Hacking the Rigol DHO800/900 Scope

[AndyBig]

I'm sorry if it sounds so harsh: This is a very very bad idea!
Like 20% of the male population, I have a red/green visual impairment. I can hardly distinguish between light green and yellow on a screen, for example, and if it's only a few pixels (as with written text, for example), I even can't tell whether something is red or green.
With the specification C1 ... it is usable, without it is useless.

Yes, I already understood that. Now I'm thinking about the idea of ​​making a switchable view - a wider panel with channel names, and a narrower one with channels designated only by colors.

[RAPo]

I was a bit playing. I saw that in fullscreen mode the measurement bleeds into the x-axis. Could that be pumped up a little higher?

[iMo]

What we should see as the noise background?
When I terminate ch4, for example, with 50ohm terminator and I look at the noise I see 340uVpp BW OFF, and 225uVpp BW 20MHz, DC at 500kSa/s and 1Mpts (after upgrade to 1.04).

Peak-to-peak is not a very meaningful or comparable metric for Gaussian-like noise because the probability distribution is unbounded. Instead, measure the standard deviation (some scopes call it AC RMS).

BW 20MHz     Avg. 39uV AC RMS, drops down to 33uV after some time (due to Avg math?)

BW OFF         Avg. 60uV AC RMS, drops down to 50uV after some time (due to Avg math?)

FYI - after the 824 mod:

DC at 500kSa/s and 1Mpts

BW 20MHz     Avg. 53.6uV AC RMS

BW OFF         Avg. 110uV  AC RMS

DC at 500kSa/s and 100kpts

BW 20MHz     Avg. 24.7uV AC RMS

BW OFF         Avg. 49.7uV  AC RMS

[AndyBig]

I was a bit playing. I saw that in fullscreen mode the measurement bleeds into the x-axis. Could that be pumped up a little higher?

Yes, I have already raised the bottom edge of the results above the X-axis. I just haven't posted this updated version yet, I want to resolve the issue with the color coding of the channels in the results, based on the fair comments made here.

[AndyBig]

FYI - after the 824 mod:
DC at 500kSa/s and 1Mpts
BW 20MHz     Avg. 53.6uV AC RMS
BW OFF         Avg. 110uV  AC RMS
DC at 500kSa/s and 100kpts
BW 20MHz     Avg. 24.7uV AC RMS
BW OFF         Avg. 49.7uV  AC RMS

Also 824 mod, input locked by 50 Ohm load.
DC at 500kSa/s and 1Mpts
- BW 20MHz     Avg. 23.604uV AC RMS
- BW OFF         Avg. 49.453uV  AC RMS
DC at 500kSa/s and 100kpts
- BW 20MHz     Avg. 23.553uV AC RMS
- BW OFF         Avg. 49.433uV  AC RMS

[iMo]

Isn't your original one the 814?? (mine is 804)..

PS: with 2.50MSa/s and higher (w/ 1Mpts and BW OFF) it is 50uV AC RMS Avg..
Below 2.50MSa/s 90-100uV AC RMS Avg..
The minimum is at 156.25MSa/s - 41.3uV AC RMS..

[AndyBig]

Isn't your original one the 814?? (mine is 804)..

Yes I have the original 814 changed to 824.

PS: with 2.50MSa/s and higher (w/ 1Mpts and BW OFF) it is 50uV AC RMS Avg..
Below 2.50MSa/s 90-100uV AC RMS Avg..

Are you closing the input with a load or a shorted connector?

[iMo]

With the 50ohm load (a terminator).

[AndyBig]

Directly at the oscilloscope input, without a probe?
Is the vertical limit 500 mV/div?

[iMo]

Sure..

[iMo]

I've returned the vendor.bin back to 804 and now I see 72uV/47uV AC RMS (no selfcal yet).. Weird..
It could be some parts degraded during the first 2weeks of operation ..
PS: tried with 4 different terminators.. the same RMS..
I will do the selfcal with the 804 and we will see..
BTW I did the mod with "-M DHO824" - hopefully I have not messed up something..

[AndyBig]

Then it's strange, I don't remember anyone reporting such a high level of intrinsic noise. The only assumption left is that something is making noise - either the power supply, or something connected to the oscilloscope (if something is connected), or something near the oscilloscope.
Have you looked at the FFT - are there any peaks that stand out?

