Siglent SDS800X HD 12 bit DSO's

[crysti]

What is the reason the SDS800X HD series oscilloscopes no longer display the time counter correctly at high frequencies? 😊

Increase the vertical resolution to 1 or 2mV div, and see if it works better for you. Using only 25% of the screen is going to limit the what the scope can do, even though you're well beyond the scopes limitations anyway at 630MHz.

Even on my SDS2000X HD, dropping the vertical resolution that low throws off the counter.

ETA: Even if you do that, it may not be able to count that high. If you sweep through the frequencies, you'll probably see a certain point that the accuracy of the counter dies. For example, on my old SDS2000X Plus, I was able to accurately count up to 841MHz. After that point, the frequency counter value dropped significantly and made no sense anymore.

The Siglent SDS2000XP is built on a 500 MHz platform. Displaying a 600 MHz signal on a 200 MHz platform represents a much greater leap in performance.

[mawyatt]

What is the reason the SDS800X HD series oscilloscopes no longer display the time counter correctly at high frequencies? 😊

We can "only" count to just above 400MHz, which is plenty good enough for a 200MHz instrument IMO

Best

[KungFuJosh]

The Siglent SDS2000XP is built on a 500 MHz platform. Displaying a 600 MHz signal on a 200 MHz platform represents a much greater leap in performance.

I understand that. It doesn't change that your counter probably won't work as well if you're only using 25% of the vertical resolution.

Mike said the 800X HD scope can count up to 400MHz. Feed a 400MHz signal and see how the counter is affected by vertical usage.

ETA: I added two screenshots of the same exact 700MHz signal. Only difference is the vertical setting on the scope.

[crysti]

I replaced the old fan with an MF60251V1-1000U-A99 and added an external temperature controller with a sensor to the middle radiator source. This way, I eliminated the bad smell from the old fan, reduced the noise, and increased the temperature range above 40°C, as specified. Previously, the air coming out of the old fan was warm. Now, the air exits at the same temperature as it enters, thanks to the fan that increases its speed only when the radiator temperature exceeds 45°C. Before I had 2.3m/s, now I have between 2.6m/s and 3.2m/s. Everything cost me about $10.  Siglent used the cheapest possible fan, with no temperature-based control whatsoever. I found the exact fan model used by Siglent available for $1 each when purchasing a minimum of 1,000 units.

[mawyatt]

Don't know if this has been posted before but here's an example of using the DSO noise floor to show how the built-in FIR filters behave.

We created a Low Pass Filter at 5KHz with F1(Orange) using CH1 noise as the input signal, then another Band Reject Filter F3(Blue) with a 10kHz to 20kHz Band Reject (Notch) also using CH1 noise as the input.

F2(Red) is the FFT of the LPF signal from F1, and F4 (Green) is the FFT of the BRF F3.

Both FFTs have the same parameters, 1kHz-100kHz decade, -60dBm Ref, 20 dB/Div, 200kSa/s, Flat-Top, 65,536pts (64k), 3Hz ∆F and 11Hz RBW.

And another with the Band Reject Filter changed to a Band Pass Filter with same Lower and Upper frequencies.

Edit: We've added another plot showing a swept 12kHz to 18kHz -80dBm signal to CH1 and the LPF and BPF results by way of the FFTs. Note the DSO sweep time changed to 10ms/div and FFTs lengths to 16, 384 pts, these were changed as shown to reduce acquisition and FFT processing time, however the filters were unchanged. The last plots show the results with the input swept from 2kHz to 4kHz and 2kHz to 15KHz respectively. Should note these were achieved using the FFT Max-Hold Mode rather than the usual Normal or Average Modes.

Best

[watchmaker]

WOW.  I just found out you can do an FFT on a Math Averaging.  VERY cool little computer.

[crysti]

What is your opinion on using low-quality probes that exhibit overshoot, seemingly extending the oscilloscope's frequency bandwidth beyond its specified limits?

[Martin72]

Regardless of the fact that it's cheap stuff:
How did you measure 80 MHz with the probe?
Who generates the 80 MHz?

[crysti]

Regardless of the fact that it's cheap stuff:
How did you measure 80 MHz with the probe?
Who generates the 80 MHz?

With LiteVNA 64 in CW FREQ mode.  To generate harmonics up to 6.3 GHz, the signal must have a rise time of approximately 56 ps.

[Martin72]

What does the signal look like when you connect a BNC cable directly to the load resistor?
I´m just curious..

[crysti]

Something like this. The displayed voltage must be divided by 10. Channel 2 is a little faster than the other channels and channel 4 is the slowest.

[Martin72]

OK, except for the amplitude, it looks the same as the one with the probe, so it can't be attributed to the probe.
Thanks!

[watchmaker]

I was tracking down a grounding issue with my scope which I think I am close to resolving (make a dedicated ckt/move USB hubs and turn off when measuring).  Did not  hurt to pull everything apart and reconfigure either.

But, as you might expect, I came up with another issue.  This involves getting an incorrect value with a switchable probe and the correct attenuation setting on the scope and the probe.  I used two fiexed 10x probes as my controls for the two conditions of switchable probe at 1x and then at 10x.

Can someone look at post https://www.eevblog.com/forum/beginners/my-esd-mat-hums/msg5914526/#msg5914526 and explain to me back in this thread what I am missing?  I expected the switchable probe to report the same value for amplitude regardless of matching attenuations settings on the scope and the probe.  BTW, I also repeated the experiment with a different 1x/10x probe.

THANKS!

[tautech]

I was tracking down a grounding issue with my scope which I think I am close to resolving (make a dedicated ckt/move USB hubs and turn off when measuring).  Did not  hurt to pull everything apart and reconfigure either.

But, as you might expect, I came up with another issue.  This involves getting an incorrect value with a switchable probe and the correct attenuation setting on the scope and the probe.  I used two fiexed 10x probes as my controls for the two conditions of switchable probe at 1x and then at 10x.

Can someone look at post https://www.eevblog.com/forum/beginners/my-esd-mat-hums/msg5914526/#msg5914526 and explain to me back in this thread what I am missing?  I expected the switchable probe to report the same value for amplitude regardless of matching attenuations settings on the scope and the probe.  BTW, I also repeated the experiment with a different 1x/10x probe.

THANKS!

Possibly a Trigger level issue.
Near the floor of a waveform you can run into hysteresis issues.
Shift it up some and see if better values return.

Also remember you are loading the signal with more capacitance with 1x probes which invokes the old adage; measurement affects the measurement. 

[BillyO]

Also remember you are loading the signal with more capacitance with 1x probes which invokes the old adage; measurement affects the measurement. 

The uncertainty principle observer effect at work here?  Increased probe loading reduces the observed voltage.

@watchmaker: What is the nature of the signal source?

[tautech]

Also remember you are loading the signal with more capacitance with 1x probes which invokes the old adage; measurement affects the measurement. 

The uncertainty principle at work here?
Increased probe loading reduces the observed voltage.

No.
The increased loading by reducing the signal brings trigger hysteresis into play when its level is so low on the waveform.

Regardless of signal strength/amplitude, trigger hysteresis can muck with results especially near the floor of amplitude.
Mostly just pressing the trigger encoder to set a 50% level just works and gets us away from all triggering uncertainty.

[BillyO]

No.
The increased loading by reducing the signal brings trigger hysteresis into play when its level is so low on the waveform.

Regardless of signal strength/amplitude, trigger hysteresis can muck with results especially near the floor of amplitude.
Mostly just pressing the trigger encoder to set a 50% level just works and gets us away from all triggering uncertainty.

If you look at the 2nd image in the post he linked to, the signal is actually reduced by a factor of 10 or so from the first image.  So, it is being measured correctly.  Or am I still missing something hiding out in the open?

[watchmaker]

Tautech and BillyO:  Thank you.  Never thought about it.  Oops!

You may have seen I was up from 3AM tracking down this gremlin.  After a nap, I got rid of a halogen lamp, which I should have known since I have to move one on my watch bench and turn it off when I use my electronic timer), screwed the mat's ground lead to the metal housing of the Amazon power strip, and moved the switch box that controls my USB stuff, and will turn it off when making measurements.

These things combined got me to a consistent floating over the mat amplitude of 20 mVACp-p which is one sixth of what I started out with this morning.  Little and a little.

For giggles I will install a dedicated ckt  for the bench from the subpanel.  Will want this anyway when I build a real room and get rid of the crates I made for moving my machinery.

Warmest thanks,

Dewey

[seavan]

Maybe try connecting load resistor directly to the bnc connector of the oscilloscope without the probe?

What does the signal look like when you connect a BNC cable directly to the load resistor?
I´m just curious..

[tautech]

No.
The increased loading by reducing the signal brings trigger hysteresis into play when its level is so low on the waveform.

Regardless of signal strength/amplitude, trigger hysteresis can muck with results especially near the floor of amplitude.
Mostly just pressing the trigger encoder to set a 50% level just works and gets us away from all triggering uncertainty.

If you look at the 2nd image in the post he linked to, the signal is actually reduced by a factor of 10 or so from the first image.  So, it is being measured correctly.  Or am I still missing something hiding out in the open?

2 things IMO....not had time to do exhaustive tests as have been travelling.

Low trigger level combined with measurements captured in Stop mode where Stop was possibly used to get a stable display.
Another possibility is a faulty 1x/10x probe switch.....

[watchmaker]

No.
The increased loading by reducing the signal brings trigger hysteresis into play when its level is so low on the waveform.

Regardless of signal strength/amplitude, trigger hysteresis can muck with results especially near the floor of amplitude.
Mostly just pressing the trigger encoder to set a 50% level just works and gets us away from all triggering uncertainty.

