Rigol's New DHO800 Oscilloscope unbox & teardown

[asmi]

Excellent, looking forward to some impressions...
And good luck sorting things out!!

Thanks! BTW you can download their DSView software and play around with it a bit in a "demo" mode even without buying anything to get a better idea of what it's like.

[ebastler]

Rigol shall [...]
Rigol may [...]

The 10 Commandments, according to iMo. 
I sm sure Rigol will appreciate your instructions...

[Fungus]

I will buy one.

Which model?

[asmi]

Which model?

I'm aiming at DSLogic U3Pro16, I see no reason to pay extra $100 for 16 additional channels as I doubt I will ever use that many. My MSO has got 16 channels, and I don't remember ever using more than 9 (for 8 bit bus + clock signal). And U3Pro16 has a dedicated clock input, so I won't even have to waste an input for it in abovementioned case.

[TurboTom]

...

Your memory is like a fish: I've said many, many, many, many, many times in this thread that the DHO924 ought to turn on the 125Mhz limiter (which it already has internally) when there's more than 2 channels enabled.

Even this measure won't help. The DHO914S that now I've got an RMA# for, that I tested (successfully  ) for aliasing, should have the 125MHz LP filter enabled all the time, but to what avail with a 3dB bandwidth of 180~210MHz? I'm pretty sure the little, symmetric LC filters directly in front of the ADC inputs (apparently 24dB/oct jobbies) are configured with their stop-band close to the ADC's fundamental nyquist frequency, so probably somewhere in the 500...600MHz ballpark. The programmable low pass filters in Rigol's front-end chips may not be any steeper than 6dB/oct and thus provide just a gentle roll-off which makes them useful for model type limiting but definitely not for providing a "brick wall" against aliasing. So, once again, it's the user's job to understand and take preventive measures against aliasing. Or live with the wobbly slopes and pre- and overshoot of signals that contain frequencies above Nyquist...

That's nothing new, we had to live with that with the DS1000Z, all channels active, and I'm pretty certain (though I cannot proove it) that it's not much different with Siglent's entry level models. What disappoints me is that Rigol took that probem still to another level with the DHO900 model in combination with the digital inputs. On the MSO1000Z (Ds1000Z Plus?) they were at least so "honest" to "steal" two analog channels if the digital inputs were in use, so they could at least maintain minimum sample rate. But of course, this doesn't look too good in the advertisments and comparison charts...

[RAPo]

Great minds think alike. DHO942s is canceled, part of the funds is going to the U3Pro16 with additional shielded wires.

[Mechatrommer]

There is no scope (including ones from Rigol) that say it is 200 MHz scope and will drop sampling below Nyquist, not mention well below BW frequency....
Solution is to call it 70 MHz scope and say it has 100/200 MHz limited mode... And limit BW when it breaks basic sampling theorem..

And Nico's GW Instek was never released that way, it was after unofficial hack. GW Instek was honest there.
What is wrong with you? DHO900 is a bad design. Why don't you want Rigol to fix it? Why do you want that your fellow Rigol users to buy badly designed scope... You're not helping Rigol.. 

"bad" design is based on your definition. a good scope must not alias? i think you didnt catch my reply to Howardlong. you cant avoid aliasing. you can feed your pure 70MHz to it and set the time scale large enough so the sample rate will reduced badly, and you get aliasing from your 70MHz. trying to add auto attenuation, will probably complicate the front end or FPGA logic cells hence increase cost... and it would not avoid confusion anyway... for the rest of us, we would like Rigol to fully open the front end BW to 1.25GSa/s divide 2.5X = 500MHz (or maybe 250MHz for 2 channels probing at 625MSa/s) like the old high end Tek/HP DSO... and later limit it ourself internally or external circuit if we need to. real circuit is not just viewing pure sine wave, there are other things like verifying risetime, gate capacitance and signal integrity on digital signal, that will not cause aliasing. if Rigol fullfill your theoritical utopia desire, then we will call it bad/crippled design, they wont be able to satisfy everybody. because this is subjective, as people call it horses for courses.

[Fungus]

"bad" design is based on your definition. a good scope must not alias? i think you didnt catch my reply to Warhawk. you cant avoid aliasing. you can feed your pure 70MHz to it and set the time scale large enough so the sample rate will reduced badly, and you get aliasing from your 70MHz.

Yep.

All these 'scopes can reduce their sampling rate on demand (usually done automatically when you change the horizontal timebase) so they can all show aliasing.

[TK]

Which model?

