Precision DC votage source

[veedub565]

Anybody have any experience or thoughts on this precision DC voltage source ?

https://www.ebay.co.uk/itm/251756273626

On the face of it it seems pretty accurate for the price. My requirement is 10v nominal, true value certified to within +/- 1.5. Ideally something which isn't going to drift around much.

 

[Gyro]

Ha, yes. There was quite a thread about that one a few years back, lots of controversy and mirth irrc, the internal construction was... interesting. 

https://www.eevblog.com/forum/metrology/calibratory-d-105-dc-precision-voltage-reference-standard/

The listing looks the same anyway.


EDIT: Post with internal photos... 

https://www.eevblog.com/forum/metrology/calibratory-d-105-dc-precision-voltage-reference-standard/msg606722/#msg606722

It was produced by former (banned) forum member Awesome14. I can't believe he's sold 280 of the things!

[veedub565]

oh dear   

Yes I'll give that a miss then.

[fcb]

+/-1.5?  If that's ppm then you've no chance outside of Fluke 734B/C or equivalent.

We make a 10V reference (see our site in my signature link), but it's +/-30ppm.

[veedub565]

Yes it's 1.5ppm although that's certified to 1.5ppm.

I need to apply 10v DC, minimum 9v and max 11v, so whatever it is within that range needs certifying to within 1.5ppm. And whatever that certified value is that's what I use. So as I understand it stability is more important than absolute accuracy. I've got access to a 3458a to certify the value of a DC source.

It's for doing a basic cal a 5700a, but I don't have a Fluke 732 DC source available.   

[TiN]

3458A is unable to certify anything without calibrated DC voltage source, as by itself it's not a standard and just annual stability of standard unit is rather meh at 8ppm/year + factory uncertainty + temperature variations. Unless you have freshly calibrated high-end voltage source, you are better off leaving 5700A as it is now. I'd also recommend full preventive maintenance of 5700A boards before they catch fire 

[Villain]

Keep in mind Fluke 5700A has 8 ppm +4uV spec for 11V range at k=2.67. 3458A absolute uncertainty is about 6. something ppm at k=2 (something like that with 002). Its gonna be a "basic basic basic" cal.


edit: TiN beat me to it..

[veedub565]

Ahh looks like I'm a bit stuffed then for the time being without a proper precision source ££££

The 5700a is currently something like 231% out of spec when I performed it's self check.

It throws up a 40dB attenuator error, but there's nothing wrong with the attenuators they work ok. Going through the ranges I get roughly the right output levels. I think it's just so far out of spec it thinks it's got a fault. So even a basic basic basic cal would bring it in a bit closer.


recapping the boards is on the list of jobs to do. I was hoping to try and clear the existing errors before starting that job.

[TiN]

Decent 732As that require some minor service popup on ePay from time to time for $1-1.5k or thereabout, if you have good calibration lab within few hours (732A have short battery life).

It is common mistake to think that PASS/FAIL self-diagnostics which has very wide margins would be tripped by minor out of spec calibration. Unit will pass self-diagnostics even with thousand % off the spec (e.g. if you format EEPROM to default factory constants). 5700A does have very extensive self-test and cal-check procedures, and they do sometimes give false fails, but in 98% of the cases when it self-test says a fault then you do have a hardware fault. Running external calibration on faulty unit can easily ruin still good constants and you will have lots of fun with bogus calibration behaviour afterwards too. Ask me how I know... 




Also you don't provide any useful information on what is exactly out of range, 231% means nothing without knowing range/function/calibration period spec.  You can generate calcheck prinout for every range/function with RS232 or GPIB. Here's the code in Python I use to read all reports after calcheck over GPIB:

def f5720_generate_reports():
    f5720_cal_report (cal_report_check_after, "CAL_RPT? CHECK")
    f5720_cal_report (cal_report_cal_after,   "CAL_RPT? CAL")
    f5720_cal_report (cal_report_const_after, "CAL_RPT? RAW")
    f5720_cal_report (cal_st_cal_after,       "CAL_SLST? CAL")
    f5720_cal_report (cal_st_shift_after,     "CAL_SLST? CHECK")
       
def f5720_cal_report(filename, cmd):
    intf.clear()
    intf.write("*CLS*")
    intf.write(cmd)
    time.sleep(1)
   
    if cfg.get('calkit', 'interface', 1) == 'vxi':
        report = intf.inst.read()          # read all chars
    elif cfg.get('calkit', 'interface', 1) == 'gpib':
        report = intf.inst.read(len=64000) # read 64k chars
    with open(filename,'ab') as od:
        od.write("%s" % report)
    od.close()

[veedub565]

And I thought I had problems fixing my 5700a 

Ok so perhaps attempting a cal at this stage with basic standards is not recommended. And we can 98% confirm a hardware fault. I couldn't see how to print out the calcheck report over GPIB, this would certainly be useful information to have. I perhaps ought to start a new thread for this 5700a project.

+/-15v OSC Supply Fault (3612) - Resolved by exchanging A12 card with one from a donor unit.

A8 Relay Fault (3819) - Swapped over the A8 Switch Matrix card with one from a donor unit, but still get the same fault. Could be unrelated to the A8 card or could be both cards were faulty. Original card was put back in.

A15 DC HV Amp Offset Fault (3107) - I haven't looked into this fault yet.

A5 0db, 10db, 20db, 30db 40db output attenuator fail - I exchanged the A5 card with one from a donor unit, and now I only have A5 40db attenuator fault. So some improvement but clearly not fixed. Looks like exchange card also had a fault. Attenuator relays seem to be switching in and out ok. Could be worth trying a wideband module cal. No external standards required

[TiN]

Swapping boards willy-nilly with unserviced power supply cards A16,A17,A18 is a very nice way to kill good boards from other "donor" unit. Perfectly fine if you want to end up with two broken units, instead of one 
A12 board must also travel together with A13 board, because calibration constants for ac/ac sensor are not updated from public calibration ACAL procedures. 

Could be worth trying a wideband module cal. No external standards required

Incorrect. You need 5790A to complete wideband module cal.

[veedub565]

Could be worth trying a wideband module cal. No external standards required

Incorrect. You need 5790A to complete wideband module cal.
[/quote]

According to the service manual

"For wideband gain calibration, the only required equipment other than that supplied as
standard equipment with the wideband option is a Type "N" female to dual banana plug
adapter (e.g., Pomona 1740)."

That is what I was referring to, but yes you do need the 5790 for complete verification.

Swapping boards willy-nilly with unserviced power supply cards A16,A17,A18 is a very nice way to kill good boards from other "donor" unit. Perfectly fine if you want to end up with two broken units, instead of one 

Yes correct, I hadn't considered this. A fault on the power supply A16, A17, A18 potentially could have caused the fault on another card. Swapping in a new card from another unit could blow the new card.

I already have two broken units. The donor unit was until recently certified working with valid cal status, but has developed a fault on the high voltage output.  This was prior to me touching it. I was hoping I might be able to make one working unit out of the two faulty ones by swapping the cards. This appears to be a bad idea though.

I swapped the the A12 card which did resolve the +/-15v osc issue. So I guess this means the cal constants will now be incorrect for that card ?


I've never used Python, so I'm not sure how to use the code to get a calcheck printout. Do I copy/paste the code into the GPIB command box ?

[TiN]

Wideband calibration have two steps, gain and flatness. Flatness require external measurement and user entry of correct value from calibrated 5790A/B or similar.

High voltage fault how? 1100V and 2.2A ranges generated by A14/A15/A16 tandem.

[veedub565]

I'm trying to provide useful information on the calcheck but I don't know how to apply the python script to make the 5700 print this information.

[try]

See here:

https://www.mikrocontroller.net/topic/398160#5945289

picture 19:
    19_teure_Ebay-Referenz_in_drei_Wochen_um_3ppm_gedriftet_-_EUR_200_fuer_die_Tonne.jpg

translates into

expensive Ebay-reference drifted 3 ppm in 3 weeks - EUR 200,- for the dustbin

[dietert1]

Maybe at three ppm of 10 V low thermal EMF wiring isn't needed. The next (real) MakerFaire is planned for 10./11. September 2022, enough time to prepare if it really happens.

Regards, Dieter

[veedub565]

Also you don't provide any useful information on what is exactly out of range, 231% means nothing without knowing range/function/calibration period spec.  You can generate calcheck prinout for every range/function with RS232 or GPIB. Here's the code in Python I use to read all reports after calcheck over GPIB:

def f5720_generate_reports():
    f5720_cal_report (cal_report_check_after, "CAL_RPT? CHECK")
    f5720_cal_report (cal_report_cal_after,   "CAL_RPT? CAL")
    f5720_cal_report (cal_report_const_after, "CAL_RPT? RAW")
    f5720_cal_report (cal_st_cal_after,       "CAL_SLST? CAL")
    f5720_cal_report (cal_st_shift_after,     "CAL_SLST? CHECK")
       
def f5720_cal_report(filename, cmd):
    intf.clear()
    intf.write("*CLS*")
    intf.write(cmd)
    time.sleep(1)
   
    if cfg.get('calkit', 'interface', 1) == 'vxi':
        report = intf.inst.read()          # read all chars
    elif cfg.get('calkit', 'interface', 1) == 'gpib':
        report = intf.inst.read(len=64000) # read 64k chars
    with open(filename,'ab') as od:
        od.write("%s" % report)
    od.close()

Still don't know how to do this, tell me then I might be able to provide "useful" information

[TiN]

You can use any GPIB inteface you are familiar with to issue CAL_RPT? * commands like shown in quoted function and readback response prinout from calibrator.

[View[+]Finder]

Ha, yes. There was quite a thread about that one a few years back, lots of controversy and mirth irrc, the internal construction was... interesting. 

https://www.eevblog.com/forum/metrology/calibratory-d-105-dc-precision-voltage-reference-standard/

The listing looks the same anyway.


EDIT: Post with internal photos... 

https://www.eevblog.com/forum/metrology/calibratory-d-105-dc-precision-voltage-reference-standard/msg606722/#msg606722

It was produced by former (banned) forum member Awesome14. I can't believe he's sold 280 of the things!

The D-105 10V reference is much better than the post history on eevblog would indicate. How do I know this to be true? Let's start with some facts about the posts:
1) The "Awesome 14" guy is the son of the old "gray-beard engineer" who built the device way back when (draw your own conclusions . . .)
2) I bought one of the $100 versions maybe 10 years ago and one of the higher precision ones (D-105) about 5 years ago
3) I had a long and informative phone call with the father/inventor when I bought the D-105 back then
4) I read all the post history that led up to the Awsome14 problem a few years back
5) In addition to the above two references, I have bought and tested pretty much all the high end (meaning $200-300) 10VDC references.
6) There are many folks on the Metrology section that know far more about the subject than I do, particularly in the area of stuff like long-term drift

OK, so my opinion about the the history of D-105 on this forum is that it was 80% about the attitude of "Awesome14" and 20% about the device's capabilities. The 80% got really nasty on both sides.

I don't know who is actually making the current D-105 and what the test environment and protocol are, so I will stop short of an opinion on the device. Based on the testing I have done over the years, I would rate my D-105 slightly better than Ian Johnston's PDVS 2 excellent adjustable voltage standard that was my prior preference.

I'll publish my results from testing over the last year when I finish.

[Gyro]

Yes there were certain strange 'religious' overtones to Awesome14's posts which certainly didn't help.

At the end of the day though, it is an epoxy packaged  REF102C in a non-hermetic plastic enclosure, together with some bog standard metal film (and even a carbon film in the one pictured) resistors, a thermistor and a couple of pots and speaker terminals. (whether it's cobbled together on a perf-board or on PCB is actually the least relevant factor).

Unfortunately it can never be better for long term and environmental drift than a non-hermetic epoxy packaged REF102C, because, even with second order temperature correction, that's all it is. There is no greybeard magic inside (even if you don't open it and let it escape).

[try]

Hello View

• Finder,

Ha, yes. There was quite a thread about that one a few years back, lots of controversy and mirth irrc, the internal construction was... interesting.  ;D

https://www.eevblog.com/forum/metrology/calibratory-d-105-dc-precision-voltage-reference-standard/

The listing looks the same anyway.


EDIT: Post with internal photos...  :palm:

https://www.eevblog.com/forum/metrology/calibratory-d-105-dc-precision-voltage-reference-standard/msg606722/#msg606722

It was produced by former (banned) forum member Awesome14. I can't believe he's sold 280 of the things!

The D-105 10V reference is much better

Better? In length? In weight? In long-term drift? in temperature coeffcient?

than the post history on eevblog would indicate. How do I know this to be true? Let's start with some facts about the posts:
1) The "Awesome 14" guy is the son of the old "gray-beard engineer" who built the device way back when (draw your own conclusions . . .)
2) I bought one of the $100 versions maybe 10 years ago and one of the higher precision ones (D-105) about 5 years ago
3) I had a long and informative phone call with the father/inventor when I bought the D-105 back then
4) I read all the post history that led up to the Awsome14 problem a few years back
5) In addition to the above two references, I have bought and tested pretty much all the high end (meaning $200-300) 10VDC references.

USD 200-300 is not high end. You do not state what you tested.

6) There are many folks on the Metrology section that know far more about the subject than I do, particularly in the area of stuff like long-term drift

Knowing is not difficult. Reading helps. Measuring drift is difficult.

OK, so my opinion about the the history of D-105 on this forum is that it was 80% about the attitude of "Awesome14" and 20% about the device's capabilities. The 80% got really nasty on both sides.

Don't try to justify anything. The "device" does not deliver the claims. See my link above. It's an overpriced scam from somebody who is not even able to reach an average level in welding. The pictures are horrible.

I don't know who is actually making the current D-105 and what the test environment and protocol are, so I will stop short of an opinion on the device. Based on the testing I have done over the years, I would rate my D-105 slightly better than Ian Johnston's PDVS 2 excellent adjustable voltage standard that was my prior preference.

Again, your undefined "better" does not contain the least bit of anything quantitative. So basically this is pure marketing talk.
Furthermore you are comparing apples and beans when comparing a fixed output reference with an adjustable one.

[Andreas]

/#msg3618917

I'll publish my results from testing over the last year when I finish.

Hello,

I hope that you have recorded the relative humidity along with your measurements.
Probably your results are optimistic since you live in a area where humidity does not change much.

Some measurements with ~40 % rH in winter and ~60-70% rH in summer can be found here:

https://www.eevblog.com/forum/metrology/tales-from-the-mouth-of-a-voltage-reference-ageing-box/msg3618917/#msg3618917

On a REF102 in DIP8 package I see ~6 ppm seasonal changes due to humidity when the reference is heated to 50 deg C every day for 1.5 hours during measurement. After a power outage additionally 6 ppm change. So 12 ppm for  a reference over the year which would be kept near room temperature.

with best regards

Andreas