Fluke 8842a LM308A Supply Voltage Above Max Rating

[Ultron81]

Troubleshooting the Precision Current Source circuit in my Fluke 8842A, and I noticed that the supply voltages for LM308A (U404) exceed the max spec given by the datasheet. The positive rail is +30V, negative rail is -5V, specs given by the datasheet are +/- 18V. Obviously it was designed like this, but I was looking into finding a replacement (LM308A is obsolete) but now I'm not sure what spec to go on.

Why would Fluke design the circuit to use supply voltages outside the spec of the op-amp?

[glarsson]

35V is less than 36V. The op amp does not care where we defined 0V.

[Ultron81]

35V is less than 36V. The op amp does not care where we defined 0V.

Oh ok thanks! So in the datasheet where it specifies +/-18V its actually 36V between the 2 rails, not a maximum of 18V on each supply pin.

[glarsson]

The amp has no ground connection so it can't tell the difference between 36V and +/-18V. The split supply is just a convenient way of getting the signal into the amplifiers active range.

[Kleinstein]

Possible (improved) replacements could be LT1008, LT1012 and maybe OPA145 (precision JFET type - would need adapter board).

[Ultron81]

Troubleshooting the Precision Current Source circuit in my Fluke 8842A, and I noticed that the supply voltages for LM308A (U404) exceed the max spec given by the datasheet. The positive rail is +30V, negative rail is -5V, specs given by the datasheet are +/- 18V. Obviously it was designed like this, but I was looking into finding a replacement (LM308A is obsolete) but now I'm not sure what spec to go on.

Why would Fluke design the circuit to use supply voltages outside the spec of the op-amp?

what did you find that resulted in the decision to replace the LM308A?

I think the LM308A is fine. The problem I'm seeing is when measuring resistance with the meter. The reading are not accurate at all, and it seems like the reference current is not changing properly with range (ex: In the 20k range, a 10k resistor measures 5.4k. In the 2k range, it measure .54k, in the 200k range, it measures 54k). It's most likely the analog switches (testing them soon), but when I order parts, I wanted to order the opamp just in case to have it on hand, but it's obsolete.

I bypassed the protection circuit and still have the same issue. Measuring the current coming out of the circuit, it measures the same in 200, 2k and 20k range - 0.0545mA (should be 1mA in 200 and 2k, and 0.1mA in 20k). 200k range it measures 0.0565mA (should be 0.01mA, and 2000k and 2M range it measures 0.0452mA (should be 0.005mA and 0.5uA).

I checked the first stage of the precision current source and that seems to check out, so it's either this opamp or the analog switches. Problem is the analog switches are a fluke part, and I cant seem to find anything close to them. I think I will try and use a LTC222 and make an adapter board.

[David Hess]

Possible (improved) replacements could be LT1008, LT1012 and maybe OPA145 (precision JFET type - would need adapter board).

The LT1008 is an exact improved replacement with external compensation.  The LT1012 and LT1097 are internally compensated and may require modifications to use.

Personally I consider 35 volts too close for a 36 volt part; they were not intended to operate at their absolute maximum ratings.  But sometimes manufacturers graded lower cost 36 volt parts for 40 or 44 volt operation and maybe Fluke did that here.

[Edwin G. Pettis]

Given your symptoms, I think it is likely that either Z401 is defective or U402/3 or the logic driving it is defective.  It is not out of the question that a resistor such as the 5Meg for instance has drifted out of tolerance.  I own an 8840A/AF which is somewhat identical to your 8842A.  The Fluke U403 is a unique design with latching circuitry on the chip, not only analog switches.  The chip can be found if you search enough on the web for a decent price but they are not that easy to find.

I had to replace Z402 when the output became unstable in mine.

[Ultron81]

Possible (improved) replacements could be LT1008, LT1012 and maybe OPA145 (precision JFET type - would need adapter board).

The LT1008 is an exact improved replacement with external compensation.  The LT1012 and LT1097 are internally compensated and may require modifications to use.

Personally I consider 35 volts too close for a 36 volt part; they were not intended to operate at their absolute maximum ratings.  But sometimes manufacturers graded lower cost 36 volt parts for 40 or 44 volt operation and maybe Fluke did that here.

LT1008 looks like a perfect replacement. I'm surprised I didn't see that when searching. Appreciate it guys!

[Ultron81]

Given your symptoms, I think it is likely that either Z401 is defective or U402/3 or the logic driving it is defective.  It is not out of the question that a resistor such as the 5Meg for instance has drifted out of tolerance.  I own an 8840A/AF which is somewhat identical to your 8842A.  The Fluke U403 is a unique design with latching circuitry on the chip, not only analog switches.  The chip can be found if you search enough on the web for a decent price but they are not that easy to find.

I had to replace Z402 when the output became unstable in mine.

Thanks for the info. I think I checked Z401, and it measured OK (I think that's the one I checked). As far as the analog switches (U402 and 403), you are right, they have a latches and a chip select pin, and are dual supply. I did find some on the internet, but I wanted to find a new solution. I was thinking of using a D flip-flop with the CS signal as the clock, and the Q going to control an analog switch IC. But, I kept searching to find a single chip solution, and came across the LTC 222 - http://www.analog.com/en/products/switches-multiplexers/analog-switches-multiplexers/single-supply-5v-44v/ltc222.html#product-overview

The problem here is Digikey and Mouser stock it in SOIC-16 package, so I need to make an adapter board to fit in the 20-pin socket. Also, it is a single supply, so I will use the +15V as the supply. I'm actually about to try it out and see if it will work in an hour or so.

[Ultron81]

Hey guys thanks for the help. The problem is the analog switch ICs (Fluke 700013). Removed the ICs, installed sockets, jumped the contacts for 20k range to test and resistance reading is spot on.

I'm going to try and replace them with the LTC222.