Siglent SDS1104X-E and SDS1204X-E Mixed Signal Oscilloscopes

[tautech]

SPI decoding works in  the latest firmware (.20) fine. It was a nice surprise that it is possible to use zoom for quick data examination of the decoded signal..............

Increasing the Transparency setting in the Display sub menu will result in not hiding traces under the Decode list.   

[Performa01]

Hi,
Somehow I thought that 1204X-E has 2 totally independent ADCs. But if one uses 3 channels (1,2,3)  then  the channel 3 will also use 500 GSa/s and 7 Mpts max although only one channel is used.

My guess would be there will be not many DSOs that use different sample rates and memory depths across their analog channels. So the Siglent scopes are in no way special in this regard.

While technically possible, mixed sample rates / record lengths would increase the already very high complexity of the firmware while providing no true benefit in practical use. We need not even look under the hood, as it starts at the UI level already. Have you noticed the information at the upper right corner of the screen?


SDS1104X-E_Screen_Sa-Rate

The current record length and sample rate is displayed there. If this would not be the same for all channels, we’d need this bit of information to be individual for each channel, which would be both space consuming and probably confusing as well. And all that for one single (rare) use case, because:

With just one channel, we can only have one record length and sample rate.
With any two channels, we’ll always have the same sample rate and record length for both, no matter whether they are within the same channel group or not.
With four channels, we’ll always get half the sample rate / record length of a single channel.

So it is really only the 3-channel scenario that would allow a mix of individual/interleaved ADC configuration.

Just look at it as a scope that only allows 1, 2 or 4 channels and gives you the option to hide one of the 4 channels when it is not needed

[borjam]

Hi,
Somehow I thought that 1204X-E has 2 totally independent ADCs.

Oh but it does !
See #1 post in this thread.

I think it would be rather confusing to have two different sampling rates on screen at the same time. After all you have a single time base for the four channels. So, no matter how much work the second converter is actually doing, it makes sense to use the same sample rate for all the signals on screen.

In this case they have done The Right Thing with the POLA™ approach (Principle of Least Surprise).

[rf-loop]

Also we need remember always bacround working waveform history buffer.
Also we need remember fast sequence mode.
These can be big hassle if diffferent channel have different amount of memory (and different max count of segments or wfms in normal mode history buffer.

Only acceptable solution is same amount of sample data length with all captured channels. Other way we may get lot of confusion and ranting how this work.
So, with 2 ADC it is 1 or 2 channel in use max 1GSa/s max 14M/channel and over 2 channel in use max 500MSa/s and max 7M/channel.

[tubularnut]

Now the signal is dropping out at 1V and 100mV settings on .20 firmware, so permanently downgrading to .12 to see how stable it is.

[17_29bis]

Now the signal is dropping out at 1V and 100mV settings on .20 firmware, so permanently downgrading to .12 to see how stable it is.

I connected all 4 inputs to the signal generator and went through the range 5mV - 50V for each channel - the signals did not drop.

Firmware 7.6.1.20
Hardware 00-03
FPGA version: 2018-01-20

[rf-loop]

It is certain without any doubt that FW7.6.1.20 have error. With some settings and use some channel trace may disappear depending V/div setting including also that it may occur some times only with some fine steps. I have not found systematic way to get this error visible but still frequently I can get this error. (If trace is visible and do not any changes it stay visible and do not disappear randomly itself so it is not this kind of random.) It is not at all depending signal. Mostly it exist after turning V/div but what is other things combination and previously changed settings - I have not foud perfect repeatable logic.

I have not seen this error any time in perevious public shared FW what was 7.6.1.12.

Siglent vacation time is 13.-22-Feb. So there is not ony reaction from Siglent about this error. But there is waiting notifications about this when they arrive to work after "New Year" holiday.

If you have any problem with this issue (is it even possible that this error do not exist in all product lots), please downgrade back to previous FW and use it until this problem is solved. Allways upraded or downgraded FW, do least one selfcal after it is enough thermal equilibrium,  after >30 minutes continuously on.

Note:
after on, cold or warm, it do frequently some internal  self adjustment procedures until it have been on enough time on  for approximately thermal equilibrium (you can see this "hick-up" on the screen trace updating very short breaks if look carefully after start) . During this period do not selfcal, it is perhaps not useful. So, take seriously this: Do selfcal after least 30minute continuously without any breakon and nothing connected to input BNC's.

