Siglent SDS2000X Plus

[ozkarah]

While I was happy to get the higher-quality PP215s, they don't really offer any performance advantage over the PP510 other than not falling apart. 

Falling apart ? 
FYI there has been a change in Siglent PP510 and PP215 probes only visible that the later have 00 marked on them.
The only change I have noticed is the slightly different compensation range in ones marked 00.

I have the 00 marked ones.

[ozkarah]

@ozkarah

Possibly a thru-line 50 ohm terminator between the vna and the probe to bnc adapter? It may make a significant difference to the test results in this case.

@Vestom I just noticed the same lack in your first image. I think you both should repeat your tests with a thru-line 50 ohm terminator (or at least use a Tee adapter with a 50 ohm terminator) and compare the result against those of the original test.

[EDIT] Oops! Apologies, Vestom for not realising that you were effectively terminating the connection with CH1 set for DC50.

Johny, you are absolutely right. Results are changed a lot after implementing a setup like Vestom or using a 50 Ohm pass-through terminator.

But still PP215 rise time (2.44ns) and imaging is much worse than Digilent 100 MHz (a pair is 29 USD) (1ns). PP510 is image is similar to PP215 similar to bdunham7's findings but rise time is calculated much better (1.18ns).
In the screenshots below reference trace is 50 Ohms cable setup.










 

[ozkarah]

I think I found the reason for the high rise time calculation.

It might be the combination of my signal source (LiteVNA64) and the probe design.

There is an overshoot after a small undershoot in the signal (marked in the red circle). My speculation is it causes a slow and long-lasting overshoot in the probe (PP510 has a similar behavior). And scope takes the top of the overshoot as the maximum point during the 10-90% calculation. Which yields it to calculate a very long rise time.


The second image shows the reference points scope takes while making the rise time calculation.



In the third image I setup measurement gates to the beginning of rise and to the point of actual signal amplitude (to compensate for the effect of the overshoot). Then rise time is calculated around 1.10ns which is close to the rise times of the 50 Ohm cable and the Digilent probe.




Now results are close but still Digilent seems to have a better rise time and imaging capability (look at the previous post). 


Do you have any suggestions to test?
 
BTW, all probes tested are LF calibrated against the 1KHz test signal of the scope.

[bdunham7]

Now results are close but still Digilent seems to have a better rise time and imaging capability (look at the previous post).

I wouldn't get too concerned about differences in calculated rise time.  I know we all like seemingly objective data, but in this case the picture tells the story.

I don't know why the Digilent probes are so good here.  Can you measure the input capacitance?

Also, something to consider is circuit loading.  Instead of just a reference, you might try and connect like I did with no cable and just a tee, then display both the reference of the 50R scope input without any other probe attached, along with the 10X probe channel and a live 50R channel.  It is possible the Digilent probes are calibrated to display a good square wave at the cost of loading the circuit more.  Or, perhaps they are the bargain of the century.

Here is a Probemaster 4910-2 200MHz 100X <3.5pF probe connected this way, note that the reference and live 50R waveforms are very nearly identical.  The displayed rise time adn waveform through the probe might not be any more fantastic than the other probes, but the circuit loading is way better.  And this is for a very benign ~1V_p-p signal with an effective 25R source impedance, virtually ideal conditions.

[2N3055]

There must be something said about these probe verifications. In order for any characterizations to make sense, they have to be repeatable and comparable to other people's measurements. That is why standards are made.

Passive probes are tested for frequency response with leveled 50Ω generator by making frequency sweep in frequency range in question. Signal has to be sinewave. On output of generator good quality 50Ω pass trough terminator is connected and then passive probe. That makes source impedance 25Ω.

Pulse response is verified with pulse generator. Pulse generator must create signal that has fast edges, but it has to have top of the signal last long enough for all equipment connected to settle. Leo's pulser has 10MHz repetition rate that gives comfortable 50ns low and high for scopes 50-100 MHz or more.   For 10-20 MHz scopes it would be better to measure with signal that has 1 MHz repetition rate or slower. 
Output of pulse generator should also be loaded with 50Ω pass trough terminator (in line) and then passive probe.

