Author Topic: Comparison of oscilloscope probes PVP2350 vs PVP3150 vs LF312 vs HV150  (Read 3350 times)

0 Members and 1 Guest are viewing this topic.

Offline CosteCTopic starter

  • Regular Contributor
  • *
  • Posts: 206
  • Country: pl
Hello guys

I made comparison between
Rigol PVP2350 (350 MHz, 1:10, 1 ns, 10 pF)
Rigol PVP3150 (150 MHz, 1:10, 2.3 ns, 10 pF)
TESTEC LF312 (150 MHz, 1:10, 2,3 ns, 15.5 nF)
TESTEC HV150 (1.5 kV, 1:100, 300 MHz, 1.2 ns, 4 pF)
All connected to the same avalanche pulse generator, 50 Ohm output, with probe-to-BNC connector. Compensated. Scope used Rigol MSO5354.
Results are confusing to me: seems fastest PVP2350 has ugliest, slowest response! HV150 has slightly higher amplitude, but this can easily be attributed to lower capacitance.

Can anybody connect PVP2350 to some fast signal source, and post result?

 

Offline nctnico

  • Super Contributor
  • ***
  • Posts: 28429
  • Country: nl
    • NCT Developments
You need to test probes from a 25 Ohm source. So with a terminator at the probe tip.
There are small lies, big lies and then there is what is on the screen of your oscilloscope.
 

Offline CosteCTopic starter

  • Regular Contributor
  • *
  • Posts: 206
  • Country: pl
You need to test probes from a 25 Ohm source. So with a terminator at the probe tip.
Why 25 Ohm? Probes are Hi-Z by definition and probing 50 Ohm source make lot of sense. Probe GND is via BNC adaptor, it can't be done much better. All probes are tested the same way. Issue is why higher bandwidth PVP2350 looks worse and slower than PVP3150?
 

Online tautech

  • Super Contributor
  • ***
  • Posts: 29809
  • Country: nz
  • Taupaki Technologies Ltd. Siglent Distributor NZ.
    • Taupaki Technologies Ltd.
You need to test probes from a 25 Ohm source. So with a terminator at the probe tip.
Why 25 Ohm?
Industry standard.  ;)

Then the only way to compare probes is to save one as a Reference waveform to compare the other against as the moment you introduce another the results change.
Don't believe me, try it and see.  ;)
Avid Rabid Hobbyist.
Some stuff seen @ Siglent HQ cannot be shared.
 

Offline CosteCTopic starter

  • Regular Contributor
  • *
  • Posts: 206
  • Country: pl
Then the only way to compare probes is to save one as a Reference waveform to compare the other against as the moment you introduce another the results change.
Don't believe me, try it and see.  ;)
No need. Differences are obvious. I could connect all 4 to generator, yet capacivtie loading would be much higher and rising edges less steep. I wonder if it is my scope or my probes and how even frequency response of PVP2350 is.

Any home method of checking it? No 350 MHz generator around, however I will make comparison to 3 GHz scope later, but only with 500 MHz probe.
 

Offline David Hess

  • Super Contributor
  • ***
  • Posts: 17427
  • Country: us
  • DavidH
Why 25 Ohm? Probes are Hi-Z by definition and probing 50 Ohm source make lot of sense.

Because probes are specified with a 25 ohm source impedance from a parallel terminated output.

Quote
All probes are tested the same way. Issue is why higher bandwidth PVP2350 looks worse and slower than PVP3150?

What does the leading edge really look like?  The time/div setting is too slow to see and I doubt the veracity of the automatic measurement.

The pulse shape is questionable.  What does it look like measured directly without a probe, and with a proper termination?
 
The following users thanked this post: CosteC

Online tautech

  • Super Contributor
  • ***
  • Posts: 29809
  • Country: nz
  • Taupaki Technologies Ltd. Siglent Distributor NZ.
    • Taupaki Technologies Ltd.
Then the only way to compare probes is to save one as a Reference waveform to compare the other against as the moment you introduce another the results change.
Don't believe me, try it and see.  ;)
No need. Differences are obvious. I could connect all 4 to generator, yet capacivtie loading would be much higher and rising edges less steep. I wonder if it is my scope or my probes and how even frequency response of PVP2350 is.
Maybe I was not clear, you can only test one probe at a time.

