New Rigol 16-bit function generators DG800/900 series

[timber23]

Thanks for posting the info!
Where do I get the binary file? Is it already in this thread and I missed it?

It is the first attachment of tv84 post.
https://www.eevblog.com/forum/testgear/new-rigol-16-bit-function-generators-dg800900-series/msg2420391/#msg2420391

It is inside the ZIP file.

[frozenfrogz]

Thank you! I completely skipped that. Maybe due to reading parts of this thread on mobile...

[tv84]

Did someone managed to get a USB to ethernet adapter working? I have a super cheap adapter from action.com here, but DG811 aka DG992 is not getting an IP-address via DHCP.

Any suggestions? I had a look on amazon's offers, but I didn't find any information about the used chipsets.

These driver modules are present:

Code: [Select]

-rwxrw-rw-  1 1000  1000      842148 2018-07-19 09:12:50 modules/r8188eu.ko
-rwxrw-rw-  1 1000  1000       19076 2018-07-19 09:12:50 modules/rtl_usb.ko
-rwxrw-rw-  1 1000  1000       50952 2018-07-19 09:12:50 modules/rtl8192c-common.ko
-rwxrw-rw-  1 1000  1000       42048 2018-07-19 09:12:50 modules/usbnet.ko
-rwxrw-rw-  1 1000  1000       32048 2018-07-19 09:12:50 modules/r8152.ko
-rwxrw-rw-  1 1000  1000       22848 2018-07-19 09:12:50 modules/usbtmc_dev.ko
-rwxrw-rw-  1 1000  1000       69052 2018-07-19 09:12:50 modules/rtlwifi.ko
-rwxrw-rw-  1 1000  1000       19936 2018-07-19 09:12:50 modules/usbtmc.ko
-rwxrw-rw-  1 1000  1000       85044 2018-07-19 09:12:50 modules/rtl8192cu.ko
usbnet.ko

Code: [Select]

.rodata.str1.4:000038E0 47 65 6E 65 53 79+.LC8            DCB "GeneSys",0         ; DATA XREF: usbnet_open+1ACo
.rodata.str1.4:000038E8 4E 65 74 43 68 69+.LC9            DCB "NetChip",0         ; DATA XREF: usbnet_open+19Co
.rodata.str1.4:000038F0 5A 61 75 72 75 73+.LC10           DCB "Zaurus",0          ; DATA XREF: usbnet_open+1BCo
.rodata.str1.4:000038F8 52 4E 44 49 53 00 .LC11           DCB "RNDIS",0           ; DATA XREF: usbnet_open+1CCo
.rodata.str1.4:00003908 41 53 49 58 00    .LC13           DCB "ASIX",0            ; DATA XREF: usbnet_open+1D8o

[ogden]

LOL. Wondering if I should grab a DG811 now before they all disappear in a puff of smoke

Out of stock in Batronix already 

[frozenfrogz]

xD Big thank you to tv84 for buying me a brand new DG992!

Worked like a charm on my DG812!

Edit: I am curious though. Is the performance actually the same as with a "true" DG992?

Sinewave from ca. 2MHz and above is heavily attenuated
Ripple on squarewave is the same as in timber23s screenshots and gets pretty unusable after that (not very square after all above 6MHz).

Would be interesting to see if the output of a genuine DG992 differs. Might be, that there is some hardware selection in the factory just like other binning. (just a guess)

[bitseeker]

Edit: I am curious though. Is the performance actually the same as with a "true" DG992?

Sinewave from ca. 2MHz and above is heavily attenuated

The output amplitude spec for all DG800 and DG900 is the same:

Amplitude (into 50Ω):
≤10 MHz: 1.0 mVpp to 10 Vpp
≤30 MHz: 1.0 mVpp to 5.0 Vpp
≤60 MHz: 1.0 mVpp to 2.5 Vpp
>60 MHz: 1.0 mVpp to 1 Vpp

There should not be heavy attenuation beginning at only 2 MHz.

Ripple on squarewave is the same as in timber23s screenshots and gets pretty unusable after that (not very square after all above 6MHz).

What are you measuring it with and what is your setup?

Would be interesting to see if the output of a genuine DG992 differs. Might be, that there is some hardware selection in the factory just like other binning. (just a guess)

From what I've seen, other Rigol products have not used any special hardware differences for software-limited functionality. I wouldn't expect them to treat this product line any differently.

[bitseeker]

Thank you, TurboTom, for the teardown pics.

Thank you, tv84, for all the work you put into feature research and for sharing them with us.

[commongrounder]

Is it confirmed that the sample frequency goes to 250ms/s with the “upgrade?.  I’m desperately trying to find some difference between the 800 and 900 series to help me justify spending $600US on my DG952.

[timber23]

I did some tests comparing a pimped RIGOL DG811 (DG992-mode) vs. a pimped Siglent SDG2042X (SDG2122X-mode).

I used the same BNC cable which I unpluged and repluged during the tests. The mesurement was taken by a RIGOL MSO5000 oscilloscope. BNC cable was terminated using 50 ohm on T-adaptor.

Rigol DG811 (pimped to DG992) set to 25 MHz square wave, 4 Vpp, 50 ohm:


Siglent SDG2042 (pimped to SDG2122X), set to 25 MHz square wave, 4 Vpp, 50 ohm:


Comparison of both signals, Siglent's trace in green colour:


Now sine wave, 100 MHz, 1 Vpp (since 1Vpp is max. amplitute of DG992), 50 Ohm.

DG811 (pimped to DG992):


SDG2042X (pimped to SDG2122X):


Comparison of both traces, Siglent in green colour:


You can clearly see that the RIGOL is much more attenuated at 100MHz than the Siglent. Furthermore Siglent's square wave is a little bit better. But SDG2042X costs 565€ (upgradeable to SDG2122X) compared to 236€ for DG811 (upgradeable to DG992). Considering the price, DG811/DG992 is the winner in my opinion.

Update: I changed the labeling of the measurements because of TurboTom's objection. I do not own a genuine DG992, nor do I have access to one. Sorry for any confusion.

[TurboTom]

Did some testing already on my "pimped" 811...

Actually, there is considerable amplitude drop above approx. 50MHz with the two channels slightly differing from each other. I only tested at 1Vpp into 50 Ohms so far and did a slow sweep from 10kHz to 100MHz. I had the Vpp measurement running and took a history record of this, see attached screenshots. If this is the result of inadequate hardware or the instrument not being calibrated over the full range has yet to be found out.

The ugly ringing on square or other waveforms with steep edges apparently is the result of the anti-aliasing filter after the DAC. This filter seems to be pretty effective with the edge at round about 115MHz. Since this means, all the higher frequency components of a square wave (which aren't present anyway at the output of a 250MSa/s DAC) are missing from the signal, the mentioned observations may just be obvious. Attached you will find two more screenshots visualizing this.

BTW, if you take a look at the "solder side" of the main PCB, you get an idea how Rigol got away without any external shielding of the AWG's enclosure...it's simply one big ground plane. I wonder how many layers this board has. It's amazing they can sell this package for round about 200EUR (plus VAT if applicable).

I found some more peculiarities but these will follow probably tomorrow night...

Cheers,
Thomas

P.S. Altogether, I consider the DG811 so far as some of the best spent money. It's not an SDG6000X (which has its shortcomings as well) but as a general purpose AWG (hacked), there's little left to wish for (not to talk about the casing design...). Hardware-wise it's close to perfect and the firmware is surprisingly complete and stable. So if you're in the market for a decent AWG for your hobby electronics workbench, with the DG811 you can't go wrong!

Edit: The output level recorded for CH2 in the attached screenshots may not be completely accurate since I later identified a dodgy adapter that i used during this test. Further tests with a power meter actually indicate that both channels perform very similar, CH2 may be even slightly better than CH1.

[commongrounder]

What about arb memory? Is there 16MB hidden in the 800 series that is made available with the “upgrade”?

[TurboTom]

What about arb memory? Is there 16MB hidden in the 800 series that is made available with the “upgrade”?

Of course -- the sample memory DRAM chip attached to the FPGA is an MT41K128M16JT-125 (128M*16) so this provides more than enough space for two channels of 16M*16 including double-buffering and what not. After having read @timber23's post I'm pretty sure the hardware of the DG8xx and DG9xx series is identical.

I may have to modify this last statement after having read @timber23's previous contributions: timber23, you specified you compared a DG992 to the Siglent SDG2000X, I assumed it was a "real" DG992 but now I'm not too sure anymore since apparently you've (also?) got an upgraded DG8xx. I guess we should be very careful and accurate with the designations of the instruments, i.e. I'ld never call my "pimped" DG811 a DG992.

We're just at the very beginning of this adventure and we cannot be sure yet that there really aren't any differences unless we had a close look at a "real" DG992. So far, my DG811 is a DG811 that "thinks" it's a DG992... 

