Tektronix P6042 Current Probe Repair

[omerd]

Hi!
I'm new here but always been a fan of Dave's videos, so I look forward to being active on this forum as well

I have got a Tektronix P6042 DC-50MHz current probe with some issues and decided to repair it.
This unit was designed in the mid 60's and it's very well constructed, almost entirely discrete-transistor based with documentation available from Tek. It is still a very useful piece of test gear as it can go from DC to 50 meg and supports currents as low as 1mA.
Forum user PA4TIM had repaired one and has very useful info on his website - http://www.pa4tim.nl/?p=3135
Original Tek manual is here - http://www.atecorp.com/ATECorp/media/pdfs/data-sheets/Tektronix-P6042_Manual.pdf

When I got my unit I noticed two things:
1) Probe cable was broken near the front panel strain relief rubber, this was an easy repair just shortening the cable and soldering the wires inside.
2) R11 missing, instead there was just a piece of wire there, and it was disconnected at one of the little studs, i.e. open circuit in place of R11.
R11 is a factory-selected resistor that has something to do with the temperature compensation for the hall effect probe. I tried using the same value as R13, a short and open... this has some effect on the drift of the output but doesn't seem to be related to the main problem I have.

The custom diff. amp looks OK, and now the whole system kind of works. I can adjust the cur/div balance circuits and play with the calibration trim pots inside while holding the tongue at the right angle 
But there is a little problem that just drives me crazy... the P6042 output is designed to be 50-ohm terminated. This is what I do - I either use my internal TDS540's termination or an external terminator, but the reading I get from the probe is about half of what it should be! and playing with the gain calibration pot has very little effect on this issue (not much dynamic range).
When not terminating the probe's output the reading seems to be correct... or at least I can somewhat calibrate the probe to give a correct reading without being 50-Ohm loaded.
I thought maybe someone hacked a 50 ohm resistor inside for some reason, and nope.
This problem exists on both low and high frequency amps inside, I get the same "half voltage" output regardless of if it's DC or say 20 MHz input (at 50 Mhz I should expect a 3db drop).
Anyway, I'm pretty sure when it says "0.1A/Div" (when scope is set to 50mV/Div), it should be one graticule line in amplitude for 0.1A of current... not for 0.2A!!

I have measured the voltage rails and tested some points along the signal path. Also inspected the attenuator and it was NOT modified in any way.
I really can't track the problem down as I don't know what signal levels to expect along the path. All I know is that the final output is no good.
The manual doesn't mention any test points along the signal path for some given reference input. Only some basic performance checks and calibration procedures, all based on final output stage.

I welcome any ideas. Thanks in advance!
I will be glad to share more info, upload photos etc...
Omer

[Alex Eisenhut]

In any case, be extremely careful with the probe itself. The slightest mechanical shock will *destroy* the probe beyond repair.

Don't drop it on the desk, not even from 1 atom up. Gently put it back into the unit. Treat it like a Fabergé Egg.

The R11/R13 resistors are part of some esoteric probe matching system that I don't know how it works.

The best place to ask for P6042 stuff is in the Yahoo Tekscopes group. Search for "x-ray" if you wanna find out how I know how fragile and unrepairable the probe is.

Have you tried degaussing the probe?

[tautech]

You will likely find much of the info you need in "Tek Probe Circuits" linked in this post:
https://www.eevblog.com/forum/testgear/dc-resistance-of-a-1x-oscilloscope-probe/msg538880/#msg538880

[edavid]

First, can you fix the topic title?  It's very confusing since the P6240 is a FET probe.

Second, half gain usually means that half the differential path isn't working.  So, I would look at the M18 output and Q22/Q24/Q29.

Or, maybe you still have a broken wire in the probe cable.

[omerd]

Hi,
Thanks for the quick replies and oops... I did get the model number wrong, thanks for correcting me on this one!
I do try to be very gentle with the probe itself, thanks for reminding me how important this is.
I always degauss the probe.
"half gain usually means that half the differential path isn't working" - shouldn't I expect to see some artifacts in the signal as well?
Now when I think about it, I do get some overshoots and undershoots when feeding a square wave, but I assumed it is due to miscalibration.
I will check some transistor gains and biases when I come home.
I appreciate your help!
Omer

EDIT: thank you tautech for the excellent reading material, lots of useful info on differential push-pull current probe amps. Great stuff!

[omerd]

Update:
I tested the probe wiring again, everything is OK.
If there was a problem with the differential amp or anything in the low frequency amp, I would have expected that the final output will still be OK when testing high frequency signals.
The signal just looks too good along the way, it really looks like the problem is in the final stages of the circuit.
The manual mentions that:

When the probe is unlocked (SW3 grounded) D128 conducts, bringing the base of Q123 up and turning the transistor on. This latches up the output of the high-frequency amplifier and prevents any display.

