Fluke 540B Thermal Transfer Standard

[quarks]

Hello all,

a few days ago I had the chance to get a Fluke 540B plus a complete accessories set of A40/A40A and A55.
I do have the instruction Manual, but there seems to be not much detail about the current shunts. Also there is mentioned a A45 Current Transfer Switch, which I do not have. Do I need the A45 or is it just convenient to have it.

Are there any users of this unit, who can share tips and tricks?
So far I could not find detailed shunt spec values. I tried to measure the resistance values, but they are odd somehow. Does anyone know these values or how to check if the shunts are in its specs?

thx
quarks

[ftransform]

What exactly does this thing do?

[quarks]

The Fluke 540B is as the name says a Thermal Transfer Standard. It is a calibration lab unit. It uses a thermocouple response method (probably one of the most accurate ways for AC) for measurement and calibration of ACV (and with the shunts also ACI). It compares/transfers AC to DC and then use a Nulldetector to match to the same exact effective values. This works in a very wide range (0.25V to 1000V (with A40/A40A 10mA up to 20A range), 5Hz-1Mhz (with A55 up to 50MHz, max. then is 50V) and down to 0.01% accuracy.

[quarks]

Are there any users of this unit, who can share tips and tricks?

So far I could not find detailed shunt spec values. I tried to measure the resistance values, but they are odd somehow. Does anyone know these values or how to check if the shunts are in its specs?

I still hope to find others here, who use this Fluke or can give some advice.

Thx
quarks

[alm]

The shunts are designed to work with a ~90 ohm load in parallel (the shunt inputs of the 540B). The voltage across the shunt would be about 0.45 V at the full rated input current with the 90 ohm in parallel. Calibrating current shunts require a precision current source (or bench supply with DMM in series). The resistance may change slightly at higher currents due to heating.

[quarks]

The shunts are designed to work with a ~90 ohm load in parallel (the shunt inputs of the 540B). The voltage across the shunt would be about 0.45 V at the full rated input current with the 90 ohm in parallel. Calibrating current shunts require a precision current source (or bench supply with DMM in series). The resistance may change slightly at higher currents due to heating.

Hello alm,
thx a lot for your reply.

I found data for the newer A40B series. Their nominal R values are
80 Ohms for the 10mA shunt
8 Ohms for 100mA
0.8 Ohms for 1A
and so on.
But my measured values seem not to corespond to that. That is why I would like to check each individual shunt if it is in spec. Do you know these values for the A40/A40A or can check if you have known good ones?

You say in parallel to the 540B it should be about 0.45V at the rated current. Why is it only "about" and not exact? If I want to check the shunt accuracy, how can I find out if the values are good?

Many thx for your help.
Bye
quarks

[alm]

The Fluke A40B shunts are designed to be used with volt meters (extremely high input impedance), unlike the A40 shunts.The specs will be very different. I'm not positive about the 90 ohm impedance of the 540B shunt input (some later models used 90 ohm, but the 540B might also be 100 ohm), but you should be able to measure it. The A40 shunts are only designed for relative current measurements (compare AC/DC). As far as I know they're only specified for flatness from DC to 100 kHz at 0.5-1x the rated current. When feeding a 5 A shunt 5 ADC current, the voltage across it should be within 0.02% of the value with 5 A 20 kHz AC. This spec includes a 540B connected in parallel. The A40B is also specified for absolute current measurements, so the DC resistance is specified. Calibration of the A40 shunts involves feeding it with an equal DC and AC current at various frequencies (using a known good shunt in series).

The A45 is just for convenience.

[amspire]

Quarks,

I have a 540B.

Had to do some repairs and mods:

1. The 540B requies a Mercury D cell - Alkaline batteries are not nearly stable enough. I replaced the D cell with a AA cell to D cell adapter  containing two alkaline AAA cells in series. In the unit, I added a 1.25V low dropout voltage regulator - it is very stable. Changed a resistor in the metering circuit to show the battery state properly. I can dig up some details if you need it, but it seemed to be the easiest and neatest solution to get a stable voltage.

2. Usually the Nicad batteries need replacing. If you have the original batteries, some cells are probably shorted by now. Pretty cheap on ebay. Use Nicads and not Nimh.