Test one and save its waveform as a reference waveform to compare the others against.

Quote
Any home method of checking it? No 350 MHz generator around, however I will make comparison to 3 GHz scope later, but only with 500 MHz probe.
You are looking for 2 things, BW and Step response and they require different tests.

For BW tests a 50 Ohm output RF gen terminated into 50 Ohms provides a 25 Ohm source.
For step response something like this, also terminated into 50 Ohms provides a 10 MHz step very suitable for tests.
http://www.leobodnar.com/shop/index.php?main_page=product_info&cPath=124&products_id=295

But, try as you might connecting 2 probes gives different results to just one probe hence capturing one and saving it to be used as the reference is necessary to compare another against.

You can see some of my tests here and in later posts:
https://www.eevblog.com/forum/testgear/playing-with-probes-guessing-game/msg4736159/#msg4736159
Avid Rabid Hobbyist.
Some stuff seen @ Siglent HQ cannot be shared.
 

Offline nctnico

  • Super Contributor
  • ***
  • Posts: 28429
  • Country: nl
    • NCT Developments
Why 25 Ohm? Probes are Hi-Z by definition and probing 50 Ohm source make lot of sense.

Because probes are specified with a 25 ohm source impedance from a parallel terminated output.

Quote
All probes are tested the same way. Issue is why higher bandwidth PVP2350 looks worse and slower than PVP3150?

What does the leading edge really look like?  The time/div setting is too slow to see and I doubt the veracity of the automatic measurement.

The pulse shape is questionable.  What does it look like measured directly without a probe, and with a proper termination?
It is probably a good idea to have a 10dB attenuator between the pulse generator and the probe + terminator so the impedance at the probe tip is 50 Ohm and reflections from the impedance mismatches of cable / pulse generator output are attenuated. Sometimes a faster edge is not better.
There are small lies, big lies and then there is what is on the screen of your oscilloscope.
 

Offline CosteCTopic starter

  • Regular Contributor
  • *
  • Posts: 206
  • Country: pl
Why 25 Ohm? Probes are Hi-Z by definition and probing 50 Ohm source make lot of sense.

Because probes are specified with a 25 ohm source impedance from a parallel terminated output.
Do you have source maybe? I would like to educate myself and replicate it if possible.

My source is avalanche generator, with 50 Ohm resistor on opposite side of BNC on picture, so maybe 2 cm from probe tip. Around 100 ps one way. I would need specific terminator to reduce this 2 cm significantly. Especially probe internal wiring is probably longer.

In mean time I will stretch edges to show them better, point about 10 ns/div is valid, despite I am mostly concerned about distorted top of impulse.
« Last Edit: May 23, 2023, 02:33:12 pm by CosteC »
 

Offline David Hess

  • Super Contributor
  • ***
  • Posts: 17427
  • Country: us
  • DavidH
It is probably a good idea to have a 10dB attenuator between the pulse generator and the probe + terminator so the impedance at the probe tip is 50 Ohm and reflections from the impedance mismatches of cable / pulse generator output are attenuated. Sometimes a faster edge is not better.

Including an attenuator will not hurt except for signal level, however if the pulse generator is properly source terminated then there will be no reflection.  In this case the load termination at the probe is not required, however leaving it out makes for a 50 ohm rather than 25 ohm source and high impedance probes are specified using the later.

 

Offline CosteCTopic starter

  • Regular Contributor
  • *
  • Posts: 206
  • Country: pl
Signal generator is avalanche transistor discharging some RG174 into 50 Ohm just behind BNC connector. As short as it can be.

'Terminating' generator does not make much sense (no line to terminate to speak of) but I made this experiment too. Amplitude dropped as expected.

Direct comparison to oscilloscope input: MSO5000 has 17 pF input + short RG174 (~40 cm) + terminator at scope input, influences trace a bit - becomes ringy.

4 probes connected simultaneously show why de-scew exists :) But each probe is different and cable lengths of probes are different, there is good 15 cm difference longest to shortest. I havent got 4 BNC adaptors unfortunatly.