I'm aiming at DSLogic U3Pro16, I see no reason to pay extra $100 for 16 additional channels as I doubt I will ever use that many. My MSO has got 16 channels, and I don't remember ever using more than 9 (for 8 bit bus + clock signal). And U3Pro16 has a dedicated clock input, so I won't even have to waste an input for it in abovementioned case.

If you ever need more than 16, you can buy an old HP/Agilent LA... they are bulky but can do 128 or more channels

[Mechatrommer]

DHO800 has it at higher (almost 180 MHz) and is also 12 Bit meaning it will alias on signal levels you would not have to worry with 8 bt scope...

So... just like many other oscilloscopes.
eg. Siglent SDS1204X-E has it at 230MHz and a 1GHz sample rate with all channels on.

says who? from reading,  SDS1204X-E has 2x ADC, turn on all channel, you get 500MSa/s. its 8-bit and double the price of DHO800. go on get it if that is what you want. try increase time scale if it can avoid aliasing, i bet not. i remember there is called hi-res capture, or ETS? to avoid such aliasing, i need to power up DHO804 again if i can find workaround. fwiw...

Great minds think alike. DHO942s is canceled, part of the funds is going to the U3Pro16 with additional shielded wires.

congratulation! thats another nearly a DHO804 cost https://www.dreamsourcelab.com/shop/ so do you now realize why we bought 804 instead of 814 or even 924s? we know Rigol's tradition for many years. ymmv.

[2N3055]

There is no scope (including ones from Rigol) that say it is 200 MHz scope and will drop sampling below Nyquist, not mention well below BW frequency....
Solution is to call it 70 MHz scope and say it has 100/200 MHz limited mode... And limit BW when it breaks basic sampling theorem..

And Nico's GW Instek was never released that way, it was after unofficial hack. GW Instek was honest there.
What is wrong with you? DHO900 is a bad design. Why don't you want Rigol to fix it? Why do you want that your fellow Rigol users to buy badly designed scope... You're not helping Rigol.. 

"bad" design is based on your definition. a good scope must not alias? i think you didnt catch my reply to Howardlong. you cant avoid aliasing. you can feed your pure 70MHz to it and set the time scale large enough so the sample rate will reduced badly, and you get aliasing from your 70MHz. trying to add auto attenuation, will probably complicate the front end or FPGA logic cells hence increase cost... and it would not avoid confusion anyway... for the rest of us, we would like Rigol to fully open the front end BW to 1.25GSa/s divide 2.5X = 500MHz (or maybe 250MHz for 2 channels probing at 625MSa/s) like the old high end Tek/HP DSO... and later limit it ourself internally or external circuit if we need to. real circuit is not just viewing pure sine wave, there are other things like verifying risetime, gate capacitance and signal integrity on digital signal, that will not cause aliasing. if Rigol fullfill your theoritical utopia desire, then we will call it bad/crippled design, they wont be able to satisfy everybody. because this is subjective, as people call it horses for courses.

It is BAD design. It is only scope that cannot satisfy (and by factor 3x less than needed) Nyquist including Hanteks, UNI-T, Owon and all other scopes that are even cheaper.

I will repeat:
If you enable all channels (which is why you bought 4 ch scope) it cannot satisfy Nyquist at ANY timebase..
If you enable MSO too, then sampling rate drops below 1/3rd of needed sample rate.. At any timebase.

I even provided images what happens with signal when you feed 1 and 16MHz squarewave with quite slow edges (2ns edges) that is slower than Atmel MCU in Arduino by factor of 2-3...

In that case scope is USELESS. Because it will show garbage on the screen.  It will be much worse than DS1000Z or DHO800 or and number of scopes that are very cheap and considered toys by many..

It is not about aliasing is not possible to happen with other scopes.
Problem is that on DHO900 you cannot make it do anything else but alias all the time, if you use 4ch  or MSO or God forbid be completely useless if you use both.. There is no setting that will NOT alias if you enable more than 2 ch and/or MSO.

So if you cannot use 4 ch or MSO, just buy DHO800 if you want to buy cheap 12 bit Rigol. That one works much better and is cheaper..



 

[wasedadoc]

.. real circuit is not just viewing pure sine wave, there are other things like verifying risetime, gate capacitance and signal integrity on digital signal, that will not cause aliasing.

You appear to be saying that aliasing is only a potential issue when viewing sine waves and that other input waveforms will not cause aliasing.  If that is indeed what you think, you are mistaken.

[Fungus]

It is BAD design. It is only scope that cannot satisfy (and by factor 3x less than needed) Nyquist including Hanteks, UNI-T, Owon and all other scopes that are even cheaper.