When I don't lock the probe, the "probe unlocked" light does go on, but I still get the output signal displayed. So something is wrong here.
I tested Q113, Q121 and Q123 (which does the "blanking" of the display), all OK. Replacing them didn't help as well.
I found that R111 was measuring ~60 Ohm instead of 47, so I replaced it.
Another 47 Ohm resistor, R123 (BTW there is a mistake in the manual, it is double-marked as R116 in the layout) was measuring 52 Ohm, replaced it as well.
D128 also tested and looks OK, but actually will never conduct!
If you look at the schematic, "probe unlocked" SW3 grounds the poor diode using R5. R5 in series with the light bulb forms a voltage divider, giving around -10 volts. On the other side of the diode I get around -7 volts, caused by R125 and R126 divider (BTW, for some reason they are around 1K and not 2K as mentioned in the manual... Arrrr).
The only thing I can think of is that the bulb isn't original, and is much lower in impedance.
That's it for now. I will measure some other components, I have a feeling some resistors have "drifted" a little bit.
Omer

[Alex Eisenhut]

How is the DC reading BTW? Is it also off?

[omerd]

DC reading is off just like AC.
I'm actually tempted to reassemble everything and keep using it without 50 Ohm termination, its accuracy isn't bad this way.
I just can't explain why the 2K resistors measure 1K, two different (also in physical form) 47 Ohm resistors measure 60 and 52 Ohm, and why the blanking circuit is grounded the way it is. Come on, it's Tektronix, there's gotta be a reason why they've designed it this way, having this incandescent-light-bulb-resistor voltage divider 
What on earth can cause resistor values to drift this bad I don't know.
If anyone has a functional unit and is willing to take a few measurements, this would be great!! hey, it's only one screw and you're in 

Omer

[Alex Eisenhut]

DC reading is off just like AC.
I'm actually tempted to reassemble everything and keep using it without 50 Ohm termination, its accuracy isn't bad this way.
I just can't explain why the 2K resistors measure 1K, two different (also in physical form) 47 Ohm resistors measure 60 and 52 Ohm, and why the blanking circuit is grounded the way it is. Come on, it's Tektronix, there's gotta be a reason why they've designed it this way, having this incandescent-light-bulb-resistor voltage divider 
What on earth can cause resistor values to drift this bad I don't know.
If anyone has a functional unit and is willing to take a few measurements, this would be great!! hey, it's only one screw and you're in 

Omer

Well it seems pretty clear that whoever had the unit before modified it because they didn't like the 50 ohm impedance. That's the thing when you buy old test gear, who knows what the previous owner did to it.

Carbon composition resistors do drift up, a 47 ohm measuring 60 ohms is something I'd expect. But a 1K won't drift to 2K unless it's burnt or filed down.

[omerd]

Could be, although usually I can tell by the solder joints and the looks of the PCB. I don't see any modern resistors or semiconductors, all look original with some of them having Tek P/Ns on them.
And what about the "probe unlocked" circuit?
Anyway I'm gonna test all components starting at the final stages against the specs, this is the best I can do for now.
Thanks for your help!
Omer

Edit: I wish I could at least trust the Tek documentation. I keep finding mistakes there. C106 should be in parallel with (R107 & 108). It is NOT. The traces just don't match the wiring net on the schematic. I have confirmed the notation is OK (unlike with R123 for example).

There is another P6042 at some junkyard not far away. Maybe I'll grab it (this one I got for free) and make one functional unit.
I have repaired lots of old (and even vintage) gear, this is the first time I encounter such problems in what looks to be a fairly simple circuit.

[Alex Eisenhut]

I took my probe apart again just to look at the little PCB that the transformer plugs to.
Then a wire popped off, I figured it broke off. But it looks like it was never properly soldered in the first place.
The solder joint on the board just looks like a grainy crater and the wire looks like in the photo.
This isn't the first time I encounter poor soldering from 1960s Tek gear.
It might pay to do a close optical inspection of your unit's solder joints.

[Alex Eisenhut]

Edit: I wish I could at least trust the Tek documentation. I keep finding mistakes there. C106 should be in parallel with (R107 & 108). It is NOT. The traces just don't match the wiring net on the schematic. I have confirmed the notation is OK (unlike with R123 for example).

It's important to match the serial number of the unit to the schematic. There's also change notices at the back of service manuals that need to be looked at.

[omerd]

Thank you Alex again for your help!
I always go over and resolder any joint that doesn't look shiny enough or if I see oxidation on component leads, although I did not take the main PCB out this time (lots of wires). Removing the bottom plate with the two screws gives access to most of the circuit.
My probe itself looks good. I have confirmed continuity all the way while twisting and playing with the cable to assure nothing is loose inside.
The only visible damage is a tiny chipped piece on the sliding half of the loop


I have re-seated the probe in its socket (as well as all transistors in the unit).
Regarding the manual - this is the only one I found and the only correction there is some typo. I will have to reverse engineer the unit a little bit.
I would love to confirm that my problem is in the final stage as I suspect. If someone can feed in a known signal, set the knob to some fixed position, and report the output right after the attenuator (on J80 with coax connected of course), that would be really really helpful. If it's about the same on my unit then the problem is with the final stage for sure.
I wish the manual could give some test points along the signal path for a known reference input signal. Unfortunately it doesn't.