After this exercise I am less concerned - PVP2350 is less smooth in response, seems more ringing while this ringing is not visible using HV150, which is 1.5 kV 300 MHz, 1:100 probe - shall see ringing invisible for 150 MHz probes.
LF312 is cheap, 150 MHz and it is somewhat visible.

1 ns/DIV overlapping each probe response is revealing differences, but was hard to catch as generator does not have stable output, which obviously is weakness of test with this generator type.

I need to make more meaningful tests. On advised 1ns/div comparison the difference is less visible, but still exists at top edge where probes give significantly different response. Any ideas how to do it?

Idea 1 - generator with more stable output like fast logic gate.
 

Online tautech

  • Super Contributor
  • ***
  • Posts: 29809
  • Country: nz
  • Taupaki Technologies Ltd. Siglent Distributor NZ.
    • Taupaki Technologies Ltd.
4 probes connected simultaneously show why de-scew exists :) But each probe is different and cable lengths of probes are different, there is good 15 cm difference longest to shortest. I havent got 4 BNC adaptors unfortunatly.
Different probes offer different propagation delays, irregardless of cable length.

Deskew is essential for accurate computed results like in Power Analysis mode where it's common to use passive, differential and current probes and each of their propagation delays need be aligned for accurate results.

Most manufacturers offer a Deskew device for connecting 2 or more probes to adjust channel skew to a minimum.
Avid Rabid Hobbyist.
Some stuff seen @ Siglent HQ cannot be shared.
 

Offline gogoman

  • Regular Contributor
  • *
  • Posts: 123
Why 25 Ohm? Probes are Hi-Z by definition and probing 50 Ohm source make lot of sense.

Because probes are specified with a 25 ohm source impedance from a parallel terminated output.

hello, what is a parallel terminated output?  :-//
 
thanks
 

Offline CosteCTopic starter

  • Regular Contributor
  • *
  • Posts: 206
  • Country: pl
hello, what is a parallel terminated output?  :-//
https://www.ti.com/lit/an/snla034b/snla034b.pdf Page 6. They have better pictures than I could make :D
It is one of methods of getting rid of reflections in long lines.

In my case there is no need, as there is no long line - maybe 2 cm. I added termination to go from 50 Ohm source to 25 Ohm source, as well as reduce amplitude and allow comparison with "bare, 1:1" oscilloscope input.
Still I need better source of reference signal - avalanche generator is too unstable.

 

Online 2N3055

  • Super Contributor
  • ***
  • Posts: 7463
  • Country: hr
What is purpose of this  exercise?  What do you try to accomplish?

Scope BW is measured by using leveled sinewave generator. Directly to 50 Ω input on scope. On scope input that is high impedance (1MΩ), an 50 Ω passthrough terminator is used to keep signal generator properly loaded. Same when measuring with a probe as a system.

For a risetime measurement you need pulse source that have (very!) flat top and pulse should be at least 10-20x wider than risetime. Same type of termination.
Using super fast edges won't help, it will only incite more resonances in probes... it is more important for an edge to be clean, linear and monotonic...
It only has to be reasonably fast. Risetimes of source and scope add as RMS so it is easy to calculate scope risetime if using known source... As an experiment you can try and use a much longer coax used as a storage transmission line on your pulser to make pulse wider.

There were several topics here where VERY detailed info, explanations etc was shared about theory and practice of passive probes.  And it was shown that (nominaly) low BW probes aren't necessarily low BW. If you have low enough source impedance..

Predominant influence to measurement by passive probes is their load capacitance. That is, they won't act as lowpass filter internally (so much) and not pass through high frequencies as much as they will "bog down" (load) signal source (the basically form an RC filter from source impedance and its shunting capacitance) and shunt signal to the ground... That is why your two probes that have same load capacitance will have similar performance...

Passive probes BW spec is more of a "use it up to this frequency" recommendation. In practice with high impedance sources, passive probes are not very "dependable to show the right thing" above some 150 MHz anyways because of the loading.... Despite saying 500 MHz on a probe..

Higher BW probes also will usually have two additional compensation pots..

