Author Topic: Repair : HP 3245A precision DC/AC source  (Read 52402 times)

0 Members and 1 Guest are viewing this topic.

Offline alanambrose

  • Frequent Contributor
  • **
  • Posts: 377
  • Country: gb
Re: Repair : HP 3245A precision DC/AC source
« Reply #75 on: December 12, 2015, 10:11:00 am »
Any thoughts on what's going on with this algorithm? Why this sequence? Is it exactly fixed or adaptive? There are obviously some checks around the extremities of the ranges - but does anyone have any thoughts on why this sequence in particular? What might the cal data do - some kind of parameterised function applied to the raw reading?
“A foolish consistency is the hobgoblin of little minds"
 

Offline alanambrose

  • Frequent Contributor
  • **
  • Posts: 377
  • Country: gb
Re: Repair : HP 3245A precision DC/AC source
« Reply #76 on: December 12, 2015, 11:16:53 am »
Ah this is also fairly curious in the cal routine:

"Reading number 45 should be taken on the HP 3458A 100V DC range. Autorange should be used for all other readings."

And reading 45 is shown as ... 0.100220V.

Errr...

They wanted the resolution truncated? There's an interaction between the 3458 cal and the 3245 cal? A tech's idea of a joke?

A.
“A foolish consistency is the hobgoblin of little minds"
 

Offline Kleinstein

  • Super Contributor
  • ***
  • Posts: 14080
  • Country: de
Re: Repair : HP 3245A precision DC/AC source
« Reply #77 on: December 12, 2015, 11:43:03 am »
For reading no. 45 it is possibly that they need it as a difference to anouther reading that uses the 100 V range. So to avoid extra errors it might be better to use the 100 V range for this low voltage too. The 3458 still has a good resolution in the 100 V range. 
 

Offline alanambrose

  • Frequent Contributor
  • **
  • Posts: 377
  • Country: gb
Re: Repair : HP 3245A precision DC/AC source
« Reply #78 on: December 12, 2015, 12:45:26 pm »
Hi,

FYI I just did an analysis of the commands in the firmware compared to those in the docs (see http://anagram.net/nuts/3245a/3245A commands analysis.xlsx ). These ones are apparently undocumented and may be interesting:

BUSYTIME
CALLIMIT
CALREAD
CALWRITE
FREAD
FWRITE
PICKACHAN
RIGGLE
SREAD
SWRITE
WIGGLE

These are there too - similar to the 3458A:

JSR
MADDR
MREAD
MWRITE

A.
“A foolish consistency is the hobgoblin of little minds"
 

Offline TiNTopic starter

  • Super Contributor
  • ***
  • Posts: 4543
  • Country: ua
    • xDevs.com
Re: Repair : HP 3245A precision DC/AC source
« Reply #79 on: December 12, 2015, 01:06:02 pm »
I thought sequence is rather usual, near full scale pos/neg on specific range, near half scale and near zero. Specific points likely chosen for AC performance cal points, by some internal math derivatives.
YouTube | Metrology IRC Chat room | Let's share T&M documentation? Upload! No upload limits for firmwares, photos, files.
 

Offline alanambrose

  • Frequent Contributor
  • **
  • Posts: 377
  • Country: gb
Re: Repair : HP 3245A precision DC/AC source
« Reply #80 on: December 12, 2015, 05:04:59 pm »
Here's an analysis of the pre-calibration error on my unit. It looks like the 1V and 10V ranges can be compensated fairly easily internally with some linear adjustments. Frankly, I'm surprised that it's so linear.



The data is here FYI:

http://anagram.net/nuts/3245a/Channel%20A%20vs%20B%20pre-calibration.xlsx

Alan
“A foolish consistency is the hobgoblin of little minds"
 

Offline TiNTopic starter

  • Super Contributor
  • ***
  • Posts: 4543
  • Country: ua
    • xDevs.com
Re: Repair : HP 3245A precision DC/AC source
« Reply #81 on: December 12, 2015, 06:36:43 pm »
Nice. Can you share program code, I can try on mine. :)
Can toss in Kei2400 as well for comparison.
YouTube | Metrology IRC Chat room | Let's share T&M documentation? Upload! No upload limits for firmwares, photos, files.
 

Offline alanambrose

  • Frequent Contributor
  • **
  • Posts: 377
  • Country: gb
Re: Repair : HP 3245A precision DC/AC source
« Reply #82 on: December 12, 2015, 07:03:00 pm »
>>> Nice. Can you share program code, I can try on mine.

