Rigol HDO1000 and HDO4000 12bit oscilloscopes launched in China

[TurboTom]

I guess the differences in RMS noise level are just normal tolerance. @Core probably got particularly lucky with his, on mine I also find the noise to be in the 23µVrms ballpark like many others, see screenshots. I also checked the 1V/div noise as @egonotto requested. And one special screenshot for our friends from Ukraine  .

[egonotto]

Hello,

thanks TurboTom, have you 50 Ohm terminators on inputs as core has?

Best regards
egonotto

[thm_w]

No protective film at all on it. I don't mind but it is unusual, companies usually like to cover everything in plastic bag/film.

Stock bandwidth measured: ~115MHz
200MHz option: 396MHz (1 channel), 320MHz (2 channel)

Anyone see with 2 channels on it seems to get way more vertical jitter? The waveform looks terrible, even at lower frequencies. Maybe an issue with my source or cable, odd it doesn't show up when on single channel though. Will try again later with a proper source.

[TurboTom]

Hello,

thanks TurboTom, have you 50 Ohm terminators on inputs as core has?

Best regards
egonotto

Yes I had both inputs (CH1 and 4) "terminated" with 50 Ohm, yet via a BNC -> SMA adapter since most of my RF stuff - and thus also the "good" terminators - is SMA.

I may check the total input voltage range of the scope at that sensitivity setting (i.e. how far the converter will linearly digitize beyond the visible vertical range) to approximate ENOB...


@thm_w: The jitter is rather aliasing due to frequency components at the edges of the input signal beyond Nyquist. This "misleads" the sinx/x waveform reconstruction engine and results in the visible jitter effect. You will find this effect on any digital scope without a proper filter in front of the sampling engine. It's probably the price we've got to pay for today's scopes where the "usable" frequency range approaches two thirds of the Nyquist frequency.

[RobbiOne]

I can't get any v2.12 FW for DHO1000 from international Rigol web site https://supportint.rigol.com.
Just the v2.11 (August 2023) from Rigol EU (Germany). Is it a special version available only for some country?

[Veteran68]

I can't get any v2.12 FW for DHO1000 from international Rigol web site https://supportint.rigol.com.
Just the v2.11 (August 2023) from Rigol EU (Germany). Is it a special version available only for some country?

Did you look here, as was posted?

https://int.rigol.com/products/detail/DHO1000

I see it there.

[David Hess]

@thm_w: The jitter is rather aliasing due to frequency components at the edges of the input signal beyond Nyquist. This "misleads" the sinx/x waveform reconstruction engine and results in the visible jitter effect. You will find this effect on any digital scope without a proper filter in front of the sampling engine. It's probably the price we've got to pay for today's scopes where the "usable" frequency range approaches two thirds of the Nyquist frequency.

Exactly, but even with a perfect anti-aliasing filter, aliasing will still be present because of mixing products between the sample clock and input signal produced by non-linearity in the digitizer.  This can be shown by using a clean sine wave source which will still produce the "wobulation".  The non-linearity is present in the digitizer itself, but also results from jitter in the sample clock.

Older DSOs tend to be worse, especially those with interleaved ADCs because of greater non-linearity in the digitizer.  On these instruments the "wobulation" goes away when random equivalent sampling is used because the equivalent time sample rate is so much higher so there can be no aliasing.  Modern instruments which lack equivalent time sampling can use averaging (HP/Agilent/Keysight recommended this in an application note), but I do not think this is as effective.

[2N3055]

@thm_w: The jitter is rather aliasing due to frequency components at the edges of the input signal beyond Nyquist. This "misleads" the sinx/x waveform reconstruction engine and results in the visible jitter effect. You will find this effect on any digital scope without a proper filter in front of the sampling engine. It's probably the price we've got to pay for today's scopes where the "usable" frequency range approaches two thirds of the Nyquist frequency.

Exactly, but even with a perfect anti-aliasing filter, aliasing will still be present because of mixing products between the sample clock and input signal produced by non-linearity in the digitizer.  This can be shown by using a clean sine wave source which will still produce the "wobulation".  The non-linearity is present in the digitizer itself, but also results from jitter in the sample clock.

