'Scope Question

[OddHarmonic]

So I brought my first two Green Pre channels into work to set the CMRR, LEDs etc... I don't have a scope yet at home, but we have a variety at work.

Tektronics TDS 2014 4-channel digital scope
Tektronics 5110 scope w/ 5A14N 4-channel amp, 5B10N Time/Base Amp
Goldstar OS-9020A

As we usually use the digital scope and inject instrumentation signals directly, I have only found one pair of probes in the equipment closet:

Tek P6109 (10X, 1.5M, 11.8pF)

I read about setting the compensation for the probes by pulling the 5V Square wave from the scope and adjusting for a flat leading edge response.

Now, setting CMRR on the Green Pre requires that you observe the adjust to get the signal down to 1-2 mV. On all three scopes with these probes, there seems to be a noise floor of about 20-30 mV, preventing me from setting the CMRR below. I used my DMM (AC V) to adjust to 2 mV, but the scopes seem useless. I know that ain't right!?

The probes are certainly questionable, as they are probably pretty old and haven't been used in years. The two analog scopes suffer from the same question. The digital scope, however, is pretty new, and used often.

Am I using the tools incorrectly, or might the probes be bad?

thanks,
andrew

 

[PRR]

> there seems to be a noise floor of about 20-30 mV

Resolve the noise. Is it signal, hum, RF, or truly random?

BTW: while you should of course check a X10 probe's calibration, it doesn't matter from bass to midrange. The R-C product is near the top of the audio band; below that, you are on the resistors which should be stable.

Check the probe's general operation. Does it really attenuate? I have seen the compensation caps blasted with high voltage and shorted.

 

[luckylocky]

Did you check the grounding?
Noise on scopes often lurks over the ground.
Make sure that your ground is clean. Maybe you'll have to ground lift the scope. For measuring use the ground provided on the probes and keep the unscreened lines as short as possible.


Cheers Patr!ck

 

[OddHarmonic]

Thanks PRR & Patr!ck,

A few answers, the noise appears to be random.

The probe does indeed attenuate correctly.

Switching to a different ground didn't do much, I will try lifting it altogether.

I checked back into the Tektronix "Scope Primer" and caught something that I missed the first time around. They don't reccommend using attenuated probes to measure signals below 10 mV Pk-Pk, which I am doing in the CMRR measurement. If I directly inject a 1K signal from my generator, I can distinguish the signal from the noise floor to about 1 mV, even a bit below. I guess the biggest malfunction was in my head, to no suprise. Although I believe setting the CMRR with my DMM will work, I suppose getting my hands on a 1X Probe would also be usefull.

Andrew

 

[bcarso]

Be aware as well that scope bandwidth is often way in excess of the signal bandwidth you are interested in. Sometimes a little R-C lowpass right before the scope is helpful.

 

[OddHarmonic]

The digital scope does have a 20 MHz lowpass which I have been using, but would it be helpful to further limit the bandwidth when dealing with the audio band?

 

[PRR]

> further limit the bandwidth

For the CMRR test: hang 0.1uFd on the output that you are scoping. That will limit bandwidth to ~17KHz.

Be SURE to take that cap off before doing any other testing! It limits small-signal bandwidth, but will really crimp large-signal performance. Above a couple volts everything will come out triangles. And of course the frequency response will taper-off badly, -1dB at 8KHz. So it is OK for CMRR tests at 20Hz-5KHz where you only have small residual output, but no good for anything else.

Actually, this might be a way to test Noise without being distracted by supersonic garbage. A 1-pole 17KHz low-pass has roughly the same bandwidth as a 20KHz brick-wall low-pass. Not close enough for precision reports, but good enough for jazz, and a lot more relevant than counting all the random noise up to 10MHz.

 

[PRR]

Also: use your dang ears! Put the 1V 1KHz (or whatever is specified) common-mode test signal in, listen to the output, turn the pot for minimum 1KHz tone. This will also tell you if "odd things" are happening, like you put 1KHz in and get 2KHz out (gross distortion, probably a half-working stage). Or if your "noise" is mostly 60Hz hum.

The ear will track a tone far below random noise. And frankly, if 1V common-mode is nulled below self-noise, that's really good enough for all but the worst situations (we had an arc-welding generator in the building today...). If the CMRR residual is above noise, the 'scope and eye is a little more precision than the ear: on a scope you can find the minimum output within a couple percent, by ear you won't hear 10% off-null. But being 10% off puts you within 1dB of the best you can do, and by definition is good-enough.

 

[CJ]

1 to 2 mv will come out fuzzy on any scope unless you put thru a 1:10 transformer with multiple shields. But then there goes your cmr adjust out the window and into the xfmr.

I would measure the gain of the amp with a signal you can see, then put the scope on the output of the amp somewhere along the line, extrapolate what the scope should read in order to have 1 mv at the input, set the generator to give you that voltage, and adjust from there.