Why is there a difference in trace width between 1x and 10x probe?

[ttodorov]

Hi everyone!

I have a Rigol DS1052Z and I was testing a simple transistor amplifier circuit using my function generator. I have been told to set the oscilloscope probe in 10x mode as to not load and distort the amplifier, which was not buffered.

I noticed that the trace on the oscilloscope screen looked very fat and ugly. So I hooked up 2 channels of the oscilloscope directly to the output of the function generator and took a screenshot to show you. The only difference between the settings is that CH1 is 1x and CH2 is 10x (both in settings and probe switch).

Can anyone explain why the blue trace looks so ugly, and if there is something I can do to fix this?

Thanks!

[wasedadoc]

Channel 2 probe's earth clip not attached?

[alm]

If you tell the scope you're using a 10x probe, it will amplify the signal a factor 10 more to compensate. So when it displays a sensitivity of 10 mv/div, then this means 10 mv/div at the probe tip, but 1 mv/div at the scope input connector. So you're seeing the noise of the vertical amplifier being set to 1 mv/div for the blue trace in your screenshot.

Basically the reason why 1x probes exist is because they allow higher sensitivity and lower noise for low level signals. At the expense of higher loading and lower bandwidth.

[ttodorov]

So there is basically no setting or a better probe that would fix this?

[Kleinstein]

Both the x1 and x10 probes have weak points. The x10 probe is more sensitive to noise of the scope input as the signal is divided down.
The x1 probe has more (especially capacitive) loading of the DUT and also usually has a quite limited bandwidth (e.g. 3-10 MHz range).  Much of the background seen with the x10 setting is from fast spikes that are reduced in amplitude by the reduced BW. The reduced BW also reduces the noise, but you of cause miss parts of the signal.

There are also active probes, that may offer a better combination of BW and noise. However these are expensive (more than the cheap scope) and need a special supply that higher end scopes may provide.

[tautech]

In circumstances like this one normally engages the 20 MHz BW filter if there is some muck of undetermined source on the waveform.

[Fungus]

If you tell the scope you're using a 10x probe, it will amplify the signal a factor 10 more to compensate.

No it doesn't. The only difference is it multiplies the numbers displayed on screen by 10.

[Fungus]

So there is basically no setting or a better probe that would fix this?

It's probably from your environment. A 10x probe has a higher impedance so it picks up a lot more radio noise.

eg.

[alm]

No it doesn't. The only difference is it multiplies the numbers displayed on screen by 10.

Oh, stop arguing semantics. While what you say is true, it does not help explain the difference between the two traces. The important difference between the two traces is that the vertical amplifier setting is different, not the number on the screen. The traces would look exactly the same if the bottom trace said 1 mv/div.

[Fungus]

No it doesn't. The only difference is it multiplies the numbers displayed on screen by 10.

Oh, stop arguing semantics. While what you say is true, it does not help explain the difference between the two traces. The important difference between the two traces is that the vertical amplifier setting is different, not the number on the screen.

The difference is that a 10x probe isn't a 1x probe so you have to manually turn up the amplifier to get the signal to be the same size on screen. The 10x probe attenuates the signal 10x, you turn up the amplifier 10x to see it on screen.

When you do that you're also amplifying the noise by 10x.

[noisyee]

Well, this is a trade-off in basically all kinds of instruments.
With input attenuation, you benefit:

• Higher input voltage range and better protection
• Lower loading effect or VSWR
• Lower distortion caused by the input circuits for the same input signal
• Possibly more bandwidth

And you sacrifice：

• Sensitivity or resolution
• Noise floor or SNR
• More likely to pick up interference for a high Z input instrument

In this case, 10mV/div with 10X has the same displayed noise floor as 1mV/div with 1X, and 10X probe is more likely to pick up interference. So the trace looks fat and fluffy.
If you are dealing with small signal, you would not want the input attenuation in most case!
If you really concern about the capacitance loading with 1X probe, try to insert a small series resistor, typically tens of Ω, between the DUT and the probe tip. This would help isolating the capacitance loading, but at the cost of lowering bandwidth.
There are some specialized active probes with little to no attenuation for this kind of work, but they are out of most hobbyists' budget.