Don't use any "T" connectors with cables going anywhere for any measurements. They create stubs in transmission line, and can influence (distort) the signal that probe sees.

That being said, there are many, many books written about humble passive probes. Input pin of the probe will present as a complex impedance (RLC) to signal node in DUT and will have it's own impedance peaks and lows and resonance (multiple actually with various Q). There will also be resultant complex impedance/resonances when probe internal impedance combines with a node that you are measuring...

That is why people give lot's of money for active probes to minimize parasitic properties of probe inputs..

In short, it is normal for one probe to have "better" behavior compared to other one with one source and worse than same other probe with different signal source.

If you expect that those measurements have merit to show "real life performance", then don't use coax adapter on the tip. Use the probe with a ground clip and see the horror then....

[ozkarah]

There must be something said about these probe verifications. In order for any characterizations to make sense, they have to be repeatable and comparable to other people's measurements. That is why standards are made.

Passive probes are tested for frequency response with leveled 50Ω generator by making frequency sweep in frequency range in question. Signal has to be sinewave. On output of generator good quality 50Ω pass trough terminator is connected and then passive probe. That makes source impedance 25Ω.

Pulse response is verified with pulse generator. Pulse generator must create signal that has fast edges, but it has to have top of the signal last long enough for all equipment connected to settle. Leo's pulser has 10MHz repetition rate that gives comfortable 50ns low and high for scopes 50-100 MHz or more.   For 10-20 MHz scopes it would be better to measure with signal that has 1 MHz repetition rate or slower. 
Output of pulse generator should also be loaded with 50Ω pass trough terminator (in line) and then passive probe.

Don't use any "T" connectors with cables going anywhere for any measurements. They create stubs in transmission line, and can influence (distort) the signal that probe sees.

That being said, there are many, many books written about humble passive probes. Input pin of the probe will present as a complex impedance (RLC) to signal node in DUT and will have it's own impedance peaks and lows and resonance (multiple actually with various Q). There will also be resultant complex impedance/resonances when probe internal impedance combines with a node that you are measuring...

That is why people give lot's of money for active probes to minimize parasitic properties of probe inputs..

In short, it is normal for one probe to have "better" behavior compared to other one with one source and worse than same other probe with different signal source.

If you expect that those measurements have merit to show "real life performance", then don't use coax adapter on the tip. Use the probe with a ground clip and see the horror then....

Thank you 2N3055 for the detailed explanations.
Unfortunately, I don't have a proper HF signal source. So I tried to use the LiteVna.

This time tried to follow your suggestions as much as I can.

I have an SDG2042X (upgraded to 120 Mhz). I used it as a signal source. Synchronized the outputs at 50 Ohm load setting. Connected scope's CH1(DC1M) to CH1 of AWG using pass-through 50 Ohm terminator at the output of AWG. CH4 (DC50) of the scope is directly connected to CH2 of the AWG.



Using 1Hz-120MHz sweep setting at 632 mVpp (~0dBm) I created a sweep signal. Analyzed the 3 probes again. The results are below:

 0-120_MHz_probe_comparison.png (91.81 kB. 1024x600 - viewed 237 times.)

[Martin72]

That is why standards are made.

"..Remember, only when everyone starts at the same time, a comparison of the times makes sense."

(J.M., german comedian)

[2N3055]

There must be something said about these probe verifications. In order for any characterizations to make sense, they have to be repeatable and comparable to other people's measurements. That is why standards are made.

Passive probes are tested for frequency response with leveled 50Ω generator by making frequency sweep in frequency range in question. Signal has to be sinewave. On output of generator good quality 50Ω pass trough terminator is connected and then passive probe. That makes source impedance 25Ω.

Pulse response is verified with pulse generator. Pulse generator must create signal that has fast edges, but it has to have top of the signal last long enough for all equipment connected to settle. Leo's pulser has 10MHz repetition rate that gives comfortable 50ns low and high for scopes 50-100 MHz or more.   For 10-20 MHz scopes it would be better to measure with signal that has 1 MHz repetition rate or slower. 
Output of pulse generator should also be loaded with 50Ω pass trough terminator (in line) and then passive probe.

