EEVblog Electronics Community Forum
Products => Test Equipment => Topic started by: entropi on March 08, 2021, 11:24:01 am
-
I have a Micsig STO1104C. I'm curious about how 1KHz reference signals are typically generated by scopes and what phenomena lead to their higher frequency components.
I've posted some traces of the compensation reference signal below. The 10X probes were compensated normally with a 1V trigger and full square wave on screen prior. I'm deliberately triggering above 1V to uncover higher frequency signal elements.
The notch within the cursors on the attached trace is what led me to look into this.
-
Not much counts at that massive Y amplitude, try at 0.5V / Div. to stay within the Y input's dynamic range.
-
Also the trigger level should be at the center of the signal. Not at the top of the edge.
-
Here it is triggered at 1V.
-
Here’s a zoomed view with a 1V trigger.
-
Here triggering just below the high frequency component within the X cursors.
What’s the likely source of this component? I’m trying to understand more the way oscilloscope reference signals are generated.
-
My scope came with four Micsig P130A 200MHz 10X probes.
After setting up the screen in the last screenshot I went back and compensated all of them again.
I found that the first probe was the worst of them, and I only got one good one.
Two of them had issues where as you pull out the trim tool the compensation would bounce back up. The first one was by far the worst with this.
All measurements were taken with a warmed up scope at approximately STP (standard temperature and pressure) in a relatively low EMI lab environment (nothing else on within several meters of the scope).
Here’s a shot of the best probe stacked with one of the worst:
-
Now using the best of the four probes I’ve gone back to looking at the high frequency component near the top of the rise. I’m now particularly interested in what might cause the variability shown within the X cursors shown here (all shots have been color graded with 10 seconds persistence). What’s a likely source of this?
-
The probe compensation output can only give a guide to step response and shouldn't be considered as the ultimate standard for such a test where every channel can be adjusted minutely differently.
The variability of the overshoot brings another factor again into your investigations as the trigger or compensation signal jitter could be the cause.
-
Now using the best of the four probes I’ve gone back to looking at the high frequency component near the top of the rise. I’m now particularly interested in what might cause the variability shown within the X cursors shown here (all shots have been color graded with 10 seconds persistence). What’s a likely source of this?
I'd start by getting the entire square wave on screen. It is possible you are overdriving the input which can cause signal distortions.
-
Get the 1kHz's top and bottom on the screen at the same time. :-/O
-
I'd start by getting the entire square wave on screen. It is possible you are overdriving the input which can cause signal distortions.
The same behavior is apparent but not as well defined when I do that.
-
That last one is getting towards normal, the next higher Y sensitivity should still fit on the height of the screen, and move the trigger level to near the middle of the waveforms amplitude.
The 1kHz It looks good enough to compensate X10 probes to me.
-
That last one is getting towards normal, the next higher Y sensitivity should still fit on the height of the screen, and move the trigger level to near the middle of the waveforms amplitude.
The 1kHz It looks good enough to compensate X10 probes to me.
Yep and for a sanity check is its only other use.
As mentioned before it's not suitable for examination of step response.
-
That last one is getting towards normal, the next higher Y sensitivity should still fit on the height of the screen, and move the trigger level to near the middle of the waveforms amplitude.
The 1kHz It looks good enough to compensate X10 probes to me.
Yep and for a sanity check is its only other use.
As mentioned before it's not suitable for examination of step response.
This.
It is important you understand the purpose for which this signal is provided - and that the quality of the signal should only be expected to fulfill that purpose. Probe compensation is not particularly demanding and sanity checks are, by their very nature, ballpark tests.
Do not expect the test signal on a scope to be low jitter, high accuracy and suitable for time dilation experiments or interplanetary navigation.
Unless there is a fault, it will be as good as it needs to be.
-
This.
It is important you understand the purpose for which this signal is provided - and that the quality of the signal should only be expected to fulfill that purpose. Probe compensation is not particularly demanding and sanity checks are, by their very nature, ballpark tests.
Do not expect the test signal on a scope to be low jitter, high accuracy and suitable for time dilation experiments or interplanetary navigation.
Unless there is a fault, it will be as good as it needs to be.
Yes, well aware it serves its purpose. I'm more curious about how they're typically generated on scopes as a design consideration and how analyzing one can lead to other insights about signals analysis..
-
This.
It is important you understand the purpose for which this signal is provided - and that the quality of the signal should only be expected to fulfill that purpose. Probe compensation is not particularly demanding and sanity checks are, by their very nature, ballpark tests.
Do not expect the test signal on a scope to be low jitter, high accuracy and suitable for time dilation experiments or interplanetary navigation.
Unless there is a fault, it will be as good as it needs to be.
Yes, well aware it serves its purpose. I'm more curious about how they're typically generated on scopes as a design consideration and how analyzing one can lead to other insights about signals analysis..
Keep the whole signal on screen and just press trigger position button to set trigger level for you.
You need to put triggering level at the point of the signal that is well defined. That is some part of clean rising edge. More vertical the better.
If you put it on the top like that, the scope cannot trigger cleanly because there is noise on the top so trigger point is not very well defined.
Just look at it, it is quite logical that if you set it to be unstable that it will be..
