EEVblog Electronics Community Forum
Products => Test Equipment => Topic started by: JP16A60 on July 04, 2016, 04:33:39 am
-
Picked up an EDC E100RC millivolt DC reference on eBay for a decent price, but calibrating it has been a frustrating experience.
Despite following known, good cal procedures I've "run out of room" on a number of the decade #1 adjustment pots (68xR20's), and many of the adjustments don't seem to "stay put", even after 4-5 iterations of the entire cal procedure.
I started to probe the myriad high-precision resistors, and it seems that almost every one of them is way out of spec:
Decade #1:
P1 resistor (19.99k, .02%) reads dead on at 19.99k
P2, P4, P6, P8, and P10 resistors (same spec as P1) all read between 15.995 and 15.996r
Question: How could five out of six resistors be *exactly* 4k out of spec? They are all very clearly marked 19.99k.
Decade #2:
2k .005% resistors all measure at ~1.6k
Decade #3:
200r .005% resistors all measure at ~160r
Decade #4:
20r .05% resistors all measure at ~16.48r
Decade #5:
2r .25% resistors all measure at ~2.04r
Has anyone ever seen anything like this before? Strangely, everything in D1-D4 seems to be at almost exactly 80% of expected values. I've measured with two different meters--one of them an HP 3456A.
I'd already made up my mind to re-cap this thing, but there's no way that I could afford to replace this many high-precision resistors!
JP
-
It could be the paralleled resistors making the reading smaller when measured online.
-
Seems your measurement setup is not good enough to deal with these.
How exactly do you measure resistances?
-
You could also try measure not the individual resistors but the resistance of a decade, how it varies as you change its setting.
-
Despite following known, good cal procedures I've "run out of room" on a number of the decade #1 adjustment pots (68xR20's), and many of the adjustments don't seem to "stay put", even after 4-5 iterations of the entire cal procedure.
P2, P4, P6, P8, and P10 resistors (same spec as P1) all read between 15.995 and 15.996r
Question: How could five out of six resistors be *exactly* 4k out of spec? They are all very clearly marked 19.99k.
Has anyone ever seen anything like this before? Strangely, everything in D1-D4 seems to be at almost exactly 80% of expected values. I've measured with two different meters--one of them an HP 3456A.
I'd already made up my mind to re-cap this thing, but there's no way that I could afford to replace this many high-precision resistors!
JP
I really doubt, that the resistors are the problem of your unit.
Anyhow, if you want to measure these resistors inside the calibrators circuit, you for sure will have (an)other resistor(s) in parallel, in this case this will be a 80k one. This may not affect the lower values as much.
To really determine these suspected resistors, you have at last to desolder the whole resistor string at one end.
Depending on the circuit of the 1st decade switch , maybe these ominous parallel 80k are always set up by the other four 20k resistors, so you would have to isolate each resistor at first.
Oh, wait a moment: Can it be, that all these resistors make a Kelvin-Varley-Divider (and not a simple linear resistor string)??
Then, at once these "odd" measurements make sense.
Make sure, that your 3456A is calibrated to much better than 0.02%, or even better than 0.005%, otherwise you won't see the error correctly.
As many trims are outside, and I'm strongly assuming all the resistors are fine, then you may first have to trim the voltage reference to its default value.
AFAIR, in these EDC standards, it's always a 1N829 based reference diode, which is divided down to a lower, even value. (Don't have this specific manual available). Therefore, the first step must be to calibrate this reference first, before you do the linearity calibration. In other EDC calibrators, it's required to trim two test points to a written-down value.
PS: Do you really follow the calibration steps inside the manual ?
PPS: If the calibration on the resistors seem to be unstable, this may as well be a problem with the zener reference circuit, maybe the switch for the different ranges.
Frank
-
I really doubt, that the resistors are the problem of your unit.
Yes, based upon the extremely consistent offset from "spec", it's now obvious to me as well that my measurement technique (and understanding of the normal, expected behavior this particular circuit arrangement) is grossly faulty.
To really determine these suspected resistors, you have to desolder the whole resistor string at one end.
I was thinking last night (after my initial post) that the only way to know for certain would be to unsolder one of the suspect resistors and measure it out-of-circuit, which would be a real pain--you've just saved me a bunch of effort!
Make sure, that your 3456A is calibrated to much better than 0.02%, otherwise you won't see the error correctly.
Yep--my 3456A was fully within factory spec as of about 20 days ago!
As many trims are outside, and I'm strongly assuming all the resistors are fine, then you may first have to trim the voltage reference to its default value.
A Zener spec of 6.1599v was written on a sticker next to the diode, and it measured pretty close to spec at 6.15916v (this was cal step #1). I was able adjust to 6.15994 without any trouble. I checked this again each time that I repeated the full cal procedure, and it remained rock-solid at 6.15994.
I had to twiddle the four main-board pots to figure out which one did what, as nothing is really marked on the board, and I do not have a complete schematic. One of the pots does nothing (it's supposed to adjust "sense zero"). Maybe this pot is bad? Sense zero was within spec, though, so not sure it mattered much.
PS: Do you really follow the calibration steps inside the manual?
Yep!
PPS: If the calibration on the resistors seem to be unstable, this may as well be a problem with the zener reference circuit, maybe the switch for the different ranges.
Can you recommend a specific brand or formulation of contact cleaner or lubricant?
Given that all of the resistors are likely still within spec (ref. my horrible measuring technique), it looks like this unit is still economically repairable, so I'm certainly willing to put a fair amount of work into this reference--do you think that it's worthwhile to disassemble the decade switches (and probably the whole front panel to do it) for a proper overhaul? Or do you think that a good application of contact cleaner/lubricant in situ would be sufficient?
How much of this instability do you think could be attributed to bad electrolytics? I haven't done a comprehensive evaluation of them, but I'd planned on re-capping the whole unit if the total cost of replacement caps was reasonable (i.e., under about $50).
JP
-
It's hard to isolate the fault, without having any clue about the circuitry.
So I don't recommend to "blindly" cleaning these switches, this is tedious and may worsen things.
There are schematics available for other, maybe similar EDC-calibrators.. often they are very similar.
Maybe you grab one of these, and try to compare with your unit.
It is not so difficult to find the fault by testing the right test points in the right settings.
If you have the manual, including calibration procedure, maybe you can put it online, so a comparison to other calibrators is possible?
It's a bit too much guessing at the moment.
Frank
-
Another separate uncertainty on my part (but which compounded my confusion and frustration):
There is a fifth physical 25-turn pot on the board (called out in the cal procedure as "pot #4"--the other four are called out as #1, #2, #2A, and #3). This is used to make the "full scale adjustment" on the millivolt range, which sounds pretty vague to me (see attached).
Do I set all decades to "10" and adjust to a value of 111.111mV?
If so, when I do this, it puts everything in the 10v range out by an obvious and measurable amount. I've tried adjusting to other values, such as 1.00000mV, 5.00000mV, 99.9999mV etc., with the same result.
I've attached the related service manual pages. Maybe I am missing some important, additional steps? My reference material on this unit is far from complete.
JP
-
So I don't recommend to "blindly" cleaning these switches, this is tedious and may worsen things.
Duly noted!
It is not so difficult to find the fault by testing the right test points in the right settings.
I'm sadly lacking in this skill, but perhaps this is a good example to learn with.
If you have the manual, including calibration procedure, maybe you can put it online, so a comparison to other calibrators is possible?
All EDC manuals that I've been able to locate:
https://www.dropbox.com/sh/kxbp54c8ban8xkb/AABxpRtEOYVievWF8v84VelQa?dl=0 (https://www.dropbox.com/sh/kxbp54c8ban8xkb/AABxpRtEOYVievWF8v84VelQa?dl=0)
I'm actually following the MVC106 manual for the cal steps, which seems to be nearly identical.
Thanks for your help on this, Frank!
-
I did calibrate and test my MV106 twice before, perhaps some of procedure and test data I covered in this article (https://xdevs.com/review/mv106_review/) may help give you some reference.
Still have the box, running 24/7, so I can test some specifics if you need.
-
For you all, dear sirs. I have a similar unit, and in my searches found the following two documents, the manual and service manual for the EDC E100RC. I unfortunately have the E100E Low Impedance MV Reference.
-
Another anomaly that I encountered when completing the linearity adjustment steps (see previously attached page 27/6.1.1) was that making specified "minor tweaks" (i.e., per the instructions), the original adjustment is thrown out of whack.
Dial 4 V on 1st decade adjust P4 exactly, look at 3 V for possible minor tweak of P4.
After making an initial adjustment (say to 4.00001v or 4.00002v), measuring and adjusting P4 for 3v (say to 3.00002), the 4v setting now produces 4.00023v. Basically, I get nowhere by following the 1st decade linearity adjustment steps as specified on page 27.
As an experiment, if I only make the major adjustments (in order: 2v, 1v, 4v, 6v, 8v, 10v), then the intermediary voltages (3v, 5v, 7v, 9v) are all off by 20-30uV. |O
JP
-
I did calibrate and test my MV106 twice before, perhaps some of procedure and test data I covered in this article (https://xdevs.com/review/mv106_review/) may help give you some reference.
Thanks, TiN.
Your procedures, documentation, and measurements are much more regimented than mine, but my overall steps mirrored those of yours. Very frustrating!
If I can think of anything specific to measure against yours, I'll certainly give you a shout.
JP
-
To anyone with more experience than me:
Would doing a comprehensive replacement of electrolytics be a better first step before diving down any further into this problem? I've fixed a lot of weird issues in other test gear just by replacing all of the old caps, but not sure if that's warranted here (at least in view of the calibration issue).
Earliest chip codes date back to 1985, so it's no spring chicken.
JP
-
The calibration instructions suggest the reference voltage should be within 100microvolts of that specified, which would be between 6.1598 and 6.1560. Do you think it is worth fixing this first?
-
I re-capped my unit, but it was initially in almost perfect agreement with my test gear, so I did not bother to (re)-calibrate it. Note that the re-capping did not affect the calibration in my unit.
-
The calibration instructions suggest the reference voltage should be within 100microvolts of that specified, which would be between 6.1598 and 6.1560. Do you think it is worth fixing this first?
Hi, rch. See my previous post:
A Zener spec of 6.1599v was written on a sticker next to the diode, and it measured pretty close to spec at 6.15916v (this was cal step #1). I was able adjust to 6.15994 without any trouble. I checked this again each time that I repeated the full cal procedure, and it remained rock-solid at 6.15994.
As far as I could tell from the documented calibration steps, I don't think that it's possible to get any more accurate than what I've done. Maybe I'm not understanding your comment correctly?
JP
-
For you all, dear sirs. I have a similar unit, and in my searches found the following two documents, the manual and service manual for the EDC E100RC. I unfortunately have the E100E Low Impedance MV Reference.
Great! Now we probably have the correct schematic!
It's very similar to the other EDC units, but it is more simple, and it has an OP37, instead of a chopper amp.
I doubt, that the capacitors might be the problem.
JP, is the PCB inside this manual similar to your unit?
Maybe you can take a photo of your units PCB, and the switches.
It's still the same linear resistor switching.. As the switches deliver 10k/ 1k / 100 ... Ohm per step, they are organized to put in parallel the different resistors, so you will always measure combined resistances, but never the single values.
So you might open the chain on one end, and measure the total resistance when switching through the different positions, and maybe also when trimming the decade 1 resistors.
If that is also rock stable, you can forget about the resistors, their switches, and the trim pots.
Frank
-
For you all, dear sirs. I have a similar unit, and in my searches found the following two documents, the manual and service manual for the EDC E100RC. I unfortunately have the E100E Low Impedance MV Reference.
Great! Now we probably have the correct schematic!
JP, is the PCB inside this manual similar to your unit?
Indeed, this is my unit!
I'll take some time to review this new document in greater detail, and will let everyone know if I can produce any different results. :-+
JP
-
The calibration instructions suggest the reference voltage should be within 100microvolts of that specified, which would be between 6.1598 and 6.1560. Do you think it is worth fixing this first?
Hi, rch. See my previous post:
A Zener spec of 6.1599v was written on a sticker next to the diode, and it measured pretty close to spec at 6.15916v (this was cal step #1). I was able adjust to 6.15994 without any trouble. I checked this again each time that I repeated the full cal procedure, and it remained rock-solid at 6.15994.
As far as I could tell from the documented calibration steps, I don't think that it's possible to get any more accurate than what I've done. Maybe I'm not understanding your comment correctly?
JP
You're right! Sorry, I missed one of your '9's!
-
SUCCESS.
(It really helps when you have the correct schematic.)
Taking a good, educated guess as to which main board potentiometer goes with which calibration step (even based upon logical and methodical cause-and-effect testing) is NOT a good idea. :palm:
A big thanks to all who offered constructive comments (and I think that all of them were), and especially to idpromnut, who posted the correct manuals.
The unit is not perfect, but it's good enough to know that it's close, it's working as designed, and that I can take it to my "real" calibration guy without him asking me why I've brought him something that was terribly broken. :o
In a couple of weeks, it will be good enough to put a real calibration sticker on it!
Next: Fluke 313A :box:
JP