Don't use any "T" connectors with cables going anywhere for any measurements. They create stubs in transmission line, and can influence (distort) the signal that probe sees.

That being said, there are many, many books written about humble passive probes. Input pin of the probe will present as a complex impedance (RLC) to signal node in DUT and will have it's own impedance peaks and lows and resonance (multiple actually with various Q). There will also be resultant complex impedance/resonances when probe internal impedance combines with a node that you are measuring...

That is why people give lot's of money for active probes to minimize parasitic properties of probe inputs..

In short, it is normal for one probe to have "better" behavior compared to other one with one source and worse than same other probe with different signal source.

If you expect that those measurements have merit to show "real life performance", then don't use coax adapter on the tip. Use the probe with a ground clip and see the horror then....

Thank you 2N3055 for the detailed explanations.
Unfortunately, I don't have a proper HF signal source. So I tried to use the LiteVna.

This time tried to follow your suggestions as much as I can.

I have an SDG2042X (upgraded to 120 Mhz). I used it as a signal source. Synchronized the outputs at 50 Ohm load setting. Connected scope's CH1(DC1M) to CH1 of AWG using pass-through 50 Ohm terminator at the output of AWG. CH4 (DC50) of the scope is directly connected to CH2 of the AWG.

(Attachment Link)

Using 1Hz-120MHz sweep setting at 632 mVpp (~0dBm) I created a sweep signal. Analyzed the 3 probes again. The results are below:

(Attachment Link)

That is nice work!
And now you can see that all probes tested have amplitude flatness inside 1 dB.  And that particular Digilent probe has very good flatness up to 100MHz at that source and load impedance...
But in reality they are all good enough, because signal as seen will be more influenced by source than anything else... As I said, try probes with a grounding clip...

[ozkarah]

There must be something said about these probe verifications. In order for any characterizations to make sense, they have to be repeatable and comparable to other people's measurements. That is why standards are made.

Passive probes are tested for frequency response with leveled 50Ω generator by making frequency sweep in frequency range in question. Signal has to be sinewave. On output of generator good quality 50Ω pass trough terminator is connected and then passive probe. That makes source impedance 25Ω.

Pulse response is verified with pulse generator. Pulse generator must create signal that has fast edges, but it has to have top of the signal last long enough for all equipment connected to settle. Leo's pulser has 10MHz repetition rate that gives comfortable 50ns low and high for scopes 50-100 MHz or more.   For 10-20 MHz scopes it would be better to measure with signal that has 1 MHz repetition rate or slower. 
Output of pulse generator should also be loaded with 50Ω pass trough terminator (in line) and then passive probe.

Don't use any "T" connectors with cables going anywhere for any measurements. They create stubs in transmission line, and can influence (distort) the signal that probe sees.

That being said, there are many, many books written about humble passive probes. Input pin of the probe will present as a complex impedance (RLC) to signal node in DUT and will have it's own impedance peaks and lows and resonance (multiple actually with various Q). There will also be resultant complex impedance/resonances when probe internal impedance combines with a node that you are measuring...

That is why people give lot's of money for active probes to minimize parasitic properties of probe inputs..

In short, it is normal for one probe to have "better" behavior compared to other one with one source and worse than same other probe with different signal source.

If you expect that those measurements have merit to show "real life performance", then don't use coax adapter on the tip. Use the probe with a ground clip and see the horror then....

Thank you 2N3055 for the detailed explanations.
Unfortunately, I don't have a proper HF signal source. So I tried to use the LiteVna.

This time tried to follow your suggestions as much as I can.

I have an SDG2042X (upgraded to 120 Mhz). I used it as a signal source. Synchronized the outputs at 50 Ohm load setting. Connected scope's CH1(DC1M) to CH1 of AWG using pass-through 50 Ohm terminator at the output of AWG. CH4 (DC50) of the scope is directly connected to CH2 of the AWG.