3. There may be corrosion on the lower PCB from the batteries. clean this up.

4. You need to get the protection trip circuit working properly. The circuit is very weird and sensitive. I had to replace one of the transistors as its leakage current was too high.  If the protection circuit is not working correctly, there is a high chance of damaging the sensor.

Once it is working properly, the 540B is a great bit of hardware. After the modes and repairs above, it can resolve to  0.001% so the 0.01% accuracy is probably conservative.

I do not have any current shunts.

It is not quick to use, as the nulling is slow, and you are meant to repeat the process 3 times for confidence. The AC source has to be extremely stable, so it is not much use at measuring typical signal sources - it is really only useful for calibrating a RMS meter with the aid of a very stable DC source, and a very stable AC source.

Richard

[quarks]

The Fluke A40B shunts are designed to be used with volt meters (extremely high input impedance), unlike the A40 shunts.The specs will be very different. I'm not positive about the 90 ohm impedance of the 540B shunt input (some later models used 90 ohm, but the 540B might also be 100 ohm), but you should be able to measure it. The A40 shunts are only designed for relative current measurements (compare AC/DC). As far as I know they're only specified for flatness from DC to 100 kHz at 0.5-1x the rated current. When feeding a 5 A shunt 5 ADC current, the voltage across it should be within 0.02% of the value with 5 A 20 kHz AC. This spec includes a 540B connected in parallel. The A40B is also specified for absolute current measurements, so the DC resistance is specified. Calibration of the A40 shunts involves feeding it with an equal DC and AC current at various frequencies (using a known good shunt in series).

The A45 is just for convenience.

thx again alm.

This is what I measured:
540B Shunt Input 90,73Ohm
and the shunts not connected to 540B
10mA Shunt 87,45Ohm
20mA Shunt 31,99Ohm
30mA Shunt 18,22Ohm
50mA Shunt 9,96Ohm
100mA Shunt 4,68Ohm
200mA Shunt 2,22Ohm
300mA Shunt 1,47Ohm
and so on.
This is not what I expected.
Do you or anyone else have these shunts and can compare the values?

[alm]

What did you expect? For example, the ~10 ohm 50 mA shunt in parallel with the 90 ohm 540B input would be 9 Ohm. 50 mA through 9 Ohm would produce about 0.45 V. Since it's only intended for comparing two different currents (see 540B manual for procedure), the exact value doesn't matter as long as it's within the input range of the 540B.

[quarks]

Quarks,

I have a 540B.

Had to do some repairs and mods:

1. The 540B requies a Mercury D cell - Alkaline batteries are not nearly stable enough. I replaced the D cell with a AA cell to D cell adapter  containing two alkaline AAA cells in series. In the unit, I added a 1.25V low dropout voltage regulator - it is very stable. Changed a resistor in the metering circuit to show the battery state properly. I can dig up some details if you need it, but it seemed to be the easiest and neatest solution to get a stable voltage.

2. Usually the Nicad batteries need replacing. If you have the original batteries, some cells are probably shorted by now. Pretty cheap on ebay. Use Nicads and not Nimh.

3. There may be corrosion on the lower PCB from the batteries. clean this up.

4. You need to get the protection trip circuit working properly. The circuit is very weird and sensitive. I had to replace one of the transistors as its leakage current was too high.  If the protection circuit is not working correctly, there is a high chance of damaging the sensor.

Once it is working properly, the 540B is a great bit of hardware. After the modes and repairs above, it can resolve to  0.001% so the 0.01% accuracy is probably conservative.

I do not have any current shunts.

It is not quick to use, as the nulling is slow, and you are meant to repeat the process 3 times for confidence. The AC source has to be extremely stable, so it is not much use at measuring typical signal sources - it is really only useful for calibrating a RMS meter with the aid of a very stable DC source, and a very stable AC source.

Richard

Hello Richard,
thx a lot for your information. This is really helpfull. I am very interested in your mod to avoid the mercury D cell.

This weekend I tried for the first time to really use it. I charged the battery, all Battery Checks (Ref./Galv/Search) show Batt ok/Overload.  So I went to Power ON and Mode Search AC and later DC. Result was, nothing was found with the range switch. I do not know what is broken. I guess it is your Point 4 and the protection circuit could be it. That would make sense, because then the relay is not switched on. And AC or DC search can not find a range signal. But right now, I do not want to search inside the box, because the seller has several more units and I can probably/hopefully get a fully working one in exchange. I also think of maybe getting a second one for spare parts. Just in case. What do you think?