Procedures:
https://www.euramet.org/Media/docs/Publications/calguides/EURAMET_cg-7__v_1.0_Calibration_of_Oscilloscopes.pdf
 

summary:

1. Modern scopes in general will have brickwall AA input filters and will not have perfect pulse response even with source directly connected to BNC on front.
2. In order to have comparable results for discussion a standard procedure has to be followed: in your case 50Ω  source parallel terminated with 50Ω terminator to scope input and probe input. Only one at a time.
3. BW cannot be determined with pulse source (risetime measurements) you need sinewave generator with 50 Ω source impedance and precise levels (or a way to independently measure calibrate level).
4. Using pulse source, a BW can be roughly estimated (roughly) but pulse source has to have certain characteristics (and you need to know your scope AA filter coefficient. It won't be 0.35 for your scope. And it will change with how many channels are on). Pulse source is used to calibrate levels and pulse response. Clean edges and flat top and pulse width at least order of magnitude wider than risetimes.
5. All of that is good and nice if search of knowledge. In practice, if you need to look at 150MHz + you need to start to think in terms that your probing is part of circuit.. A 10x passive 50 Ω probe made from 450 Ω resistor and a piece of coax will do better job and load circuit much less at 200 Mhz than a 10pf 500 MHz 10MΩ passive probe...
6. Which brings us to the point that in that case you should have bought a scope that has 50Ω inputs if these kinds of signals and measurement is what you need. I personally would not buy a 200MHZ+ (less than 1.5 ns risteime) scope without 50Ω inputs...
"Just hard work is not enough - it must be applied sensibly."
Dr. Richard W. Hamming
 
The following users thanked this post: nctnico, luudee

Offline CosteCTopic starter

  • Regular Contributor
  • *
  • Posts: 206
  • Country: pl
What is purpose of this  exercise?  What do you try to accomplish?

For a risetime measurement you need pulse source that have (very!) flat top and pulse should be at least 10-20x wider than risetime. Same type of termination.

Modern scopes in general will have brickwall AA input filters and will not have perfect pulse response even with source directly connected to BNC on front.
2. In order to have comparable results for discussion a standard procedure has to be followed: in your case 50Ω  source parallel terminated with 50Ω terminator to scope input and probe input. Only one at a time.
3. BW cannot be determined with pulse source (risetime measurements) you need sinewave generator with 50 Ω source impedance and precise levels (or a way to independently measure calibrate level).
4. Using pulse source, a BW can be roughly estimated (roughly) but pulse source has to have certain characteristics (and you need to know your scope AA filter coefficient. It won't be 0.35 for your scope. And it will change with how many channels are on). Pulse source is used to calibrate levels and pulse response. Clean edges and flat top and pulse width at least order of magnitude wider than risetimes.
5. All of that is good and nice if search of knowledge. In practice, if you need to look at 150MHz + you need to start to think in terms that your probing is part of circuit.. A 10x passive 50 Ω probe made from 450 Ω resistor and a piece of coax will do better job and load circuit much less at 200 Mhz than a 10pf 500 MHz 10MΩ passive probe...
6. Which brings us to the point that in that case you should have bought a scope that has 50Ω inputs if these kinds of signals and measurement is what you need. I personally would not buy a 200MHZ+ (less than 1.5 ns risteime) scope without 50Ω inputs...
Thank you @2N3055!

My purpose is to understand why PVP2350 seems to have worse response than 'slower' PVP3150 . Maybe it is better, but I do not understand something...
Probe is mandatory for most measurements I do, this is why I need to understand and trust my probes.
Either it is digital signal I cannot load with 50 Ohm probe, or power signal where voltage is too high. While, as you say, it is impossible to not impact measured circuit, I would like to minimise loading and impact.
And yes, I know perfectly well how low impedance 1:10 probe has at 100 MHz. For this reason I got PVP3150 over older RP2200. 10 pF is better than 17 pF

Scope 50 Ohm input itself is not very useful for me... I know and use all trickery of 1:10 50 Ohm probes, if needed. Bare 50 Ohm

Anyway, I need new pulse generator, one with flat, rectangular top, even when loaded with ~10-20 pF. I will test it with some better scope. Hope 2.5 GHz will be enough.
 
