TDS3054, TDS5054 & MSO4104, Pulse Voltage Reading Comparison

[daveyk]

Interesting Comparison.  All three scopes are ISO17025 Calibrated/Certified.  The same custom 100:1 scope probe is being used with all three scopes.  All three scopes were on four >4 hours.

My "Golden Standard" is the TDS3054 with a TDS3AAM Advanced statistics module.

Please see the attachment. 

I am most disappointed in my MSO4104.  Its voltage readings are ~2%-~2.8% lower than either the TDS3054 or the TDS5054.  When I use these scopes, my recorded readings of the pulses should be within 5% of the previous year's readings, and within a +/-10% of "ideal" reading as published by the equipment manufacturer.  The MSO4104 (my favorite new baby) is already starting out with almost 3% differences from the MFG "golden standard" TDS3000 series that their spec is based on.

I had a friend (recently passed) that swore scopes were just funny money reading voltages and pulse widths. No two would read the same.  Heck you get different readings between channels.  He was a huge fan of his 1GHz BW Leckroy.   He swore by it, but still complained that we could not get the same readings on the same instrument.  He said the MFG was wrong to set the standard using a Tek TDS3000 series. Anyway, I wish he were still with us to see this study I did this afternoon.  We could talk for hours on this alone.   

Anyway, I though I would share it with everyone here as an FYI.

[2N3055]

That is normal.
Did you read datasheets for these scopes?
Also did you check attenuation of probes?
For really critical work they have to be calibrated and verified before measurements..

There are scopes with tighter vertical accuracy specs (for instance SDS2000X HD has 0.5% vertical accuracy).

But I agree that it is important to discuss and raise awareness of this.
Nice measurements.

[daveyk]

Thank you.  What is looks like is a baseline offset on the MSO4104 on the 58.0v range.  I ran the compensation calibration routine (took 20 minutes, made no difference).   On the MSO4104, it matched the TDS5054 if I use P-P rather than MIN).  Hmmmm... the MSO4104 has an SPC error on some vertical ranges.

[daveyk]

"Also did you check attenuation of probes?"

Same probe used on all three scopes.  It is a custom 100:1 probe.  The front end (probe end) is basically a 40dB attenuator feeding a 1 meter coax with a 50 ohm terminator on the Scope's end. 

Like the probe or not, it is used by the MFG and it is buildable consistent.   Get 10 Tek Probes, all compensated properly, and all ten will read different pulse voltages and different PW readings.

The MSO4104 has a different way to tell it what type of probe attached. Instead of selecting 100:1, you tell it's attenuation.  I think there is also a different way too.  I need to look at that and see if it effects the baseline offset.

[2N3055]

I don't doubt you do nice work. What I wanted to convey is fact that if you take a very nice (expensive)  Tek probes datasheets, they don't even specify amplitude accuracy there.
I wouldn't be surprised that your custom made probe has more accurate attenuation.
As for word "calibration" I used it in classic metrology way: " verification of accuracy by measuring well know quantity".  i.e. checking scope against accurate voltage source and see what it says.
Both those scopes have best accuracy spec of 1.5-2% and as I said for Tek probes I couldn't find that spec but should not be much more than that.
But in general scopes are no better than 3% all things accounted. Unless you use one of those scopes (like one I mentioned) that have stated accuracy better than that. My 2 12bit Siglents are pretty much in range of nice 4.5digit DMM as far as resolution and accuracy (both do much better than spec as it seems).
My 16Bit Picoscope is even better (that one has guaranteed 0.25% DC accuracy and does better than 0.06%. That Picoscope is main reason that I never bought dedicated thermal RMS meter because it does better job than most. And with those two 12bit scopes I get true RMS meter up to 1GHz BW...

But I digress. Your numbers for accuracy are actually decent. Not perfect but well in spec. Better than that, you need to check if those scopes of yours will let you type in a custom attenuation factor of the probe. With that you can fine tune amplitude accuracy by measuring reference source, and if measurement is done soon after setup in thermally stable environment you can do much better. And after measurement you check reference source again. If nothing moved, you can safely presume your measurement is much more accurate than spec.

[tautech]

There's a little project for you Sinisa, demonstrate the new probe check feature and how it improves system accuracy. 

[daveyk]

"But I digress. Your numbers for accuracy are actually decent. Not perfect but well in spec. Better than that, you need to check if those scopes of yours will let you type in a custom attenuation factor of the probe. With that you can fine tune amplitude accuracy by measuring reference source, and if measurement is done soon after setup in thermally stable environment you can do much better. And after measurement you check reference source again. If nothing moved, you can safely presume your measurement is much more accurate than spec."

I was wrong about the MSO4104, you tell it that it has a 100X probe.  It's the TDS5054 that you can do the custom attenuation.  That is a complex scope to use and program.  It is not straight forward and easy to use like most digital scopes, but it does have a lot of versatility. Measuring positive offshoot with it is also rather finicky.  You can not use the same settings as the other two scopes to get an accurate reading.  If fact, I have yet to figure out how to match the TDS3000 series on offshoot with either the TDS5054 or MSO4104.  The TDS3000s always seemed to read too high of positive offshoot.  Oh well...

With the MSO4104, I ran the SPC several more times.  It doesn't get any better than what you see in the PDF file.  Looking at a -296v signal, it has a +6volt offset, which doesn't show up as much of all on the display.  If I read the P-P, it does match the other two scope fairly well.  With those relatively clean pulses that works.  Get one with ringing on it and that doesn't work.  Read MAX and add that to the MIN reading and you go it just like the PP.  Again, I am guessing that will also have an issue with a ringy signal.

Besides the slight offset, it is still a nice scope to use and easy to use for trouble-shooting or calibrating.  Plus it's big beautiful display!!  I got it for a good price on ShopGoodwill - lol

[2N3055]

There's a little project for you Sinisa, demonstrate the new probe check feature and how it improves system accuracy. 

Thanks for mentioning . SDS2000X HD I have here has automatic probe check function that checks both probe compensation and real probe attenuation. If probe is outside a spec for a bit, it will adjust.
Problem is that my probes are doing such a good job "unfortunately" so you couldn't see scope adjust anything.
But here is a stable measurement of 3V DC from my AOIP voltage calibrator, where I found a probe that was a "bit out" (0.4%) and was adjusted to 9.96:1 and is stable showing 3.004 with 0.13% error.
0.4% was already in spec (that is for scope only 0.5%) and was already excellent for probe/scope combined errors.
I have few probes that at bog standard 10X atten settings show less than 0.1% DC error....
Also few Testec probes (very inexpensive good probes) are all better than 0.2% combined with scope..

So scope and probes are doing very good job.

[daveyk]

Siglent does have some nice stuff.  That probe check/calibration feature seems really innovative and cool.  With my experience with Siglent scopes (somewhat low end SDS1204E-E) was that the measurement versatility was not there.  I literally begged them for the ability in the user interface to be able to set the mid-reference points for pulse width measurement and really never received any response other than from the local rep saying that he entered the suggestions. I was also told to read in all the waveform data, and calculate the PW based on the 10% down (90%) points.  <Sigh> At that time, I couldn't figure out how to read that data.  I am sure I could do it now, but I am not revisiting that.   I would loved to have bought a higher end Siglent, but I had no trust the I could do that on the higher end models since it seemed like a "new" and "interesting" idea to them.  Both Agilent and Tek had that capability going back to their earliest digital scopes.  I had sent a question asking that about their higher end scopes and never received a response from them.  So, I went with old models of Tek Scopes.  Another reason is that all my industry specs seems based around the Tek TDS3000 series. Siglent scopes are as sexy as they come though.

Looking at the baseline offset of the MSO4104 makes me wonder if a calibration house that knows how to actually calibrate (tweak) the scope (maybe Tektronix themselves) knows how to calibrate the baseline offset calibrator, could fix that offset, or if that is normal for the MSO4000 series?  I get ISO17025 accredited calibrations and data on my equipment, but for $150, I think TMI more, or less, is sticker changers.  Their data books are great and comply with ISO17025, but rather than perfecting the instrument, it is good if it falls in-between X&Y.  I have a work around for it now in my utility software, but being a little OCD, it's hard to ignore - lol.

[Kleinstein]

It is not just the accuracy of the scope, but also on how accurate the 50 Ohm termination is and how the proble compensation is set. This could especially be a problem with a scope that needs an external termination.

[jonpaul]

suggest you repeat this process with fast rise cal gen. Leo Bodnar 40 ps and TEK PG506 or equiv.

Use no probe, just 50 ohm précision coax direct gen>>>scope 50 ohm

That eliminates the unknown probes and unknow generators from the test.

Jon

[daveyk]

Thank you.  It is the same probe and "Pulser" used with each scope.  The problem with the MSO4104 is the baseline compensation calibration routine does not work as well in the higher amplifier voltage ranges
.  Whether that is normal with the MSO, or an issue with his particular one, I do not know. 

PP reading is within .5% of the TDS3054 and TSD5054 MIN reading. 

This baseline offset may be why the MSO4104 always shows 0% overshoot, the way it is with the very clean pulser I am using now.  I also am not sure if the method I was taught, by the MFG, to measure that overshoot with the TDS3000 series is correct or not.  The procedure is to change the sweep to 1ms and acquire to envelope. This pulse, btw, has a PRF of about 750Hz.   You do that with the TDS3054, or TDS5054 and you can read 3%, but it also doesn't seem accurate; almost taking a reading while anti-aliasing.  You do that to the MSO4104, it still shows a perfect 0% overshoot (with it what you see on-screen).    So, I am thinking, of the three scopes, the MSO4104 is the only one correctly reading the over-shoot - lol.