(Attachment Link)

Using 1Hz-120MHz sweep setting at 632 mVpp (~0dBm) I created a sweep signal. Analyzed the 3 probes again. The results are below:

(Attachment Link)

That is nice work!
And now you can see that all probes tested have amplitude flatness inside 1 dB.  And that particular Digilent probe has very good flatness up to 100MHz at that source and load impedance...
But in reality they are all good enough, because signal as seen will be more influenced by source than anything else... As I said, try probes with a grounding clip...

Thank you 2N3055. I am very surprised that a very cheap (25 USD/pair) 100MHz probe outperforms these PP215 200 MHz probes that came with the scope (which is supposed to have a good match). And, even a 100 MHz probe from the same brand (PP510) performs better.

Actually, my main concern is if there is any possibility that the probes are from a faulty batch or a kind of clone.

If someone here has the same setup I would love to see the results.

[bdunham7]

If someone here has the same setup I would love to see the results.

You did a lot of work for that nice picture, I took the easy way out and used noise w/ FFT.  Same AWG and same setup--50R terminator + BNC adapter on one channel, just a cable and the scope set to 50R on the other. 

The noise function on the SDG2042X isn't exactly flat, but it is easy to compare the deviation between the traces.  I threw in a few other random probes as well.

[Performa01]

I have an SDG2042X (upgraded to 120 Mhz). I used it as a signal source. Synchronized the outputs at 50 Ohm load setting. Connected scope's CH1(DC1M) to CH1 of AWG using pass-through 50 Ohm terminator at the output of AWG. CH4 (DC50) of the scope is directly connected to CH2 of the AWG.
...
Using 1Hz-120MHz sweep setting at 632 mVpp (~0dBm) I created a sweep signal. Analyzed the 3 probes again.

Nice measrements!

Here come some remarks:

Contrary to popular believe, probe bandwidth is nothing like e.g. scope bandwidth. There is no clear 3 dB corner frequency, where the probe output begins to fall at a constant rate. Essentially, probe bandwidth only means that the manufacturer is not willing to guarantee anything for frequencies above.

It has been demonstrated quite often by now, that even a humble 100 MHz probe can outperform a 500 MHz probe on certain instruments and in certain circumstances.

As has been pointed out already, the true performance characteristics in any practical setup will be determined by many parameters, of which the specified probe bandwidth is certainly the least important. The input capacitance is about the only obvious quality criterion – for practical use – not for the industry-standard test using a 25 ohms source.

What is important is the frequency compensation. All probes have a compensation trimmer for the LF (low frequency) compensation. This is for adjusting for best pulse edge fidelity with the internal 1 kHz square wave calibration signal. But this only affects the probe gain at higher frequencies without changing the frequency response. In other words, if you ignore everything below about 1 MHz, then the low frequency compensation is nothing but a gain control.

A passive probe is a rather complex design, and it needs more compensation measures to work properly at high frequencies. This is because of the input impedance of the DSO, which is certainly much more complex than just 1 megohm in parallel with 17 pF. So you can have either probes that are optimized for the oscilloscope model or the universal ones, that provide two or even three trimmers for optimizing the frequency response on different instruments.

Your measurements show that the HF-compensation happens to be just perfect for the combination of SDS2000X+ and the Diligent probe, while it is rather mediocre for the low end Siglent probes – up to the point where the rise time measurements are seriously impacted.

The reason for this would be that the PP510 and PP215 have been designed for the SDS1000X-E series DSOs, where they both perform pretty good. See reply #56 in the thread linked below:

https://www.eevblog.com/forum/testgear/siglent-sds1104x-e-in-depth-review/msg1434665/#msg1434665

You can also see from my tests that the performance of the PP510 and PP215 is within about 0.5 dB up to 100 MHz and pretty much identical above that. Even though my tests have been carried out with early samples of these probes, I can’t believe that performance has changed significantly since then. Maybe you could repeat your test, but instead of using the 1 kHz square wave for calibrating the probes, you could try to use a 100 MHz sine wave and adjust both probes for 0 dBm response in your scenario. After that, the results should be pretty much identical for both probe models.