There are many good beginners books on Internet on usage of scope, including explaining what trigger is..
-
Yes I’m interested in that noise and weirdness near the top. That’s what I’m asking about.
I set my trigger levels normally and compensated my probes normally before starting this line of questioning and posting this thread. I didn’t think people would get so caught up on thinking I’m asking about probe compensation fundamentals.
-
If you want to take a close look at a signal you'll need to do some math. The key is to make sure the input amplifier isn't overdriven otherwise you'll get distortions in the signal.
First you need to know the amplitude of the signal. You are likely using a 1:10 probe so the oscilloscope gets 1/10th of the signal's amplitude. Next you need to know the offset ranges of the oscilloscope for various V/div settings. From this you can calculate the minimum and maximum voltages per V/div setting. Now you have a list with amplitude / offset ranges versus V/div setting.
As a last step you need to take the amplitude of the signal and choose a V/div setting that the full amplitude of the signal fits in half the offset range. This means that when you move the signal down so the top is (vertically) in the middle of the screen the input amplifier is not overdriven and thus you won't get any distortion.
This may be hard to understand without pictures; if you Google for 'oscilloscope vertical offset' you can find more information.
-
Yes I’m interested in that noise and weirdness near the top. That’s what I’m asking about.
If there is so much gain in the analogue front end that any part of the front end saturates, then there are no guarantees as to how the front end will behave until it is fully out of saturation. That's why you are recommended to keep all of the waveform visible on the screen.
The standard technology that avoids such problems is analogue sampling, often implemented as diode rings. The principle is that the parts that can saturate are only connected to the input when they are guaranteed not to be saturated.
You may benefit from understanding the contents of a couple of Jim William's classic application notes...
Jim Williams: LT AN120 “1ppm Settling Time Measurement for a Monolithic 18-Bit DAC” http://cds.linear.com/docs/en/application-note/an120f.pdf (http://cds.linear.com/docs/en/application-note/an120f.pdf)
Jim Williams: LT AN47 “High Speed Amplifier Techniques” http://cds.linear.com/docs/en/application-note/an47fa.pdf (http://cds.linear.com/docs/en/application-note/an47fa.pdf)
-
Another bit of weirdness. I walked away for a while with a probe still hooked up to the scope signal. I came back to find this (attached). It looks to me like low frequency loss and phase shift like the wave is going through a differentiator.
A bug on the scope? Still more curious how scopes generally generate these waves.
Edit: Resetting scope didn’t fix this. Turned out the probe failed! And I thought that was the good one. Not having good luck with the Micsig probes.
-
Looks like you lost a direct connection and have capacitive coupling somewhere along the line.
Check everything is firmly seated.
-
Thanks. Turned out the probe was the culprit.
-
If you want to take a close look at a signal you'll need to do some math. The key is to make sure the input amplifier isn't overdriven otherwise you'll get distortions in the signal.
First you need to know the amplitude of the signal. You are likely using a 1:10 probe so the oscilloscope gets 1/10th of the signal's amplitude. Next you need to know the offset ranges of the oscilloscope for various V/div settings. From this you can calculate the minimum and maximum voltages per V/div setting. Now you have a list with amplitude / offset ranges versus V/div setting.
As a last step you need to take the amplitude of the signal and choose a V/div setting that the full amplitude of the signal fits in half the offset range. This means that when you move the signal down so the top is (vertically) in the middle of the screen the input amplifier is not overdriven and thus you won't get any distortion.
This may be hard to understand without pictures; if you Google for 'oscilloscope vertical offset' you can find more information.
Some good resources:
https://www.picotech.com/library/application-note/using-analog-offset-to-maximize-oscilloscope-resolution (https://www.picotech.com/library/application-note/using-analog-offset-to-maximize-oscilloscope-resolution)
https://www.youtube.com/watch?v=-eccL4oLVM4 (https://www.youtube.com/watch?v=-eccL4oLVM4)
-
If there is so much gain in the analogue front end that any part of the front end saturates, then there are no guarantees as to how the front end will behave until it is fully out of saturation. That's why you are recommended to keep all of the waveform visible on the screen.
The standard technology that avoids such problems is analogue sampling, often implemented as diode rings. The principle is that the parts that can saturate are only connected to the input when they are guaranteed not to be saturated.
You may benefit from understanding the contents of a couple of Jim William's classic application notes...
Jim Williams: LT AN120 “1ppm Settling Time Measurement for a Monolithic 18-Bit DAC” http://cds.linear.com/docs/en/application-note/an120f.pdf (http://cds.linear.com/docs/en/application-note/an120f.pdf)
Jim Williams: LT AN47 “High Speed Amplifier Techniques” http://cds.linear.com/docs/en/application-note/an47fa.pdf (http://cds.linear.com/docs/en/application-note/an47fa.pdf)
Thanks! Those are exactly the sort of thing I was hoping for.
-
Here's some of the 1kHz on the no knobs version, difficult to see how much clreaner they are.
Channel 1: https://ibb.co/dMAG7v
Channel 2: https://ibb.co/bPtEua
Channel 3: https://ibb.co/fQPyLF
Channel 4: https://ibb.co/b4pXfF