Author Topic: Repair : HP 3245A precision DC/AC source (Read 52723 times)

deadlylover · « **Reply #50 on:** December 02, 2015, 06:38:18 pm »

The same thing happened to my 3457A, same mains filter too, those Schaffner made in Switzerland ones are cursed. After a cosy life of 100V mains, mine only took 5 seconds to blow when exposed to the full wrath of 240V.

Your lab is going to be smelly for weeks.

free_electron · « **Reply #51 on:** December 02, 2015, 08:08:30 pm »

awww , maaaaaan. that 'paffner' puked its guts all over that pretty gpib interface.

THe problem in those paffner things is they used those amber colored foil RIFA caps that are notorious to go kaboom...

alanambrose · « **Reply #52 on:** December 02, 2015, 08:14:13 pm »

Q'n is what other eqpt has those dodgy parts inside?

TiN · « **Reply #53 on:** December 03, 2015, 04:17:15 am »

Whoa.. I got Delta EMI filter already, but was thinking to replace it later. Now I hear you, and will swap it tonight.

No I did not have cal program ready yet, as got carried away with more issues in 3458a instead

Can you post few more photos of your analog side boards, I wonder if any difference there as well.

alanambrose · « **Reply #54 on:** December 03, 2015, 12:00:56 pm »

>>> Can you post few more photos of your analog side boards

OK here goes, will put up the highres versions on my website. Bit of a mech design bodge getting the B channel out as you need to move the metal deck due to interference between the support posts and the backplane connectors, duh. Also, it's mounted so you can't get at the component side without taking it out. I can see why, but I think they could have come up with a better solution. Have not compared it detail yet, but both channels look identical ... including a 2nd vref. As they we're clearly borrowing a lot of parts from the 3458A already, I wonder why they didn't use just one 3458A vref for both channels? Noticed one bodge wire on the digital btw. C900 looks tricky to source as it's got a screen connection - not sure how important that is. Plan to write a 3245 / 3458 cal program presently.

Alan

alanambrose · « **Reply #55 on:** December 03, 2015, 12:02:17 pm »

A side...

alanambrose · « **Reply #56 on:** December 03, 2015, 12:03:56 pm »

B side...

alanambrose · « **Reply #57 on:** December 03, 2015, 12:25:43 pm »

Spot the difference...

alanambrose · « **Reply #58 on:** December 03, 2015, 01:32:56 pm »

Large images fyi:

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

BTW I notice there's no internal shields on the 3245.

A.

TiN · « **Reply #59 on:** December 03, 2015, 02:34:17 pm »

Thanks.

As of reference, I think they saved the effort of characterizing and sorting LTZ's for one more instrument, and it's requirements for "only" 6.5 digit generator is not required. If you really interested, we can do comparisons in drift/stability/noise whatever with yours 3245A and mine LTZ-modded. You have two analog outputs to compare

But i think i need get 3458a stable first...

P.S. That D7000 of yours screams for more attention, it can produce much better images...

I downloaded all, and had a dejavu moment on few photos. Open EXIF, hm, D800, good camera there.. Then realized it's mine photo

alanambrose · « **Reply #60 on:** December 03, 2015, 04:40:04 pm »

>>> P.S. That D7000 of yours screams for more attention, it can produce much better images...

Yes agree, v. lame images - I noticed that I had some better images on my phone - and, no excuses, as I have a nice macro lens that I couldn't be bothered to get out...

Ho ho, about including some of your images, apologies - I just copied over what I though was interesting from my \3245a\images folder - didn't notice that some of the backgrounds were different.

>>> we can do comparisons in drift/stability/noise

Good idea, I'm going to get a measurement set-up sorted out soonish ... atm I can't get ezgpib to talk to anything from win 7 64, so might revert to writing some code myself.

A.

TiN · « **Reply #61 on:** December 03, 2015, 05:01:56 pm »

I'd suggest getting RPI so we can bodge some python GPIB sorcery. It's better to have standalone setup to do long-term measurements, so whatever you do on your PC would not upset ppm-level stuff

P.s. I use Win2008R2 64bit (it's based on 7 x64) with EZGPIB, it works here okay (with NI GPIB USB pod though, not Agilent you have).

P.s. no problems about pic, you can take them surely.

alanambrose · « **Reply #62 on:** December 06, 2015, 05:10:38 pm »

OK FYI these are the results I'm getting:

3245A setting (V) 0 0.01 0.1 1 10
sd (ppm of 10V range) 0.083 0.069 0.18 0.13 0.339
sd (ppm of reading) 69 18 1.3 0.339

This is with a 3458A at NPLC=1 on the 10V range, no auto zero. I measured the 3458A with a short at 0.053ppm of range.

A.

TiN · « **Reply #63 on:** December 06, 2015, 05:17:54 pm »

Paffnerr is out of the way now...

Also replaced all electrolytic caps on A6 PCBA, and while doing so, spotted a bit of QC control issue:

Little guy's lead was not formed properly.

alanambrose · « **Reply #64 on:** December 06, 2015, 08:40:42 pm »

Nice, btw I'm assuming the block of foam you show on the original ref came with the machine? - there was none on mine.

TiN · « **Reply #65 on:** December 07, 2015, 12:40:01 am »

I added that

alanambrose · « **Reply #66 on:** December 07, 2015, 11:17:05 am »

Any stats guys here who understand accuracy of noise calculation i.e. how many samples you need to get a given confidence level on the noise figure?

I see e.g. this:

http://web.eecs.umich.edu/~fessler/papers/files/tr/stderr.pdf

...and it looks like it converges fairly quickly.

Alan

TiN · « **Reply #67 on:** December 08, 2015, 02:10:12 pm »

50 samples would be enough, based on results I saw so far.
Likely will be best to take 60 samples, and disregard first 10 (as some instruments may have some initial offset from range switch operation).

3roomlab · « **Reply #68 on:** December 08, 2015, 03:53:16 pm »

Quote from: TiN on December 08, 2015, 02:10:12 pm

50 samples would be enough, based on results I saw so far.
Likely will be best to take 60 samples, and disregard first 10 (as some instruments may have some initial offset from range switch operation).

ah ok, i get what the sterr.pdf meaning now you said 50. that plot on page 2 fig.1

dom0 · « **Reply #69 on:** December 09, 2015, 02:29:23 pm »

Quote from: deadlylover on December 02, 2015, 06:38:18 pm

those Schaffner made in Switzerland ones are cursed.

To be fair it only affects the older ones with the RIFA MP2? caps. New Schaffner filters are as good quality as it gets.

Dr. Frank · « **Reply #70 on:** December 09, 2015, 03:26:43 pm »

Quote from: dom0 on December 09, 2015, 02:29:23 pm

Quote from: deadlylover on December 02, 2015, 06:38:18 pm
those Schaffner made in Switzerland ones are cursed.

To be fair it only affects the older ones with the RIFA MP2? caps. New Schaffner filters are as good quality as it gets.

Do you, or anybody else know, from which datecode on these Schaffner filters have been improved?

Frank

VK5RC · « **Reply #71 on:** December 10, 2015, 02:31:46 am »

Re Schaffners, some other blogg postees have commented that it seems to be a problem of
-old equipment (old IEC filter)
- humid environment
- infrequent use (build up moisture in cap)
- then switch on failure- quite quickly
Occurrence in NOS treated in a similar manner.
Looking for a replacement for some older HP gear, Schaffner seems to be one of the better brands. I plan to think of them like electrolytics, need an eye and not infrequent replacement.

alanambrose · « **Reply #72 on:** December 11, 2015, 10:36:25 pm »

Ah FYI the DCV cal sequence attached from page 4-6 of the cal manual. Answers on a postcard...

alanambrose · « **Reply #73 on:** December 11, 2015, 10:55:20 pm »

Errr...it doesn't want to play - try this...

TiN · « **Reply #74 on:** December 12, 2015, 04:31:33 am »

This is what I had (run manually):


EEVblog Main Site	EEVblog on Youtube	EEVblog on Twitter	EEVblog on Facebook	EEVblog on Odysee

Author Topic: Repair : HP 3245A precision DC/AC source (Read 52723 times)

Share me