Hacking the Rigol DHO800/900 Scope

[AndyBig]

I think 1.4 GSa/s or 1.6 GSa/s is more real limit of overclocking FPGA and DDR3 memory.

I think that raising the clock of the FPGA even by 10-15% will be problematic.
That is, maybe some oscilloscopes will work, but it is unlikely that it will be stable.

[Fungus]

The DHO8/9 and DHO1000 hardware looks the same, but the DHO1000 is 2GSa/s and the DHO8/9 is only 1.25GSa/s, so it would be nice to switch to 2GSa/s sampling as well

It's not just the ADCs, it's the FPGA, the RAM, many dozens of places in the firmware that do math based on the sample rate, etc.

[AndyBig]

It is enough that the 1000 and 800/900 have completely different FPGAs. This immediately puts an end to any analogies between these oscilloscope lines.

[LEER333]

The main reason is that the FPGA configuration has been reduced. thank you

[mrisco]

Next version will have a docked result panel (can be undocked to the former view) and a surprise I can overcome all obstacles

[shapirus]

I wonder if it's feasible to make it display the math traces in the main window.

[mrisco]

I wonder if it's feasible to make it display the math traces in the main window.

I try to do it, but there is a lot of changes to do in the apk and maybe the precompiled binary library draws inside of a window with a predefined id. In other models that is possible, but fight with a binary library may not be worth it.

[RAPo]

@mrisco and @AndyBig,

Both of you are making this scope so much better, thanks for that.
My question is: are you two joining forces, because I sure would like to see all your great adaptions in one package!

[Fungus]

I wonder if it's feasible to make it display the math traces in the main window.

I try to do it, but there is a lot of changes to do in the apk and maybe the precompiled binary library draws inside of a window with a predefined id. In other models that is possible, but fight with a binary library may not be worth it.

I think that would be very difficult. You might even need to have the math window open anyway to be able to put the trace in a different window.

[mrisco]

I could open the window and hide it, but the problem is the trace drawing itself.

[iMo]

FYI - 804 modded to 824, no selfcal after mod.
50ohm terminator on CH4.
Waveform view.

BW OFF       49.4uV Avg. AC.RMS  (+12uV against 804)
BW 20MHz   24.1uV Avg. AC.RMS  (the same as the 804)

[AndyBig]

BW OFF       49.4uV Avg. AC.RMS  (+12uV against 804)

Of course, because the bandwidth has expanded almost 3 times

[iMo]

Yep, the noise looks ok.
The model is 824 and time base 2ns.

BUT - I still see only 25M mem depth and the options - see below.
What went wrong?
vendor.bin generated by -o -M DHO824

[AndyBig]

BUT - I still see only 25M mem depth and the options - see below.
What went wrong?
vendor.bin generated by -o -M DHO824

This is because you have two channels active - channel 2 serves the trigger and channel 4 shows the signal.

[iMo]

The same with trigger at the same channel.. 25MB max..
OK, the counter and measurement on a diff channel have to be OFF, and I get 50MB.. 

What about the options? Is that ok?

PS: why the counter enabled on a different channel limits the max mem depth ? 

[AndyBig]

What about the options? Is that ok?

Yes

PS: why the counter on a different channel limits the max mem depth ? 

Any access to the channel puts it into full operation, so it takes both the sampling frequency and memory.

[AndyBig]

My question is: are you two joining forces, because I sure would like to see all your great adaptions in one package!

mrisco and I have incompatible economic models )

[iMo]

PS: why the counter on a different channel limits the max mem depth ? 

Any access to the channel puts it into full operation, so it takes both the sampling frequency and memory.

Hmm, a weird design.. - their counter looks like a 6digit counter gated by 1sec (perhaps 10 lines in verilog). Has nothing to do with sampling into the mem. Looks like a bug to me..

[AndyBig]

Hmm, a weird design.. - their counter looks like a 6digit counter gated by 1sec (perhaps 10 lines in verilog). Has nothing to do with sampling into the mem. Looks like a bug to me..

It seems to me that all operations on the channel - trigger, counter, measurements, etc. - are carried out programmatically based on the ADC readings accumulated in memory.
I agree that this is not the best solution, but apparently they did it this way to reduce the cost of the oscilloscope.

[RAPo]

I was afraid of that answer.
So now I have a decision to make ....

My question is: are you two joining forces, because I sure would like to see all your great adaptions in one package!

mrisco and I have incompatible economic models )

[mrisco]

FYI - 804 modded to 824, no selfcal after mod.
50ohm terminator on CH4.
Waveform view.

BW OFF       49.4uV Avg. AC.RMS  (+12uV against 804)
BW 20MHz   24.1uV Avg. AC.RMS  (the same as the 804)

When I see the original interface of the DHO, I wonder what happened in the UI design engineering team to release that result bar layout!
It is not to continue the layout of the other models because, for the DHO1000/4000 series it is different (due to the bigger screen) so it is only laziness to adapt the UI properly to a lower resolution.

Let's see, this is with the stock application:



And this is with the extended layout:




In the promotional pictures of the DHO800 series, the result bar had a better layout and was docked to the right:

[mawyatt]

We are attempting to use the DHO814 for Transistor Curve Tracer utilizing the XY Mode. Really like the display that includes the X and Y numeric labels, but can't get the display to show enough brightness in XY mode (we have brightness set to max), it's fine in regular Time Domain mode tho.

The results are very faint as shown.

The setup is an AWG used to sweep (0-8V) the X Axis thru a Resistor (Collector 100Ω) and the DSO CH1 connects to Resistor, the AWG other channel is stair-stepped (8 Steps 0-10V) thru a Resistor (Base 50KΩ). Another Resistor (Emitter 10Ω) and DSO CH2 is connected across this resistor to display Current on the Y Axis.

You can see the Siglent DSO displays the result with good brightness while the DHO814 we can't get to display with sufficient brightness.

BTW wish the Siglent would include the XY display numerics like the Rigol, they are nice

Any help is from the Rigol gurus is appreciated.

Best

[Fungus]

We are attempting to use the DHO814 for Transistor Curve Tracer utilizing the XY Mode. Really like the display that includes the X and Y numeric labels, but can't get the display to show enough brightness in XY mode (we have brightness set to max), it's fine in regular Time Domain mode tho.

The results are very faint as shown.

Turn up the persistence?

[mawyatt]

First thing we tried, makes no difference in XY Mode, but does work in normal TD Mode?

Best

[AndyBig]

In the promotional pictures of the DHO800 series, the result bar had a better layout and was docked to the right:

Honestly, I like the measurement panel on top of the signal window better. When it doesn't make the signal window shrink