HP34401A Ohms function problem

[Dr. Frank]

I think I rather overlooked Q211. I figured since source and drain where shorted it was impossible to test in circuit and didn't check between gate and source/drain. This measures 80Ω in both directions on resistance and 0.08V on diode test which presumably cannot be correct. A leak/short would pull Q204 base to ground which would presumably be enough to shut down the array? Initially I will just remove Q211 to see whether that makes a difference but I don't think this would explain the absence of current from Q202? Still, its another item to sort out. I do have some J390's to hand but the operating voltage is 25V rather than 40V for the MMBF4117.

I will have to give the Q202 FET issue some thought and maybe order those NOS MMBT5460s from Israel. At least then we know its the correct part.

It's strange that the voltage at the Drain of Q202 (pin 2) is exactly 0V, maybe something is still shorting it to AGND. Q203, Q204 and Q211 might be short, so if you remove Q211, it should work.

In the BOM of the 34401A and the 34970A, there's definitely a 2N5461 required, as the 2N5460 has 1mA I_DSS only.
The cutoff voltage should be maximum 4.5V, better 2V only, to be selected from the batch, otherwise the FET will not be able to operate properly in the 10MOhm and 100 MOhm  range.

Frank

[WaveyDipole]

Q211 is uses as a low leakage diode.  It does not absultely have to be a JFET.  A low leakage diode would work as well. They just selected the 2N4117 because it is sopposed to be tested for low leakage. A different FET may not be tested as well.  For a substitute one could consider the collector - base diode of the jmall BJT like 2N3904 or BC548.

For a curde test to see if the rest is OK one could also consider a PNP transistor instead of Q202: it should also operate the current soruce, just not very accurate.

From the datasheet the Idss of the 2N4117 is only -30-90μA and Igss -10pA compared to the J309's -2-25mA and 1nA. so I take your point about the part in this position having to be a low leakage part! Thank you for pointing it out. Makes sense given the tiny currents involved, especially on the high resistance ranges.

I do have a question though. You mention the base-collector junction specifically. The circuit board shorts the S and D pads as per circuit diagram with G opposite:

  G
-----
S--D

If a BJT were substituted as you suggest, it would have to have B on one side and C + E on the other.

  B
-----
C--E  (or E--C)

If soldered in directly, both junctions would be in play so in this case, the E pin would presumably need bent up out of the way or cut off but either way disconnected leaving only the B-C junction in circuit? Also, am I right in thinking that B would connect to the G pad? My thinking is that for an NPN transistor the base connects to the P material, as does the gate for an N-channel FET?

So far, I have checked the datasheet for the 2N3904 and the SOT23 pinout has C at the top so might not be a suitable physical fit. I can't seem to find the BC548 in a SOT23 package but so will continue to search for something that might fit. You mention that another FET may not have been tested as well, but, in principle,  is another N-Channel FET with similar Idss/Igss characteristics likely to be suitable?

It's strange that the voltage at the Drain of Q202 (pin 2) is exactly 0V, maybe something is still shorting it to AGND. Q203, Q204 and Q211 might be short, so if you remove Q211, it should work.

Q211 has been removed and confirmed leaky between G and S/D. However, unfortunately there is still no output from Q202.

In the BOM of the 34401A and the 34970A, there's definitely a 2N5461 required, as the 2N5460 has 1mA I_DSS only.
The cutoff voltage should be maximum 4.5V, better 2V only, to be selected from the batch, otherwise the FET will not be able to operate properly in the 10MOhm and 100 MOhm  range.

Frank

Thank you once again. Embarrassingly, I had noted down and subsequently incorrectly referred to Q202 throughout as an MMBF5460. Having just re-checked the BOM list after your last post, I now realise that it should, in fact, be an MMBF5461. The only explanation I can think of is that I had mis-read the 0 from the line below, besides my mind is very much distracted at the moment with caring for an elderly parent who has been in and out of hospital several times over the last few days. Whatever the case, this has led me down the the path of a part with a slightly different specification.

