Low Cost PCB's Low Cost Components

Author Topic: Noob: DAC linearity in the DP832  (Read 4034 times)

0 Members and 1 Guest are viewing this topic.

Offline LaurentR

  • Frequent Contributor
  • **
  • Posts: 370
  • Country: us
Noob: DAC linearity in the DP832
« on: January 09, 2015, 07:38:30 AM »
Hi all,
since I got my DP832, I noticed the output was always a few mV off, as measured by both an external DMM and the internal ADC. That's still in spec but I got curious so I decided to script a sweep on Ch1 Volts to see the big picture.

The setup ran at cold ambient (~15C), so somewhat out of calibration spec but the overall shape of the curve is still similar to what I saw at higher ambient.

I don't have much experience with DAC linearity, but the result still surprised me. Ignoring static offset, there is still about 8mV of range in variation (in spec for the DP832), but I can't explain the abrupt changes, especially around 10V. I assume the plot is a combination of calibration curve-fitting and actual DAC response...

BTW, the DP832 internal ADC agreed in the general shape of the curve.
 

Offline wiss

  • Frequent Contributor
  • **
  • Posts: 485
  • Country: ch
Re: Noob: DAC linearity in the DP832
« Reply #1 on: January 09, 2015, 09:59:22 AM »
But, you can only set it to even 10 or 1 mV, right?
 

Online TopLoser

  • Supporter
  • ****
  • Posts: 1477
  • Country: gb
Re: Noob: DAC linearity in the DP832
« Reply #2 on: January 09, 2015, 10:27:12 AM »
Sounds like it's related to your DMM changing ranges at 10v?
 

Offline LaurentR

  • Frequent Contributor
  • **
  • Posts: 370
  • Country: us
Re: Noob: DAC linearity in the DP832
« Reply #3 on: January 09, 2015, 10:33:13 AM »
But, you can only set it to even 10 or 1 mV, right?

Right. This particular run is stepping through every even 10mV value from 0 to 30V. That's already 3000 steps... I am planning to try a smaller window with 1mV steps (using the high res option), but first I was wondering if anybody could explain what is going on with this one.

Zooming on this one, the vertical noise already looks like the DAC has somewhere between 1/4 and 1/2 mV steps (which would be more or less consistent with a 16 or 17b range over 30V).
 

Offline LaurentR

  • Frequent Contributor
  • **
  • Posts: 370
  • Country: us
Re: Noob: DAC linearity in the DP832
« Reply #4 on: January 09, 2015, 10:47:59 AM »
Sounds like it's related to your DMM changing ranges at 10v?

I thought it might be the case and ran some sanity checks around this, but I'll have another more serious look.

OTOH, the huge peak and dip happen before 10V and the DMM stays in the 10V range until 12V (I am stepping up).
And the internal ADC of the DP832 agrees with the curve (I am not sure what the precision on the ADC is - accuracy is something like .01% + 5mV). There is some slow drift compared to the DMM across the range, but it definitely sees the same relative peaks and dips.
 

Offline wiss

  • Frequent Contributor
  • **
  • Posts: 485
  • Country: ch
Re: Noob: DAC linearity in the DP832
« Reply #5 on: January 09, 2015, 11:10:51 AM »
What DMM is it?
 

Offline LaurentR

  • Frequent Contributor
  • **
  • Posts: 370
  • Country: us
Re: Noob: DAC linearity in the DP832
« Reply #6 on: January 09, 2015, 12:12:53 PM »
What DMM is it?

34461A. This plot is 0.5sec delay in SW between a status read of the DP832 after the voltage change and the DMM READ? trigger and 10PLC on the DMM (supposed to get a 10uV of resolution on the 10V range). I tried various other parameters on subsets (like 2sec and 100PLC) and it's all very similar.
 

Offline LaurentR

  • Frequent Contributor
  • **
  • Posts: 370
  • Country: us
Re: Noob: DAC linearity in the DP832
« Reply #7 on: January 11, 2015, 02:55:50 AM »
Did a full 1mV sweep. Nothing groundbreaking on the weird peaks and dips. However, did confirm that the DAC output precision is ~520uV (see histogram), which is consistent with a 16b value over ~34V.
 

Offline thmjpr

  • Regular Contributor
  • *
  • Posts: 75
  • Country: ca
Re: Noob: DAC linearity in the DP832
« Reply #8 on: January 13, 2015, 05:40:45 PM »
Here is some quick data using 0.1V steps 1s delay and only the internal ADC to read back. 0.5s delay had some data loss.
Some spot checks agree that the ADC readings are at least relatively accurate. Might try with proper DMM readback in future.

I manually calibrated the unit ~6months ago in an attempt to remove the offset. Its a bit worse now it seems, but I did not record original data. I wonder if the DAC reference is drifting internally (ADR381?).

Looking at various DAC cal points: 8.4, 9.8, 12.1 for example, they don't seem to correspond to jump locations. I would assume rigol is using a basic linear interpolation between the cal points.

To improve this they potentially could single step through each voltage value to build an offset table (~60kB). Or employ some digital feedback to adjust DAC up/down until the ADC readback matches set voltage.
 

Offline LaurentR

  • Frequent Contributor
  • **
  • Posts: 370
  • Country: us
Re: Noob: DAC linearity in the DP832
« Reply #9 on: January 14, 2015, 05:37:22 AM »
Thanks!

I ran a home-written automated calibration based on the calibration DIY in the  DP832 thread http://www.eevblog.com/forum/testgear/rigol-dp832-firmware-updates-and-bug-list/180/ and as a result:
* The DAC behavior hasn't really changed. It looks a lot like yours and my old one. Strangely, even after calibration, the error is not centered on 0 at all. I get very little undershoot (going to ~ -1mV) and mostly overshoot (to ~ +9mV).
* During the calibration, the DAC is way off (by about 500mV) when stepping through values.
* The ADC does pretty nicely after calibration. With ambient matching my calibration ambient, the ADC goes from about -0mV to +1mV, well within specs. There is noticeable ambient temperature drift.
* During calibration, it seems pretty close, so there is not much adjustment due to calibration.
 

Offline smgvbest

  • Supporter
  • ****
  • Posts: 348
  • Country: us
    • Kilbourne Astronomics
Re: Noob: DAC linearity in the DP832
« Reply #10 on: January 14, 2015, 03:02:55 PM »
@LaurentR
Would you be willing to share your calibration script?
Sandra
(Yes, I am a Woman :p )
 

Offline LaurentR

  • Frequent Contributor
  • **
  • Posts: 370
  • Country: us
Re: Noob: DAC linearity in the DP832
« Reply #11 on: January 14, 2015, 03:17:21 PM »
@LaurentR
Would you be willing to share your calibration script?

Absolutely. I am cleaning it up and will post it in the DP832 thread in the next few days. It's a straight port of ted572's instructions (from TooOldForThis's work).
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf