HP 3465A DMM repair

[bd139]

Quick repair on this one. Obtained from a hamfest. Unloved and uninterested. 4.5 digit manual range bench multimeter from HP. This one circa 1979. Felt sorry for it so here it is.



Of course when I got it, it was broken. Drifting on all ranges and a severe offset towards one polarity on every measurement. Checked the power supply and reference voltages and all good. So  I spent some time staring at the wrong version of the service manual. Of course there are a couple of revisions of this DMM and the components are in different places. Eventually found the correct one, which was the really badly scanned one from Keysight. Ugh.

First diagnostics concentrated on the auto zeroing circuit. This was eliminated after grounding the input FET and getting sensible zeroing. The front end was buzzed out and impedances checked. All good. Hmm. Eventually I was poking around the Hi-Z parts of the input chain and one of the input clamping diodes fell to bits!



As you can see the leg has disintegrated. When this is out of circuit it behaves properly! The diode does look like a diode out of circuit but clearly has some leakage problems going on.

This is an HP 1901-0586 which is completely undocumented and expensive to get hold of. The service manual specifies it as a general purpose diode 30V 25mA, so pretty much anything will do here apparently. Well that's not good enough for this as it's very high impedance IMHO and it was clearly affected by leakage of some sort. I'm not paying for an original diode so after doing some research and simulations of the clamp in LTspice, the B-E diode in a 2N3904 is sufficient for the job so I will snip the collector off one and solder it in, then carefully clean it and re-test.

That I will do tomorrow.

Calibration is slightly off which is probably my fault for turning the wrong pot inside when I was checking out the amp balance.

[Caliaxy]

I was poking around the Hi-Z parts of the input chain and one of the input clamping diodes fell to bits!
As you can see the leg has disintegrated. When this is out of circuit it behaves properly!

Fortunate accident

I'm not paying for an original diode so after doing some research and simulations of the clamp in LTspice, the B-E diode in a 2N3904 is sufficient for the job so I will snip the collector off one and solder it in, then carefully clean it and re-test.

Nice way to replace a hard-to-find low-leakage diode. I read about it, never got to try it myself. Low leakage PN4117s connected as diodes (S and D tied together as the cathode, G as the anode) are cited to work better, but if 2N3904 is good enough, why not. Let us know how it behaves. I have a handheld DMM drifting on DCV only, was planning to work on it. Now I'm inspired

[bd139]

I actually looked at JFETs, but not the that one. That specifies 1pA leakage but apparently you get around 4-5pA on a 2n3904 B-E junction which is probably fine for this job. The auto-zero will kill any small persistent offsets off from leakage. The problem I had was it was dragging it heavily down. I probably don't have the time or energy to work out leakage here so it'll be a cut and try approach.

And I just discovered I have no 2N3904s around so this will be a few days off. And on to the HP 427A sitting here. That uses current injection into a FET to kill offsets as well. More fun which will probably be broken!

[MrYakimovYA]

Nice old equipment! I do like it! Some time ago I bought an old GDM-8145 for $25. It is completly working! Sorry for offtop) Manual range switching is fantastic. Sometimes it is more convinient than auto ranging or pushing button "Manual" on autoranging devices.

[mawyatt]

Would think you should get lower leakage and higher breakdown from the B-C junction rather than the B-E junction of the 2N3904.

Best,

[Caliaxy]

I actually looked at JFETs, but not the that one. That specifies 1pA leakage but apparently you get around 4-5pA on a 2n3904 B-E junction which is probably fine for this job.

Would think you should get lower leakage and higher breakdown from the B-C junction rather than the B-E junction of the 2N3904.

According to this plot in the Art of Electronics, the leakage current of the B-E junction of 2N3904 is <100fA (up to ~80mV). Not bad at all... It's also smaller than the leakage current of the B-C junction at the same forward voltage. The breakdown voltage is indeed 10x higher for B-C, per the data sheet (60V vs. 6V).



(leakage forward current vs. applied voltage)

[bd139]

Good find with that reference. Didn’t even think to look in the bible for this one

[bd139]

Nice old equipment! I do like it! Some time ago I bought an old GDM-8145 for $25. It is completly working! Sorry for offtop) Manual range switching is fantastic. Sometimes it is more convinient than auto ranging or pushing button "Manual" on autoranging devices.

Indeed. I generally tend to prefer manual range meters myself as well.

[BU508A]

As you can see the leg has disintegrated. When this is out of circuit it behaves properly! The diode does look like a diode out of circuit but clearly has some leakage problems going on.

This is an HP 1901-0586 which is completely undocumented and expensive to get hold of. The service manual specifies it as a general purpose diode 30V 25mA, so pretty much anything will do here apparently. Well that's not good enough for this as it's very high impedance IMHO and it was clearly affected by leakage of some sort. I'm not paying for an original diode so after doing some research and simulations of the clamp in LTspice, the B-E diode in a 2N3904 is sufficient for the job so I will snip the collector off one and solder it in, then carefully clean it and re-test.

A proper replacement for the input protection diodes would be the PAD1. But they are not cheap, I can tell you!
Mouser is asking ca. 15 Euro per diode.

Attached datasheet from Siliconix.

[bd139]

That’s what I paid for the entire meter 

2N3904 it is 

[Vince]

BD, this thread is lacking in the annotated schematic department... this thread needs more illustration please !

[MrYakimovYA]

BD, this thread is lacking in the annotated schematic department... this thread needs more illustration please !

There are plenty of circuits diagrams, theory, figures, descriptions ( ): http://hparchive.com/Manuals/HP-3465A-Manual-SNprefix-1546.pdf

[bd139]

If only it was that easy. The revision of the DMM I have is not covered by that manual and the component placement and front end are slightly different.

The only copy is the Keysight one and that's awful quality scan:

[mawyatt]

Would think you should get lower leakage and higher breakdown from the B-C junction rather than the B-E junction of the 2N3904.

According to this plot in the Art of Electronics, the leakage current of the B-E junction of 2N3904 is <100fA (up to ~80mV). Not bad at all... It's also smaller than the leakage current of the B-C junction at the same forward voltage. The breakdown voltage is indeed 10x higher for B-C, per the data sheet (60V vs. 6V).



(leakage forward current vs. applied voltage)

Interesting, would not expect the B-E junction to be lower, even in forward direction!! However, suspect the protection diode in the meter is used in reverse direction and the B-C junction should have lower leakage in reverse as well as higher breakdown.

How is the protection diode used in the meter?  This will determine which PN junctions in the 2N3904 are best for the intended meter use.

Best,

[bd139]

Purely coincidence but I just posted the front end and clamp in my previous post

CR9 is the faulty diode.

[xrunner]

BD, this thread is lacking in the annotated schematic department... this thread needs more illustration please !

Did you check that option when you paid for your subscription? 

[Vince]

BD, this thread is lacking in the annotated schematic department... this thread needs more illustration please !

Did you check that option when you paid for your subscription? 

Pretty sure I did ! 

[Vince]

BD, this thread is lacking in the annotated schematic department... this thread needs more illustration please !

There are plenty of circuits diagrams, theory, figures, descriptions ( ): http://hparchive.com/Manuals/HP-3465A-Manual-SNprefix-1546.pdf

Yes I know what a SM is thanks... but the point of a repair thread is to explain and show people what there is to show to understand and see what's going on. If it's just about a one liner  " I replaced a diode have a good day", well might as well not create the thread to begin with 

[Vince]

If only it was that easy. The revision of the DMM I have is not covered by that manual and the component placement and front end are slightly different.

The only copy is the Keysight one and that's awful quality scan:

Wow crap scan indeed, can't even see which one is CR9 ! 

Looks like the one on the up side of the input path, going to the negative 7V rail.... OK... so you were wrong, your offset repair DOES pertain to my old Rochar Nixie DVM, as this is general purpose stuff and my meter has something similar a s well

[bd139]

BD, this thread is lacking in the annotated schematic department... this thread needs more illustration please !

There are plenty of circuits diagrams, theory, figures, descriptions ( ): http://hparchive.com/Manuals/HP-3465A-Manual-SNprefix-1546.pdf

Yes I know what a SM is thanks... but the point of a repair thread is to explain and show people what there is to show to understand and see what's going on. If it's just about a one liner  " I replaced a diode have a good day", well might as well not create the thread to begin with 

To be fair I did say "one of the input clamping diodes" which is pretty easy to locate if you have any interest in the front end of the DMMs.

Sometimes it's better to leave breadcrumbs and a journey because the people who don't follow the breadcrumbs as a rule aren't going to be satisfied with the destination either

[bdunham7]

This is an HP 1901-0586 which is completely undocumented and expensive to get hold of. The service manual specifies it as a general purpose diode 30V 25mA, so pretty much anything will do here apparently. Well that's not good enough for this as it's very high impedance IMHO and it was clearly affected by leakage of some sort. I'm not paying for an original diode so after doing some research and simulations of the clamp in LTspice, the B-E diode in a 2N3904 is sufficient for the job so I will snip the collector off one and solder it in, then carefully clean it and re-test.

The original HP part appears to be a "low-leakage PIN diode" according to some descriptions, but it also might be a JFET in disguise.  I believe someone else has already pointed out an appropriate replacement from InterFET.  In this case, I think it is just the low reverse leakage that matters, not much else.  However, looking at the circuit you posted, I think a reverse-biased 2N3904 BE junction won't quite cut it as the circuit as shown could have as much as 8 volts reverse bias across one diode before the other one clamps.  Having the reverse-biased BE junction go avalanche might actually sort of work in an unintended way, but I'm not sure it is a good long-term plan.

The CB junction might be a better bet, otherwise the PN4393 JFET from Central Semiconductor is something I've used in this sort of application with excellent results.  At 4 volts they would likely leak <1pA.  And you should check to make sure that the resistors and zeners in the biasing circuit for these clamps are all working properly as well.

[bd139]

Good points there. Actually I just noticed there was an error in the LTspice simulation I did of the clamp. Just running my test cases through it again now... 

[Kleinstein]

There are also afforable low leakage diodes BAV199 available in a SMD case. Not tested for super low current, but typical leakage current < 10 pA.

[Caliaxy]

In this case, I think it is just the low reverse leakage that matters, not much else.  However, looking at the circuit you posted, I think a reverse-biased 2N3904 BE junction won't quite cut it as the circuit as shown could have as much as 8 volts reverse bias across one diode before the other one clamps.  Having the reverse-biased BE junction go avalanche might actually sort of work in an unintended way, but I'm not sure it is a good long-term plan.

Totally agree, the B-C junction would be the appropriate one here (for 2N3904), as both diodes are can get reverse biased at a voltage higher than the reverse breakdown voltage of the B-E junction...

[TimFox]

After the introduction of silicon planar transistors, you should always assume that the B-E reverse breakdown voltage is much lower than the B-C breakdown voltage, even for high-voltage V_CEO devices.