Yeah, yeah.

What If I'm into HAM radio and Arduinos?

I get a 200Mhz, 2-channel scope for my radio work and a 4 channel 'scope for Arduino work.

It's not BAD design. You really think they don't know about it and made a mistake? It's a deliberate design/marketing choice.

[Howardlong]

For the avoidance of doubt, for PC based LAs I use both the Analog Discovery and the Digital Discovery extensively: while the AD in particular is often deemed a "hobbyist" tool, in practice I wouldn't be without it. Both are worthy of bench positions, or, more accurately, worthy of hand baggage inclusion for field work - in fact I have three ADs for exactly this reason!

That explains a lot. Take a look at DSLogic U3Pro https://www.dreamsourcelab.com/product/dslogic-series/ This is a good example of what modern PC-based LA can do. And that is only using USB3 5Gbps, so with 10 Gbps you can double all numbers.

Yes, I'm aware of this unit. The specs on the Digital Discovery and the this are quite similar, same memory and 20% higher sampling speed like-for-like. I've been tempted before because of the probing it offers and USB 3 SuperSpeed would make longer captures faster, although I don't see any evidence that it offers SuperSpeed+.

How does the software compare to Waveforms? Waveforms is very much a key reason to use the Discoveries. I haven't seen any in depth reviews of these DSLogic U3Pro units, it seems to be a Chinese sourced device?

[2N3055]

It is BAD design. It is only scope that cannot satisfy (and by factor 3x less than needed) Nyquist including Hanteks, UNI-T, Owon and all other scopes that are even cheaper.

Yeah, yeah.

What If I'm into HAM radio and Arduinos?

I get a 200Mhz, 2-channel scope for my radio work and a 4 channel 'scope for Arduino work.

It's not BAD design. You really think they don't know about it and made a mistake? It's a deliberate design/marketing choice.

It is BAD design. It is design that is broken as a scope (if you violate Nyquist, digital scope DOES not work properly, mathematical fact) if you use all channels and MSO.  They could DISABLE use of 2CH when MSO is used and limit BW..

They know damn well what are they doing. It is not a mistake. It is decision for specifically type of users like you who think half made product is OK if price is right. They made what they market as cheapest 200MHz 12 bit 4ch MSO scope on the market. And when people point out it is not all of that at the same time answer is "well, what can you expect for the price you paid....."..

Literally go buy MSO5000. It is bit noisy but honest product. If you need 12bit analog you can get DHO800. It is very simple scope, but it is 12 bit and is also honest product and for even better price.

Make note here: despite of my preferences I'm not saying buy Siglent. Buy other Rigol, just not DHO900...
I have criticized Rigol for the state of development of their software. But on DHO800/1000/4000 they can potentially fix all software problems. I'm sceptical how much and how fast but that option is on the table, because DHO800/1000/4000 don't seem to have design problems with hardware (that we know of). DHO900 problems cannot be fixed by software. It is bad/wrong hardware design.
So I have no problems if people buy DHO800/1000/4000 because in theory they can be fixed with time if people are willing to wait.
But whoever buys DHO900 buys scope that has hardware specification problems. It will never be fixed.

[Mechatrommer]

What If I'm into HAM radio and Arduinos?
I get a 200Mhz, 2-channel scope for my radio work and a 4 channel 'scope for Arduino work.
It's not BAD design. You really think they don't know about it and made a mistake? It's a deliberate design/marketing choice.

the problem is, he is not HAM, so i think its wise to make this thread non-repetitive, otherwise aliasing will occur. here attached of Leo Bodnar pulse, i wish to see shorter rise time, no aliasing guaranteed!

[switchabl]

What makes you think you can trust the rise time measurement in the presence of aliasing?

EDIT: Ok, I guess we are back with the 804 (and single channel). I was assuming this was still about

real circuit is not just viewing pure sine wave, there are other things like verifying risetime, gate capacitance and signal integrity on digital signal, that will not cause aliasing.

[Fungus]

the problem is, he is not HAM

Not necessarily "HAM" but anything where you want to connect some coax and have maximum possible bandwidth.

It is BAD design. It is only scope that cannot satisfy (and by factor 3x less than needed) Nyquist including Hanteks, UNI-T, Owon and all other scopes that are even cheaper.

As noted earlier: I can make your favorite oscilloscope alias with a 1kHz sine wave. All I have to do is fiddle with the settings.

(nb. I've actually done this...)

If I suspect a signal is aliasing on my DHO900? I can turn a channel or two off and look at the signal with 500Mhz or 1Ghz sample rate. It only takes a second. If the frequency jumps around then, yep, it was aliasing.

[Fungus]

What makes you think you can trust the rise time measurement in the presence of aliasing?

Simple: He only has one channel enabled!

[2N3055]

What If I'm into HAM radio and Arduinos?
I get a 200Mhz, 2-channel scope for my radio work and a 4 channel 'scope for Arduino work.
It's not BAD design. You really think they don't know about it and made a mistake? It's a deliberate design/marketing choice.

the problem is, he is not HAM, so i think its wise to make this thread non-repetitive, otherwise aliasing will occur. here attached of Leo Bodnar pulse, i wish to see shorter rise time, no aliasing guaranteed!

I'm talking (and I said many many times) about DHO900.
I said DHO800 will work fine...

But you have to push your agenda..
Good for you.

[2N3055]

the problem is, he is not HAM

Not necessarily "HAM" but anything where you want to connect some coax and have maximum possible bandwidth.

It is BAD design. It is only scope that cannot satisfy (and by factor 3x less than needed) Nyquist including Hanteks, UNI-T, Owon and all other scopes that are even cheaper.

As noted earlier: I can make your favorite oscilloscope alias with a 1kHz sine wave. All I have to do is fiddle with the settings.

(nb. I've actually done this...)

If I suspect a signal is aliasing on my DHO900? I can turn a channel or two off and look at the signal with 500Mhz or 1Ghz sample rate. It only takes a second. If the frequency jumps around then, yep, it was aliasing.

You can also smash it with a hammer...It will also make oscilloscope malfunction..

Any digital scope can be made to alias.
But Rigol made only scope that is guaranteed that it will alias if you try to use it for what it is.

But hey, Mecahtrommer and you are fighting so hard for poor Rigols right to sell bad design to others...
More power to you..
Whoever thinks they know better should buy whatever they want.

I said my peace..

[nctnico]

Sorry, but you are making way too much fuss about this. My good old Tek 2230 sampled at 20Ms/s while it had a 100MHz bandwidth. And if you look a bit harder you can probably find more modern examples as well. Nothing is hidden by Rigol but the scope will not work at multiple extremes of the specifications; the user will have to choose. Nothing new here and personally I like equipment that lets me choose which extreme of the specs I need. I have a 60V 50A power supply on my desk. But it won't do both at the same time as the power is limited to 1000W. It still is a good power supply because sometimes I need the high voltage, sometimes I need the high current.

[Mechatrommer]

I'm talking (and I said many many times) about DHO900.
I said DHO800 will work fine...

point understood, thats why i tend to stop arguing... you probably right, but imho DHO900S is just a bit step up of DHO800 (50Mpts and 200MHz uncrippled) with some toy LA added. but when you activate LA and all channels and expect it to show no aliasing, i already said much earlier, the right tool is probably in $2-5K range. and at that setup, i'm not sure what you are expecting to see on analog channels while probing digital bus. you must be doing complex project, then again, its advisable to get $2-5K MSO. you cant expect too much from shared ADC front end... but the good thing imho is built in monitor and some functionalities of LA, no need PC. ymmv.

What makes you think you can trust the rise time measurement in the presence of aliasing?

Simple: He only has one channel enabled!

its not about how many channels enabled, its about being aware at what sample rate you are probing, at every dso setup you've made. and know whether aliasing is possible when probing your dut. how do i know no aliasing in this case? i only rely on report for the Pulser module that has been tested with sampling (ETS) head/scope by Leo Bodnar.. other than that, there is not much i can tell, when we exceed nyquist limit, we can only guess... its that simple.

[Mechatrommer]

DHO800 has it at higher (almost 180 MHz) and is also 12 Bit meaning it will alias on signal levels you would not have to worry with 8 bt scope...

So... just like many other oscilloscopes.
eg. Siglent SDS1204X-E has it at 230MHz and a 1GHz sample rate with all channels on.

...i remember there is called hi-res capture, or ETS? to avoid such aliasing, i need to power up DHO804 again if i can find workaround...

got it, its called "Peak" acquisition mode... DHO800 has it like the previous DS1000Z/E it removes aliasing. https://www.tek.com/en/support/faqs/what-acquisition-modes-are-available-most-tektronix-oscilloscopes

[switchabl]

Be aware that peak detect is only effective if the ADC (and peak detector) is running fast enough. The same is true for high-res and random sampling. On the 924 this would not be the case anymore with 4 channels enabled. A switchable (analog) low-pass filter would certainly help, as would ETS (for periodic signals). I do not know if either is available on the Rigol.