(pity, there is so many nice things in new FW and it have this kind of error)

[17_29bis]

I wonder what criteria was used to to choose deskew time limit for 1204X-E. Right now it is +/- 100ns which seems to be quite small. 

I was playing with a modified UNI-T 210E (added an oscilloscope output, the mod that has been done by many who do not have access to expensive current probes, i know that  UNI-T 210E has very limited bandwidth (around 3 kHz)  but for my current applications/experiments  that's ok) and noticed that even if I set one channel deskew to -100ns and another channel deskew to +100ns I still could not get two signals (voltage/current) match. It would probably have worked if the deskew range had been be bigger and then I could use the math  function for  power measurements.

[17_29bis]

I cannot believe that the included probes PP215 200MHz are what the label says due to their poor performance. I have a very old probe (around 27 years old, all labels are long gone due to extensive usage,  I used it when I was an E.E.),   I  also have a couple of P6100  100MHz probes I bought at AliExpress for a different oscope. Today I was probing some signals on 8 bit MCU using the standard probes and did not like the signal shape. So, I decided to compare all those probes and that is what I got (I was mostly paying attention to the rise/fall times)

All probes have been properly calibrated and used in 1:1 mode.

a) 27 year old probe - 20 ns  - yellow trace;
b) P6100  100MHz Chinese probe - 26 ns - blue trace;
c) P215 200Mhz - 33 ns - green trace;

The result speaks for itself.

[tautech]

The result speaks for itself.

They do.

Inexperience !
1x probe usage on a circuit sensitive to capacitive loading when 10x setting need be used.

Get the pamphlet/datasheet for each probe and examine input capacitance spec for 1x and 10x....then result will speak for itself.

Little traps like this await the unwary. 

[Performa01]

I cannot believe that the included probes PP215 200MHz are what the label says due to their poor performance. I have a very old probe (around 27 years old, all labels are long gone due to extensive usage,  I used it when I was an E.E.),   I  also have a couple of P6100  100MHz probes I bought at AliExpress for a different oscope. Today I was probing some signals on 8 bit MCU using the standard probes and did not like the signal shape. So, I decided to compare all those probes and that is what I got (I was mostly paying attention to the rise/fall times)

All probes have been properly calibrated and used in 1:1 mode.

Sorry to hear you’re unhappy with the PP215, because they are more than adequate in my book.

Probes performance is specified for x10 mode and this is what manufacturers focus on. Higher end scopes usually come with x10 probes exclusively, this even applies to the SDS2kX already.

Probe x1 mode generally is an emergency solution for tasks, where we want highest possible sensitivity whereas bandwidth and capacitive load are irrelevant. So this mostly comes down to probing power supply rails for ripple and noise. It is absolutely not recommended to probe digital buses in x1 mode, because the high capacitive load will affect rise time and propagation delay of the signals there as well. In analog circuits, carefree use of x1 probes can severely affect circuit performance and even lead to unwanted oscillations.

The data sheet for PP215 clearly states 6MHz bandwidth and up to 120pF input capacitance in x1 mode.
The actual bandwidth is much better than that, >9MHz (measured), which results in a rise time of <38ns (calculated). Yes, especially older probes tend to have more bandwidth up to 16MHz in x1 mode and there were (and still are) spezialized x1 probes that can go up to some 30MHz, but that does not change the fundamental problems of x1 probing as mentioned in the previous paragraph.

Here’s the frequency response graph for the PP510, which has identical specs to the PP215 in x1 mode.


SDS1104X-E_Probe_x1_BW

With the high input sensitivity of 500µV/div for the SDS1000X-E it should almost never be necessary to use x1 probes anyway.

Btw, Here are the specifications for the Keysight N2889A 350MHz Probe (much more expensive, you bet!):

•  Bandwidth: DC to 350MHz (@10:1), DC to 10MHz (@1:1)
•  Risetime: 1 nsec (@10:1), 35 nsec (@1:1) (10%-90%)
•  Attenuation ratio: 10:1/1:1
•  Input resistance: 10 M? (@10:1), 1 M? (@1:1 when terminated to 1M?)
•  Input capacitance: 11 pF(@10:1), 60 pF (@1:1)
•  Maximum input: 300 VRMS (or >400Vpk) CAT I/II (@10:1), 150 V RMS CAT I/II (@1:1)
•  Cable length: 1.3 m

[TK]

It clearly is a loading problem.  In the 4th image where you have all 3 probes connected at the same time, all 3 traces look very similar to me.

[17_29bis]

TK, tautech & Performa01 in particular (I really appreciate your detailed explanation but I do hope you copied/pasted the part of/all the text instead of typing it)  thanks for answering. But I am still believe that enclosed probes (PP215) are poor performers comparing to my other probes (even one having lower bandwidth) in this test. I repeated it again this time using all (properly calibrated) probes in 1:10 mode.

The previous result (1:1):

a) 27 year old probe - 20 ns  - yellow trace;
b) PP6100  100MHz Chinese probe - 26 ns - blue trace;
c) PP215 200Mhz - 33 ns - green trace;

The second test (1:10):

a) 27 year old probe - 10 ns;
b) PP6100  100MHz Chinese probe - 10 ns;
c) PP215 200Mhz - 22 ns;

The result still speaks for itself and  PP215  will be the last probe I choose from those 3 probes. And no, I did not say that PP215  unusable, but I am saying that in this particular test PP6100 100MHz  probe outperformed it.


 

[rf-loop]

TK, tautech & Performa01 in particular (I really appreciate your detailed explanation but I do hope you copied/pasted the part of/all the text instead of typing it)  thanks for answering. But I am still believe that enclosed probes (PP215) are poor performers comparing to my other probes (even one having lower bandwidth) in this test. I repeated it again this time using all (properly calibrated) probes in 1:10 mode.

The previous result (1:1):

a) 27 year old probe - 20 ns  - yellow trace;
b) PP6100  100MHz Chinese probe - 26 ns - blue trace;
c) PP215 200Mhz - 33 ns - green trace;

The second test (1:10):

a) 27 year old probe - 10 ns;
b) PP6100  100MHz Chinese probe - 10 ns;
c) PP215 200Mhz - 22 ns;

The result still speaks for itself and  PP215  will be the last probe I choose from those 3 probes. And no, I did not say that PP215  unusable, but I am saying that in this particular test PP6100 100MHz  probe outperformed it.

c) PP215 200Mhz - 22 ns;

This result is really strange! Something is now wrong. Probe or use of probe.
Also PP6100 100M what I found is around same ballpark if look 10:1 input capacitance. So these load your signal around same.
It feels very strange if you really measure normal digital bus what are still quite low impedance typically or your probe is broken. Difference between PP6100 can not be so high, when look rise time from what ever source if all is ok when we talk this kind of very slow risetimes.

Typically probe test (characterizing probe)  need do using 50ohm output impedance terminated with 50ohm feed thru and probe tip connected to this feed thru termination using probe tip adapter for avoid GND inductance.

Can you show exactly your test setup (enough good picture) and with exatly with this setup, and then also exatly with this setup image from scope screen and explained all settings.

Do you have any reliable signal source what have known pulse rise time and this source impedance, example some known function generator)

[17_29bis]

Just wanted to add a couple of things to Performa01's review of web server, the page 156, SDS1104X-E Review 07.pdf
This is very basic piece of software, indeed. Something is definitely wrong when it comes to using 1:10 probes.
For example:
From your browser connect to the oscope, press "Refresh" button to read the latest data from the device. In my case this is:



Don't change anything either on the oscope or in the web interface, instead  just press  "Apply" and look at the scope / the screen - V-Scale has changed from 2V to 20V for all channels, i.e. was multiplied by 10



Another annoyance is that after you press "Apply" the channel adjustment type automatically changes from coarse to fine. So in order to quickly change vertical scale of all channels back  (on the oscope) you would have to set the adjustment type to coarse and after that  use the vertical scale knob.

[Performa01]

The second test (1:10):

a) 27 year old probe - 10 ns;
b) PP6100  100MHz Chinese probe - 10 ns;
c) PP215 200Mhz - 22 ns;

The result still speaks for itself and  PP215  will be the last probe I choose from those 3 probes. And no, I did not say that PP215  unusable, but I am saying that in this particular test PP6100 100MHz  probe outperformed it.