So the PP510 and PP215 probes have been optimized for the SDS1000X-E series and are certainly not the preferred solution for higher class oscilloscopes like the SDS2000X Plus. Yet in practical terms this hardly matters because … see next paragraph.

All this is just an academic discussion, because in any real-world scenario a passive high impedance probe like the ones discussed here isn’t going to be very useful for high frequencies anyway. Input impedances in the realm of 10 pF mean that any probing above 100 MHz requires a source impedance below 160 ohms, just to keep the measurement error below 50%. Once you’re at source impedances that low, a passive low impedance probe (which essentially is just a resistive match to a 50 ohms coax line) usually is the much better solution.

[Vestom]

All this is just an academic discussion, because in any real-world scenario a passive high impedance probe like the ones discussed here isn’t going to be very useful for high frequencies anyway. Input impedances in the realm of 10 pF mean that any probing above 100 MHz requires a source impedance below 160 ohms, just to keep the measurement error below 50%. Once you’re at source impedances that low, a passive low impedance probe (which essentially is just a resistive match to a 50 ohms coax line) usually is the much better solution.

This! And when probing digital signals the bandwidth of the edges is easily beyond 100MHz and the signal will be distorted by the capacitance of your passive probe.

Once had to find a troublesome glitch on a 16 MHz clock signal (AC logic), which disappeared when attaching the venerable P6139A 8pF 500MHz probe. Had to use a 50 Ohm BNC cable with a resistor in series at the tip to see it (remember to put the scope in 50 Ohm mode). That was before we got the hugely expensive active differential probes good for several GHz 

I can highly recommend making your own cheap Lo-Z probe, if you need to see signals beyond 100 MHz. I should probably myself make some small PCBs with a 2.5k in series, a 51R in parallel and SMA connector for cheap high speed x100 Lo-Z probes...

[bdunham7]

Once had to find a troublesome glitch on a 16 MHz clock signal (AC logic), which disappeared when attaching the venerable P6139A 8pF 500MHz probe. Had to use a 50 Ohm BNC cable with a resistor in series at the tip to see it (remember to put the scope in 50 Ohm mode). That was before we got the hugely expensive active differential probes good for several GHz 

I can highly recommend making your own cheap Lo-Z probe, if you need to see signals beyond 100 MHz. I should probably myself make some small PCBs with a 2.5k in series, a 51R in parallel and SMA connector for cheap high speed x100 Lo-Z probes...

I've had issues like that with as low as ~4 MHz clocks.  In addition to homemade, keep an eye out for the <1pF Tektronix P6156 (and the rare 20X attenuator) and the older ProbeMaster 100X 4910-2 which are <3pF.

[Johnny B Good]

@Vestom,

 There's no need to add a 50 ohm resistor across the co-ax at the Lo-Z probe end if it's terminated in a matching impedance at the scope end. You'll be halving your already attenuated signal voltage for no real benefit otherwise. A 50 ohm transmission line,  terminated with a purely resistive 50 ohm load will look just like a 50 ohm resistor

 Matching impedances only really matter when joining lengths of transmission line or something like a passive filter designed to function with a specific impedance load or source. In this case, the impedance of your source can be anything you care to choose (a 2.5KR resistor in your case  ), including Lo-Z voltage sources and Hi-Z constant current sources.

 You can simply use whatever attenuating series resistor is needed in this case - the tricky bit is dealing with keeping the probe's ground connection as short as possible to minimise 'ground bounce' and the effective area of the ground loop minimised to reduce stray noise ingress.

[Vestom]

There's no need to add a 50 ohm resistor across the co-ax at the Lo-Z probe end if it's terminated in a matching impedance at the scope end. You'll be halving your already attenuated signal voltage for no real benefit otherwise. A 50 ohm transmission line,  terminated with a purely resistive 50 ohm load will look just like a 50 ohm resistor

In theory you are absolutely right. However, the scope is not a perfect 50 ohm termination but is typically slightly capacitive (specified to 17pF for SD2kX+) yielding a slight mismatch and a small reflection, which the source termination will dampen. Previously, I have just used a leaded resistor soldered to the end of the cable (without source termination), but did then also see some artifacts, which could be replicated in spice. But I guess, experimentation will show, whether a small PCB and a source termination provides any real benefit.  At least using a small PCB as a solder-in probe tip connection would actually be nice.

the tricky bit is dealing with keeping the probe's ground connection as short as possible to minimise 'ground bounce' and the effective area of the ground loop minimised to reduce stray noise ingress.

