Fluke 5440B warmup time

[retroware]

A buddy of mine and myself both recently acquire a 5440B. When powered on, his unit starts about 50ppm low and warms up to the specified output voltage in about 30 minutes. My unit (which happens to be the B/AF version) starts 500ppm low and takes closer to an hour to warmup. Once warmed up, it appears to be stable but I haven't done any long term measurements.

Is this wide range in startup voltage/warmup time to be expected? I'd be curious to hear from other 5440 owners.

[quarks]

I am not a owner, but have other calibrators and suggested warmup time of them is 4h, but none of them is this far off.
My guess is, that is not what it should be.

[Dr. Frank]

Well, I've got a 5442A, which is identical to the 5440 in aspect to the reference and ovens.

My 3458A was already running, so I powered the 5442A up, and it was low, -12ppm @ 10V.

I don't have a warm-up diagram, so I'd guess that the 5442A would go further down a bit, and then up and overshooting until it would stabilize to < 1ppm after about 1/2 .. 1h until these bulky ovens are warmed up. http://www.amplifier.cd/Test_Equipment/other/Fluke5442A.html

It takes about 4h to stabilize, as < 0.2ppm stability over 24h is possible. One nights time before taking measurements is preferred, anyhow.

B/AF instruments may be designed for special purposes or special customers, but that sounds very strange; 500ppm excursion seems to be much too much.
I'd guess, that this doubled SZA263 reference would NEVER have the possibility to make such big swings if it's working oK. These RefAmps  are usually trimmed for low T.C., see according thread about the SZA263, here in this section. That means, maybe 2ppm/K T.C. w/o oven, and as the oven runs on about 55°C, that would limit this swing to about 60ppm, that's within the observation you have seen on your first 5440A.

If you have guts, you might open the reference and make sure, that it's really the SZA based reference. Maybe you find something obviously damaged inside.
Don't be afraid, it's a very rugged Standard, and should be stable enough, to survive w/o any changes. 

If it looks like inside mine, I would guess, that there's something fishy with the RefAmp circuitry.
The oven might stabilize its output anyway, but a stability measurement, against a 3458A for example, will quickly discover the possible culprit, as said, < 0.2ppm stability to be expected.

Frank

[ManateeMafia]

It wouldn't hurt to take a measurement of the power supply test points. The manual lists 13 steps for the in guard supply including the specs.

When I received mine I also reseated all the boards and inspected them for damage and foreign debris. You never know how long these things sit before they are sold. 

[retroware]

Thanks for the info. It sounds like I have a repair job ahead of me

If people don't mind, I'll use this thread to note any progress in diagnosing the issue. Otherwise, I could move it to the repair thread (where it maybe should have been in the first place).

[lowimpedance]

Thanks for the info. It sounds like I have a repair job ahead of me

If people don't mind, I'll use this thread to note any progress in diagnosing the issue. Otherwise, I could move it to the repair thread (where it maybe should have been in the first place).

I don't think anyone will mind, This sub forum is the best place for this type of gear. (IMO)
The more repair info accumulating the better.

[retroware]

Well, I managed to figure out why my 5540 started 500ppm low from a cold start.

I first verified that the issue was indeed the reference by monitoring the VREF test point. I then removed the reference and powered it up on the bench. By simply touching one of the references (U3), one could move the output 100pm in a minute or so.  The U6 reference seemed relatively stable.

On the 5440, each SZA263 has two tempco resistors setting the current for the refamp. On my unit, the two resistors feeding U3 had the values of 5.36K whereas the two resistors feeding U6 had the values of 150K. This seemed odd in that while I'd expect some variation between refamps I wouldn't expect that much.

I then attempted to figure out the correct value for the U3 refamp by heating it up, seeing whether it had a positive or negative coefficient and adjusting the resistor accordingly. I was able to determine that the correct value was around 145K or so (my environmental control was a SMD hot air rework gun and hence not particularly precise).  The other thing I observed was that if I heated U6 with a hot air gun, I saw about a  90ppm difference going between 70 and 100 degrees F.  This also seemed a bit high.

So at this point I'm trying to decide what resistors to use for U3 as they are obviously incorrect. I'm also a bit suspicious of the 150K x 2 = 300K resistors in U6.  At this point a buddy of mine had opened his 5440 up to see what resistor values were in his unit. For both U3 and U6 the two resistors had the values: 5.11K and 118K, the sum of which was close to the 145K resistance that I thought my references needed.  The other thing to note is that both references used the same value collector resistors.  Taking a look at the picture of the 5440 reference on the amplifier.cd site showed very similar values: a 133K resistor and a 8.87K resistor.