You are right that the PP215 are certainly not the best probes money can buy; just look at the input capacitance, which is specified 16–20pF. Compare that to the Keysight N2889A specifications, where it is just 11pF – at 5 times the price, that is. If Siglent is going to ship the DSO with probes of that quality level and the price of the 4-channel model increases by $400,- because of that, most folks will be very unhappy and in fact, essentially no one would really appreciate that. The bells and whistles is what sells, not inconspicuous quality probes. The original probes are cheap and will serve most folks well and the ones who really need or want something better can always invest some serious money to buy whatever they want whenever they want.

It’s not that easy though and price is not the ultimate performance indicator, as the probe should be a good match for the rather complex input impedance of the scope. For instance, I have tested a bunch of probes on the Siglent SDS2304X, including a Keysight N2843A 500MHz probe, which happened to perform very similar to the old bulky (nameless!) 300MHz probes that shipped with the SDS2000 (without X), whereas I clearly prefer the characteristics of the Siglent SP2030A (shipped with the SDS2304X) over the Keysight – for that particular scope, that is. After all, the SP2030A extends the scope ±3dB bandwidth up to 450MHz! So one should be careful and keep in mind that higher price and/or more specified bandwidth does not automatically mean better performance … but more prestige at the most

Back on topic - PP215. The 200MHz Siglent scopes have a specified (calculated!) rise time of 1.8ns and since the actual bandwidth is more like 240MHz, the real rise time is 1.4~1.5ns (measured!) accordingly. The rise time of the probes has to be added, but is not specified for the PP215. Since these are only 200MHz rated, we have to assume their rise time to be also some 1.8ns, which already means a total of about 2.4ns rise time. Then the signal source sure has some rise time too, which needs to be added as well (of course this is always the same so it could be ignored when just comparing probes).

What’s more important is probe input capacitance and source impedance. There we could get some unlucky combination that makes the probe look worse than it normally is – but by no means in the realm of double-digit nanoseconds when probing low impedance circuit nodes!

From previous tests I know that the PP215 does not limit the frequency response of a 200MHz frontend as in the SDS1202X-E. It performs pretty much the same as a direct coax connection and we get >240MHz bandwidth. From that measurement, its rise time has to be better than 1.5ns.

Just to be sure, I did a quick test right now to verify the performance of the PP215. Signal source for all following tests is a pulse generator with precisely 1ns rise time, repetition rate 1kHz, pulse width 50ns, amplitude 600mVpp, with a 50ohm through termination directly fitted to its output. Probe is connected via its BNC adapter accessory, ground lead is disconnected of course.

First the PP215 on the SDS1202X-E it was shipped with:


PP215_x10_SDS1202X-E_1ns_Risetime_T

We get 2ns rise time and doing the math, knowing the signal source rise time to be 1ns and assuming the SDS1202X-E has 1.4ns, then there is just 1ns left for the PP215. Not bad at all…

Now let’s verify this on a faster scope, the SDS2304X, same test setup otherwise:


PP215_x10_SDS2304X_1ns_Risetime_T

Doing the math again. We measured 1.75ns total rise time (toggling between 1.7 and 1.8 ), scope and signal source both have 1ns and there’s once again just a round single nanosecond left for the PP215.

Finally I wanted to know what does it look like with the (rather nice) 300MHz SP2030A probe, shipping with the SDS2304X?


SP2030A_x10_SDS2304X_1ns_Risetime_T

It looks definitely different with less bad ringing. Now we get 1.6ns total rise time, minus 1ns for scope and signal source each, there is only 750ps left for the SP2030A. Not a significant difference on paper, but in practice, as the screenshots are illustrating.

In general, dealing with transition times close to 1ns, matters start getting a little tricky

[Performa01]

Just wanted to add a couple of things to Performa01's review of web server, the page 156, SDS1104X-E Review 07.pdf
This is very basic piece of software, indeed. Something is definitely wrong when it comes to using 1:10 probes.

Well, I only had a brief look at the web server as you can tell from my review. I use it regularly for pulling screenshots from the scope, but nothing else. So I’m not at all surprised that it still has some issues.

