terminal to the circuit on a proto board.

Can you work around the low signal level by operating the sig-gen and oscilloscope at, say,  250mV, then use a simple resistive divider to feed your circuit?

I hadn't thought of this and it may be something to try. This got me thinking though; something I forgot to mention in the OP was that the sig-gen is connected via an op-amp configured as a differential amplifier - not directly to circuit ground. If I play with the input impedance of my diff amp can I get a reduction of noise?

[w2aew]

It is quite likely that your signal generator's output is *not* floating - the shell of the BNC might likely be connected to GROUND.  So, if you connect it to a non-ground referenced differential input, you're going to be tying one side of the diff input to ground - probably not what you intended.  There are some function generators whose outputs are floating and will work this way - you just have to check to see if your's is one of them.

And on your scope, the probe ground IS connected to GROUND, so you can't probe your differential input directly... ...for the same reason.

[Andy Watson]

something I forgot to mention in the OP was that the sig-gen is connected via an op-amp configured

That throws something else into the  mix - not all op-amps are happy when driving into capacitive loads - like say a length of coaxial cable. Try isolating the output of the op-amp from the cable with resistor - 100-200 ohms should do.

[Ahrenp]

It is quite likely that your signal generator's output is *not* floating - the shell of the BNC might likely be connected to GROUND.  So, if you connect it to a non-ground referenced differential input, you're going to be tying one side of the diff input to ground - probably not what you intended.  There are some function generators whose outputs are floating and will work this way - you just have to check to see if your's is one of them.

And on your scope, the probe ground IS connected to GROUND, so you can't probe your differential input directly... ...for the same reason.

You are absolutely right and this was an oversight on my part. I will have to rethink the way I am doing this and determine if my signal generator is floating.

something I forgot to mention in the OP was that the sig-gen is connected via an op-amp configured

That throws something else into the  mix - not all op-amps are happy when driving into capacitive loads - like say a length of coaxial cable. Try isolating the output of the op-amp from the cable with resistor - 100-200 ohms should do.

Sorry, I may have worded that poorly. The diff amp isn't driving the cable, the cable is driving the diff-amp input to the circuit.