Rigol DHO804 Test and Compare Thread

[Martin72]

Hi folks,

No no, I don't have the scope yet, I also have to wait about 2 weeks for it....
This is already a placeholder or should give the possibility to consider it in advance, what I should test everything.
And compare, because I still have other scopes here....
First my own, quasi as a reference, the SDS2504X HD.
Then from work the Rigol DS1054Z and the Siglent SDS1104X-E.
The predecessor and the current competitor, so to speak.
Other existing measuring equipment:
- SDG2122X 120Mhz function generator.
-R&S SML01 1.1Ghz signal generator
-Tektronix A6302/AM502 current clamp combination
- Demoboards with several signals (and decoder functions) from Batronix and Siglent
- Deskew fixture from Siglent
- Analog 10Mhz test oscillator for ENOB determination
- Ultra low distortion 1khz sine generator (still to be tested)
- Leo Bodnar pulser 40ps risetime

So think about something, I'll get back to you here when the Rigol has arrived.

[tautech]

USB keyboard and mouse performance in the UI compared to 2kX HD.
Also webserver UI performance with PC mouse, keyboard, scroll wheel etc.

[Njk]

Fan noise measurement. That'll be real challenge

[tautech]

Fan noise measurement. That'll be real challenge

Not.

Piss easy with a phone app like this:
Audio Spectrum Analyzer dB RTA
https://apps.apple.com/nz/app/id1281873790

Dozens of them on the www, some vastly better than others.

[Fungus]

Piss easy with a phone app like this:

An app won't tell you how whiny or annoying a sound is, only how loud it is.

USB keyboard and mouse performance in the UI compared to 2kX HD.

Mouse: It works
Keyboard: It doesn't

Well, there goes Martin72's shiny new topic... 

[tautech]

Piss easy with a phone app like this:

An app won't tell you how whiny or annoying a sound is, only how loud it is.

Really

About time you did some research instead of spouting about shit you have no experience with.
Audio spectrum analyzers give frequency and level measurements just in case you didn't know.

Their measurement only need be qualified by the distance to microphone measurement and I'm very sure Martin already knows how these things work.

[hbozyq]

Piss easy with a phone app like this:

An app won't tell you how whiny or annoying a sound is, only how loud it is.

Really

About time you did some research instead of spouting about shit you have no experience with.
Audio spectrum analyzers give frequency and level measurements just in case you didn't know.

Their measurement only need be qualified by the distance to microphone measurement and I'm very sure Martin already knows how these things work.

I guess the real problem is can you trust those result?
I bet that no one is calibrated.

[tautech]

Piss easy with a phone app like this:

An app won't tell you how whiny or annoying a sound is, only how loud it is.

Really

About time you did some research instead of spouting about shit you have no experience with.
Audio spectrum analyzers give frequency and level measurements just in case you didn't know.

Their measurement only need be qualified by the distance to microphone measurement and I'm very sure Martin already knows how these things work.

I guess the real problem is can you trust those result?
I bet that no one is calibrated.

They don't need be for comparative measurements or would you rather trust Dave's ear to judge fan noise ? 

[Fungus]

About time you did some research instead of spouting about shit you have no experience with.
Audio spectrum analyzers give frequency and level measurements just in case you didn't know.

Please tell us what a whiny/annoying spectrum looks like compared to a nice relaxing one.
This needs to be precisely defined if the results are to be valid.

[tautech]

About time you did some research instead of spouting about shit you have no experience with.
Audio spectrum analyzers give frequency and level measurements just in case you didn't know.

Please tell us what a whiny/annoying spectrum looks like compared to a nice relaxing one.
This needs to be precisely defined if the results are to be valid.

Oh FFS Fungus, must you be spoon fed ? 
Did you not follow this link posted above:
https://apps.apple.com/nz/app/id1281873790

[Fungus]

Please tell us what a whiny/annoying spectrum looks like compared to a nice relaxing one.
This needs to be precisely defined if the results are to be valid.

