EEVblog 1501 - Rigol HDO4000 Low Noise 12bit Oscilloscope Unboxing & First Impression

Rigol's new HDO4000 Series ultra low noise 12bit oscilloscope with Centaur chipset.
Unboxing, first impression, and noise measurements.

Support the EEVblog on:

Buy anything through that link and Dave gets a commission at no cost to you.

Donate With Bitcoin & Other Crypto Currencies!
#ElectronicsCreators #Rigol #Oscilloscope

[EEVblog]

5:50 I totally missed that bias is actually different to position offset. You can actually DC bias the input to any level required which is pretty cool.

[dm-brhv]

We got a new HDO4000 late Dec-22. We work with high-voltage power supplies and use a x100 probe to check the ripple voltages around 700-800V. Using our scope, we found that over 100VDC input (while AC coupled) the scope generates some proportional DC offset on the display. At 800V, the offset is so high that is impossible to check the ripple voltages around 200 mVpp using the AC coupling mode. It is currently with Rigol in the US to try and figure out the issue. I have a strange feeling that it may be some design issue, and that it is present on all the 4000 series scopes. The same HV probe works perfectly with our other scopes in the lab (Agilent, Lecroy etc.). Maybe Dave can perform some testing on his side?

[ptamog]

That average is not stopping. Its a sliding window of 1000 samples. Those stats are configerable and very nice.

[AmericanLocomotive]

It still blows my mind that Keysight are still the only ones who have figured out how to fully decouple the UI from the underlying data acquisition and math computations to keep responsiveness up. It just seems so janky that everything stops updating, and the whole UI slows down to what seems to be ~8 FPS when you move a waveform around. Especially considering a 10 year old Agilent/Keysight 3000 series keeps updating, and the UI remains silky smooth when moving things around.

We live in an era where even cheap bargain phones have 8 core processors. There's no excuse for slow janky UIs, especially when Keysight figured this out over a decade ago.

[bertblankenstein]

In all the excitement the scope did not get taken apart.

[padmanabhaprasannasimha]

This is a beautiful scope. I don't need anything more than 50MHz bandwidth for what I do but I need low noise, 12 bits and more channels. 8 channels would be nice. 16 would be perfect.

[vincei]

12 bits ... hmmm. Sounds nice. The wavy dots logo thingy next to the scope probe inputs looks kinda janky. If the base scope can be hacked to 800Mhz then this is definitely a winner ...

[AngDavies]

Bandwidth limit would reduce noise, just the thermal noise at 20mhz bandwidth corresponds to 4 microvolt rms noise absolute minimum and scales as the square root of bandwidth.

17 microvolt is actually fairly close to the absolute limit

[TheDefpom]

Rigol need to start sending me stuff to review… I’ve never reviewed a Rigol device, done plenty of Siglent stuff… cough cough 😉🤓

[movaxh]

I think the statistics top at 1000 count, because it is doing windowed statistics. That is it is showing avg, min, max, dev for the last 1000 measurements. This is actually pretty advanced feature, as most osciloscopes only show avg, min, max, dev since the stat reset. Ability to do windowed statistics actually require extra memory to do so. I am sure if you click on the measurements settings, you can change it down or up, or enable infinite count. In data sheet it says even "Statistical times settable". There is probably an upper limit (but I hope they have infinity, as that requires no extra memory), but probably common options like 100, 1000, 10k, 100k are available, as that can all be handled easily in few megabytes of memory in the SoC. It is not super complex feature, but even mid range scopes from R&H or Keysight, do not have that. Nice.

[todayonthebench]

To be fair, there is a lot of useful applications for a 12 bit scope.
But it would be nice if its DC gain accuracy were better than 2%. At least it is better than the DS1054z with its 4%.
Though, the R&S RTB2004 is down at 1.5%

Not that DC gain accuracy is the most critical spec for oscilloscopes. They don't replace multimeters by a long shot.
But sometimes measuring voltages accurately is important, and it would be nice to see a scope with a DC gain accuracy better than 1%. Even if it hurts bandwidth.

[albertluna]

Great video, greetings from Dominican Republic an electronics enthusiast!!!

[fullmoon]

Excited for the teardown, thanks Dave!

[gork]

The increase in sample rate corresponds to roughly an additional bit (ENOB) so comparing to the R&S is should be about 3 bits (8x) higher resolution measurement. The same goes to the 20mhz bw limit when increasing mV/div. The massive oversampling gives an insane boost to the dynamic range. I’d like to know the ADC they are using; I would be curious to build a radio with it if it’s a COTS part. Also wish you would have fed it some signals closer to its bw limit or at least above 20mhz… If you show it again, please consider roughing it up a bit more.

[davidgriffin]

10:45 Why is there a discrepancy between VRMS and ACRMS? For a frequency of f=7.1428MHz, a time period of t=1/f=140ns a Vpp = 185.02μV, giving Vp=92.51μV; VRMS =√2/2Vp=65.41μV (not 77.72μV as displayed) and discretising for n=100 (and i=1 to n) to get Vi = V(f, ti)=Vpsin(2πfti) then using the statistical formulae ACRMS gives a value of ACRMS=65.41μV agreeing perfectly with the simple hand calc; am I missing something here?

[AlexTaradov]

It stops as 1000 because it wants to use only 1000 last samples for averaging.

[OneBiOzZ]

Clearly some software bugs to work out in software but i am rather impressed and surprised by how well they implemented everything

[maxcode]

The declared 12 bit (or 16 bit in "High Resolution mode") is an illusion. Look at (always) the ENOB spec. ENOB is > 8 bit for HDO4000...so it's 8 bit!!! Never trust label attached by marketing people!!!!

[cavemaneca]

It's always great to see them improving their hardware and software, but I don't think this scope is really worth it unless you have a really niche application.

[nhand]

Classic RIGOL, affordable hardware and buggy software.