Rigol's New DHO800 Oscilloscope unbox & teardown

[TomKatt]

Please watch this three times before posting anything like that again:

OT - that video is the absolute best explanation of bit depth I have ever encountered.  Although the example is based on audio signals, it equally applies to any AD/DA signal process such as modern digital oscilloscopes.

Thanks much for that link !!!

[NE666]

Considering that scopes are not precision measurement devices, and seem to be most useful for visually identifying signal issues and time based analysis, I have to admit that the whole 12bit vs 8bit thing has me confused. 

It may help to consider that very often, we're using oscilloscopes to make *relative* measurements i.e. how does one signal change with respect to another e.g. filter performance, rather than against a fixed external reference.  Higher resolution (accompanied by an appropriately low noise floor) allows us to be more certain about that change.  Consider the difference between two cheap plastic office rulers, one marked in mm, the other only in cm.  Both will probably fail miserably on absolute accuracy when compared to the SI meter reference.  However the former allows us to be much more certain about how large or small the length of one object is, relative to another.  Which comes into play for accuracy of things like frequency response curves, FFT etc.

However, in the context of this particular thread and the product range under discussion, I'm inclined to agree - few people will be purchasing these purely on the grounds that they are 12-bit scopes and that being all they can accept.  It's a combination of price point, form factor, and 12-bits being not inherently worse than 8. And as others have already said, there are still compromises made for this.  The horizontal resolution of the 8-bit MSO5000 series is 8x times that of the DHO800/900 for the analogue channels and ~2x for the LA section, albeit a ~30% price premium.

 

[TomKatt]

Considering that scopes are not precision measurement devices, and seem to be most useful for visually identifying signal issues and time based analysis, I have to admit that the whole 12bit vs 8bit thing has me confused. 

...Which comes into play for accuracy of things like frequency response curves, FFT etc.

I had overlooked measurements based on sample data beyond simply displaying the waveform on the screen.  I was hung up on bit depth vs screen resolution, where the benefit might be relegated to zooming into the waveform.  Indeed, further analysis such as FFT etc would likely improve with additional resolution.

Thanks everyone for the insight.

[mwb1100]

However, in the context of this particular thread and the product range under discussion, I'm inclined to agree - few people will be purchasing these purely on the grounds that they are 12-bit scopes and that being all they can accept.  It's a combination of price point, form factor, and 12-bits being not inherently worse than 8. And as others have already said, there are still compromises made for this.

For sure!

One thing that seems to often get overlooked when talking about the relative value of the DHO800 is that it is in the price range very close to the entry level scope such as the DS1054Z.  When the DHO800 is widely available, only the absolutely most price sensitive buyer should even think about the DS1054Z.  Even aside from the 8 bit/12 bit difference, the DHO800 offers quite a bit over the DS1054Z:

  - touch screen
  - much more compact
  - VESA mount
  - as good or better in all specs (?) compared to DS1054z
  - probably hackable to 200MHz (I think this has already been done?)
  - web-based remote control
  - multi-pane windows

There might be reasons someone would consider the SDS1014X-E (Bode plot, LA and AWG options, arguably better decode, WiFi), but the DHO800 should still stack up favorably for many - if not most - people who would look at the SDS1014X-E.

I'm sure Rigol will be selling a ton of DHO804's

[dreamcat4]

One of the things I hope that the early adopters will try to more fully explore is the mobile / battery powered / isolated (but not floating, tied to DUT reference PD).

And see how useful that is. For automotive and also for PSU or other power oriented applications. (In combination with diff probes too when applicable).

The reason I say this is because to justify this scope's place / existence for users who already own the higher models DHO1000, or DHO4000. Which are not capable of being battery powered.

So in that context then it's becomes more a question of whether adding an 800/900 (as extra scope) is really justifiable. To get out for those specific scenarios. Makes sense?

[edit] BTW remembering that this would then also be a capability for budget users that isn't available with the competitor scope SDS1014X-E. Which might also be a value to some? ... in particular if you also (even as a budget user) maybe wants to works on vehicles, PSU or other things? Makes sense?

[madires]

The latter is the benefit of 24bit audio, especially for classical music (if done right). At the same time it's not hard fit less demanding music into 16 bits without clipping, but some studios manage to screw up even that (loudness wars). More bits can help if used properly.

Please watch this three times before posting anything like that again:

That's what I've written about 16bit vs 24bit audio:

The latter (dynamic range) is the benefit of 24bit audio, especially for classical music (if done right). At the same time it's not hard fit less demanding music into 16 bits without clipping, but some studios manage to screw up even that (loudness wars).

And that's the conclusion of a higher resolution in general:

More bits can help if used properly.

And you made that into: madires said that a higher resolution for audio is always better. But that's not what I've meant. Please read again!

[Fungus]

I'm inclined to agree - few people will be purchasing these purely on the grounds that they are 12-bit scopes and that being all they can accept.  It's a combination of price point, form factor, and 12-bits being not inherently worse than 8.

Check out the pics Simone posted in the "Pre sales" thread.

Who wouldn't want one...?

(Yeah, I know, there's a lot of curmudgeons here...  )

[2N3055]

Even DHO4000 has only 2% ...

You can probably do better than that if you warm it up properly and do a self-cal just before the measurement. 

Well I can attest that my SDS2000X HD is better than 0.1% but I did not talk about that .

Siglent is guaranteeing  0.5% or better. On every single one.
Rigol is guaranteeing  2% for same circumstances.

It is about instrument class, components used, quality control and standing behind your product.

Nothing is wrong with Rigol's 2%. It's plenty enough for most uses.
But when people compare prices, those are also details that might be interesting for fair comparison.

Because of that, I literally use my multimeters much less for basic voltage measurements. I use scope as a multimeter..

[Fungus]

That's what I've written about 16bit vs 24bit audio:

The latter (dynamic range) is the benefit of 24bit audio, especially for classical music (if done right).

There might be some benefit for recovering from a bad recording setup but there's no benefit for playback.

The latter (dynamic range) is the benefit of 24bit audio, especially for classical music (if done right). At the same time it's not hard fit less demanding music into 16 bits without clipping, but some studios manage to screw up even that (loudness wars).

[/quote]

"Less demanding" music can fit into 8 bits.

This site has some audio tests that might interest you: https://www.audiocheck.net/blindtests_16vs8bit.php

[TomKatt]

There might be reasons someone would consider the SDS1014X-E (Bode plot, LA and AWG options, arguably better decode, WiFi), but the DHO800 should still stack up favorably for many - if not most - people who would look at the SDS1014X-E.

I'm sure Rigol will be selling a ton of DHO804's

Agreed.  I think an influencing factor for those who frequently use multiple channels at high frequencies will be the DHO800's single shared ADC that reduces sampling rate as additional channels are enabled.  Though I think consumers have been spoiled with modern scopes that provide so much value for the money we tend to purchase 4 channel models when really most people only need 2.  I got my Siglent SDS1104X-E over the 2 channel version primarily because the price difference for the additional 2 channels was rather insignificant in the scheme of things, and some features like Bode Plot were only available on the 4 channel models.

[Fungus]

I think consumers have been spoiled with modern scopes that provide so much value for the money we tend to purchase 4 channel models when really most people only need 2.

I wouldn't say "most" - we live in a very digital world.

I got my Siglent SDS1104X-E over the 2 channel version primarily because the price difference for the additional 2 channels was rather insignificant in the scheme of things

Exactly!

Me? I can always find uses for extra channels, eg. for software debugging on my Arduino (I can toggle IO pins to show which code path was taken, if a variable passes a threshold value, etc...). I also use the 'scopes pulse width measurements for timing sections of code, interrupt handler response times, etc. It's a big help when optimizing.

I can't say I've ever needed more than four channels but having only two would be a big handicap.

[UK]

Considering that scopes are not precision measurement devices, and seem to be most useful for visually identifying signal issues and time based analysis, I have to admit that the whole 12bit vs 8bit thing has me confused. 

However, in the context of this particular thread and the product range under discussion, I'm inclined to agree - few people will be purchasing these purely on the grounds that they are 12-bit scopes and that being all they can accept.  It's a combination of price point, form factor, and 12-bits being not inherently worse than 8. And as others have already said, there are still compromises made for this.  The horizontal resolution of the 8-bit MSO5000 series is 8x times that of the DHO800/900 for the analogue channels and ~2x for the LA section, albeit a ~30% price premium.

Should we really take the MSO5000 lineup as a compromise? 8x times higher horizontal resolution but such a high floor noise and 4x worse vertical resolution than DHO800/900.

[TomKatt]

I think consumers have been spoiled with modern scopes that provide so much value for the money we tend to purchase 4 channel models when really most people only need 2.

I wouldn't say "most" - we live in a very digital world.

I got my Siglent SDS1104X-E over the 2 channel version primarily because the price difference for the additional 2 channels was rather insignificant in the scheme of things

Me? I can always find uses for extra channels, eg. for software debugging on my Arduino (I can toggle IO pins to show which code path was taken, if a variable passes a threshold value, etc...). I also use the 'scopes pulse width measurements for timing sections of code, interrupt handler response times, etc. It's a big help when optimizing.

I can't say I've ever needed more than four channels but having only two would be a big handicap.

This is where a low cost logic analyzer can help as well.  Something like the DS Logic Plus can be a better investment than adding the LA option to many scopes, and even an $8 clone with Sigrock can sometimes provide more functionality than the LA add-on.  If you need shared timebase, you can run the scope trigger to the external LA.  I find my DS Logic much easier to work for serial decoding than what can be captured on the scope screen of my Siglent.  You could easily use an external LA to time things like uc io pins.

But certainly more channels are always better if the price isn't much different - you will always find a use for them.  And also why the DH800 not offering these options doesn't have to be a dealbreaker.

[mzzj]

There might be reasons someone would consider the SDS1014X-E (Bode plot, LA and AWG options, arguably better decode, WiFi), but the DHO800 should still stack up favorably for many - if not most - people who would look at the SDS1014X-E.

I'm sure Rigol will be selling a ton of DHO804's

DHO914S probably gives SDS1014X-E run for the money: SDS1014X-E +WFG+software is only little bit cheaper than DHO914S
SDS1014X-E +WFG+logic module+2x software is actually more expensive(1197 eur) than DHO914S+logic module. (1186eur) 

[nctnico]

Though I think consumers have been spoiled with modern scopes that provide so much value for the money we tend to purchase 4 channel models when really most people only need 2

That depends on what you are doing. If your work involves microcontrollers or generic digital stuff then having at least 4 channels is a must.

[NE666]

Should we really take the MSO5000 lineup as a compromise? 8x times higher horizontal resolution but such a high floor noise and 4x worse vertical resolution than DHO800/900.

In the real world, the noise floor of the 5k is of no consequence for many common problem domains.  For example, if you're mainly working with microcontrollers, high speed logic or inherently noisy digital environments such as automotive/industrial CAN and LIN. It's *probably* also of no real consequence to the average hobbyist and their Arudino either, save for it being repeated on these forums ad nausea.

That's not to say that the DHO is bad, or worse, or better.  Just a different proposition, at a different price point.

Horses for courses.

[nctnico]

Let me kill that dream: at some point you will want to measure some analog signals or just measure the level of a digital signal and in those situations noise just gets in the way of getting work done. Been there, done that. If the same money can buy you a lower noise oscilloscope, then buy the lower noise oscilloscope.

[iMo]

Does it mean with the single ADC chip inside the sampling rate 1.25Gs will drop to ~300Ms per channel when using all 4 channels? What would be the impact when hacked to 200MHz BW, for example?
Or the sampling rate is hackable too?

[NE666]

Let me kill that dream: at some point you will want to measure some analog signals or just measure the level of a digital signal and in those situations noise just gets in the way of getting work done.

Hasn't happened yet but if it should, there will be a DHO800 hanging off my VESA mount.

[mwb1100]

Does it mean with the single ADC chip inside the sampling rate 1.25Gs will drop to ~300Ms per channel when using all 4 channels? What would be the impact when hacked to 200MHz BW, for example?
Or the sampling rate is hackable too?

The 250MHz DHO924 samples at 312.5 MSa/s for 3 or 4 channels enabled.

[Veteran68]

I'm still a Siglent fan and my SDS1104X-E still fits my modest needs, but if Siglent doesn't drop the 1000HD line in NA soon with similar or better features at a comparable price point (like within 120%), I may be looking to add my first piece of Rigol gear to the collection. If I do I'd want the 900S series to get all the possible hardware options (plus, I'm apparently in the minority by really liking the black case, though it's not going to match my other white gear). I'm not averse to hacking -- my 1104 is fully upgraded -- but here given the price difference is only $100 MSRP between the 914S and the 924S, I think I'd probably just bite that bullet, assuage my conscience and peace-of-mind, and spring for the fully loaded 924S and be done with it. It'll be doing my part to help Rigol subsidize all you pirates. 

[Martin72]

Does it mean with the single ADC chip inside the sampling rate 1.25Gs will drop to ~300Ms per channel when using all 4 channels? What would be the impact when hacked to 200MHz BW, for example?
Or the sampling rate is hackable too?

All models are having the same ADC.
Top model DHO4000 have two ADCs, min/max Samplerate is 1/4GSa/s.
DHO1000 have one ADC, min/max Samplerate is 500M/2GSa/s , which makes sense when only one is for all channels.
But DHO800/900 also have one ADC, the same ADC - So why don´t they have the same samplerates like the DHO1000 ?
Different FPGA, 800/900 are using a different one comparing to the 1000/4000.
Maybe this is the reason.

[iMo]

Does it mean with the single ADC chip inside the sampling rate 1.25Gs will drop to ~300Ms per channel when using all 4 channels? What would be the impact when hacked to 200MHz BW, for example?
Or the sampling rate is hackable too?

The 250MHz DHO924 samples at 312.5 MSa/s for 3 or 4 channels enabled.

Hmm, that is interesting. From some discussions in past I learned (perhaps I misunderstood something) that the sample rate has to be at least 4-5x higher than the BW (and say only 2.5x for >1Ghz BWs)..

[2N3055]

Does it mean with the single ADC chip inside the sampling rate 1.25Gs will drop to ~300Ms per channel when using all 4 channels? What would be the impact when hacked to 200MHz BW, for example?
Or the sampling rate is hackable too?

Sampling rate is not hackable (or at least nobody managed to do it so far.. because nobody really has it yet)...
ADC is specified at 1.25GSp/s, it might be that it is hard clocked as such, or maybe even that maybe ADC could but FPGA could not cope with data rate. Until there are thousands of them "in the wild" and some capable people work on it, we wont know.
So far, based at what we know now, even when hacking it I would not "open" BW to more than 100MHz. Even at 100 MHz set, BW is probably more than 100MHz, and then signal does not just die suddenly. It gets progressively weaker.
Problem is that with 8Bits, signal soon drops below ADC resolution and you are all right. With 12 bits, that doesn't happen until higher BW, so you would have more aliasing than similar 8bit scope.  In 4ch mode aliasing start at 156,25 MHz.
If they are clever, they decrease BW to 100MHz MAX automatically in 4CH mode...

[2N3055]

Does it mean with the single ADC chip inside the sampling rate 1.25Gs will drop to ~300Ms per channel when using all 4 channels? What would be the impact when hacked to 200MHz BW, for example?
Or the sampling rate is hackable too?

The 250MHz DHO924 samples at 312.5 MSa/s for 3 or 4 channels enabled.

Hmm, that is interesting. From some discussions in past I learned (perhaps I misunderstood something) that the sample rate has to be at least 4-5x higher than the BW (and say only 2.5x for >1Ghz BWs)..

It depends on type of antialiasing filter used . If filter is really sharp, 2,5X is starting point where it works. If you are using more traditional filter with Gausian response (slower roll off), then 4X or 5X might be needed.