The following users thanked this post: 2N3055

Online 2N3055

  • Super Contributor
  • ***
  • Posts: 7463
  • Country: hr
What is purpose of this  exercise?  What do you try to accomplish?

For a risetime measurement you need pulse source that have (very!) flat top and pulse should be at least 10-20x wider than risetime. Same type of termination.

Modern scopes in general will have brickwall AA input filters and will not have perfect pulse response even with source directly connected to BNC on front.
2. In order to have comparable results for discussion a standard procedure has to be followed: in your case 50Ω  source parallel terminated with 50Ω terminator to scope input and probe input. Only one at a time.
3. BW cannot be determined with pulse source (risetime measurements) you need sinewave generator with 50 Ω source impedance and precise levels (or a way to independently measure calibrate level).
4. Using pulse source, a BW can be roughly estimated (roughly) but pulse source has to have certain characteristics (and you need to know your scope AA filter coefficient. It won't be 0.35 for your scope. And it will change with how many channels are on). Pulse source is used to calibrate levels and pulse response. Clean edges and flat top and pulse width at least order of magnitude wider than risetimes.
5. All of that is good and nice if search of knowledge. In practice, if you need to look at 150MHz + you need to start to think in terms that your probing is part of circuit.. A 10x passive 50 Ω probe made from 450 Ω resistor and a piece of coax will do better job and load circuit much less at 200 Mhz than a 10pf 500 MHz 10MΩ passive probe...
6. Which brings us to the point that in that case you should have bought a scope that has 50Ω inputs if these kinds of signals and measurement is what you need. I personally would not buy a 200MHZ+ (less than 1.5 ns risteime) scope without 50Ω inputs...
Thank you @2N3055!

My purpose is to understand why PVP2350 seems to have worse response than 'slower' PVP3150 . Maybe it is better, but I do not understand something...
Probe is mandatory for most measurements I do, this is why I need to understand and trust my probes.
Either it is digital signal I cannot load with 50 Ohm probe, or power signal where voltage is too high. While, as you say, it is impossible to not impact measured circuit, I would like to minimise loading and impact.
And yes, I know perfectly well how low impedance 1:10 probe has at 100 MHz. For this reason I got PVP3150 over older RP2200. 10 pF is better than 17 pF

Scope 50 Ohm input itself is not very useful for me... I know and use all trickery of 1:10 50 Ohm probes, if needed. Bare 50 Ohm

Anyway, I need new pulse generator, one with flat, rectangular top, even when loaded with ~10-20 pF. I will test it with some better scope. Hope 2.5 GHz will be enough.

Thank you for the answer. I wasn't trying to be patronizing just to make a summary of all relevant facts at one place.

As an experiment you could try replacing transmission line discharge coax on your avalanche gen with a longer one... Pulse width is defined by pulse propagation time in coax..

A passive 50Ω 10x probe will have DC load of 500 Ω but order of magnitude lower loading at higher frequencies.. Most modern low voltage logic will happily drive it .... As I said, it is not hard, try it. You need 450 Ω resistor and a piece of coax. Terminate at scope side with 50Ω pass through terminator. A noninductive trough hole 0.25W resistor works well enough for experiment... You can even try 950Ω for 20x probe with even lower loading... For looking at logic levels a bit of noise is not that important as pulse fidelity...
"Just hard work is not enough - it must be applied sensibly."
Dr. Richard W. Hamming
 

Offline David Hess

  • Super Contributor
  • ***
  • Posts: 17427
  • Country: us
  • DavidH
4. Using pulse source, a BW can be roughly estimated (roughly) but pulse source has to have certain characteristics (and you need to know your scope AA filter coefficient. It won't be 0.35 for your scope. And it will change with how many channels are on). ...

Could I get a list of DSOs, or DSO manufacturers who do that so that I can avoid them?  Having the bandwidth and transient response change with the number of channels in operation sounds like fun.
 

Online tautech

  • Super Contributor
  • ***
  • Posts: 29809
  • Country: nz
  • Taupaki Technologies Ltd. Siglent Distributor NZ.
    • Taupaki Technologies Ltd.
4. Using pulse source, a BW can be roughly estimated (roughly) but pulse source has to have certain characteristics (and you need to know your scope AA filter coefficient. It won't be 0.35 for your scope. And it will change with how many channels are on). ...

