OWON SDS7102 - Measured frequency beyond rating

[holozip]

Hi guys,

I'm very green when it comes to test and measurement equipment, and bought myself an entry level OWON SDS7102 to learn the ropes with. Whilst testing a Ham-It-Up converter, I tried measuring a 125Mhz crystal just for laughs, under the assumption the 100Mhz scope wouldn't manage to trigger.

To my surprise it managed it, and successfully measured the frequency pretty accurately. To get a reasonable trace I ended up setting the scope all the way down to a 2ns timebase, and only a 1k sample length

I'm struggling with the concept of how a scope with a 100Mhz front-end can accomplish this, am I misunderstanding what the 100Mhz limit really is, or do manufacturers build in some wiggle room?

[Rerouter]

the 100Mhz rating is essentially the -3db point, as the scope front end acts like a low pass filter, (essentially your 1V signal is filtered down to ~0.7V at 100Mhz) as you keep increasing frequency this voltage will drop further,

Now yes this is a digital scope, but equally the sampling stage cannot simple record at only 100Mhz or you would get horrible aliasing, e.g. a single point per waveform, so you would have no hope to tell dc from ac or sine from square, so in most cases they will sample at-least 3 times as fast as the bandwidth (3 points is about the minimum per waveform to figure out what its meant to look like,

Equally depending on the model the trigger / counter if implemented in hardware can work far far beyond the rated bandwidth of the scope, e.g. my old analog tektronics rated at 200Mhz will still trigger and display something up to almost 550Mhz,

[holozip]

Ah, I see - so whilst just getting the frequency of something like an oscillator is doable, if I were after accurate results for measuring the actual voltage/amplitude it wouldn't be so great given the filtering behaviour.

At the settings I mentioned in my previous post, I think the sample rate was 500MS/s, so around 4 points per waveform - even at that rate I noticed that the resulting waveform looked more like a sine wave than the nice clean square wave I'd have expected at lower frequencies.

Thank you for taking the time to explain - it's much appreciated and I feel a lot happier having half a clue as to whats going on

Kind regards,
Gareth

[ConKbot]

Ah, I see - so whilst just getting the frequency of something like an oscillator is doable, if I were after accurate results for measuring the actual voltage/amplitude it wouldn't be so great given the filtering behaviour.

At the settings I mentioned in my previous post, I think the sample rate was 500MS/s, so around 4 points per waveform - even at that rate I noticed that the resulting waveform looked more like a sine wave than the nice clean square wave I'd have expected at lower frequencies.

Thank you for taking the time to explain - it's much appreciated and I feel a lot happier having half a clue as to whats going on

Kind regards,
Gareth

Simple reason you only saw a sine wave. What is a square wave? A fundamental frequency and a whole bunch of odd harmonics packed together.  Whats the 3rd harmonic of your 125 MHz square wave signal? 375 MHz, wayyyy beyond what the scope can do, and at a point where the anti-aliasing filter(or just the frequency response of the front end if its not a specific AA filter)  has it attenuated into the noise floor.  So the only thing you capture is the fundamental component of the signal, or your 125 MHz sine wave. 




As you can see from that picture, the 3rd harmonic helps, but the rise time is pretty ambiguous, the 5th harmonic being added makes it a bit more square, so youre talking about over 625 MHz of bandwidth to capture a high fidelity square wave at 125 MHz. Got a GHz scope handy? :p

[holozip]

Thanks ConKbot,

This prompted an evenings worth of fun just playing with different frequencies and seeing how well they were represented by the scope. Turns out that much beyond 30MHz is almost indistinguishable as a square wave with the rudimentary stuff I had to hand to test with.

Given the 3rd harmonic would be 90MHz in this case, its now making perfect sense - and what I expected after digesting your post.

Setup:

• Raspberry Pi as a variable signal generator:
  http://blog.riyas.org/2014/01/raspberry-pi-as-simple-low-cost-rf-signal-generator-sweeper.html)
• OWON 7102 hooked up to GPIO pin 4
• 100MHz probe, using x10 attenuator mode

The interesting thing for me out of all this is learning that just because my scope says 100MHz, it doesn't means its capable/useful for measuring 100MHz square waves. Your post also prompted some further reading around harmonics - interesting and useful stuff.

It was also interesting to look at the attenuation when approaching the limit, it really is dramatic. I'd love to graph this characteristic for different frequencies to get a profile of the OWON.

If I can get the USB capturing working properly (something else to play with) then I'll try and post this.

Thanks again for your replies guys.

Kind regards,
Gareth

[marshallh]

Also please throw out your ground clip lead on your probe. Its basically useless for anything digital.
To probe anything past maybe 10mhz and actually see what's going on with the signal you must reduce the loop area of the connection (No 4 inch ground clips)
Otherwise it adds an additional filtering effect which may explain what you're seeing.

[holozip]

Thanks Marshallh, I'm really glad I asked here now instead of trying to just wing it; I would never have realised that ground loop inductance has such a detrimental effect.

Your reply prompted some more reading for me. I'll try and find a better way to probe / reduce the ground loop length and see if I can get cleaner traces without as much overshoot.

Thanks again,
Gareth

[Rick Law]

Ah, I see - so whilst just getting the frequency of something like an oscillator is doable, if I were after accurate results for measuring the actual voltage/amplitude it wouldn't be so great given the filtering behaviour.

At the settings I mentioned in my previous post, I think the sample rate was 500MS/s, so around 4 points per waveform - even at that rate I noticed that the resulting waveform looked more like a sine wave than the nice clean square wave I'd have expected at lower frequencies.

Thank you for taking the time to explain - it's much appreciated and I feel a lot happier having half a clue as to whats going on

Kind regards,
Gareth

Gareth, 2 points to note, (not to be confusing with data-points of sample)...

1. Sample rate may change according to the time/division.  Make sure you are using the right number instead of the max sample rate.

2. The method of interpolation is important when data-points per cycle is small.  Typically default is sin(x)/x.  So your scope pretends what it should be is a sin-wave and interpolate accordingly.  So you see a perfect sin wave.  If possible, switch interpolation to a line, that will give you some additional info such as where the data-points are.

Rick

[w2aew]

Hi guys,

I'm very green when it comes to test and measurement equipment, and bought myself an entry level OWON SDS7102 to learn the ropes with. Whilst testing a Ham-It-Up converter, I tried measuring a 125Mhz crystal just for laughs, under the assumption the 100Mhz scope wouldn't manage to trigger.

To my surprise it managed it, and successfully measured the frequency pretty accurately. To get a reasonable trace I ended up setting the scope all the way down to a 2ns timebase, and only a 1k sample length

I'm struggling with the concept of how a scope with a 100Mhz front-end can accomplish this, am I misunderstanding what the 100Mhz limit really is, or do manufacturers build in some wiggle room?

You may want to view these two videos I did on oscilloscope BW.  The first is using an analog scope:


And the second is showing a digital scope:

[rf-loop]

Hi guys,

I'm very green when it comes to test and measurement equipment, and bought myself an entry level OWON SDS7102 to learn the ropes with. Whilst testing a Ham-It-Up converter, I tried measuring a 125Mhz crystal just for laughs, under the assumption the 100Mhz scope wouldn't manage to trigger.

To my surprise it managed it, and successfully measured the frequency pretty accurately. To get a reasonable trace I ended up setting the scope all the way down to a 2ns timebase, and only a 1k sample length

I'm struggling with the concept of how a scope with a 100Mhz front-end can accomplish this, am I misunderstanding what the 100Mhz limit really is, or do manufacturers build in some wiggle room?

Owon SDS7102  is named as 100MHz. It is promised in specs that risetime (10% - 90%) is less or equal 3.5ns.
In practice, every SDS7102 what I have tested (and I have tested lot of) have -3dB point around 150MHz to 170MHz when used 50ohm cable and external 50ohm feed thru termiantion in scope input. (exept 2mV/div range what have highly reduced BW and there -3dB "niminal" point is around 20MHz. (in practice +/- 5MHz)
If use probe with 1x setting it need note also that probe bandwidth is very higly reduced and in practice if want even some level accuracy it is better to stay well under 5MHz.

With single channel in use and if selected 10M memory it kan keep 1GSa/s down to 0.5ms/div horizontal speed.

With FFT it can use up to 500MHz (note signal attenuation over 100MHz frequencies)

here can also find some tests:

http://owon.freeforums.org/some-tests-with-fft-t8.html

http://owon.freeforums.org/some-tests-with-new-sds7102v-version-2-6-and-2-8-2-t10.html

also here (partially same but something more. http://owon.freeforums.org/

This is main forum but due to some old issues with this forum there is "spare" forum owon.freeforums.org  but also without registration there can see images)

There is now lot of not updated information due to very busy times with some projects.

Example that today FW version Owon waveform update rate is much better than previously.

Also there can find samplerate tables for different Timebase (time/div) and memory settings.

----

Of course probe GND wire is too long for fast rising (or high frequency) signals.
It may be around 180nH.
(straight wire, 0.3mm diameter and 150mm lenght inductance is around 200nH
with 100MHz its reactance is 125ohm, and with 200MHz of course then 250ohm.
This is not resistive, it is inductive reactance and then there are affecting also capacitances... result is example "ringing" with fast edges)