Older DSOs tend to be worse, especially those with interleaved ADCs because of greater non-linearity in the digitizer.  On these instruments the "wobulation" goes away when random equivalent sampling is used because the equivalent time sample rate is so much higher so there can be no aliasing.  Modern instruments which lack equivalent time sampling can use averaging (HP/Agilent/Keysight recommended this in an application note), but I do not think this is as effective.

Siglent scopes have dot mode where they practically perform RIS (random interleaved sampling), although it is not marketed as such. If signal is triggering fast enough you get nice waveform reconstruction on repetitive waveforms, even when severely undersampled. So they don't market ETS or RIS but it still works as such.

[core]

Indeed, noise level is very low.

1mV/div, 2 channels, 50 Ohm terminators on inputs, BW 20MHz / 200MHz, memory auto / max.
For me ch 1 have the highest bias, and ch 4 the lowest.

Hello,

Can you please make the measurements with 1 V/div instead of 1 mV/div?

Best regards
egonotto

Yes, see attached. BW 20MHZ / 200MHZ. 50 Ohm terminators.
About the same as TurboTom.

[David Hess]

...

Older DSOs tend to be worse, especially those with interleaved ADCs because of greater non-linearity in the digitizer.  On these instruments the "wobulation" goes away when random equivalent sampling is used because the equivalent time sample rate is so much higher so there can be no aliasing.  Modern instruments which lack equivalent time sampling can use averaging (HP/Agilent/Keysight recommended this in an application note), but I do not think this is as effective.

Siglent scopes have dot mode where they practically perform RIS (random interleaved sampling), although it is not marketed as such. If signal is triggering fast enough you get nice waveform reconstruction on repetitive waveforms, even when severely undersampled. So they don't market ETS or RIS but it still works as such.

That is effectively what HP/Agilent/Keysight recommended for DSOs which lack equivalent time sampling.  The problem is that even in dot mode, (sin x)/x reconstruction must still be used to determine the trigger point, so the aliasing moves the trigger point around corrupting the result.  Averaging helps, but it does not produce as accurate a result as equivalent time sampling.

As a practical matter, the difference should be irrelevant.  In both cases the bandwidth of the instrument is insufficient to accurately represent the signal.

[rf-loop]

...

Older DSOs tend to be worse, especially those with interleaved ADCs because of greater non-linearity in the digitizer.  On these instruments the "wobulation" goes away when random equivalent sampling is used because the equivalent time sample rate is so much higher so there can be no aliasing.  Modern instruments which lack equivalent time sampling can use averaging (HP/Agilent/Keysight recommended this in an application note), but I do not think this is as effective.

Siglent scopes have dot mode where they practically perform RIS (random interleaved sampling), although it is not marketed as such. If signal is triggering fast enough you get nice waveform reconstruction on repetitive waveforms, even when severely undersampled. So they don't market ETS or RIS but it still works as such.

That is effectively what HP/Agilent/Keysight recommended for DSOs which lack equivalent time sampling.  The problem is that even in dot mode, (sin x)/x reconstruction must still be used to determine the trigger point, so the aliasing moves the trigger point around corrupting the result.  Averaging helps, but it does not produce as accurate a result as equivalent time sampling.

As a practical matter, the difference should be irrelevant.  In both cases the bandwidth of the instrument is insufficient to accurately represent the signal.

Display mode dots, interpolation x
( I have reduced memory for get enough low samplerate for demonstrate this)

In this image Zoom window sample interval is 2 div (10ns)

Rise time here is <0.5*sample interval. I can not see big problems with trig position...

Of course can say that it is not accurately trig'd but...  how big problems you see there...  and this is not ETS.
There on screen is sequential acquisitions randomly interleaved (overlaid)

However, this method has limitations, although it is useful in some situations. Most of the time, our real-time sample rate is sufficient because the full 2GSa/s can be kept up to 10ms/div (SDS2000X Plus)
But as can see trigger position is still ok even when in this case sample interval is 2 div in this Zoom window (4000x Zoom ). Digital trigger engine after ADC can still handle this quite well.

[core]

...

Older DSOs tend to be worse, especially those with interleaved ADCs because of greater non-linearity in the digitizer.  On these instruments the "wobulation" goes away when random equivalent sampling is used because the equivalent time sample rate is so much higher so there can be no aliasing.  Modern instruments which lack equivalent time sampling can use averaging (HP/Agilent/Keysight recommended this in an application note), but I do not think this is as effective.

Siglent scopes have dot mode where they practically perform RIS (random interleaved sampling), although it is not marketed as such. If signal is triggering fast enough you get nice waveform reconstruction on repetitive waveforms, even when severely undersampled. So they don't market ETS or RIS but it still works as such.

That is effectively what HP/Agilent/Keysight recommended for DSOs which lack equivalent time sampling.  The problem is that even in dot mode, (sin x)/x reconstruction must still be used to determine the trigger point, so the aliasing moves the trigger point around corrupting the result.  Averaging helps, but it does not produce as accurate a result as equivalent time sampling.

As a practical matter, the difference should be irrelevant.  In both cases the bandwidth of the instrument is insufficient to accurately represent the signal.

Display mode dots, interpolation x
( I have reduced memory for get enough low samplerate for demonstrate this)

In this image Zoom window sample interval is 2 div (10ns)

Rise time here is <0.5*sample interval. I can not see big problems with trig position...

Of course can say that it is not accurately trig'd but...  how big problems you see there...  and this is not ETS.
There on screen is sequential acquisitions randomly interleaved (overlaid)

However, this method has limitations, although it is useful in some situations. Most of the time, our real-time sample rate is sufficient because the full 2GSa/s can be kept up to 10ms/div (SDS2000X Plus)
But as can see trigger position is still ok even when in this case sample interval is 2 div in this Zoom window (4000x Zoom ). Digital trigger engine after ADC can still handle this quite well.

Very interesting, indeed. Thanks !

All we can get with sinc is below. Rigol doesn't like x and dots.

[Fungus]

Anyone see with 2 channels on it seems to get way more vertical jitter? The waveform looks terrible, even at lower frequencies. Maybe an issue with my source or cable, odd it doesn't show up when on single channel though. Will try again later with a proper source.

The problem is trying to trigger at 2ns/div horizontal when you have 1GS/sec sample rate.

[rf-loop]

...

Older DSOs tend to be worse, especially those with interleaved ADCs because of greater non-linearity in the digitizer.  On these instruments the "wobulation" goes away when random equivalent sampling is used because the equivalent time sample rate is so much higher so there can be no aliasing.  Modern instruments which lack equivalent time sampling can use averaging (HP/Agilent/Keysight recommended this in an application note), but I do not think this is as effective.

Siglent scopes have dot mode where they practically perform RIS (random interleaved sampling), although it is not marketed as such. If signal is triggering fast enough you get nice waveform reconstruction on repetitive waveforms, even when severely undersampled. So they don't market ETS or RIS but it still works as such.

That is effectively what HP/Agilent/Keysight recommended for DSOs which lack equivalent time sampling.  The problem is that even in dot mode, (sin x)/x reconstruction must still be used to determine the trigger point, so the aliasing moves the trigger point around corrupting the result.  Averaging helps, but it does not produce as accurate a result as equivalent time sampling.

As a practical matter, the difference should be irrelevant.  In both cases the bandwidth of the instrument is insufficient to accurately represent the signal.

Display mode dots, interpolation x
( I have reduced memory for get enough low samplerate for demonstrate this)

In this image Zoom window sample interval is 2 div (10ns)

Rise time here is <0.5*sample interval. I can not see big problems with trig position...

Of course can say that it is not accurately trig'd but...  how big problems you see there...  and this is not ETS.
There on screen is sequential acquisitions randomly interleaved (overlaid)

However, this method has limitations, although it is useful in some situations. Most of the time, our real-time sample rate is sufficient because the full 2GSa/s can be kept up to 10ms/div (SDS2000X Plus)
But as can see trigger position is still ok even when in this case sample interval is 2 div in this Zoom window (4000x Zoom ). Digital trigger engine after ADC can still handle this quite well.

Very interesting, indeed. Thanks !

All we can get with sinc is below. Rigol doesn't like x and dots.

Of course you get around same but as you can see in my image samplerate was 100MSa/s  when you use 2GSa/s.
Now try to do same using 100MSa/s but same signal. Or if not possible set oscilloscope and signal so that sampling interval is least 2x longer than risetime.  Whole thing was just this... sampling speed far below fNyquist (risetime).


Naturally this below using 2GSa/s as is your image for same signal is just easy.. all can do it.
Sinc, and 2GSa/s. (yes but this is not 12bit scope)
As can also see there is now 100M bandwidth, so risetime bit reduced.. (because 8 bit scope run in 10bit mode)

[RobbiOne]

Inside that web page:

https://int.rigol.com/products/detail/DHO1000

I just see a tech features of the scope and 3 links for manual, datasheet and demo request.
No firmware or download link.

[ebastler]

Inside that web page:

https://int.rigol.com/products/detail/DHO1000

I just see a tech features of the scope and 3 links for manual, datasheet and demo request.
No firmware or download link.

Pro tip: Try the "Software & Firmware" section. Scroll down until you get to the light-grey background, then there's a menu bar (tab selector).

[thm_w]

Anyone see with 2 channels on it seems to get way more vertical jitter? The waveform looks terrible, even at lower frequencies. Maybe an issue with my source or cable, odd it doesn't show up when on single channel though. Will try again later with a proper source.

The problem is trying to trigger at 2ns/div horizontal when you have 1GS/sec sample rate.

Yeah, though even if I drop the source signal frequency by half to compensate for 1GS/s vs 2GS/s, signal is still not clean. As explained above there are probably higher frequency components causing issues, so I need to find a cleaner sine source more suitable for testing.

Additional hardware filtering options would be useful, the MSO5000 had 100MHz and 200MHz filters. In this case we know at least a 100MHz filter could be implemented (since the ~100MHz bandwidth option exists).

edit: maybe some filter is applied with 2ch on, since 3dB is dropping ~80MHz? But it seems not ideal.

[core]

Yeah, though even if I drop the source signal frequency by half to compensate for 1GS/s vs 2GS/s, signal is still not clean. As explained above there are probably higher frequency components causing issues, so I need to find a cleaner sine source more suitable for testing.

Additional hardware filtering options would be useful, the MSO5000 had 100MHz and 200MHz filters. In this case we know at least a 100MHz filter could be implemented (since the ~100MHz bandwidth option exists).

edit: maybe some filter is applied with 2ch on, since 3dB is dropping ~80MHz? But it seems not ideal.

Indeed, an additional 100MHz option as filter can be useful.
But we can send SCPI commands to switch between 70/100/200MHZ.

[Antonio90]

I'm wondering how hard would it be to do FFT with Matlab's signal processing suite, extracting the waveform data form the scope.
Maybe even bode plot with the instrument control and/or audio apps?
Could someone extract some kind of waveform from his scope and post it? I don't have a scope at hand right now, but I do have a laptop and some spare time to play around with it.

[voltsandjolts]

This has been done several times in the past for older Rigol models.
Maybe the HDO1000 wfm format is compatible with one of those implementations already, dunno.
e.g.
https://github.com/scottprahl/RigolWFM

This was quite a good utility for the 1052E, with good window options for fft:
http://meteleskublesku.cz/wfm_view/

Rigol seem uninterested in fft features such as averaging or peakhold, so a PC app is likely gonna be the only way to get those nice features. But it's a slower and clunkier than having it built-in of course. Hello..Rigol..are you listening.............<crickets>

[Martin72]

By the way, the 40% off discount for most of the DHO1000 models is still valid until end of this year.

[Veteran68]

By the way, the 40% off discount for most of the DHO1000 models is still valid until end of this year.

Not in the US. They expired in the first few days of December. Back to regular price at RigolNA and Saelig, anyway.

[Martin72]

Depending on the import duties, it might be worth ordering in the EU.

[Jacon]

...Could someone extract some kind of waveform from his scope and post it?...

Done already for FFT from DHO800 here:
https://www.eevblog.com/forum/testgear/rigol-dho804-test-and-compare-thread/msg5214702/#msg5214702

[voltsandjolts]

Yes, lots could be achieved with a bit of python on the PC side. There is also possibility of hooking into gnu-radio and using the Fosphor fft and spectrogram plotting.

But since the scope runs on android, and folks have already demonstrated side loading other applications, someone much smarter than I, might be able to craft an android app that controls and reads scope data over lxi (locally) and plots fft etc with nicer features than the rigol plot. There may not be the system resources for it though.