Cheers,
Thomas

[timber23]

What about arb memory? Is there 16MB hidden in the 800 series that is made available with the “upgrade”?

timber23, you specified you compared a DG992 to the Siglent SDG2000X, I assumed it was a "real" DG992 but now I'm not too sure anymore since apparently you've (also?) got an upgraded DG8xx. I guess we should be very careful and accurate with the designations of the instruments, i.e. I'ld never call my "pimped" DG811 a DG992.

Sorry for the confusion. I modified my post for clarity.
@commongrounder : You said that you own a DG952. Maybe you can perform some test, which we can compare our "pimped DG811s" to.

Maybe someone has an idea for a test-setup which can distinguish a sample rate of 250 MSa/s from 125 MSa/s.

[TurboTom]

Sample rate is 250M on the "DG811+++" otherwise it wouldn't be possible to output 100MHz at all and we should observe bad aliasing above 62.5MHz output frequency. I'm quite certain that even the original DG8xx sereis samples at 250M, maybe arbitrary waveforms are limited to 125MSa/s, doubling each sample. It should also be possible by checking when (i.e. at what frequencies) the instrument outputs a "clean" square wave without edge jitter. These frequencies would need to be whole-number factors of the sampling rate, so by figuring out a few it should be possible to deduce on the latter. Will look into this more closely in the evening.

Edit: Added the screen shot of my sample rate test. I used an arbitrary waveform with as many steep edges as possible for that. The reason is that the "controlled edge technology" that Rigol calls "SiFi2" is not operational in Arb mode. This means, slopes are defined by just a single sample interval, also recognisable by the slightly faster rise/fall times and (even) more ringing. Now I selected a frequency at which the signal cannot be reproduced by full integers of the sampling frequency, resulting in edge jitter. I measured the time between the two traces that are shown when the scope "averages" over several scans and calculate the frequency which should resemble actual sample rate. And voila, 250MHz!

[idolclub]

HOW TO CHANGE RIGOL DG800/DG900 MODELS

1. Get a blank formatted USB disk (capacity >= 1GB)

2. Write signature in the disk with the following command (in linux):

        sudo dd if=DG800_sardinha.bin of=/dev/sda1 bs=512 seek=2000000

Amazing !!  DG811 free upgrade to DG992. 

[frozenfrogz]

I am going to go through this: DG900 performance verification guide and see what gives.

[commongrounder]

I’m willing to perform tests with my DG952, if given the exact procedure.  I’m curious whether paying a premium for the 900 series got me more than a black cabinet color!

[Kodovo]

Out of stock in Batronix already 

Wow, there is now up to 40 days lead time for the orders.  Placed my today.

[timber23]

I am going to go through this: DG900 performance verification guide and see what gives.

Thank you for this Guide. I performed the Harmonic Distortion Test using my Siglent SSA 3021X. My pimped DG811 (aka DG992) passed all scenaios on both channels. Please find attached the resulting screenshots.

[tv84]

I performed the Harmonic Distortion Test using my Siglent SSA 3021X. My pimped DG811 (aka DG992) passed all scenaios on both channels. Please find attached the resulting screenshots.

Nice work and good news!

[bitseeker]

That's looking good, timber! How does it look at 100 MHz, just to complete the range?

[timber23]

That's looking good, timber! How does it look at 100 MHz, just to complete the range?

As you requested, here are the results for 100MHz. This test is not part of RIGOL's "DG900 Performance Verification Guide". Furthermore I add the fall/rise-time and overshoot test. I performed the tests with respect to the beforementioned guide.

It is funny that my channel 2 has less overshoot than channel 1, because channel 2 was disabled before unlock...

Just for completeness: This is how my measurement setup looks like.

[bitseeker]

That's looking good, timber! How does it look at 100 MHz, just to complete the range?

As you requested, here are the results for 100MHz. This test is not part of RIGOL's "DG900 Performance Verification Guide".

Thanks. Looks good. Since the spec table in the guide said less than -35 dBc for any frequency over 40 MHz, I just had to see if it could make it all the way to 100 MHz. It certainly does.

Furthermore I add the fall/rise-time and overshoot test. I performed the tests with respect to the beforementioned guide.

Does the rise/fall time measurement change much if you use a faster timebase? I'd get more slope visible on screen (i.e., instead of a vertical line) for measurement purposes.

[timber23]

Does the rise/fall time measurement change much if you use a faster timebase? I'd get more slope visible on screen (i.e., instead of a vertical line) for measurement purposes.

No, it doesn't change much. I adjusted the timebase to 1µs in the previous test, because it is specified in the guide. Now I changed it to 10ns. Please find attached the resulting measurement.

[bitseeker]

OK, cool. Thanks for checking, timber. I feel more confident seeing it measured this way.