Hi Tin,

Sure, I'll just tidy it up a bit so you don't think I'm a complete hacker :)

Having decided there was nothing of value in the existing cal - here's the post-cal results. Error is ~5-10uV on the 1V range, +-20uV on the 10V range. I ran the cal a second time and there was no improvement. Allow for minus ~1-2uV error on the 3458A (measured with a short) even though I did autocal the DCV beforehand. Only 0.1C variation over the <1h it took to do these runs.





data is here:

http://anagram.net/nuts/3245a/3245A post-calibration A vs B.xlsx

Alan
“A foolish consistency is the hobgoblin of little minds"
 

Offline alanambrose

  • Frequent Contributor
  • **
  • Posts: 377
  • Country: gb
Re: Repair : HP 3245A precision DC/AC source
« Reply #83 on: December 15, 2015, 09:53:19 am »
>>> Nice. Can you share program code, I can try on mine.

Hey TiN,

OK alpha-quality app here:

http://anagram.net/nuts/3245A/calibration/

- together with example output. It does the performance check as in the cal manual, the calibration sequence, and another pre/post cal check working from +-1uV to +-10V. For DCV only right now - it will be easy to add DCI.

It comes up with a calibration 'headroom' number i.e. how far away on average the tests are from the specified error e.g. if the spec says max 30uV error and the actual error is 6uV then the 'headroom' is 80%. As you'll see from the example output, ~80% is actually the kind of number I'm getting.

Let me know if you find any quirks/bugs. I would be interested in seeing the output of your runs.

Alan
“A foolish consistency is the hobgoblin of little minds"
 

Offline alanambrose

  • Frequent Contributor
  • **
  • Posts: 377
  • Country: gb
Re: Repair : HP 3245A precision DC/AC source
« Reply #84 on: December 21, 2015, 05:58:24 pm »
OK more cal data - this time DCI:





This suggests to me that my box was never calibrated during its life as it was so far out, and it seems easy to pull it way into spec. Cal headroom on DCI is into the 90% i.e. on average it is coming in more than 10x better than the official spec.

Data is here if anyone cares:

http://anagram.net/nuts/3245A/calibration/3245A post-calibration A vs B.xlsx

Alan
“A foolish consistency is the hobgoblin of little minds"
 

Offline TiNTopic starter

  • Super Contributor
  • ***
  • Posts: 4543
  • Country: ua
    • xDevs.com
Re: Repair : HP 3245A precision DC/AC source
« Reply #85 on: December 22, 2015, 12:55:58 am »
I care.
Perhaps using logarithmic scales would be more easier to read for DCI? Worth a try, but would need bit of inversion/sign fiddling for negative side...
YouTube | Metrology IRC Chat room | Let's share T&M documentation? Upload! No upload limits for firmwares, photos, files.
 

Offline alanambrose

  • Frequent Contributor
  • **
  • Posts: 377
  • Country: gb
Re: Repair : HP 3245A precision DC/AC source
« Reply #86 on: December 22, 2015, 12:58:49 pm »
Oh yeah I forgot, it was you who got me into this in the first place :)

Max DCI error is 5nA < 1mA (5ppm) and ~1nA < 0.1mA (10ppm). I've updated the DCV graphs also. There seems to be a ~+5-10uV offset on my unit on the lower ranges.



“A foolish consistency is the hobgoblin of little minds"
 

Offline TiNTopic starter

  • Super Contributor
  • ***
  • Posts: 4543
  • Country: ua
    • xDevs.com
Re: Repair : HP 3245A precision DC/AC source
« Reply #87 on: January 19, 2016, 04:15:45 pm »
>>> Nice. Can you share program code, I can try on mine.

Hey TiN,

OK alpha-quality app here:

http://anagram.net/nuts/3245A/calibration/


Hm, seems you removed app from folder?
I got around finally getting VISA back on my Win PC, so wanted to give it a whirl, but no exe anymore.
Can you upload it once again ? :)
YouTube | Metrology IRC Chat room | Let's share T&M documentation? Upload! No upload limits for firmwares, photos, files.
 

Offline TiNTopic starter

  • Super Contributor
  • ***
  • Posts: 4543
  • Country: ua
    • xDevs.com
Re: Repair : HP 3245A precision DC/AC source
« Reply #88 on: January 19, 2016, 04:40:17 pm »
I found it, running now.

Seem there is little bug, if frequency a bit higher than 60Hz, program throws error:

Code: [Select]
00:35:27 DMM LOCK OFF sent
00:35:27 DMM LOCAL sent
00:35:27 *** Connection closed ***
*** Exception - invalid line frequency '60.0056348' ***

I have to rerun it 5-10 times to catch that 59.9x Hz dip to start.

Here's data I got:


Smoke check log
First performance verification test
Precal test
Calibration log
Postcal log
Post cal perf test log


I'd be happy to test current too, if you have version with it.



Also another idea, perhaps results will be bit better if measurements at 3458A would be done with multiple samples, let's say 10-20 samples each step, and ignore first and last 3 samples. Just to make sure it all settle nice and tidy. Maybe run average for remaining samples too. I rerun perfcheck few times, and results were bit different each time.
« Last Edit: January 19, 2016, 07:16:07 pm by TiN »
YouTube | Metrology IRC Chat room | Let's share T&M documentation? Upload! No upload limits for firmwares, photos, files.
 

Offline alanambrose

  • Frequent Contributor
  • **
  • Posts: 377
  • Country: gb
Re: Repair : HP 3245A precision DC/AC source
« Reply #89 on: January 19, 2016, 11:17:53 pm »
Well v. nice accuracy - do you think this is the result of your LTZ mod? Did you figure out how the LTZ translates to better cal accuracy - not sure I get that yet?

Will fix the 60Hz thing and supply the DCI version shortly.

Alan
“A foolish consistency is the hobgoblin of little minds"
 

Offline alanambrose

  • Frequent Contributor
  • **
  • Posts: 377
  • Country: gb
Re: Repair : HP 3245A precision DC/AC source
« Reply #90 on: January 20, 2016, 12:00:59 am »
OK still alpha standard code but V0.2 here:

http://anagram.net/Nuts/3245A/calibration/app/V0.2/

As you know you can't do DCI cal without DCV so it'll beep at you when it wants the leads changed over on the meter.

A.
“A foolish consistency is the hobgoblin of little minds"
 
The following users thanked this post: philipz, manupthehills

Offline Kleinstein

  • Super Contributor
  • ***
  • Posts: 14080
  • Country: de
Re: Repair : HP 3245A precision DC/AC source
« Reply #91 on: January 20, 2016, 04:32:59 pm »
Using more smaples at a single point is a good idea - though I would expect the 3458 to give a good reading rather fast. I would still look at the first samples too - they may point to possible settling time after a step in settling. Especially after a large step the voltage can take some time - possibly even more than 10 s to reach it's final value. The most obvieous case would be coming from a high voltage to a a rather small value (e.g. 1 V). Possible source for this could be thermal effects in dividers and dielectric absorbtion.

There is absolutely no need to ignore the alst 3 measurements - if you would do that, no need to take that data.
 

Offline alanambrose

  • Frequent Contributor
  • **
  • Posts: 377
  • Country: gb
Re: Repair : HP 3245A precision DC/AC source
« Reply #92 on: January 24, 2016, 11:28:52 am »
Hi Kleinstein,

Could you explain what you are referring to? The cal executable I provided has a 'SETTLETIMEINMS' parameter (which defaults to 1,000 ms) and a meter 'NPLC' parameter which defaults to 100 - so there's an adjustable wait time to allow the 3245A to settle and an adjustable amount of meter averaging - so the user can (a) experiment and (b) decide for themselves. The official cal procedure is written as though the cal is done by hand (and doesn't suggest a set amount of wait time, but I think, leaves it up to the user to judge) and specifies on page 4-5 'RESET, ACAL, and NPLC 100 on the HP 3458A'.

>>> There is absolutely no need to ignore the alst 3 measurements - if you would do that, no need to take that data.

Sorry I don't see where the 'last 3 measurements are ignored' - are we looking at the same thing?

Alan
“A foolish consistency is the hobgoblin of little minds"
 

Offline TiNTopic starter

  • Super Contributor
  • ***
  • Posts: 4543
  • Country: ua
    • xDevs.com
Re: Repair : HP 3245A precision DC/AC source
« Reply #93 on: January 24, 2016, 03:43:12 pm »
Alan, he was referring to my post with that suggestion.
What I meant is to use average of few readings already after settle time elapsed, with few first readings disregarded. I often see that voltages measured are not accurate to last bits if just first reading is taken, even NPLC 100. Sounds fishy, but it could play a bit at very last digits of accuracy.

Even after 2 recals using 1s of settling time, I still cannot get 10.00000 reading from 3458A, while gen set to 10.00000. Have to set it 10.00002 or so, to get exact 10V. It's nitpicking, but hey, we doing this for fun here, right? :)

Also I don't think LTZ have anything to do with better calibration accuracy, but I can't prove it, since I never tried it with original LM399 in came condition test.
« Last Edit: January 24, 2016, 03:44:54 pm by TiN »
YouTube | Metrology IRC Chat room | Let's share T&M documentation? Upload! No upload limits for firmwares, photos, files.
 

Offline TiNTopic starter

  • Super Contributor
  • ***
  • Posts: 4543
  • Country: ua
    • xDevs.com
Re: Repair : HP 3245A precision DC/AC source
« Reply #94 on: January 24, 2016, 05:26:14 pm »
YouTube | Metrology IRC Chat room | Let's share T&M documentation? Upload! No upload limits for firmwares, photos, files.
 

Offline alanambrose

  • Frequent Contributor
  • **
  • Posts: 377
  • Country: gb
Re: Repair : HP 3245A precision DC/AC source
« Reply #95 on: January 24, 2016, 09:50:47 pm »
>>> Alan, he was referring to my post with that suggestion. What I meant is to use average of few readings already after settle time elapsed, with few first readings disregarded. I often see that voltages measured are not accurate to last bits if just first reading is taken, even NPLC 100. Sounds fishy, but it could play a bit at very last digits of accuracy.

Ahhhhh - I didn't see that text underneath the graph, now I understand.

OK will add a count of readings to ignore at the beginning and also a count of readings to average.

93% headroom - very nice - will try a recal on mine.

Alan
“A foolish consistency is the hobgoblin of little minds"
 

Offline klaus11

  • Supporter
  • ****
  • Posts: 156
  • Country: 00
Re: Repair : HP 3245A precision DC/AC source
« Reply #96 on: January 25, 2016, 10:58:50 am »
TiN, it would be feasible and safe to make a board of Amplification (option 002). For units without 002
HP3458A, HP3245a, Keithley 2000, Fluke 87V, Rigol DP832, TEK TDS5052B, HP33120A
 

Offline alanambrose

  • Frequent Contributor
  • **
  • Posts: 377
  • Country: gb
Re: Repair : HP 3245A precision DC/AC source
« Reply #97 on: January 25, 2016, 11:12:36 am »
>>> TiN, it would be feasible and safe to make a board of Amplification (option 002). For units without 002

That's interesting - mine doesn't have the 002 channel. Care to explain further?

>>> I still cannot get 10.00000 reading from 3458A

Well a bit ambitious given the specs of the 3245A - but 'nuts' is in our name right :)

A.
“A foolish consistency is the hobgoblin of little minds"
 

Offline TiNTopic starter

  • Super Contributor
  • ***
  • Posts: 4543
  • Country: ua
    • xDevs.com
Re: Repair : HP 3245A precision DC/AC source
« Reply #98 on: January 25, 2016, 11:55:11 am »
Alan, I'm sure your read my article, aren't you? :) I cover this right in the intro.
You have 3245-001 (option 001, which is two identical +/-10V 0.1A channels), while I have 3245-002 (option 002, on which instead of board B there is x10 power amplifier board, to boost +/-10V to +/-100V, purely analog).
YouTube | Metrology IRC Chat room | Let's share T&M documentation? Upload! No upload limits for firmwares, photos, files.
 

Offline alanambrose

  • Frequent Contributor
  • **
  • Posts: 377
  • Country: gb
Re: Repair : HP 3245A precision DC/AC source
« Reply #99 on: January 25, 2016, 12:21:59 pm »
Ah yes, I know you have 100V option and I have 2x10V.

I was interested in what Klaus had in mind re a non-HP 002 option. It would need to go in a separate box for me as there's little room once you have 2x10V boards installed. Easy enough (well maybe) to build the option from the schematic, but maybe it can be improved / extended.

On another subject ...  we have a sense of the 3458A and 3245A tempcos now. Well, in both cases, the fan is a simple always-on low-tech thing. So one easy way to reduce drift would be to run the fan from a PID controller. I think to do this well, we might characterise the main components responsible for the tempco a little and then site the sensor(s) near those. What do you think?

A.
“A foolish consistency is the hobgoblin of little minds"
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf