Test Equipment Anonymous (TEA) group therapy thread

[beanflying]

That is 0.7 of one count in a minute of drift if you use only that snip of @Meds picture. My thoughts would be it is 'ok' for a sanity check and if there was figures for maybe 10 minutes+ showing that it had stopped that drift downward it would be ok as a number to read at 5.5 digits. I haven't checked the Siglent sheet at all but chances are that measurement is well within spec for the meter (most likely X% +-1-3 LSD?).

To use it to Calibrate from to that LSD you would want minimum some stats with averaging over days measured with that same meter to say the reference is good to X uncertainty. To then maybe use on another 5.5 digit meter and even then you really want to be 5 or 10X better than the meter you are trying to Cal with it which is the rabbit hole problem. At some point you have to declare enough or play with a Calibrated 3458A and then look at if it drifts or not

[med6753]

Hi Med,

some points to rant about (dont take it personal):

- the bias current of the 723s difference amp is not balanced. One one side, you just connected it directly to the reference, the other input comes from the divider.
- The reference should be filtered.
- Yes, you are using low tolerance resistors (0.1%, 20ppm ?). But the trimmer resistor is too large in value and probably has a much worse tempco.
- Maybe a precision trim (parallel trimmer from VOut to ground, tap connected to 723 neg. input via a high-value resistor) is better here.
  Trim range is smaller and more accurate, DC resistance can be balanced with the input resistor at the reference side.
- some prestab because of limited 723 PSRR is also in order.

Not taking it personal at all. This circuit is a direct copy of a voltage reference published in a 1982 article in Radio-Electronics magazine. The only thing I did over the years was change the 1% tolerance resistors to 0.1%. That's it. BTW, the 723 was made by TI.

Here's the 1V drift over time. The Siglent was warmed up for 1 hour.

At initial power on.


After approx 15 minutes


Finally settles here after 30 minutes or so. I can adjust it back up to 1.000V exactly but after a few months drifts again.


Edit.....this is the reason I now have the much more accurate and stable 190mV, 1.9V, and 19V references.

[Wolfgang]

Hi,

was your 723 a normal CN or a J (military) type ? I found that this makes a lot of difference regarding drift and stability.

https://electronicprojectsforfun.wordpress.com/silly-circuits/silly-circuits-a-heated-lm723-reference/
https://electronicprojectsforfun.wordpress.com/silly-circuits/silly-circuits-a-heated-lm723-reference/lm723-long-term-stability-results/

[Inverted18650]

Now you ll are talking my language, lets get VOLT_NUTS in this BEEEOOOTCH!

edit**more to come..gotta run a quick errand to prep for incoming bad weather.

[med6753]

Hi,

was your 723 a normal CN or a J (military) type ? I found that this makes a lot of difference regarding drift and stability.

https://electronicprojectsforfun.wordpress.com/silly-circuits/silly-circuits-a-heated-lm723-reference/
https://electronicprojectsforfun.wordpress.com/silly-circuits/silly-circuits-a-heated-lm723-reference/lm723-long-term-stability-results/

It's a 723CN.

[med6753]

@med you are well into the voltnuttery hole watching trend plots of VRefs about seals the deal 

Next step LM399 Vref for more stability then realise you cant measure its drift without 6 1/2 digits .........

I think the 2 AD584-M references I have use the LM399 but I'm not sure. And according to everything I have here to measure them they exhibit almost no drift.

For now my rabbit hole is stopping at 5.5 digits.     

[Wolfgang]

Hi,

was your 723 a normal CN or a J (military) type ? I found that this makes a lot of difference regarding drift and stability.

https://electronicprojectsforfun.wordpress.com/silly-circuits/silly-circuits-a-heated-lm723-reference/
https://electronicprojectsforfun.wordpress.com/silly-circuits/silly-circuits-a-heated-lm723-reference/lm723-long-term-stability-results/

It's a 723CN.

Last question: How long has it been running ?

[med6753]

Hi,

was your 723 a normal CN or a J (military) type ? I found that this makes a lot of difference regarding drift and stability.

https://electronicprojectsforfun.wordpress.com/silly-circuits/silly-circuits-a-heated-lm723-reference/
https://electronicprojectsforfun.wordpress.com/silly-circuits/silly-circuits-a-heated-lm723-reference/lm723-long-term-stability-results/

It's a 723CN.

Last question: How long has it been running ?

I don't keep it powered on. It is battery operated for portable use, such as take to a Hamfest to check a multimeter prior to purchase, or I can power it from an external source. I would typically power it up and let it stabilize for 30 minutes or so. Check/adjust voltage as required with a known good DMM. Check what I wanted then shut it off. 

[beanflying]

I think the 2 AD584-M references I have use the LM399 but I'm not sure. And according to everything I have here to measure them they exhibit almost no drift.

For now my rabbit hole is stopping at 5.5 digits. 

The AD584 is the clue They are one of the reasonable cheapies and are more than up to sorting out junk like these Victor 86B's I have on my Coffee Roaster (logging USB outputs) The meters are from the factory horrid and out by 10-30mV at 10V which on the roaster doesn't matter as you are looking at relative Temperatures to preceeding roasts so 'accuracy' doesn't matter but consistency does.

I have a voltage logging job I wanted to use them for so time to see about tweaking them closer to the truth and sort out what the pots available are as there is no Calibration info out there. No need to breakout anything more serious for these 3 1/2 digit ones 

[med6753]

Well yea, they don't measure exactly 10V. They have stickers on the back indicating their voltage. For the 10V one of them measures 9.99691V and the other 10.00096V. From what I've been able to determine that's correct. What's important to me is that they are stable with no drift as far as I can measure. And that seems to be the case. But I supposed once you get into 6.5 digit territory that might not be the case. 

[Inverted18650]

I read and got caught up the topic seems beaten to death by the time I joined. SAD FACE. The negative temp co is certainly interesting.

If you are interested in building another VRef, I have a VRE3050B that was donated from another forum user than I can send your way. I have a few ADR4540BRZ and LM4140ACM. I ran them all through the LTC2400CS8, 24-bit ADC. I tried a couple different high end, zero-drift op amps for voltage following and buffering but I have none of those leftover. Plenty of cool details in the Meteorology thread about the variety of op amps and results vs. cost.

I also snagged a bunch of beautiful Caddock thin film resistors that are 5 ppm/C (most are better) and couple of the divider networks as well. I haven't done much with yet. If you want to build a new unit, let me know. I sent a handful of the Caddock units to xDevs.com for testing and he put the results on his website.

here is the datasheet for the VRE3050B:
https://www.apexanalog.com/resources/products/vre3050ds.pdf

here is a link for the Caddock resistors test results:
https://xdevs.com/article/caddock_news/

[beanflying]

I have actually just remeasured mine carefully after about a year of use 5 or 6PPM of drift was all I got so they are more than up to 4 1/2 over a few years and a good sanity check on your Siglent.

If you felt like it run them for an hour each and save the logs then check them over time you should then be able to see how they track relative at least to the Siglent as it gets some age on its own reference too.

[med6753]

I read and got caught up the topic seems beaten to death by the time I joined. SAD FACE. The negative temp co is certainly interesting.

If you are interested in building another VRef, I have a VRE3050B that was donated from another forum user than I can send your way. I have a few ADR4540BRZ and LM4140ACM. I ran them all through the LTC2400CS8, 24-bit ADC. I tried a couple different high end, zero-drift op amps for voltage following and buffering but I have none of those leftover. Plenty of cool details in the Meteorology thread about the variety of op amps and results vs. cost.

I also snagged a bunch of beautiful Caddock thin film resistors that are 5 ppm/C (most are better) and couple of the divider networks as well. I haven't done much with yet. If you want to build a new unit, let me know. I sent a handful of the Caddock units to xDevs.com for testing and he put the results on his website.

here is the datasheet for the VRE3050B:
https://www.apexanalog.com/resources/products/vre3050ds.pdf

here is a link for the Caddock resistors test results:
https://xdevs.com/article/caddock_news/

Appreciate the thought but I have no intention at this time of rebuilding that old reference since I now have the recently built 190mv, 1.9V, 19V, 190V, and soon to be completed 500V references.

[med6753]

I have actually just remeasured mine carefully after about a year of use 5 or 6PPM of drift was all I got so they are more than up to 4 1/2 over a few years and a good sanity check on your Siglent.

If you felt like it run them for an hour each and save the logs then check them over time you should then be able to see how they track relative at least to the Siglent as it gets some age on its own reference too.

These 2 AD584's have ALOT of hours on them. And I've noticed no degradation in their output. But I have noticed that the more hours I put on the Siglent the faster it warms up and it's overall accuracy appears to be better and stabilizing.   

[Inverted18650]

Nothing old about these chips. I used them to build a damn nice Arduino 5.5 digit DMM, and a calibration unit that can output 50VDC (all based off the Scullcom Hobby Projects, YouTube channel; I just tweaked them a bit). I also recreated his DIY DC load and had a great time with it. It became the base model 18650 tester I used for about 8 months.

I think I am going to be good for a while (as far as CAL UNITS go), so the offer is extended to anyone who wants to build a fun project and see where they can take it. Like I said, the blueprints and all credit go to the Scullcom Hobby YouTube channel, it will guide you seamless through the project. Some great HOMEMADE TEA there.

[Inverted18650]

I have actually just remeasured mine carefully after about a year of use 5 or 6PPM of drift was all I got so they are more than up to 4 1/2 over a few years and a good sanity check on your Siglent.

If you felt like it run them for an hour each and save the logs then check them over time you should then be able to see how they track relative at least to the Siglent as it gets some age on its own reference too.

These 2 AD584's have ALOT of hours on them. And I've noticed no degradation in their output. But I have noticed that the more hours I put on the Siglent the faster it warms up and it's overall accuracy appears to be better and stabilizing.

I didnt look closely at the circuit (at all really, guess I should go do that), is it ovenized as well?

[med6753]

I have actually just remeasured mine carefully after about a year of use 5 or 6PPM of drift was all I got so they are more than up to 4 1/2 over a few years and a good sanity check on your Siglent.

If you felt like it run them for an hour each and save the logs then check them over time you should then be able to see how they track relative at least to the Siglent as it gets some age on its own reference too.

These 2 AD584's have ALOT of hours on them. And I've noticed no degradation in their output. But I have noticed that the more hours I put on the Siglent the faster it warms up and it's overall accuracy appears to be better and stabilizing.

I didnt look closely at the circuit (at all really, guess I should go do that), is it ovenized as well?

Which one? The 723 reference that I built many years ago is not. Neither are the references I just built. And the AD584's aren't either.

[Inverted18650]

Yeah I just looked. Shit man it works though...As long and you can trim the unit after it warms up and it holds that value for a few years, it worked damn well.

[Inverted18650]

So I got my HP 3245A today ladies and gents. I actually got out of bed a bit earlier and opened it up, powered it up, and played with it just long enough to post a couple pics on the forum. (been depressed again and struggling) But this unit is much more than I expected. Not cheap, but damn, its so packed full of features and just the ARB GEN functionality sent my brain into happy mode.

[beanflying]

Good 'Nough for 'stralia 

Going on to a 12V AGM charge and discharge job so they will do the job nicely. 27 Degrees in the shack AD584 is -1 PPM from when I tested it and even with three 10 Meg meters hanging off it the voltage stayed rock solid to 0 PPM change to just the Agilent swinging off it.

[med6753]

Yeah I just looked. Shit man it works though...As long and you can trim the unit after it warms up and it holds that value for a few years, it worked damn well.

Yep, that original 723 reference did a damn good job keeping the 3.5 digit crew honest. But as I started collecting 4.5 digits I knew it was barely adequate. And for the Siglent SDM3055 and the Fluke 8810A forget it.

The key is to keep written records and do cross checks. That way you'll know immediately if someone has got a problem. When I did the recent calibrations I discovered that the 20K and 200K ohm range on the Fluke 8600A is suddenly out of spec. It was OK last year. That guy has been a problem child anyway and has needed 2 other repair actions on the DCV. So I have to rip into again and figure out what the issue is.
 

[Inverted18650]

Good 'Nough for 'stralia 

Going on to a 12V AGM charge and discharge job so they will do the job nicely. 27 Degrees in the shack AD584 is -1 PPM from when I tested it and even with three 10 Meg meters hanging off it the voltage stayed rock solid to 0 PPM change to just the Agilent swinging off it.

27 degrees! hum...Ill trade ya. 3 here right now.

No denying the AD584 hangs tough...shit now you can get a prebuild unit on evilBay for $3USD that is plenty good enough for 3.5 digits.

[beanflying]

The very cheapies seem to be more of a mixed bag over the LiPo/Acrylic KK Moon types. Mine turned up without any readings on the case and that seems to be far more common compared to when they first came out in that style like @Med got.

In most cases it saves me having to fire up any of the better Calibration gear and it just sits on the shelf for bench use which in truth is all I need in 95%+ of cases when teamed with the Agilent.

Agilent II the 33461A has left the Sellers so new insanity checking will likely commence Monday when it doesn't agree with the 34401A

[Mr. Scram]

So I got my HP 3245A today ladies and gents. I actually got out of bed a bit earlier and opened it up, powered it up, and played with it just long enough to post a couple pics on the forum. (been depressed again and struggling) But this unit is much more than I expected. Not cheap, but damn, its so packed full of features and just the ARB GEN functionality sent my brain into happy mode.

Being depressed is no bueno, but sometimes it just takes you for a ride. If you're ever in a bad place, be sure to reach out.

[GregDunn]

Quote from: med6753 on Today at 09:29:15 pm

Well yea, they don't measure exactly 10V. They have stickers on the back indicating their voltage. For the 10V one of them measures 9.99691V and the other 10.00096V. From what I've been able to determine that's correct. What's important to me is that they are stable with no drift as far as I can measure. And that seems to be the case. But I supposed once you get into 6.5 digit territory that might not be the case.

I have bought 2 different models of the AD584 reference and both of them seem to be spot on with the marked values.  I'm sure others can chime in with examples where they aren't, but so far good luck with mine.  And yes, they're nice and stable even though I don't power them up constantly.  It'll be interesting to compare them with branadic's LTZ1000 reference once I get that lovely HP 3456A.