Oh FFS Fungus, must you be spoon fed ? 
Did you not follow this link posted above:
https://apps.apple.com/nz/app/id1281873790

It's a spectrum analyzer (for a device not everybody has).

OK, please tell us what a whiny/annoying spectrum looks like compared to a nice relaxing one.

ie. How does that amazing app help us put a neat numerical score on each fan?

[Martin72]

Their measurement only need be qualified by the distance to microphone measurement and I'm very sure Martin already knows how these things work.

Maybe it's no surprise but I have a calibrated microphone at home.
And access to a ISO calibrated Class 1 SPL meter..

[tautech]

It's a spectrum analyzer (for a device not everybody has).

No it's not.
Audio analyzer that does LF which you can find online for many smartphones not just fruity phones and tablets.

An annoying audio spectrum might be heard by the young and not the old therefore annoying is NOT universal.
Do some homework Fungus just like I had to a few months back to sort an issue I could not hear.

[tautech]

Their measurement only need be qualified by the distance to microphone measurement and I'm very sure Martin already knows how these things work.

Maybe it's no surprise but I have a calibrated microphone at home.
And access to a ISO calibrated Class 1 SPL meter..

Not that its needed for a comparative test. 
And a measuring tape I hope.

Anyways, sure lots better than Dave's ear.

[Njk]

Piss easy with a phone app like this:
Audio Spectrum Analyzer dB RTA

App is not a problem. The problem is to collect a relevant and stable data. For that, an acoustically representative model of typical work place will be required. With the microphones located into artificial head, etc. Environment is a critical thing. With existing technologies such as Audyssey, the results are relevant as long as the environment remains the same. When something changed, all the measurement needs to be performed again. And I'm not sure a single model will be enough (e.g. a server room is very different from a bed room). It's a lot of work to design and build that models.

[2N3055]

Whiny fans are just that: instead of producing wideband noise, they make sound that has peaks in spectrum that allows our ear to recognize tones.
If you have any tone between 400-4000Hz where our ears are most sensitive, you will hear annoying whine (tone). Simple as that. It can also be at other frequencies, but between 400-4000Hz we are very sensitive.

As long as you do it the same for all of them you will have good relative measurements. Then compare spectrum to how it sounds to you and you will quickly find what the "whine" is all about.

I did just that to few devices. One good example of device that is not completely silent but has very "noise like" sound is Keysight MSOX3000T. Just a whooshing, not very silent but with no distinguishing tones so no whining.. Excellent acoustic design.

Best one is, of course, Siglent SDS2000X HD. Pretty much completely silent. Large, low speed fan, and cooling based on low pressure/low speed airflow volume.
Siglent SDSD6000H12 Pro OTOH is quite louder. But that one needs to expel much more thermal energy out of the case... It is still not annoying because there are no large peaks inside so you forget about it, but after few hours, when you switch it off you suddenly realize how quieter the room is. Of course, that is my very quiet lab, in open space office not so much.

DHO800 has miniscule fan that pretty much guarantees whine and noise by itself. Offending tones will be easily measured.

[2N3055]

Hi folks,

No no, I don't have the scope yet, I also have to wait about 2 weeks for it....
This is already a placeholder or should give the possibility to consider it in advance, what I should test everything.
And compare, because I still have other scopes here....
First my own, quasi as a reference, the SDS2504X HD.
Then from work the Rigol DS1054Z and the Siglent SDS1104X-E.
The predecessor and the current competitor, so to speak.
Other existing measuring equipment:
- SDG2122X 120Mhz function generator.
-R&S SML01 1.1Ghz signal generator
-Tektronix A6302/AM502 current clamp combination
- Demoboards with several signals (and decoder functions) from Batronix and Siglent
- Deskew fixture from Siglent
- Analog 10Mhz test oscillator for ENOB determination
- Ultra low distortion 1khz sine generator (still to be tested)
- Leo Bodnar pulser 40ps risetime

So think about something, I'll get back to you here when the Rigol has arrived.

Martin,

in my opinion, testing the new DHO800 should be more about usability and how it can be used to measure than the basics of analog performance. It is already proven that Rigol made decent 12 bit converter and that it has low noise front end.

What is not clear is:
- overload recovery.
- distortion (two tone test)
- BW in sampling restricted modes (that is for DHO900 or DHO900 hacked 800) - what happens with 250 MHz BW when you drop to 312.5 MSps/s
 - What is utility of UltraAcquire mode, i.e. what can you do with it except replay it on screen. Does it have any functionality to compare it to other scopes segmented mode where all measurements, math decoding etc. works.
 - In a similar manner, what is utility of Record/replay mode, i.e. what can you do with it except replay it on screen. Does it have any functionality to compare it to other scopes segmented mode where all measurements, math decoding etc. works.
- Is FFT still half implemented where you cannot control directly FFT parameters? What can be controlled and how it impacts capability to measure
-It has 4 ch of math. What are capabilites? For instance, on Siglent you have arbitrary formula math with multiple sources (including reference, channels or other math channels). Micsig has something similar, slightly less capable. The way I see it on DHO800 math channel is single operator per channel (FFT and filtering are some of the operators) but that is it. If you want more sophisticated expression, you need to nest more math cahnnels. So if you want AxB than lowpass filter anything more than 10kHz and then do FFT of that, you need to nest 3 math channels. On SDS2000X HD (and incoming SDS100X HD) you can combine that expression (and even more) in single of the 4 math channels. This is not a show stopper, just to make sure we know how it works. 4ch of nested math on very inexpensive scope is still better than MSOX3000T that has same thing but only 2.....Just to put it in perspective..

I'm sure as you start using it you will find more  both good and bad stuff...

[Performa01]

What is not clear is:
- overload recovery.

Since this depends on the input buffer design and the crossover frequency particularly, it is not that easy to make a fair comparison. Chances are you test this at certain conditions (vertical gain, offset, signal frequency, percentage of overload) that make one scope look much better than the other, but there might be other conditions, where it is exactly the other way round.

Especially for a cheap scope like the DHO804, which has to make do with a single attenuator stage (by contrast, even bottom end SDS1000X-E has two attenuator stages!), things will only be comparable at a setting where no attenuator kicks in on either of the contenders. But even that could be unfair, because that point might be very different for a single or double attenuator design. And of course, the higher the sensitivity (lower V/div), the better the overload recovery will be. So it might be safest to do it at maybe 50 mV/div uniformely...

- distortion (two tone test)

People are often tempted to use the "Wave Combine" function of the SDG for this. Unfortuanetely, this generates some 3rd oder intermodulation products itself - at least at higher frequencies. So for this test, we need to use the two outputs of the SDG individually and have a resistive wideband power-combiner to create the test signal. This gives only 6 dB isolation between channels, but in my experience this is sufficient and no additional attenuators for the individual channels are required.

[mawyatt]

People are often tempted to use the "Wave Combine" function of the SDG for this. Unfortuanetely, this generates some 3rd oder intermodulation products itself - at least at higher frequencies. So for this test, we need to use the two outputs of the SDG individually and have a resistive wideband power-combiner to create the test signal. This gives only 6 dB isolation between channels, but in my experience this is sufficient and no additional attenuators for the individual channels are required.

Recall the Wave Combine on the SDG2XXX is pretty good, vaguely remember we measured such on the SSA3021X+, but may not be good enough for a 12 bit DSO. WRT to using separate AWG outputs, adding a 6~10dB pad in each AWG output before the resistive combiner should eliminate any potential IMD between the AWG channels, and still have enough signal level at the test DSO input.

Anyway, this new low end 12 bit DSO series from Rigol is creating lots of interest, and many like ourselves are anxious is see how well it behaves and performs. Rigol took a huge risk in attempting a 12 bit chip-set, especially after the disappointing 8 bit chip-set & front end performance (noisy), evidently learned from such and have produced apparently a good low noise front end to supplement the 12 custom ADC chip-set. The various tests mentioned by 2N3055, especially if performed by knowledgable unbiased folks will be highly beneficial in getting grips on just how good a bargain these new DSO will be

Best,

[TimFox]

I recently ordered a DHO914S, expected sometime in November.
I will see how well it works for harmonic distortion measurements, analyzing the "Monitor" output of my -hp- 339A analyzer (distortion products after removing the fundamental).

[Martin72]

Not that its needed for a comparative test. 
And a measuring tape I hope.
Anyways, sure lots better than Dave's ear.

I could take this to the extreme....
With a structure born noise test.
No, I will do the classical test.
Once SPL, which first reveals only the level.
Then with an RTA to capture the spectrum - and actually only that is interesting.
Because whether something is perceived as annoying depends on the frequency and that can only be captured with an RTA.
Then you have to accept that the actual test procedure does not say much.
Usually you don't have such conditions, but the scope is on the table and immediately there are boundary surfaces that reflect/amplify the sound of the fan.
I will play through the normalized case, i.e. as free standing as possible with few early reflections, then the practical case.
For the tests, I benefit from the fact that we regularly perform airborn noise tests at work and I measure loudspeakers in the room at home.

[Performa01]

Recall the Wave Combine on the SDG2XXX is pretty good, vaguely remember we measured such on the SSA3021X+, but may not be good enough for a 12 bit DSO. WRT to using separate AWG outputs, adding a 6~10dB pad in each AWG output before the resistive combiner should eliminate any potential IMD between the AWG channels, and still have enough signal level at the test DSO input.

Just to know the truth once and for all - and to have it documented, I've done a couple of tests to finally conclude that the internal "Wave Combine" feature unsurprisingly would be barely adequate even at low frequencies. Not to mention the catastrophy at high frequencies (Reply #460):

https://www.eevblog.com/forum/testgear/siglent-sdg6000-series-awg_s/msg5093877/#msg5093877

[Njk]

I'm not sure about the generalization. The fact that it does not work well with one particular instrument does not mean the whole idea of the internal combining is compromised. Perhaps generally it'll be wise to try the internal combining first and to take a more complex approaches only if necessary.

[Martin72]

Hi,

The scope arrived today, I had picked it up right after work - And of course had to switch on once....
Optical impression, it is "terribly" small, really tiny, as if half was forgotten.
And it smells like melting plastic, but you get used to it.
Turn on...
The fan is now not super loud, but has an annoying frequency, I had the scope now but not in the quiet study.
Boot time: 1min20sec. Edit: Time for the first boot, after this it´s about 50sec. ...
First technical impressions:

-Only one "normal" USB port, which I consider a big minus.
I could always swap between Bodnar and USB stick for the pictures earlier.
- Display is very sharp, you notice the higher resolution immediately, very pleasant, BUT:
The display itself is very reflective and not super bright, rather somewhat "dimmed". Edit: Now it seems brighter.
-Bodnar pulser connected via 50Ohm resistor, risetime is 3.6ns which makes something around 100Mhz, not bad.
Or if you put in the formula 0.45, you get exactly 125Mhz...Aha.
But I will check this correctly.
-Bodnar, the second:
Memory on automode, when the sample rate drops, the signal is already visibly changed.
As soon as a third channel is active, the sample rate has already reached the minimum of 312.5.
When you enable Ultraacquire, things get weird, but more on that later.
Oh, and it takes a long time until the scope reports the USB stick as "connected". OK, not sooo long..
That's it for now, more soon.

Martin

[ebastler]

Boot time: 1min20sec.

  That's even longer than the 47 seconds or so which Dave had reported! Maybe this was just the first initialization for the brand-new unit? Does it get any faster on subsequent boot cycles?

The display itself is very reflective and not super bright, rather somewhat "dimmed".

The photos show that you clearly need new (different) ceiling lights to work with that scope.

-Bodnar, the second:
Memory on automode, when the sample rate drops, the signal is already visibly changed.

Makes sense that one sees the effect of capturing fewer harmonics with the reduced sampling rate. But why does the curve appear noisier too? Is that an effect of phase fluctuations between the incoming pulses and the sampling clock?