Quite frankly, I cannot be motivated to put a lot of effort in testing things I have no use for. I’ll point Siglent to your posting though.

[17_29bis]

Can you show exactly your test setup (enough good picture) and with exatly with this setup, and then also exatly with this setup image from scope screen and explained all settings.
Do you have any reliable signal source what have known pulse rise time and this source impedance, example some known function generator)

rf-loop thanks for your willingness to help but before you start spending your time let me repeat the test with other PP215 probes that came with my oscope (at the moment I have only one).
I don't have a calibrated signal generator with knows rise/fall time, I just used a 16 MHz MCU, the output pin has a  1k  pull-up resistor.

Just to be sure, I did a quick test right now to verify the performance of the PP215. Signal source for all following tests is a pulse generator with precisely 1ns rise time, repetition rate 1kHz, pulse width 50ns, amplitude 600mVpp, with a 50ohm through termination directly fitted to its output. Probe is connected via its BNC adapter accessory, ground lead is disconnected of course.

Performa01  thanks for the very extensive and detailed test, I think it's much more trustworthy than mine.

[josip]

Performa01, thank you for details with pictures, but one is missing SP2030A_x10_SDS1202X-E_1ns_Risetime_T.png 

[rigol52]

Performa01, thank you for details with pictures, but one is missing SP2030A_x10_SDS1202X-E_1ns_Risetime_T.png 

Probably you need to click on it again - it open without problem with me here.

[Performa01]

Performa01, thank you for details with pictures, but one is missing SP2030A_x10_SDS1202X-E_1ns_Risetime_T.png 

Excellent request!

It turned out that this is a perfect example to back my warnings about using random probes with your instrument, just because they have a higher bandwidth rating and/or are coming from a higher prestigious brand.


SP2030A_x10_SDS1202X-E_1ns_Risetime_T

The SP2030A is a 300MHz probe that clearly performs better than the PP215 when tested on the scope it belongs to – the SDS2304X. But on the SDS1202X, rise time gets even worse: 2.3ns vs. 2.0ns! (The transition time measurements are flickering between 2.2 and 2.4ns)

How can that be?

The most important (and expensive) part of any scope probe is the cable, that should be low capacitance and needs to have some well defined resistance for the inner conductor (around 200 ohms) in order to damp the resonance effects from the ill-terminated characteristic cable impedance. The optimum cable resistance depends on the circuit details of the scope input and on the SP2030A it quite obviously is just right for the SDS2304X, but a tad too high for the SDS1202X-E.

Some people like to replace their multimeter probes immediately after purchase with “sexier” ones and that is perfectly fine (except that I never did it because I have no use for these impractical probes at all in a lab, be they sexy or not). But for a scope, you should not replace the original probes with random ones without a second thought, just because they look sexier or have a higher bandwidth rating – only inexperienced folks do that. You might end up with an unpleasant surprise if you take the time to actually measure the performance of the probe/scope combination. Of course there’s always a chance that it actually fits well, but you should be able to verify that beforehand.

As a verdict, even though the SP2030A is a much nicer and higher quality probe (x10 only!) with index pin and 300MHz bandwidth rating, even though it performs beautifully on a SDS2304X, cannot be recommended as a performance upgrade over the PP215, at least not for the 200MHz versions of the SDS1000X-E series scopes.

[Performa01]

I don't have a calibrated signal generator with knows rise/fall time, I just used a 16 MHz MCU, the output pin has a  1k  pull-up resistor.

You write you use a MCU output pin with 1k pull-up resistor in your circuit. I do hope that output is not configured as a single ended open drain driver and the pull-up is just there to ensure a logic high level when that MCU GPIO pin is configured as input.

Because otherwise we have 1k source resistance for the rising edge. You could easily prove that by looking at the falling edge – if this is much faster, than it is actually an open drain output.

1k ohm is already a very high impedance when it comes to high frequencies (or short rise times for that matter).

Just for fun, I turned into a scope probe designer for a little while and designed a x10 probe for the SDS1202X-E, with 1.2 meters cable length and about 15pF input capacitance. Here is the circuit:


Probe x10_50_25 CD

The most important part is the lossy cable, here is the data:

Resistance = 186 ohms
Inductance = 300 nH
Capacitance = 120 pF

As can be seen, the circuit already contains the standard test setup: Signal generator with 50 ohms source impedance and 50 ohms through termination directly on its output, resulting in a total source impedance of 25 ohms, as seen by the probe.