[iMo]

..I will do the selfcal with the 804 and we will see..

So, I did the selfcal and let it running for some time - the values are now at the level as it were 18days back.. Weird..

[iMo]

It could easily be (my bet actually) they are selecting/binning the input AFE/ADC chips (or other components there) on noise and thermal coefficients. For the 804 models they might use the lowest grade.

With higher speeds/BW settings (like when modding to the higher models) the 804 components get noisier and/or more prone to the TC of the various offsets.

Therefore, imho, they allow the 250uV lowest V/div with the higher models (because they use chips/components with the lower noise and less TC). And all looks the same on the pcb of all the models, incl. the values of components.. 

PS: you may try an another experiment which may support my suspicion - simply let run the scope for couple of hours and make the selfcal. Then let the scope cool down for a day or so and after the powering it on look at the 500uV range (DC) with 50ohm terminator in the BNC. Look at what the initial offset and noise will be and how it moves/changes with time (as the scope heats up)..

[AndyBig]

Therefore, imho, they allow the 250uV lowest V/div with the higher models (because they use chips/components with the lower noise and less TC).

The physical lowest vertical limit for all models is 1 mV/div. 500 and 200 µV/div is software scaling.

[iMo]

Fungus published here an adb script/batch for changing the vendor.bin he is using by clicking on it (off his PC).
Would it be possible to do it directly in the scope (no adb needed)??

[ebastler]

It could easily be (my bet actually) they are selecting/binning the input AFE/ADC chips (or other components there) on noise and thermal coefficients. For the 804 models they might use the lowest grade.

Hmm... I struggle to imagine Rigol going to that length in a mass-manufacturing process. (Or Siglent for their competing entry-level scopes, for that matter.) Given that they do not even specify the noise in the datasheet, why would they bother? 

[iMo]

The selection could be ("is") done by the chip manufacturer (or a party which makes the packaging of the dies), it may take couple of seconds, full automated. They may do it because their chips are their custom made design and they certainly want increase yield/profitability for any price..
Mind the zillion megatons of fake chips/parts coming from that area via ali/ebay - and they indeed bother to mess with the faking of couple of cents worth parts..

[Fungus]

Nah. "Binning" is largely a myth.

The percentage of chips that would fit that category ("works at one speed but not another") would be tiny and simply not worth the logistical problem it would create. It's better to improve your process and produce more working chips instead.

They don't manufacture separate batches of each model, they manufacture "working PCBs" and put in different SD cards depending on the incoming orders.

Tests will be pass/fail and the bad chips thrown out.

(dons flame-proof undies).

[Fungus]

Higher bandwidth always means higher noise. That's just the way it is.

Noise:bandwidth ratio is probably non-linear as you go towards the limits of a device.

[Fungus]

nb. "Binning" does happen in things like multi-core CPUs where individual cores can be defective and the chip is designed for disabling of individual cores by soldering little configuration resistors to the back of it.

By "speed-grade" though? Not as much as people imagine.

[iMo]

Their custom chips were not designed for "1.25GSa/s", sure. They targeted at least 2GSa/s. So they had/have issues with their design or/and technology, thus, imho, they do the binning.
Moreover I doubt the companies like Rigol (a rather small company) might afford itself the selecting of an already populated board like "pass/fail(trashed)".
Anyhow, yes, a higher BW needs a different settings of the AFE and its noise and drift will most probably increase..

[iMo]

FYI - my small hw mod.. 

[Njk]

The physical lowest vertical limit for all models is 1 mV/div. 500 and 200 µV/div is software scaling.

BTW, in Rigol 1054Z, which uses 8-bit ADC, I once tried to enable the 500 uV/div scale. It's interesting the noise at that scale is not higher than at 1 mV/Div. Both the scales use digital zoom and at 500 uV/Div the trace looks more step-like, but the noise level is the same, perhaps due to careful SW processing.

[ebastler]

BTW, in Rigol 1054Z, which uses 8-bit ADC, I once tried to enable the 500 uV/div scale. It's interesting the noise at that scale is not higher than at 1 mV/Div. Both the scales use digital zoom and at 500 uV/Div the trace looks more step-like, but the noise level is the same, perhaps due to careful SW processing.

If you are referring to the noise quantified in e.g. µV rms, why would it be any different? It's the exact same data, acquired with the same amplification setting. The data are just rendered differently (namely zoomed-in) on the screen.