Oh, so true! This is why it is important to consider placement of test points when doing PCB layout - or having the luxury of RF connectors like MMCX or UFL for test points.

[Performa01]

However, the scope is not a perfect 50 ohm termination but is typically slightly capacitive (specified to 17pF for SD2kX+) yielding a slight mismatch and a small reflection, ...

Thankfully not true. If the input capacitance of 17 pF would still be present even in 50 ohms mode, it could hardly be useful at frequencies above 100 MHz.

This is also the reason why an external terminator  on a high-Z scope input can never replace the genuine 50 ohms input mode.

It is still true that the 50 ohms input mode is not perfect. Yet the VSWR stays below 1:1.5 at the specified bandwidth (500 MHz in case of the SDS2000X Plus).

[gf]

It is still true that the 50 ohms input mode is not perfect. Yet the VSWR stays below 1:1.5 at the specified bandwidth (500 MHz in case of the SDS2000X Plus).

Is that "good enough" for a Z0 probe without additional 50 Ohm termination of the cable's source end?

[bdunham7]

Is that "good enough" for a Z0 probe without additional 50 Ohm termination of the cable's source end?

It is if you have reasonable expectations...

[Martin72]

Today I prepared the scopes for sending away.
I´ve decided to let them make a self-cal before packaging.
So I got two scopes on the desk, both on.
All three are new, but the last arrived is "newer"...

Left scope:

Software 1.3.9R6
U-Boot 5.0
FPGA 2021-07-16
CPLD 03
Hardware 02-05

Right scope:

Software 1.3.9R10
U-Boot 5.3
FPGA 2022-02-08
CPLD 05
Hardware 04-05

Nothing really surprising...except one thing.
The self-cal routine takes siginificant longer on the newer one.
I´ve didn´t start them at the same time, maybe 15...20seconds delay, but it takes much longer than the 20s until the newer one was finished.

[Peter_O]

My SDS2104X plus, bought January 2021 does not start any more.

When pressing the power button at the front, the button lights up, but no fan, no screen, no further button lights.
Same after disconnecting all connection or trying different power cord, which are double checked and known good anyway.
Pressing lit button for some seconds makes it to unlight again.

I found the hint and pressed "Math" repeatedly during power on, but with no effect.
https://siglentna.com/operating-tip/oscilloscope-hardware-reset/

Yesterday it run as usual, normal power off and shut down.

Any ideas?

[bdunham7]

It should still be under warranty, so I wouldn't muck around with it too much.

[Peter_O]

Sure. Bought at a listed dealer.

[Vestom]

Thankfully not true. If the input capacitance of 17 pF would still be present even in 50 ohms mode, it could hardly be useful at frequencies above 100 MHz.

This is also the reason why an external terminator  on a high-Z scope input can never replace the genuine 50 ohms input mode.

It is still true that the 50 ohms input mode is not perfect. Yet the VSWR stays below 1:1.5 at the specified bandwidth (500 MHz in case of the SDS2000X Plus).

Hmm. Yes, that must be true, 17 pF is rather high and only about 94 Ohms at 100 MHz . I wonder what tricks they do in the front end to reduce the capacitance and improve the match in 50 Ohms mode? 

I also looked up the ~20 year old Tek TDS3054, which is specified for 13 pF, and is bit disappointed that a 15-20 year newer scope has such a high capacitance 
(Though, it is mostly an issue when not using compensated probes...)

[Martin72]

Lecroy WR9054 17pf, Tek MSO4 13pf, Keysight 3000 14pf....Nothing uncommon.

[Vestom]

RTB2000: 9pF - it can be done 
(But that scope does not have a built in 50 Ohm selector, which is arguably more important...)