Could I get a list of DSOs, or DSO manufacturers who do that so that I can avoid them?  Having the bandwidth and transient response change with the number of channels in operation sounds like fun.
Typically any DSO that shares the 1 ADC, those with dual ADC's are better providing the user has the smarts to use the 2 to best effect and best is with an ADC for each channel.
Shared memory also has an impact with less sample points and hence a greater reliance on accurate interpolation.

Truth is most are affected in some manner however most modern DSO's plainly display sampling rates and memory use for each timebase setting so informing the user so they can use that info to best advantage.
Avid Rabid Hobbyist.
Some stuff seen @ Siglent HQ cannot be shared.
 

Offline CosteCTopic starter

  • Regular Contributor
  • *
  • Posts: 206
  • Country: pl
Thank you for the answer. I wasn't trying to be patronizing just to make a summary of all relevant facts at one place.
No problems.
As an experiment you could try replacing transmission line discharge coax on your avalanche gen with a longer one... Pulse width is defined by pulse propagation time in coax..
There is 10-11 ns of coax inside now. You suggest longer delay will flatten top and improve "rectangularity" of rising edge top?
A passive 50Ω 10x probe will have DC load of 500 Ω but order of magnitude lower loading at higher frequencies.. Most modern low voltage logic will happily drive it .... As I said, it is not hard, try it. You need 450 Ω resistor and a piece of coax. Terminate at scope side with 50Ω pass through terminator. A noninductive trough hole 0.25W resistor works well enough for experiment... You can even try 950Ω for 20x probe with even lower loading... For looking at logic levels a bit of noise is not that important as pulse fidelity...
I use them for logic sometimes, particularly in test systems where 50 Ω mux is available but not 1 MΩ mux :)

In this case I want to understand PVP2350 behaviour. Looking for nice rectangular pulse generator :) maybe there is something of-the-shelf.

Could I get a list of DSOs, or DSO manufacturers who do that so that I can avoid them?  Having the bandwidth and transient response change with the number of channels in operation sounds like fun.
All? Even LeCroy 7300 has two two-channel ADCs. Either 4x10 GSps either 2x20 GSps (CH1, CH3). This is very common architecture. You just need to know what you are doing. I do not think BW or response changes really - sampling changes so you can loose details. However some details are byproduct of AA filters, when your signal is too close to them.

Here is one of my doubts - maybe PVP2350 is better but "distortion" I see is AA filter, as pulse is fast, contains components around 350 MHz or even higher. Maybe PVP3150 smooth those HF components out.

If you will get ancient Tektronix MSO2024B you will get 4x1 GSps always, yet Rigol MSO 5350 devastates this ancient tektronix with 8/4/2 GSps and 200 times larger memory and ton of other features, which better display is most visible (pun intended). Tek has only lower noise...
 

Offline nctnico

  • Super Contributor
  • ***
  • Posts: 28429
  • Country: nl
    • NCT Developments
In this case I want to understand PVP2350 behaviour. Looking for nice rectangular pulse generator :) maybe there is something of-the-shelf.
Try the calibrator output on your oscilloscope. It is intended to adjust probes with and many can be used as pulse generators as well (for TDR measurements for example).
There are small lies, big lies and then there is what is on the screen of your oscilloscope.
 

Offline CosteCTopic starter

  • Regular Contributor
  • *
  • Posts: 206
  • Country: pl
In this case I want to understand PVP2350 behaviour. Looking for nice rectangular pulse generator :) maybe there is something of-the-shelf.
Try the calibrator output on your oscilloscope. It is intended to adjust probes with and many can be used as pulse generators as well (for TDR measurements for example).
Och no... Rigol MSO5000 probe calibrator has 3.2 us rise time  :-DD DS1054Z has 3.1 us.
Build in generator offers 10.5 ns rise time and "rectangle" is not very rectangular at 15 MHz.

Some scopes like LeCroy 7300 has option of 1 kHz or 1 MHz rectangle generation for probe compensation - this would possibly work, but not with 1 k$ class of equipment. Some generators have 10 MHz clock output - yet I doubt it will be very rectangular, rather super low jitter.
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf