Restoration : Wavetek 4920/4920M teardown, repair and usage cases, come join!

[TiN]

Now my humble lab welcomes those rare green-glow Datron/Wavetek magical pieces. No, not another 8.5-digit Datron 1281... 

We talking AC-nuttery this time, time to open new pandora box for me. Yes, it's Wavetek 4920! And not just any, but military version of it, 4920M. 
What's different from "regular" 4920? From what I found out , M-version have next differences:

* Both N connectors for input, instead of N for WB and banana for regular
* No low mV ranges option
* More scarce menu system and few differences in features

I hence summon all Datron/Wavetek 4920 and 4920M owners to join me in this ACV-nuttery journey begin.. I know you folks are out there, hiding gems behind DC gear! 

There was recent sale of 5pcs Wavetek/Datron 4920M. Instead of waiting and hoping that some voltnut from EEVBlog will snatch one and show us it's insides, I decided to close ebay shopping year 2016 for me with this last bit, which happened to have more recent Wavetek-only branded 4920M. Is it a bust for 592$ or a "winner-winner chicken dinner"? Well, in worst case it's just another box with LTZ1000 reference inside and nice glowing VFDs 

After some waiting on GSP shipping and delivery (which is great btw, packing on GSP is top notch!) box just arrived today. Excited? Hell, yea.

Will it be another 220+ hours repair project, like rusty 3458A? Perhaps even more so, as I have not much experience in AC stuff, don't have any thermal converters, and only calibrated measurement piece of ACV/ACI I have is Keithley 2001 with Tek cal from 2014. My 3458A had never ran SCAL, nor 2002's completed AC calibration. Anyhow, stay tuned, will be fun! 



Break the rule "take it apart" and actually power this unknown black cat to see if she's alive.



VFD is unusual nice and emerald green. I actually dig it, looks way better contrast than Keithley's/3458A. Big digits too! I think it's same exact front panel as used in other 1271/1281/4950 Datrons.



Last cal January 2015? Really? Too good to be true, I say.



And calibration sticker says 2009. Sticker on the rear tells to replace battery in 2012, so it's 4 years overdue or someone did not bother to update sticker?

Running internal self-test reveals only one error: 4053, WB calibration validity..



No other issues reported. Running WB Wlinearity self-cal (lucky me, it's the only calibration available which does not require fancy AC sources and gear).



After ~10 minutes, self-test pass without single error..



Is it like a fairy-tale and one of those repairs where nothing was broken?

Only one way to find out, connect ACV input and see what's out there.
* Set 5.66ACV 50Z on my trusty HP 3245A source, which gives pretty close to 2V with 50ohm load. Frequency for test is 100e3 Hz



2.122V? huh? 

What HP 3458A has to say about it?



1.9984V! But we never calibrated HP DMM for AC.



1.9962V from Keithley... Now i'm sorta believe those two folks more, not the Wavetek... Uh, oh.. 

I ran manual step tests via Pathway section as well.



All steps pass with range stated in service manuals (Thanks to ManateeMafia for scans!), except one :
Test 4035 (PA22) : AC RMS Conversion - Linearity (10V)
Expected in manual result: 2.350 < x < 2.450 V
Unit reported : 0.242792
This one sounds somewhat important don't you think?

So I tested various input signals 1kHz from Time Electronics 9823 calibrator on various ranges.

2V on 10V range gives 2.007988V (K2001 reports 2.00053)
same 2V on 30V range gives 3.30198V. What? 



Similar problem on 200mV range 



Few more steps:





Next step - you already know it 

[TiN]

Removed top cover. Ta-daaa...



Digital board and it's SMPS supply is located on rear top side, analog ADC circuitry board is located in front of it.
Wideband module fully enclosed in metal can box on the side of it.



Volt-nuts easily spot the magical parts - Datron main reference module with Linear LTZ1000 + Vishay foil network resistor in hermetic TO-package and traditional LT1013 in hermetic can as well.
It's same module used in Datron 1271/1281's. Lonely LM399 is used as auxilary reference to check drift and output levels of main one. AFAIK it's used only for self-test procedure.



More parts around.



Processor is good old friend, Motorola 68000.

Firmware lives here:



And battery bodged up by someone else already. Still reads 3.04V, so it's sorta ok. Will replace anyway...



And none of those calibration keylock trash, usual plain switch, easy! (maybe that's not native though?)





They are not joking, everything screwed down, harness tied and locked in place, overall very nice construction.



Analog board PSU is linear, using jellybean 78xx and 79xx regs + LM337. Pretty common. Capacitors are subject to be replaced ASAP



Digital board talks with analog world only via 5 optocouplers and FARNELL TESTED ribbon cable



Let's remove front panel cover.



Well it was like this, I swear. Nothing superglue can't fix though Also check those ties around connector, so FP keyboard interface cable would not pop out unexpectedly. Nice attention.



And - What's that screw doing there? It's just stuck to mylar. And I don't see any place where it could come from, it's too long!



Oh well..

Let's see some real stuff now.

WB module



Well, copper rigid coax, thank you very much. No that cable junk here.
AC 50 ohm load is nice too, nothing less than array of 3xx ohm Vishay S102C foil resistors. Make sure that load is stable!



Then we have some DIP packages around and output circuitry..



And surely, we have TVC cell under that metal can, nicely protected and soldered tight all around. I'm not brave enough for now to remove the shield for sake of photo



Now, assemble everything back in (wearing gloves at all times, yes) and let's see what waits for us on bottom side of the instrument.

And there are: dual transformers, very similar to 12x1 Datrons and fancy Fluke 8508A.



AC board with bunch of packages and MAT transistor arrays.





Plenty of free space too:



And famous "Paffner" time bomb. Will order new filter with caps too. I don't want fireworks!



Teflon material PCB! What's there? ACV main attenuator network and floating guarded preamp! Why Teflon? Because it provide very high insulation and good stable frequency responce.



Carefully remove the board to see the fancy voodoo...



Oho-ho.... Isn't that worth every penny paid? Look at those monster Vishay (AE back then) foil resistors!



They are just gorgeous..



Well. I can't write anymore, TBD

[TiN]

And here goes post with stuff I did not figured out yet 

[pelule]

VFD is unusual nice and emerald green. I actually dig it, looks way better contrast than Keithley's/3458A. Big digits too! I think it's same exact front panel as used in other 1271/1281/4950 Datrons.

Yes, that's true - Datron used same front paneö/display board for 4920, 1271 and 1281 and I guess for some other tools also.

After ~10 minutes, self-test pass without single error..

Out of my experiance the tool need to warm up >2 hours before running the selftest - neded to get stable temperature.

As the selftest runs fine, the basic operation is ok, thus I assume some recap is needed.

[TiN]

I'm aware of usual precision gear warm-up times It was sitting ~3 hours, but no warming up would cause 1.3V difference of same input signal in two ranges 
Unless that's is "normal" for these AVMS, which I doubt!

[quarks]

Bookmark
Looking foward to see the inside

[ManateeMafia]

I will get out my 4920 (non M version) for some comparison tests.

They should be nearly identical in most respects barring any options. I just need to get a suitable source set up 

Of note: my unit runs a POST at power on including ADC linearity. No warmup time required but the limits are probably pretty wide.

[Vgkid]

Went to see about reading a manual for this, and TiN's site is throwing an error :/ .

[TiN]

Of note: my unit runs a POST at power on including ADC linearity.

Same here, it does short version of selftest once powered.

There is plenty space to add some extra PCBs into chassis if desired too No noisy fans too, which also means no dust in unit. 
Ordering capacitors and Delta mains filter tomorrow at DigiKey. 

TiN's site is throwing an error :/ .

What error?

[TiN]

Uploaded pics in second post. Beauty! If anyone have more information about magic stuff there - do tell! 

[doktor pyta]

Great stuff
Do You have part of the schematic showing what is inside the metal can ?
Some day I will show pictures of my vintage thermal transfer standard...

[dacman]

A Cal Validity error means it's past its due date.  It's not a hardware nor measurement error.  My guess as to why the readings are off is that someone adjusted it that did not have the proper standards.  One clue is that the calibration label dates do not match the dates in memory.

[lowimpedance]

Nice bit of kit, in good shape inside but whats up with the outside !............. never ceases to amaze me the treatment such precision gear get. Well it is the 'M' version  .
Which also goes some way to explaining the 'spare' screw, just in case another gets lost you know  .
Anyway its still a winner. 

[ManateeMafia]

Simple test of 2v @1kHZ  on 10V range 2.000270 VAC

same signal on 30V range = Error Ur

It seems the fw will not measure below a certain range input value (10%?) but yours is off enough that you don't get that message.

[zlymex]

Enjoying the show, even for reasons of very nice and big photos, as always.

[dacman]

A footnote for the specifications for the unit is that the specifications are valid from 30% to 110% of range.  It has been my experience that it will under range at about 10% of range, although, 0.99 Vac to < 15.0 Vac is supposed to be a valid input for the lower calibration point on the 30 Vac range with 10 Vac being nominal.

It appears that the Ur point is 9% of range (2.7 on the 30 Vac range, and 0.9 on the 10 Vac range).

[Vgkid]

TiN's site is throwing an error :/ .

What error?

It was throwing a 404 error earlier, but it is working now.
That attenuator board is rather dreamy. Interesting to see the hermetic network in a sea of hermetic resistors.
Interestingly it appears that the meter is using the rather ubiquotous squaring + exponential amplifier in the process of rms conversion. But the more interesting side does not involve this...

[zlymex]

I got three questions:
- Are those big fat hermetic resistors also from Alpha Electronics just like the smaller ones?
- Was Pickering Electronics(marked on the black relay) related to John R. Pickering(of Datron) in any way?
- Are all those plastic foil resistors from Vishay?

[TiN]

1. I think so, BOM lists them as Alpha as well.
3. Did not check every single one of them, but ones I looked at are Vishay S102C's. 

Bit more reversing on TVC in WB module. It's manufactured by US company Best Technology which is still seem to be around, and have insulated UHF-type hermetic thermal converter used here. One used in 4920M is listed as insulated 5.0mA type, likely S.7 type, as TVC DC output amplifier is configured to provide signal gain of 867 (1 mV output from TVC will be amplified into 867 mV) by LT1028 ultra-low noise amplifier. Gain is set by fixed 8.66 k and 10  R203/R204 resistors.
Higher voltages, like 10mV would cause opamp to saturate due to it's +11VDC and -19VDC power supply. Unless my math is messed up somewhere.

[acbern]

At a theoretical 10mV output the TVC would already be destroyed. They are very sensitive, I cooked one just by a short switch spike of my Datron 4808, although I was aware of the potential issue. So should you do any tests, be very carefull. Also, not sure the S7 would be a correct replacement, more usual are 200ohms/V types, although of course this may be totally different here.

[TiN]

Here are my DUMP? constants output on 10V range:

4920M log
DUMP? Range 10.0 V, 0 = 001,+0.3000000E+01,+0.3000000E+01
DUMP? Range 10.0 V, 1 = 002,+1.0000000E+01,+1.0000000E+01
DUMP? Range 10.0 V, 2 = 003,+0.0000000E-40,+0.0000000E-40
DUMP? Range 10.0 V, 3 = 004,+0.5000068E+05,+0.8776189E-01
DUMP? Range 10.0 V, 4 = 005,+0.2000027E+06,+0.8704789E-01
DUMP? Range 10.0 V, 5 = 006,+0.5000067E+06,+0.8533230E-01
DUMP? Range 10.0 V, 6 = 007,+1.0000137E+06,+0.8916546E-01
DUMP? Range 10.0 V, 7 = 008,+0.0000000E+00,+0.0000000E+00
DUMP? Range 10.0 V, 8 = 009,+1.0000001E+00,+1.0818163E+00
DUMP? Range 10.0 V, 9 = 010,+0.9765597E+00,+0.0000048E+00

And here are good ones from calibrated 4920, read by fellow ManateeMafia.

Good 4920 log
DUMP? Range 10.0 V, 0 = 001,+0.3000000E+00,+0.4175232E+00
DUMP? Range 10.0 V, 1 = 002,+1.0000000E+00,+1.3915722E+00
DUMP? Range 10.0 V, 2 = 003,+0.9999817E+05,-0.2553497E-02
DUMP? Range 10.0 V, 3 = 004,+1.0000093E+04,-0.8196756E-04
DUMP? Range 10.0 V, 4 = 005,+0.2000019E+05,-0.3578179E-03
DUMP? Range 10.0 V, 5 = 006,+0.5000044E+05,-0.2259891E-02
DUMP? Range 10.0 V, 6 = 007,+1.0000088E+05,-0.8410558E-02
DUMP? Range 10.0 V, 7 = 008,-0.0001296E+00,+0.0002198E+00
DUMP? Range 10.0 V, 8 = 009,+1.0012110E+00,+1.1019141E+00
DUMP? Range 10.0 V, 9 = 010,+0.9765555E+00,+0.0000049E+00

Those three first values on the right shows me that calibration is lost 100%, as they are defaulted now.
Quick set-cal to Keithley 2001's reading allowed 4920M to show correct value, and updated constants as:

4920M log after pre-calibraiton
DUMP? Range 10.0 V, 0 = 001,+0.2997650E+01,+0.3263413E+01
DUMP? Range 10.0 V, 1 = 002,+1.0000350E+01,+1.0880032E+01
DUMP? Range 10.0 V, 2 = 003,+0.0000000E-40,+0.0000000E-40
DUMP? Range 10.0 V, 3 = 004,+0.5000068E+05,+0.8776189E-01
DUMP? Range 10.0 V, 4 = 005,+0.2000027E+06,+0.8704789E-01
DUMP? Range 10.0 V, 5 = 006,+0.5000067E+06,+0.8533230E-01
DUMP? Range 10.0 V, 6 = 007,+1.0000137E+06,+0.8916546E-01
DUMP? Range 10.0 V, 7 = 008,+0.0000000E+00,+0.0000000E+00
DUMP? Range 10.0 V, 8 = 009,+1.0000001E+00,+1.0818163E+00
DUMP? Range 10.0 V, 9 = 010,+0.9765597E+00,+0.0000048E+00

So, no need to worry about saving calibration NVRAM on this box. 
Seems like cal data was reset/lost, but functionally it have no problems (remain to be confirmed, it could have excessive noise/drift/whatever problems, not diagnosed by self-test).
Now we face bigger question - how to calibrate this unit..

[acbern]

I also bought one of these. Mine worked, said CAL invaild and it also seemed reading incorrectly. Battery was ok, replaced it anyhow (without loosing any data). Then I started to calibrate ist, using a Datron AC calibrator, just for the sake of checking. What was strange is that when doing the CAL sequence where you need to apply specific frequencies at each voltage step, it does calibrate, and is fine at these frequencies, but when I apply e.g. 100kHz instead of 200 (where it was spot calibrated), it is off by quite a bid (several percent). That is wrong.
So my recommendation is you do a CAL with an AC source based on K2002 or 3458A (preferred; SYNC mode) and check what happens when you test the frequency range. Maybe there are other issues, just as the ones I see. If not, it may be worthe sending it to CAL (if you cannot do it yourself). I use a Ballantine 1605 for this (if I ever get there).

[TiN]

Same seller? I see same thing. But that's expected, as flatness and other special cal constants are lost too. Neither of my 2002s or 3458 had AC calibration, so most trusted is K2001 which had full cal almost 3 years ago. I can calibrate HP 3245A (sources up to 20Vpk-pk 1Mhz or 200Vpk-pk 100kHz) but it's nowhere near required accuracy.

So in the end I'll likely poke local cal labs if they take this 4920M for full cal+adjustment, as I don't see much of other cost-effective choices here.

[pelule]

Have tried to upload the doc to KO4BB, but failed.
Thus I uploaded them in my Dropbox:
 - Datron 4920M_service_manual_Vol-1 & 2
 - Datron 4920 AVMS Users Handbook
 - Datron 4920 Design, Application and Performance
 - Datron 4920 Calibration

If someone have interest, here is the link (will be valid until Sunday).
https://www.dropbox.com/sh/8y662mm9ckdzzat/AACGJvTBm1Nxm1hD_TcC-vrha?dl=0

[lowimpedance]

Have tried to upload the doc to KO4BB, but failed.
Thus I uploaded them in my Dropbox:
 - Datron 4920M_service_manual_Vol-1 & 2
 - Datron 4920 AVMS Users Handbook
 - Datron 4920 Design, Application and Performance
 - Datron 4920 Calibration

If someone have interest, here is the link (will be valid until Sunday).
https://www.dropbox.com/sh/8y662mm9ckdzzat/AACGJvTBm1Nxm1hD_TcC-vrha?dl=0

The 4920M vol 1 and 2 are on KO4BB but not the others. Maybe have another go with those, anyway thanks for sharing.

[acbern]

Same seller? I see same thing. But that's expected, as flatness and other special cal constants are lost too.

No, thats not expected. You calibrate 50kHz, 200kHz, 500kHz and 1MHz, and inbetween it should be flat. Mine is not. So something else is wrong too. If you have that issue as well, then you should not go to a CAL lab yet. So I recommend you calibrate based on your K2001, and then apply test signals between calibrated spot frequencies, just to check.

[zlymex]

One of my Chinese friends also bought one from the same seller(the $481 one). He hasn't got it yet but seeing from the photo of the transshipment the unit is a 4920 instead of 4920M

[TiN]

pelule
You are the man. Thanks.
I rehosted all files on 4920M article (in works).

zlymex

He hasn't got it yet but seeing from the photo of the transshipment the unit is a 4920 instead of 4920M

Maybe his got millivolt option then?
Invite your friend to join us here  EEVBlog metrology section would not be even remotely as good without your contributions 

acbern
Ok, I'll check more in details with some python snake help.
Perhaps 10Hz - 1MHz in 10kHz steps on 1V,3V,10V ranges would suffice?

[acbern]

TiN,
actually if you want to make that many steps, thats fine, but for starting, 50k or 100k steps is probably more than enough. I see this when calibrating at the 4 given steps (10V; 50, 200, 500, 1000kHz) and testing inbetween (1kHz, 100kHz)... So even works with just a few manual checks, even no SW needed initially.

[pelule]

The 4920M vol 1 and 2 are on KO4BB but not the others. Maybe have another go with those, anyway thanks for sharing.

Uuups. I have checked KO4BB first, but didn't found them. Anyway, a 2nd place is better than none.

[zlymex]

Uuups. I have checked KO4BB first, but didn't found them.

The 4920 manuals are not under Datron or Wavetek sub-directory, have to search.......

[zlymex]

zlymex

He hasn't got it yet but seeing from the photo of the transshipment the unit is a 4920 instead of 4920M

Maybe his got millivolt option then?
Invite your friend to join us here  EEVBlog metrology section would not be even remotely as good without your contributions 

Thanks TiN. 4920(without M) probably got the millivolt option, he is even glad when he knows the non-M version. Some of my friends visit EEVBLOG often, but they seldom/never speak here probably because of the language.

[acbern]

It turns out the 4920 had not only lost its special cal constants but also all of its normal routine calibration parameters (even though the battery was good, maybe someone deleted it actively). That looks as if it explains the deviations at different frequencies. So doing routine cal and then special call seems to result in correct behaviour.
What is strange though is that in the 30V and 100V range, where the 1MHz routine cal is made at 20V only (not its maximum, due to limitations of calibrators), I need to use SET to set the voltage to 20V, otherwise it calibrates at erroneous voltages. This however is not in the documentation, and the other levels (e.g. in the 30V range at 10V and 30V 1kHz) are recognized automatically. Can anybody confirm?

[TiN]

That what I was referring to, all constants are lost, so full special+normal recalibration is required. I don't have source capable to generate 20V 1MHz, but I'd expect your assumption is correct, based on manual description of calibration procedures and function.

[TiN]

Today's haul. Guess what will be happening this weekend?



Also spot the special type of the part on the photo for the free bonus!

[Vgkid]

I'm curious to know, what you use that oled display for.

[Dr. Frank]

Today's haul. Guess what will be happening this weekend?

I already know, what's going to happen, so I'm out of this contest..

But that RF connector stuff looks like mine, when I'm doing precise AC/DC transfers. 

Frank

[TiN]

Dr.Frank

The only catch though, that there will be no AC/DC transfers happening, as there is no TVC's anywhere near my reach.
It will be AC/AC stuff purely. More to come soon

Vgkid
I'm curious about that too. Saw it on the shelf during check-out, couldn't resist not to grab one.

I'll cut the suspense on the special type - it's those red capacitors, Wima MKC. Didn't expect to see them in stock, I heard polycarbonate film caps are out of production long ago. So to feed the greed, bought 10pcs to collect dust on my shelves

[lowimpedance]

Sorry for the thread spam...but..
 Everyone should have a stash of carbs in the cupboard  .
Quick peek at some well aged ones in my pile (not stale yet i don't think !).
And a bonus pic of a 'fat' styrene next to a carb   (also need a good styrene stash to share the dust too!)

[TiN]

We want to know the story behind metal box with array of switches 

[lowimpedance]

Ahh.... some things are best kept secret 

[TiN]

Time to hit the road. Mobile calibration lab, lol. 3458 and 4920 are about to see some 360km/h train love. 36kg , just below 40kg limit .



Maybe USA has ebay, but Taiwan have it's pros too, it's small and no need airplanes to visit other city.

I hope I'll be allowed to take few photos of calibration mayhem planned to happen

[TiN]

Phew, 4920M is back home now...



mini Cal-fest TW was a blast, very happy with this weekend. Spent 20 hours or so cooking python snakes, talking to calibrators and Fluke AVMS, and learning 4920M cal quirks.
I now sorta understand why Tektronix wanted 7000$ for 4920M cal.

Did not had enough time to run WB cal or extra steps, so I'll still have to calibrate few things, but I can do those at home using my 3245A and 3458A w/Swerlein's algorithm for precise AC.
Most difficult high voltage ranges are calibrated now. I ran both low-level calibration and user calibration as you may remember, cal rom was erased in this unit.

I'll post more info once I get everything prettified.

Btw, 4920 is very capable instrument. During second user-cal check, I was watching for readings, they match well <10ppm on each range/freq (1kHz,1MHz) to 5790A. So these old gems are truly capable.

[lukier]

Did not had enough time to run WB cal or extra steps, so I'll still have to calibrate few things, but I can do those at home using my 3245A and 3458A w/Swerlein's algorithm for precise AC.

I googled Swerlein's algorithm, read the paper and googled if he published some other interesting papers, but found much darker history instead:

http://www.thedenverchannel.com/news/homeowner-arrested-after-explosives-found-in-longmont
http://www.dailycamera.com/ci_16600312

[pelule]

Here are some "Swerlein's algorithm" documents. It may help.

[Dr. Frank]

I've salvaged the Swerlein program, an executable file, and already sent TiN.
Whoever else is interested, may contact me
Frank

[plesa]

Here are some "Swerlein's algorithm" documents. It may help.

Thanks for sharing. Is there any reason why this algorithm is not part of firmware 3458A?

[TiN]

I think 4920 are underrated instruments. During calibration readings to 5790A were more close (<10ppm usually) than I expected. 

[Vgkid]

Good job on the fix, how long did it take for you to cal the 4920?
One of the local labs has a 792A, and a 5790A.

[TiN]

Well, its a difficult question.
Techically calibration is ~30min for user calibration and about 2 hours for manufacturing one, however this is with automated setup with Python snakes and using 5720+5790A. If one after best accuracy, TVC or 792A would make things much slower with nV-meter. I'd plan for at least 2 full days, if you have working software to do all steps.
Also its picky so some steps may not pass, few multiple runs would be needed.

You also need 5700/5720 and 5725 booster to calibrate 300V and 1kV ranges, as base 57xx cannot go >219.9999V for >1kHz. On positive side, you can leave those ranges out , as calibration steps are per range/freq.

I heard there may be software available somewhen next year to do such 4920 cal

[ManateeMafia]

I hope so, looks like my cheap calibration option is finished  . Now I will have to ship it to Taiwan.

[TiN]

Too good for too cheap never lasts. So NIST TVC is in the remote distant plans.

[acbern]

One thing to add, and TiN has realized it anyhow, e.g. even the 90day uncertainties in the 100 to 500kHz frequency range of the 5720, where most of the calibration takes place (1MHz ratio is not much better), are about a factor of 10 as big as those of the 4920M. So the Fluke is totally inadequate to properly calibrate the 4920M. You really need a Fluke 792 or a Balantine 1605. (Or, if time is not an issue, and you have a lot of tranquilizer and the reverse switch setup in your tool box, a set of calibrated TVCs). And the 792 has no GPIB, so lots of manual work (the Ballantine has; a very underrated device...)

[CalMachine]

One thing to add, and TiN has realized it anyhow, e.g. even the 90day uncertainties in the 100 to 500kHz frequency range of the 5720, where most of the calibration takes place (1MHz ratio is not much better), are about a factor of 10 as big as those of the 4920M. So the Fluke is totally inadequate to properly calibrate the 4920M. You really need a Fluke 792 or a Balantine 1605. (Or, if time is not an issue, and you have a lot of tranquilizer and the reverse switch setup in your tool box, a set of calibrated TVCs). And the 792 has no GPIB, so lots of manual work (the Ballantine has; a very underrated device...)

I recently purchased a Datron 4950 I plan on bouncing my 4808 off of to cal my 4920 when it comes due!  It will be my first time self calibrating it...  My most recent cal came from Fluke. no one else would touch it lol.

[TiN]

Hope you have nanovoltmeter and nanovolt source, proper TVCs, and wiring. Would be very fun to see the worklog and photos of DIY 4920 cal.
I'm still processing data and writing on mine...

[CalMachine]

Hope you have nanovoltmeter and nanovolt source, proper TVCs, and wiring. Would be very fun to see the worklog and photos of DIY 4920 cal.
I'm still processing data and writing on mine...

The nano voltmeter I use here in the lab is a Keithley 181.  As far as TVCs go, my 4950 should provide a higher degree of accuracy than TVCs.  Really, the best I can see myself getting is around a 2:1 TUR, if that, as the 4920 is insanely accurate.

[TiN]

Any teardowns of 4950 planned? It would be joy to see insides.

Different TVCs provide different accuracy. Best of them do <10ppm which is hard to beat.

[CalMachine]

Any teardowns of 4950 planned? It would be joy to see insides.

Different TVCs provide different accuracy. Best of them do <10ppm which is hard to beat.

Hmm really!?  Could you give me a source to some of these TVCs?  Most, if not all, I've seen have got a best spec of like +/ 0.0095%  I'd greatly appreciate it. 

I'm not sure if/when a taredown of the 4950 will happen.  We just recently purchased it and we are still searching for a place to provide a 17025 accredited cal.  It will probably end up going to fluke and then sent to England for cal, like our 4920.

[ManateeMafia]

@CalMachine

Try contacting Keysight in Loveland. I spoke to them a few months ago and they told me they are can calibrate MFC's. They can also characterize each of the functions and ranges. I didn't think to ask them about my D4920 but a D4950 shouldn't be much more difficult than a F5720A. It will be a matter of getting a copy of their Scope of Accreditation and determining if it fits your needs.

[CalMachine]

@CalMachine

Try contacting Keysight in Loveland. I spoke to them a few months ago and they told me they are can calibrate MFC's. They can also characterize each of the functions and ranges. I didn't think to ask them about my D4920 but a D4950 shouldn't be much more difficult than a F5720A. It will be a matter of getting a copy of their Scope of Accreditation and determining if it fits your needs.

I would think a Datron 4950 would be the hardest cal to find... as it's the 'calibrator of calibrators' and has some of the tightest specs out there.

[ManateeMafia]

I would think very few labs have the capabilites to calibrate a calibrator. The 4950 was touted as being able to calibrate 4.5 - 8.5 digit meter calibrators so the spec's can vary and the 4808 is certainly one of the best ever built.
I found Loveland's Scope of Accreditation and they have a full setup but it looks like Fluke would be a better choice.

http://www.keysight.com/upload/cmc_upload/All/KeysightLovelandCertScope-ANAB-V006.docx.pdf
http://us.flukecal.com/support/accreditations
 
Looking at the datasheet for the 4950, the transfer stability is only listed for 30 and 90 days. Do you plan on sending it once per year or on 90 day intervals?

[TiN]

Any teardowns of 4950 planned? It would be joy to see insides.

Different TVCs provide different accuracy. Best of them do <10ppm which is hard to beat.

I'm not sure if/when a taredown of the 4950 will happen.  We just recently purchased it and we are still searching for a place to provide a 17025 accredited cal.  It will probably end up going to fluke and then sent to England for cal, like our 4920.

Teardown sounds logical to me, if I would have 4950. Replace battery with fresh one, check boards, replace electrolytic caps, IEC mains filter, etc, save old calibration dump before sending to cal. There is no fun if you send box to cal, pay big money for it and unit loose this calibration after 2 months because of 2$ battery go poof.

[lukier]

I would think very few labs have the capabilites to calibrate a calibrator. The 4950 was touted as being able to calibrate 4.5 - 8.5 digit meter calibrators so the spec's can vary and the 4808 is certainly one of the best ever built.

I wonder why Fluke dropped 4808 then. For 7001 reference the argument was that it wasn't as good as 732B.

Glaswegians can get it cheap it seems, suspiciously cheap (and on craigslist?):
http://glasgow.craigslist.co.uk/bfd/5835760255.html

If I had more spare cash I would be tempted, but it's far away from London and I don't think dealing on Craigslist is particularly safe.

[CalMachine]

I would think very few labs have the capabilites to calibrate a calibrator. The 4950 was touted as being able to calibrate 4.5 - 8.5 digit meter calibrators so the spec's can vary and the 4808 is certainly one of the best ever built.
I found Loveland's Scope of Accreditation and they have a full setup but it looks like Fluke would be a better choice.

http://www.keysight.com/upload/cmc_upload/All/KeysightLovelandCertScope-ANAB-V006.docx.pdf
http://us.flukecal.com/support/accreditations
 
Looking at the datasheet for the 4950, the transfer stability is only listed for 30 and 90 days. Do you plan on sending it once per year or on 90 day intervals?

Yeah, we sent one of our 4808s to Tektronix and we're getting ready to send the other one to Keysight's Loveland facility...  Fluke won't touch 4808s anymore, which is a shame    I have yet to hear back from fluke on whether or not they will calibrate the 4950.  I plan on sending the 4950 in every 90 days for calibration.  Our 4808s, the 4950, 4920, Fluke 732B, and Datron 4910 will all be our main standards we will send out every 90 days for the first 2 years or so.  Then I think we're going to back down to every year and track the stability and drift of the units.

Any teardowns of 4950 planned? It would be joy to see insides.

Different TVCs provide different accuracy. Best of them do <10ppm which is hard to beat.

I'm not sure if/when a taredown of the 4950 will happen.  We just recently purchased it and we are still searching for a place to provide a 17025 accredited cal.  It will probably end up going to fluke and then sent to England for cal, like our 4920.

Teardown sounds logical to me, if I would have 4950. Replace battery with fresh one, check boards, replace electrolytic caps, IEC mains filter, etc, save old calibration dump before sending to cal. There is no fun if you send box to cal, pay big money for it and unit loose this calibration after 2 months because of 2$ battery go poof.

I will very much do that!  Good thinking.  You can expect some pictures in the next month or 2.

I would think very few labs have the capabilites to calibrate a calibrator. The 4950 was touted as being able to calibrate 4.5 - 8.5 digit meter calibrators so the spec's can vary and the 4808 is certainly one of the best ever built.

I wonder why Fluke dropped 4808 then. For 7001 reference the argument was that it wasn't as good as 732B.

Glaswegians can get it cheap it seems, suspiciously cheap (and on craigslist?):
http://glasgow.craigslist.co.uk/bfd/5835760255.html

If I had more spare cash I would be tempted, but it's far away from London and I don't think dealing on Craigslist is particularly safe.

They dropped the 4808 to push their 5700 series I think.   And yeah... 1500 lbs?  That's extremely cheap... it could be hot

[TiN]

Fluke dropped 7001,1281,4808 instruments because it's a lot of work and overhead to support two lines of similar products, so they'd rather focus on their single line of own designs, which are 732,8508,55xx/57xx instead. Well expected move, actually.

You can expect some pictures in the next month or 2

Looking forward. 

[e61_phil]

Sorry, but I have to ask an offtopic question: What exactley does a Datron 4950 do? I had a look at the manual which says it is a transfer standard. My understanding of a transfer standard is a unit like the Fluke 732A/B or something like that. The manual looks like the 4950 is a kind of multimeter with very tight specs. Is this right? Or is there anything more special?

[Dr. Frank]

Any teardowns of 4950 planned? It would be joy to see insides.

Different TVCs provide different accuracy. Best of them do <10ppm which is hard to beat.

Hmm really!?  Could you give me a source to some of these TVCs?  Most, if not all, I've seen have got a best spec of like +/ 0.0095%  I'd greatly appreciate it. 

Try these: https://www.nist.gov/node/1108526/sri-6002-multi-junction-thermal-converter

That's not that expensive.. AC/DC transfer uncertainty of these primary TVC standards is between 2..10 ppm for frequencies from 10 Hz up to 500kHz:
https://www.nist.gov/calibrations/voltage-measurements-calibrations#533

Appropriate switching box and nV amplifier required..

An actual SRM 6002.b even may have AC/DC uncertainty of << 10ppm up to 1MHz 
Frank

[acbern]

Or you buy a TVC from the bay and have it calibrated. Cost me less than half of the price of the NIST unit. But then you still just have one voltage range. So the problem just starts there. Still no way around something like a Fluke 792 or a Ballantine 1605 or a set of TVCs to do a ladder calibration. Doing that with a set of TVCs is painfull, takes forever, lots of manual work, so I gave it up after 3 converters, and moved on with the 1605.

[TiN]

What exactley does a Datron 4950 do?

It's the multimeter, but it's scale is very limited to "transfer values". So unlike regular meter, which on 10V range goes from -10V to -10V + overrange, 4950 does only -1 to +1V, +9 to +11V and +18V to +19.5V, and negative ranges like this. Since these values are usual cardinal points for calibration of generic DMMs and MFCs, hence the 4950 name as transfer standard. Also this is why specifications are so tight and different, as it's not absolute accuracy specification, but transfer specifications.

E.g. 4950 calibrated to 10VDC with absolute uncertainty 2.5ppm. Using 4950 to measure 10V after would allow you to know absolute value within 2.5ppm + 1.5ppm = 4ppm total, within 30 days and +/-1C from calibration temperature.
Using 3458A calibrated to same source would give you 2.5+2.65 ppm = 5.15 ppm total, within 1C over 90 days, or 2.5+0.1ppm = 2.6 ppm total within 10 minutes, +/-0.5C from ACAL.

Datron/Wavetek 4950 User's Manual.

[VintageNut]

You are the mayor of the volt-nut village. Congratulations!

What exactley does a Datron 4950 do?

It's the multimeter, but it's scale is very limited to "transfer values". So unlike regular meter, which on 10V range goes from -10V to -10V + overrange, 4950 does only -1 to +1V, +9 to +11V and +18V to +19.5V, and negative ranges like this. Since these values are usual cardinal points for calibration of generic DMMs and MFCs, hence the 4950 name as transfer standard. Also this is why specifications are so tight and different, as it's not absolute accuracy specification, but transfer specifications.

E.g. 4950 calibrated to 10VDC with absolute uncertainty 2.5ppm. Using 4950 to measure 10V after would allow you to know absolute value within 2.5ppm + 1.5ppm = 4ppm total, within 30 days and +/-1C from calibration temperature.
Using 3458A calibrated to same source would give you 2.5+2.65 ppm = 5.15 ppm total, within 1C over 90 days, or 2.5+0.1ppm = 2.6 ppm total within 10 minutes, +/-0.5C from ACAL.

[CalMachine]

Sorry, but I have to ask an offtopic question: What exactley does a Datron 4950 do? I had a look at the manual which says it is a transfer standard. My understanding of a transfer standard is a unit like the Fluke 732A/B or something like that. The manual looks like the 4950 is a kind of multimeter with very tight specs. Is this right? Or is there anything more special?

What TiN said!  It's used to transfer cardinal values within +/- very few ppm to other calibrators.

What exactley does a Datron 4950 do?

It's the multimeter, but it's scale is very limited to "transfer values". So unlike regular meter, which on 10V range goes from -10V to -10V + overrange, 4950 does only -1 to +1V, +9 to +11V and +18V to +19.5V, and negative ranges like this. Since these values are usual cardinal points for calibration of generic DMMs and MFCs, hence the 4950 name as transfer standard. Also this is why specifications are so tight and different, as it's not absolute accuracy specification, but transfer specifications.

E.g. 4950 calibrated to 10VDC with absolute uncertainty 2.5ppm. Using 4950 to measure 10V after would allow you to know absolute value within 2.5ppm + 1.5ppm = 4ppm total, within 30 days and +/-1C from calibration temperature.
Using 3458A calibrated to same source would give you 2.5+2.65 ppm = 5.15 ppm total, within 1C over 90 days, or 2.5+0.1ppm = 2.6 ppm total within 10 minutes, +/-0.5C from ACAL.

Datron/Wavetek 4950 User's Manual.

The 4950 is far superior in every other aspect than the 3458A (with opt 02), when it comes to calibrating a calibrator.  The 10 DCV range is fairly close in specs as you show but, for resistance, DCI, and especially ACI and ACV the 4950 is significantly better.

Any teardowns of 4950 planned? It would be joy to see insides.

Different TVCs provide different accuracy. Best of them do <10ppm which is hard to beat.

Hmm really!?  Could you give me a source to some of these TVCs?  Most, if not all, I've seen have got a best spec of like +/ 0.0095%  I'd greatly appreciate it. 

Try these: https://www.nist.gov/node/1108526/sri-6002-multi-junction-thermal-converter

That's not that expensive.. AC/DC transfer uncertainty of these primary TVC standards is between 2..10 ppm for frequencies from 10 Hz up to 500kHz:
https://www.nist.gov/calibrations/voltage-measurements-calibrations#533

Appropriate switching box and nV amplifier required..

An actual SRM 6002.b even may have AC/DC uncertainty of << 10ppm up to 1MHz 
Frank

I'm going to look into these!! Thanks you

[Pipelie]

attach the firmware of datron 4920M

[TiN]

Thanks. Uploaded:

Pipelie 4920M's 1.0, 12.07.1990 A binary
Pipelie 4920M's 1.0, 12.07.1990 B binary

With help of simple python snake:

Code: [Select]

# xDevs.com combiner tool
import os

print "Combine tool for 2 FW ROMs into 1 256KB ROM, for Wavetek 4920M"
with open('4920M-40093S-1.0A-12.7.90.BIN','rb') as b:
    with open('4920M-40093S-1.0B-12.7.90.BIN','rb') as a:
        with open('w4920-1_0_12_07_1990.BIN','wb') as x:
            for cnt in range(0, 65536):
                x.write ("%c" % a.read(1) )
                x.write ("%c" % b.read(1) )
                x.write ("%c" % a.read(1) )
                x.write ("%c" % b.read(1) )

Combined human-readable with little-endian

[Pipelie]

Thanks. Uploaded:

Pipelie 4920M's 1.0, 12.07.1990 A binary
Pipelie 4920M's 1.0, 12.07.1990 B binary

With help of simple python snake:

Code: [Select]

# xDevs.com combiner tool
import os

print "Combine tool for 2 FW ROMs into 1 256KB ROM, for Wavetek 4920M"
with open('4920M-40093S-1.0A-12.7.90.BIN','rb') as b:
    with open('4920M-40093S-1.0B-12.7.90.BIN','rb') as a:
        with open('w4920-1_0_12_07_1990.BIN','wb') as x:
            for cnt in range(0, 65536):
                x.write ("%c" % a.read(1) )
                x.write ("%c" % b.read(1) )
                x.write ("%c" % a.read(1) )
                x.write ("%c" % b.read(1) )

Combined human-readable with little-endian

Thanks, here is the ROM of my friend's  4920.

[TiN]

Thanks, this one is good, but it says 4920M in the code, not 4920.
Combined, LE dump.

[Pipelie]

Thanks, this one is good, but it says 4920M in the code, not 4920.
Combined, LE dump.

It is from 4920, not 4920M. perhaps they share the same firmware. i don't know how to look into the firmware...
you are right, it's  the firmware of 4920M.
I got my unit (4920M) three years ago from eBay Australia.
I try to update 4920M's firmware to 4920 with new ROM. it is working,but revealed 4003 error. no further test for now.
---------------------------------------------------------------------------------------------------------------------------
a couple days ago, I find some pictures of 4920, and notice there is difference in A-D ASSEMBLY between 4920 and 4920M.
turn out the 4920 I said before is actually 4920M, except the front panel.
 

[Pipelie]

here are some photos, you can see,  the front panel of unit say it's 4920, but the label on rear panel reveal 4920M.
 

[Pipelie]

and some photo of 4920, you can compare to the 4920M, as far as i can tell, there is difference in A-D ASSEMBLY.

[TiN]

These are photos of two different meters, right?

Two N connectors are M-version only AFAIK. Perhaps someone replaced front mylar from 4920 

[Pipelie]

These are photos of two difference meters, right?

Yes ,they are  photos of two difference meters.

and   photos I posted at 06:26:52 PM come from my friend, He have a Datron 4920.

[TiN]

Seems there is more to it than just WB/mV-option module, so adding mV-functionality (or WB to usual 4920) wouldn't be as straight-forward, as some hoped.
How is repair going for you?

I will be testing my 4920M soon, once 3458B arrives (enroute now). It's sitting cold all that time so far  .

[CalMachine]

What's weird is, the 4920M comes default with the wideband module (AFAIK), where as the 4920 has a discrete wideband module (the 4922).

[TiN]

Having bouncing around modules for wideband ain't reliable enough for military, man .
I wouldn't complain though, mV is easier to get by DIY means than shmick tested 50MHz TVC module..

[CalMachine]

Having bouncing around modules for wideband ain't reliable enough for military, man .
I wouldn't complain though, mV is easier to get by DIY means than shmick tested 50MHz TVC module..

Yeah, you've got a good point there.  and that image is awesome! hahaha


Aww snap! Those Navy Metcal stickers!  I see those on the regular, as the US Navy is one of our biggest customers.

[CalMachine]

I heard there may be software available somewhen next year to do such 4920 cal

I wonder who would be releasing such software 

I am attempting to try to write some code to utilize our 4920 and 4920M to get automated testing done.  This thread has definitely helped a lot!  I'll be looking into those Swerleins algorithms too!  Really intriguing stuff here...

[Pipelie]

Seems there is more to it than just WB/mV-option module, so adding mV-functionality (or WB to usual 4920) wouldn't be as straight-forward, as some hoped.
How is repair going for you?

I will be testing my 4920M soon, once 3458B arrives (enroute now). It's sitting cold all that time so far  .

after replaced the SRAM(m5m5256) and a bunch of electronic capacitor,  the unit back to the working condition, but every time I turn it on Running internal self-test reveals  one error: 4053, WB calibration validity. I try to Run the WB Wlinearity self-cal, but didn't Successful.
any suggestions are welcome.

[ap]

It looks like your item needs calibration. This is not always that easy, because it seems the different cal routines sometimes influence each other. TIN I think had the same experience. So a full cal may be needed.

[Pipelie]

Thanks for the reply,
I'm looking forward to read TiN's  article about 4920's calibration.
couldn't fix my 4950 this weekend, but GOOD NEW, My 4920 pass all self test without error,when the room temperature around 20C. no more error 4053.

[ap]

What did you do to fix the 4053-error?

[TiN]

My poor box died, just from sitting unpowered on the desk.
4014 error, -19V supply for A/D. It's simple LM337 there , what could go wrong? 

[Pipelie]

My poor box died, just from sitting unpowered on the desk.
4014 error, -19V supply for A/D. It's simple LM337 there , what could go wrong? 

It mostly due to the cap near to LM337( connect to the output of LM337), I have this -19v supply problem before when  I repaired my 4920.  desolder it and test the cap or test it on board.

[Pipelie]

What did you do to fix the 4053-error?

run the wideband calibration and don't forget to update the calibration date once a year at least.

[TiN]

Pipelie right on money. Replaced 1uF 100V caps, and thing is alive again. 
Lesson: don't ignore little electrolytic caps during repairs on old instruments 

[chinapp]

anyone know how to upgrade the mV option of 4920? I upload the FW of the 4920,it's NO is 400978,ver 1.14,I can find the strings of the mV option in the FW,maybe a decompile can find something.I am a newbie on this

[TiN]

Don't understand what you mean. If you have 4920 it already have mV option installed, no?

[chinapp]

NO, my 4920 has no mV option, when I press the mV, it said :1014 error,no mV option fitted
 
I opened the aliminium box beside the analog board ,it should be the mV option ,there are some ICs  not fitted (see attachment)

,I wonder is there some way to upgrade the mV option

[TiN]

Interesting. I didn't think there are 4920 without mV option.
I don't have 4920 so can't tell what's there missing parts are, perhaps somebody who repaired 4920 could help.
My 4920M has only wideband option in place instead of mV.

[chinapp]

yes, I checked the FW of 4920 and 4920m,seems have much differences,4920M FW has some strings of 1281

the 1281 can use the short link to choose the options, did not find  short link for 4920 to choose option

[ManateeMafia]

I think I see a blown trace on the left side of the pcb just between the two resistor packs.

[TiN]

It looks cut on purpose, not blown

[chinapp]

is it means fix the cut trace can make the mV option  on ?

[TiN]

Making meter "detect" mV option will not magically add missing components, no? 

[chinapp]

seems no more info on this ............

[meandeev]

Hello,
was there somebody in and removed some components?!
Attached you can find a photo of the mV-option of my 4920 and also the 400978-2.5 firmware of this device (Sticker says 2.5, inside the firmware you can find the number 2.05)

[chinapp]

thanks so much for your pic

it seems not a big problem, could you give a view of the five  transistors? can not get the model of them,and also the foil resistor beside the fuse

[chinapp]

recently research update >>>

the main board of none mV option 4920 is different to the mV option added 4920,so it seems there is no way to update