If you look at the 2nd image in the post he linked to, the signal is actually reduced by a factor of 10 or so from the first image.  So, it is being measured correctly.  Or am I still missing something hiding out in the open?

2 things IMO....not had time to do exhaustive tests as have been travelling.

Low trigger level combined with measurements captured in Stop mode where Stop was possibly used to get a stable display.
Another possibility is a faulty 1x/10x probe switch.....

Not a faulty 1x/10x switch.  Tried another pp510.  For those who follow:

My concern was the amplitude of the noise present when I left the probes on the mat.  Initally it was 120 mVAC p-p.  The main culprit was a bad mat ground and my USB hubs.  Pics below of the Hub on and off.  I get to below 20 mVACp-p with the hub off.

But now I would like to know what is an acceptable level of noise when the probe is unloaded? Much of my confusion (?) is that I have no idea of the proper expectations.

As a statistician, I abhor noisy data.  This is why I was playing with averaging a signal and then doing an FFT on that average a couple days ago.  I would rather fix the issue than apply bandaids.

I do know when I had 120mVACp-p that I had difficulty with some simple low level signals.  I have not had the chance to go back since I totally recleaned my electronics area; but I will.

Not sure of the purpose of connecting a resistor directly into the BNC connector is, but the third shot shows the resulting signal.

THANKS

[BillyO]

Having the unconnected probes just lying on your mat is not a good way to measure your noise floor.  They will just behave like antennas and pick up whatever is floating through your local EM field.  Connected to any kind of "real" source, they won't pick up that noise in and of themselves.  The DUT might pick it up and send it to the scope, but that is a different matter and a part of the results of testing teh DUT.  To test noise floor, look at the noise with the probes unattached and measure StDev or Avg.

Below are screen shots of three of my scopes.

1) SDS1204X-E (an 8-bit scope)
2) SDS2504X-P (8-bit mode)
3) SDS2504X-P (10-bit mode)
4) SDS824X-HD (a 12 bit scope)

[tautech]

My concern was the amplitude of the noise present when I left the probes on the mat.  Initally it was 120 mVAC p-p.  The main culprit was a bad mat ground and my USB hubs.  Pics below of the Hub on and off.  I get to below 20 mVACp-p with the hub off.

But now I would like to know what is an acceptable level of noise when the probe is unloaded? Much of my confusion (?) is that I have no idea of the proper expectations.

As a statistician, I abhor noisy data.  This is why I was playing with averaging a signal and then doing an FFT on that average a couple days ago.  I would rather fix the issue than apply bandaids.

Noise is something we must work with and get to understand so not to misinterpret real measurements.

A few screenshots are attached, the 1st with no probe connected and later with probe added which is our normal signal attachment method.
Instantly we see the probe along with the scopes sensitivity doing its job where we need basic understanding what it is and why we see it as probing technique can do wonders for the integrity of signals. 
For low impedance signals it normally doesn't but we always should be mindful it might.


A few posts back I mentioned, every measurement affects the measurement in some way and we also need to know if and when that might matter.


Following screenshots show mains noise, instantly recognisable as the 50 Hz sinewave of our 50 Hz mains, yours will be 60 Hz.
PNG 4 is the probe reference lead hooked to the probe grabber.
The last shows a probe draped over a USB socket in a power board where we can clearly see the higher amplitude and typical SMPS noise.

Also notice my screenshots don't show the Save menu as I instead use the more convenient Save button on the front panel which also provides the smaller PNG screenshot files.

[temperance]

The main culprit was a bad mat ground and my USB hubs.

I find that strange because an ESD mat has a very high resistance. Depending on the type of mat you have the top of the mat measures over a GOhm and the bottom on most mats will measure around 80...100 K while a scope input is 10MOhm, forming a voltage divider.

I found this for you made by Dave:

But now I would like to know what is an acceptable level of noise when the probe is unloaded? Much of my confusion (?) is that I have no idea of the proper expectations.

With the probe shorted, nothing. With the probe open on the table even the orientation matters.

Most noise problems in measurements are caused by noise coupled trough Y type capacitors present in switching power supplies between rectified GND and the output GND of a switching power supply.

To give you an idea, FesZ Electronics has this nice presentation:

Because of this I use power supplies with "normal transformers" when switching power supplies generate to much noise. Other sources of noise are computers (especially if you connect some debugging tool to your circuit), LED lighting power supplies, or just any other switching power supply in your lab. Some of those low cost power bricks can be "seen" at the other side of the room.

[watchmaker]

BillyO,

Thanks, I had already done the floor noise exercise.  The issue is ambient noise.

I had no idea if this impacts measurements, but in other areas I know, ambient noise contaminates the data.


BUT, I think you just told me I am worried about the health of unicorns.  THANK YOU!.

Warmly,

Dewey