Either way, with your hints I will check for the weak points, when I talk to the seller.

About the 0.01% (and your resolve 0.001%) that would be just great, because I so far have nothing this accurate in ACV (my very best meter is only 0.025%).

I am sure I have stable enough sources for DC (Valhalla/Burster/Knick) and I hope my Valhalla 2703 is as good as it is suppose to be. I also have a nice Fluke 5220A for DC/AC currents up to 20A, which I hope and feel will be good enough.

thx  a lot
quarks

[quarks]

What did you expect? For example, the ~10 ohm 50 mA shunt in parallel with the 90 ohm 540B input would be 9 Ohm. 50 mA through 9 Ohm would produce about 0.45 V. Since it's only intended for comparing two different currents (see 540B manual for procedure), the exact value doesn't matter as long as it's within the input range of the 540B.

Hello alm,

thx again and now I got it.

I was missleaded because I expected corresponding values (lets say 10mA shunt is 90Ohm, then 100mA expected to be 9Ohm) like documented in the A40B (see picture). But it was not nearly like this and therefore I wondered if any of my shunts is damaged or just is out of spec.

Now I see your point and did the calculation 90Ohm//shunt value * I = around 0,45V. That looks ok now. But I still wonder why there are no details about the spec values.

bye
quarks

[alm]

But I still wonder why there are no details about the spec values.

Because these were only intended for relative current measurements. If you look into the manual, the procedure to measure an unknown AC current is to feed the unknown AC current through the shunt, zero the meter, and then feed it an adjustable DC current and adjust the current source until the meter reads zero again. The value of the shunt is irrelevant as long as its constant over time and frequency and the resulting voltage is within range of the meter.

The A40B are very different. They are intended for absolute current measurements in combination with a DMM, so their DC resistance has to be accurately specified for them to be of any use.

[amspire]

Hello Richard,
thx a lot for your information. This is really helpfull. I am very interested in your mod to avoid the mercury D cell.

Here is a description of my mods.

I wanted to do it in a way that was reversible. Here is the back panel now after I went a bit mad with my Brother P-Touch labeller:


First the batteries. The original Saft batteries were leaking and about 5 cells were shorted. I replaced then with 2.4AH NiCd Sub-C cells with tabs from ebay (something like $30). I put Kapton tape on the terminal strip with the live 240V.



I replaced the D-cell Mercury battery with a AA cell to D cell adapter that I had modified to hold two AAA cells in series. You could just remove the D cell holder (it is no loner much use) and replace it with a dual AA cell holder. You will find non-mercury batteries just drift way too much, so I went for my 3V battery + 1.25V regulator option.



The regulator has to go after the power switch, so I put it on a small prototyping board wrapped in Kapton tape and mounted it on the 23 position divider switch.



The circuit consisted of surface mount parts - a LM1117 1.25V regulator, a 1uF cap and a resistor across the output. I have forgotten what value, but just check the data sheet for the minimum stable load. The two wires with the 1.25V come in to two ajacent pins so the + connection now goes to the regulator board, and the regulator output goes to the + connection on the switch.

Now that the reference battery is 3V instead of 1.2V, the battery check resistor needs to be changed. Add 1K8 in series with the existing 180 ohm resistor.



The protection board is below the NiCd batteries, so if they have been leaking at all, this board and wires going to the boards may need some cleaning and repair. Q303 and Q304 are the critical transistors. If you hear a relay click the moment you switch from BATTERY CHECK - SEARCH to ON, then the protection circuit is turning on. Unplugging Q303 and Q304 will stop the protection circuit from switching.



It is a very weird circuit, in that the silicon transistors switch with base voltages at about 0.4V - this means the collector currents are sub-microamps. Any leakage in the transistors or on the board will trigger the protection board.  I removed the sockets so I could really clean the board and sockets, and I still had to replace the transistors with new devices - the old ones had become leaky.

I used a MPSA56 for Q303 and a 2N3904 for the Q304. I picked these as they were on hand and they worked.

[BravoV]

Ah.. finally, really glad to hear you're back Richard !

Nice photoshoots & thanks for the teardown, and I can see you've labeled it so tidy & nicely. 

[quarks]

Hello Richard,
thx a lot for your information. This is really helpfull. I am very interested in your mod to avoid the mercury D cell.

Here is a description of my mods.

Hello Richard,

thx a lot for your details.

I now have checked several units from my seller. All already have this Battery Mod (see picture, looks like an original Fluke mod) inside.
That's the good news. Bad news is, not one is fully working. All have dead NiCd  cells, some do not work at all and some show the same defect as mine. But I identified two where "AC + DC Search" works. But still I cannot really measure with them. Because the Galvanometer is going strait out of scale  when I use Mode DC Transfer and then test GALV Momentary switch, no matter how exact I match the DC Voltage to the AC.

Is it possible to cross test the A54-2 in different 540B units without damage?

Do you have a idea what to do next?

thx
quarks

[amspire]

I didn't know about the official Fluke battery mod, but it looks like they have done something similar. They have a couple of resistors to boost the voltage from 1.25V to 1.35V to match the mercury battery, but it is not really necessary.

If the NiCd batteries are dead, I am not sure how you can test any of the units. The NiCd batteries are the voltage regulators for the unit, and so if the battery packs are dead, the supplies will be all wrong or at least noisy. I never try and use the 540B while it is plugged into the mains myself.

The key thing is that the sensor must be OK - if it is blown, the unit will be worthless. It is probably OK to swap the A54 modules, but I have no experience in that.

Here is my guide to checking the sensor.

Get two multimeters, including one with a decent mV range that can read to 0.1mV.

Find the ohms range on the other multimeter that outputs the most current. In some multimeters, the Diode range has a higher current then the low ohm range. It must be less then 5mA but if it can do 0.5 mA to 2mA, that should be fine.

The first 3 steps to check the sensor are done with the 540B OFF.

1. Put the second ohm meter across the shunt connector on the A54. On my meter, it reads 87.8 ohms. I do not know the tolerance, but 80 to 100 ohms is probably OK. If it is open of above 100 ohms, the sensor may be burnt out and the whole A54 module will be doomed.

2. Unplug the HIGH FREQUENCY CONVERTER plug on the bottom left of the 540B and put the sensitive multimeter on a fixed mV Range between the lower pin on the socket (black lead) and the right upper pin on the socket.

You should get about 0.4mV/mA of current from the multimeter connected to the shunt input.

You can press the Polarity switch on the A54 and it should stay the same. If the multimeter on the shunt is disconnected, the mV multimeter should settle on 0.0 mV


3. Put the HIGH FREQUENCY CONVERTER plug back in.

4. Switch the 540B ON and on the A54, to 0.5V and DC TRANSFER. the resistance between the DC input pins should be slightly higher then the shunt ohm reading. If it stays open circuit then either relay 701 (the transfer enable relay) is not on, or the protection circuit (relay 801) has tripped. If the protection circuit is faulty - like a leaky transistor, pressing the PROTECTION DISABLE switch does nothing.

Now to check the galvanometer. If you have dud batteries, the supplies may not be adequate, even if the meter says they are OK. If you are running off mains, there may be a heap of 50/60Hz ripple on the supplies with bad batteries.

5. Switch the 540A ON and the SENSITIVITY to HIGH. The GALV toggle should be in the center OPEN position.  The A54 settings do not matter - it can be off. You should be able to zero the galvanometer with the inner know on the SENSITIVITY control. It should  not be drifting. If you cannot, then there is a problem with the galvanometer circuit or the supply to the galvanometer circuit.

6. Set the SENSITIVITY to LOW and the three REFERENCE ADJUST knobs all fully clockwise. Switch  the GALV toggle to LOCK. The galvanometer should stay at 0.

7. turn the COURSE REFERENCE ADJUST one click to the left. The galvanometer should move left to about -10. If this works, then the galvanometer circuit is probably OK.

8. If a unit passes all the above tests, it is probably serviceable, so it is just a matter of looking for the least corrosion from the batteries, and do the sniff test on the A54 to make sure there is no smell of burnt resistors. If you can pull out the A54 and remove the white plastic cover on the range selector, all the better. It does include a lot of custom wire wound resistors, and I would hate to have to replace them.

If there is the sign of battery leakage or corrosion, some of the wires the the battery module and to the lower board my be corroded at the ends, and you will have to strip the wires back to an uncorroded spot and resolder. But this is all work that is achievable.

The two things you want to avoid are a dud sensor, and a dud galvanometer circuit (it just looks very difficult to debug).

Good Luck.

Richard.



If this tests works, then at least the sensor is probably good.

[quarks]

Hello Richard,

that is really great help. I will first try to make a working NiCD set out of what I have and then try out what you suggest. I will come back with the results.

Thx very  again
quarks

[amspire]

I wouldn't mix old and new NiCds myself. If some batteries discharge much quicker then others, you will just end up damaging the other cells.

It looks like it costs a bit over $30  to get a new set. (For example ebay #170823485972). Capacity does not matter. You will not manage to flatten the batteries with any rational amount of useage, so there is no advantage in going for the highest capacity batteries. 1600mAH would be fine.

Cleaning up the leakage from the old pack, an building a new pack is a fair bit of work. Not something I would want to do as a temporary job.

Richard.

[quarks]

Hello Richard,

I have a NiCd set assembled, just to be able to do the tests you suggested (on one test unit).
This test unit is one who worked in RANGE AC/DC SEARCH, but as mentioned before, the GALVANOMETER jumps right out of range when I do switch GALV on the DC Transfer. No matter how accurate I try to adjust the DC source.

Here are the results:
1. shunt is about 90Ohms
2. High Frequency Converter Test shows 0.33mV/mA
4. DC INPUT resistance is 93Ohms (RANGE 0.5 and DC TRANSFER)
5. SENSITIVITY GALV ZERO is ok and no drift
6. perfect 0, no movement
7. COARSE one to the left shows -7

Do you think this is in tolerance?

thx
quarks

[amspire]

Hello Richard,

I have a NiCd set assembled, just to be able to do the tests you suggested (on one test unit).
This test unit is one who worked in RANGE AC/DC SEARCH, but as mentioned before, the GALVANOMETER jumps right out of range when I do switch GALV on the DC Transfer. No matter how accurate I try to adjust the DC source.

Here are the results:
1. shunt is about 90Ohms
2. High Frequency Converter Test shows 0.33mV/mA
4. DC INPUT resistance is 93Ohms (RANGE 0.5 and DC TRANSFER)
5. SENSITIVITY GALV ZERO is ok and no drift
6. perfect 0, no movement
7. COARSE one to the left shows -7

Do you think this is in tolerance?

thx
quarks

Sounds really good. I think you can go with this 540B.

[quarks]

Sounds really good. I think you can go with this 540B.

Hello Richard,

my original 540B (Q305 was defect) is now working as the other two. Now 3 units passed the tests you suggested.
But sill none of them is usable for transfer, the GALVANOMETER jumps out of range when I do DC Transfer.
Do you have any idea what I can check next?

thx
quarks

[amspire]

Sounds really good. I think you can go with this 540B.

Hello Richard,

my original 540B (Q305 was defect) is now working as the other two. Now 3 units passed the tests you suggested.
But sill none of them is usable for transfer, the GALVANOMETER jumps out of range when I do DC Transfer.
Do you have any idea what I can check next?

thx
quarks

Are you using the wrong method? It looks like the galvanometer works. Perhaps the Search calibration is way out and you are putting in too much volts. Put 0.8V in on the 1V DC range. Turn the COARSE adjustment fully antclockwise and then clockwise  with DC Transfer/ LOW range. Does the galvanometer flip from one side to the other? if so, there will be a null somewhere between the extremes.

Richard.

[quarks]

Hello Richard,
just tried it and it is flipping but no zero is possible.
Bye
quarks

[amspire]

Are you doing it right? If it is flipping, then there will be a null in the middle.

Are you starting on low? Set all reference knobs anticlockwise. The galvanometer will be hard left. Turn the coarse till the meter goes right and then click back to the left one click so the meter is to the left again. Turn the center knob till the galvanometer is slightly to the left of center, and then zero with the right knob. You can now increase sensitivity until you you a null on high.

The voltage source you are measuring has to have a short term stability of 1 part in 10-4 or better, otherwise you will never get a null.