[jonpaul]

Tektronix Circuits Concept books Oscilloscope Probe Circuits

See Tekwiki and W140

j

Jon

[bdunham7]

Can anyone explain why the blue trace looks so ugly, and if there is something I can do to fix this?

The 1X probe will have much higher input capacitance and much lower bandwidth--perhaps 150pF and 3-5MHz.  The 10X probe will have better specs and thus is more capable of picking up high frequency noise.  You can maybe 'fix' this by engaging the BW limit function and filtering out everything over 20MHz.  They still won't look quite the same, but that should make a difference.  You can also try using a much faster timebase, perhaps 1µs/div or less, so see what exactly that noise is. 

Also consider that the 10X probe setting at 10mV/div means that the scope is actually using an input scale of 1mV/div, which is probably its maximum sensitivity.  This means that the noise floor will be higher and the trace will be fatter, just not quite as fat an noisy as what you are seeing.  There's no easy fix for that and this is why scope front-end noise performance is so important.

[Fungus]

You can maybe 'fix' this by engaging the BW limit function and filtering out everything over 20MHz.

Also try turning on "High Res" mode.

[ttodorov]

Thank you all for providing your answers!

The BW limit on CH2 was enabled - I did not notice any difference. I don't know if I made it clear enough before, but the probes were clamped directly to the coax cable outputting the signal from the function generator (Rigol DG1022A). I used some cut-off resistor leads to make the signal and ground connections respectively. So there were no additional components and/or power supplies connected to the probes. The termination on the generator output was set to HighZ since I don't think that the cheap Rigol DS1054Z has 50 ohm input impedance matching.

I also turned off everything else around me that I could, even the lights. The only thing I did not think to turn off was my WiFi access point in the same room, so I have no idea if that can produce enough interference to be picked up by the 10x probe like that. Up to now I have dealt mostly with GPIO or PWM signals where 1x probe was always enough. This is the first time for me dealing with an amplifier circuit and thus I noticed the behavior only now...

[tautech]

Never discount it's a SMPS somewhere close by, maybe even in the next room as walls aren't made of lead and RF doesn't obey walls.
Keep switching things OFF until you find the culprit which then should be given flying lessons. 

[wasedadoc]

Never discount it's a SMPS somewhere close by, maybe even in the next room as walls aren't made of lead and RF doesn't obey walls.
Keep switching things OFF until you find the culprit which then should be given flying lessons. 

Even the SMPS in the 'scope.

[Wallace Gasiewicz]

Here is what I would do, you probably have done most of this.
Plug in both of your probes and let them lie on the bench, get traces with 1X and 10X settings.Connect the ground lead on probes to the tip and get readings on 1X and 10X. This makes a RF "sniffer".
Then put them near the SG with the SG ON and repeat.
I think you have already done this with the probes attached to the SG,Just to be complete:  are your probes matched to your scope input impedance.

This might help locating the point of interference.If you see interference with the probes not attached to the SG the interference is in the ether.
If you see interference with the probes near the turned on SG then you have a culprit.You can also move the probes near to the scope.

Ethernet cables can also produce RF. I have observed this at about 14.2 MHz.If you see interference off the air, with the probes used as a sniffer, you might wish to characterize this interference using the FFT.
I think you are using 1X-10X switched probes. They can have bad characteristics and IMHO, not usually as good as "straight" 10X or straight 1X probesI am far from an expert on probes, but I have had problems similar to this with probes that switch from 1X to 10X.

[2N3055]

What you see is difference in BW of probes mostly, with some little more noise of the scope...
But sharp peaks are really there. Enabling BW limit is hiding what crap PSU really sprays...

BW limit at 20MHz is used when you do measurements to compare with other PSU that are measured at 20MHz BW.

You also need to NOT use ground clip but a very short ground spring (or something of that sort).
That will make sure that your probe is not an antenna but really measures at certain point.

[sonpul]

Do not use clamps or long conductors to measure weak signals. You can put a special spring on the probe, perhaps it is included with the oscilloscope. If not, then you can do it.
Look at the last two waveforms. The second one is with a spring. The latter with a clamp and a grounding conductor with a crocodile. Generator with high output impedance.