Then it dawned on me what might have happened.  The unit should have been assembled with one 5.36K and 150K resistor PER refamp, not two 5.36K resistors for U3 and two 150K resistors for U6. So I swapped one of the 5.36K resistors for the 150K resistor, powered the unit on and observed a much reduced warmup drift. I'm seeing about 25ppm low to start which is in line with my buddy's unit and Dr Franks. It warms up to within 1ppm of the output voltage in about 30 minutes.

I'm guessing that the unit was built this way given that there was absolutely no indication that the board had been reworked.  The fact that the reference is ovenized helped to coverup the fact that the reference was incorrectly constructed.

[Dr. Frank]

What a great story and a fine analysis...

QA failed at FLUKE for your unit, obviously.

These instruments often have an issue, if they show up on the 2nd hand market.

Now, that you solved it, you'll have for sure a very good instrument, maybe needs a re-calibration.

You didn't take pictures of your reference board, containing that placing error, or afterwards, haven't you?

Frank

[retroware]

What a great story and a fine analysis...

QA failed at FLUKE for your unit, obviously.

These instruments often have an issue, if they show up on the 2nd hand market.

Now, that you solved it, you'll have for sure a very good instrument, maybe needs a re-calibration.

You didn't take pictures of your reference board, containing that placing error, or afterwards, haven't you?

Frank

Unfortunately, I did not take a picture and now it is all buttoned up.  But there was one curiosity: U3, R27 and R19 all had a black smudge on them like someone took a marker to them.  In the amplifier.cd pictures and my buddies unit, only one resistor and one of the SZA263's had a smudge on them. R21 and R19 were both 100K, Dale T9, 25ppm resistors, R29 and R27 were 5.36K, Dale T1, 100ppm resistors.

I'd love to get it calibrated but I'm scared to get a quote.  My current level of volt nut insanity is limited to a number of 6.5 (K2000, 34401, 3456, etc) and near 7.5 digit (Datron 1071, Fluke 8506) DMMS. The 5440B is my most precise piece. My first step might be to get my Fluke 731B calibrated and then do a transfer from that. I also have a Fluke 720 KVD to do a complete external cal (skipping maybe the 1KV scale as I don't want to drop 10W into my 720), realizing that this isn't as accurate as having a Fluke 752A.  While I realize it won't bring the 5440B into any sort of traceable calibration, I'm hoping it will at least get it into the ballpark.

[Tony_G]

I just had one of my 5440s up at Fluke in Everett for a basic Cal - US$557 + tax

TonyG

[e61_phil]

My first step might be to get my Fluke 731B calibrated and then do a transfer from that. I also have a Fluke 720 KVD to do a complete external cal (skipping maybe the 1KV scale as I don't want to drop 10W into my 720), realizing that this isn't as accurate as having a Fluke 752A.  While I realize it won't bring the 5440B into any sort of traceable calibration, I'm hoping it will at least get it into the ballpark.

A 10V source and a Nullmeter should be enough to calibrate the 5440B. I played a lot with the unit at work before calibration. The 5440B will calibrate all ranges with only 10V.

I paid about 300€ for the calibration with data. (It was only a verification with a 3458A calibrated against a PTB traceable 732A just before the measurements)

[Dr. Frank]

A 10V source and a Nullmeter should be enough to calibrate the 5440B. I played a lot with the unit at work before calibration. The 5440B will calibrate all ranges with only 10V.

2.2V and 220mV will have to be calibrated by means of a reference divider or KVD, as these are not auto-calibrated by the 5440B.

Also, if the instrument was far off before, I assume that a complete calibration according to the manual is required.
The autocal process is described as only be capable to make small corrections (but this, to a high degree of accuracy).

The 720A may be used at 1kV without fear, as it's an autocal primary ratio standard, and any drifts from such high wattage will be safely re-calibrated afterwards.
Anyhow, it's not well suited to make precise 10:1 and 100:1 transfers. 100V => 10V transfer is about 1.2ppm precise, 1000V => 10V about 6.6ppm, if I calculated everything correctly.
Therefore, a 752A would be much better.

But maybe, these levels of accuracy are sufficient as an initial step, so that the 5440B may afterwards autocal itself to better limits by its internal Hamon type process.

Frank