Unfortunately, I have the same problem with MMBF5461 as I had with MMBF5460 in that both show as no longer manufactured and I can't find the SMD package anywhere. It is available as a NOS TO-90 package, but again, the J176 is readily available.

So, the bottom line is that I now need to do my homework and find and order something suitable and low leakage in place of the 2N4117 and get a suitable substitute for the 2N5461.

Have you tried another range?
the new FET seems not to work properly, as its Gate voltage is too low, in conducting mode.
have you selected the FET somehow, for minimum 2..3mA for example?

this CC circuit might have a parasitic modus operandi, which will depend on the jFET characteristics. maybe you select your FETs for different pinch off and zero GS current

Are we talking about a paper based excecise, i.e. looking at datasheets, or some kind of testing?
If datasheets, then what parameters am I looking for?
The all seem to have a range for both Idss and Igss, e.g. the one I chose has Idss between -1.5mA to -20mA.
I anticipated that 2..3 mA would be covered within that 1.5-20mA range?
Does this involve some kind of testing to select a particular FET from the 10 that I received in the batch?
Probing loose SMD parts is particularly problematic as they tend to ping off in all directions never to be seen again....

[WaveyDipole]

Farnell still have a stock of 2SK545 which has a higher min Idss (50uA vs 30uA) but has a similar max Idss (95uA vs 90uA). The Igss is much higher though at 500pA vs 10pa so I am wondering whether it is a suitable replacement for the MMBF4117? It seems to be the only other FET I have found so far with Idss measured in uA rather than mA.

[Kleinstein]

Using a BJT as a substiture low leakage diode, the gate of the 4117 would correspond to the base of a NPN or collector of a PNP to get the same direction.
The BC548 is the To92 case version - the SOT23 equavalent is BC848. Other small type should be similar.
From the usual pinout a PNP would likely be easier.

An alternative would be a low leakage diode BAV199 - not tested by the manufacturer, but typical very low leakage.

MMBF4117 (or 4118, 4119 - just with higher threshold) are usually still available relatively cheap.  However I am not  sure if all manufacturers actually test them for very low gate leakage.

[WaveyDipole]

Using a BJT as a substiture low leakage diode, the gate of the 4117 would correspond to the base of a NPN or collector of a PNP to get the same direction.

Thank you for the confirmation and further detail concerning the PNP device.

The BC548 is the To92 case version - the SOT23 equavalent is BC848. Other small type should be similar.
From the usual pinout a PNP would likely be easier.

Ah, different number. Thanks. I see that these, along with the previously mentioned MMBT3904s are readily available.

An alternative would be a low leakage diode BAV199 - not tested by the manufacturer, but typical very low leakage.

Again, these seem to be readily available.

MMBF4117 (or 4118, 4119 - just with higher threshold) are usually still available relatively cheap.  However I am not  sure if all manufacturers actually test them for very low gate leakage.

Have been really struggling to find MMBF4117 anywhere. They come up as "No longer manufactured". I did find a NOS supply in the USA on eBay, but the cost was for 50-off and with considerable postage and import duty costs. The total cost would come to somewhere near 50GBP. If you or anyone is aware of any other source I would be interested to know.

The diode seems the most favourable and convenient in the circumstance. It fits the profile and has two diodes in series with the common at pin 3 which corresponds to the G position. It would be just a matter of soldering pins 2 (cathode) and 3 (anode) and cutting or lifting pin 1 (anode) to leave it disconnected.

[Kleinstein]

Mouser still has some mmbf4117 in stock and is supposed to get new ones in august, with no sign of an EOL. Similar the european site TME.eu still has some in stock at some 25 cents, but maybe 10 GBP for postage.

For the low leakage diodes one has a typical / expected leakage level and a test limit that can be quite a bit higher, as testing to very low current is slow and thus cost money.
The original 2N4117 are supposed to be tested to a 10 pA (or similar) limit. For the mmbf4117 they still give the same limit, though I am no so sure they actually test them all at the low price. The SK545 may be a very similar chip, just not tested to the same limit but a much higher 500 pA. Still chances are low to find one with a leakage current higher than 50 pA.

For the expected / typical leakage current I would expect the BAV199 to be lowest - after all it is made just for low leakage. The mmbf4117 is probably similar.
The BJTs leakage is likely slightly higher, though very likely still good enough ( e.g. < 10 pA). For a quick test one can measure the CE leakage, which is higher by a considerable factor.

There are also special low leakage diode tested to low pA levels (e.g PAD5) - but these are expensive an internally they may be 2N4117.

[Dr. Frank]

I've successfully repaired my DAQ34970A: https://www.eevblog.com/forum/repair/hp-34401a-error-612-613-615-617-618-619-621/msg3571485/#msg3571485
AD706 was broken, as usual.

You will definitely need a junction FET for Q211, as the BC diode of a bipolar transistor usually breaks down (zener) at about 5V.. but the current source may deliver up to +13V to the transistor array.
I used a certain FET to improve the Ohm mode protection circuit of the 121GW multimeter, which could withstand 25V. Please search for that thread..
I assume, that you could also use a p junction FETs, when reversing the polarity inside the circuit.

Frank 

[WaveyDipole]

Yey! I have it working! 

Additional FETs arrived today so the J177 previously soldered into position Q202 was replaced with a J176 (MMBFJ176) as was originally suggested back in your posts on the 10th May. Since I am still waiting on the arrival of MMBF4117s, Q211 is currently not present in circuit.

On power up, the meter performed its boot-up routine as usual and the resistance function was now working and the continuity/diode test function was now returning sensible results as well. I measured a handful of resistors from ohms to meg-ohms all of which returned appropriate results. The test current levels for each range were checked and seem to be in the right ballpark.

There was still one error present:

620 - AC rms zero failed

Checking the code in the service manual, it states "The internal residual noise of the ac section is measured and checked against a limit of –10 mV to 70 mV at the output of the rms-to-dc converter."

The meter was on the bench, opened up and shields had not been replaced. Also the LED bench light, which is known to generate some noise, was turned on. It occurred to me that perhaps the meter was picking up EMF from the environment and sure enough, when the TEST ALL sequence was performed again with the lamp turned off and the case covering the chassis, the test sequence returned a PASS! Success!

As it turns out then, this repair required a little more than just replacing the AD706. In addition, at least 3 transistors in the array Q203-Q210 had failed along with the FETs Q202 and Q211. Fortunately it looks like U101 was still fine.

On the Fluke 187, the 100Ω 1% tolerance resistor being measured in the photo returned a reading of 99.91Ω. The Brymen 869s returned a reading of 99.79. The reading on the 34401 falls between the two which looks sensible. At this point I have only carried out 2W measurements but once the MMBF4117s arrive, the repair can be completed and the meter tested thoroughly on all ranges in both 2W and 4W modes.

Thank you Dr Frank and Kleinstein for helping me get to this point. Initially ordering an incorrect FET notwithstanding, I am nevertheless thrilled to have the meter fully operational again.

[Dr. Frank]

Hello WaveyDipole,

it's always a pleasure  to do such remote failure analysis, which by itself is not easy, due to limitations in communication and lack of direct measurement investigation..
This time it was even more thrilling, as more components were affected by a real HV overload, and not only this boring AD706 deterioration  .

So, congratulations and have fun with your fine piece of equipment! 

Frank 

[WaveyDipole]

Just as a final postscript, the MMBF4117 arrived today and has been installed. The Ohms function was re-tested and continued to work as expected. The ohms source current was measured again to make sure installation of the new FET had not affected anything and the measurements were found to be correct on all settings. The meter was then re-assembled and a final self-test performed which returned a PASS. Lastly, all functions were briefly checked and found to be working correctly. The repair is now complete.