Of course we could use this same probe and connect it to a circuit node that represents an unterminated 1k ohm source impedance. It would look like this:


Probe x10_50_1k CD

Now let’s compare the frequency response of these two scenarios:


Probe x10_50 FR

What a difference. My probe design performs well up to almost 300MHz within +/- 1dB when probing a 25 ohm source impedance node. The -3dB bandwidth is more than 400MHz, equivalent to <1ns rise time. For 1k ohm, we barely get 10MHz -3dB bandwidth, which means some 35ns rise time.

The question remains, why do the other probes perform more than twice as good?

They could have a lower input capacitance, but this could not explain the huge difference. Even the Keysight probe with only 11pF input capacitance and assuming the PP215 actually had 20pF (which I really doubt, 16pF is much more likely) could not turn 22ns into 10ns.

The actual difference in capacitive loading has to be lower anyway, as there is some capacity in the circuit already – The GPIO port pin of a typical MCU alone can have up to 10pF. Then add the board traces and everything that is connected to them. With just 10pF circuit capacitance, the difference in capacitive loading would be 30pF/21pF worst case and the Keysight probe would measure at least 15ns when the PP215 measures 22ns.

So there must be something else going on…

[rf-loop]

Sidenote:
Many good and more expensive probes are really different animals than some Rigol-Siglent-Gratlent-Whatlent scopes standard accessory ones.

They have also more adjustments than just this usual "LF" compensation what everyone turn as needed.

Attached example about quite simple and not so bad class HP probe. (part of instructions)

More deep adjustments can do also in many probes. But these adjustments must not do without needed equipments.
Some different passive probes have enen more adjustments for compromise highest, middle and low freq and very fast edge response and then also adjustment for DC level error. Some probes are quuite tricy to adjust for perfect response with just one individual scope and just individual channel because in final game all are bit different, not only installed compoonents inside scope but also bit different paarasitics in scope, cable and probe tip and adjustment box.

And we do not forget Tektronix special probe cables what of course have some "impedance" but also inner wire have designed resistance what is also part of damping ringing.
Then expensive probes are also mechanically very different. Example some older Tek and HP passive probes are made for repair. All spare parts was available... tips, cables etc--- all replaceable.

--------------

About this previous @17_29bis test.

OpenCollector type output with 1k pullup. This explain more than enough this result with P215.
1kOhm  and if think roughly around 100pF 1:1 probe setting. It acts like single RC LP filter.  LP cut-off freq is 1.6MHz without parasitics etc calculation because they are unknown and just only for imagine what is going on. But they do here things only even more bad.
Also pulse response to up direction is slower than down due to fact that C charge/discharce is not symmetrical (To up this passive 1kOhm and in this case this active down component in your MCU what may have quite low impedance but sure faster than up with 1kOhm pull up resistor. )

---------------
Nice probe work @Performa01

[tautech]

They could have a lower input capacitance, but this could not explain the huge difference. Even the Keysight probe with only 11pF input capacitance and assuming the PP215 actually had 20pF (which I really doubt, 16pF is much more likely) could not turn 22ns into 10ns.

Little side note:
PP215 10:1 input capacitance is listed on the probe brochure and in Siglent websites as 13 or 14 pF......not 16pF

OpenCollector type output with 1k pullup. This explain more than enough this result with P215.
1kOhm  and if think roughly around 100pF 1:1 probe setting. It acts like single RC LP filter.

Exactly 

[Performa01]

They could have a lower input capacitance, but this could not explain the huge difference. Even the Keysight probe with only 11pF input capacitance and assuming the PP215 actually had 20pF (which I really doubt, 16pF is much more likely) could not turn 22ns into 10ns.

Little side note:
PP215 10:1 input capacitance is listed on the probe brochure and in Siglent websites as 13 or 14 pF......not 16pF

Well, then that's Siglent's fault. 

I've got my data from there "SIGLENT_Probe_Datasheet V1.5" which I've downloaded just a week ago (see attachment).

But you are right: The little card that comes in the probe bag lists the input capacitance as 14pF