Omer

[Alex Eisenhut]

There's a little known piece of info that's only included when you get a spare transformer, you should only use water to clean the surfaces of the transformer assembly.
Or Freon...

[omerd]

Just a sec, I'm getting my Freon 11 bottle to clean the probe...

Playing with the current/div balance knob I have noticed strange behavior of the signal which worsen when the unit got warm. It turned out Q54 was slowly dying. At first it measured OK with DMM on diode mode, now it's completely gone 
Replaced with PH2907 as I don't have 2N2905 right now. I will anyway have to curve-trace some transistors and find a "match" for best results. I don't care about thermal drift right now... hate this old fashioned can package.
So it seems both differential paths on low frequency amp are OK now, high-frequency amp is OK, attenuator is OK... but I still get half voltage when terminating the line.
I have replaced a few resistors that were more than 10% off.
It's probably something so stupid that when I'll find it I'll bang my head on the wall 
Omer

[omerd]

I've been replacing a few more passives, cleaning up the board and going over some solder joints.
Then I tried a full calibration. While playing with the low frequency loop gain pot, Q44 and Q45 failed. I replaced them and now things look much better:
1) High frequency amp is working very well. When feeding a high frequency signal (say, >100KHz) the output is perfectly fine, while the gain pot is around half way turned. Couldn't ask for more!
2) Low frequency response is still a little off. This is best seen when feeding s square wave - it "overshoots" and "undershoots" to the correct amplitude, but then decays to a lower amplitude signal.




(Note: the scope isn't 50 Ohm terminated internally, I use external terminator this time)
Playing with the pots on the final stage does very little effect regarding this issue.
I would still very much appreciate some reference data from a known-working unit!
Omer

[tautech]

Are you aware Tek used a link between Probe Cal and Gnd for checking their current probes.
Check your scope manual to see if your model supports this.
Sure it will be only at 1 KHz. but there should be a specified current for that setup.
That will help you check amplitude is correct.

[omerd]

I know my output amplitude is NOT correct, why check?
I use a signal gen terminated with 50 Ohm resistor as my wire-under-test current loop. CH1 is showing the voltage across the resistor.
Problem is now only with DC and low frequencies. There are various components along the signal path to compensate and "combine" the high and low frequency signals to a smooth output.
Something is still wrong but I'm on the right track, I feel.
Omer

[tautech]

I know my output amplitude is NOT correct, why check?

Because the Tek current probe check setup is a calibrated accurate pulse (assuming your scope supports it) and expected gain can then be verified along the signal path.
Might help you pinpoint the remaining faults ?

[Alex Eisenhut]

Are you aware Tek used a link between Probe Cal and Gnd for checking their current probes.
Check your scope manual to see if your model supports this.
Sure it will be only at 1 KHz. but there should be a specified current for that setup.
That will help you check amplitude is correct.

Witness the power of my 547!

[omerd]

Following the manual you need a pile of vintage Tek gear to calibrate the probe
I do keep an old 465B ("Any good lab should have an analog scope", right?), I think it has a current cal loop on the front.
What I meant to say is that my probe is still way off in DC and low frequencies. There is clearly something wrong there and it's not a matter of tweaking a pot or two.
Say I use the Tek scope current calibration loop, where did you find information about signal levels other than the final output? Did I miss something..?
To pinpoint the problem: DC and LF response is poor. DC is attenuated just like before, I don't get much more than half of the expected amplitude.

[edavid]

Say I use the Tek scope current calibration loop, where did you find information about signal levels other than the final output? Did I miss something..?

No, you are right.

To pinpoint the problem: DC and LF response is poor. DC is attenuated just like before, I don't get much more than half of the expected amplitude.

Bad Hall element? 

[omerd]

Bad Hall element?

I hope not.. It does work, it is sensitive and noise free.
Maybe it's yet another drifty resistor causing all the problems, who knows.
Omer

[omerd]

Found some time to get back to this probe...
Setting the high frequency correctly is fairly easy. The final trim pots can modify the signal by orders of magnitude. When getting closer to DC this final stage does nothing, so I went back to the Hall sensor and the first stages.
Now, I feel a little stupid that I didn't notice it before, but the schematic does give some voltage test point levels around the custom differential amp. Why couldn't they put the letter "V" following the number? Anyway, the outputs should float on 8 volts, which they do, and the input leads (pins 2 and 10) should be at 0.6 volts. There are even wires going to some nice test socket (pins 1 and 3).
According to the manual, the current flowing through the Hall sensor should be about 20mA and is right about there.
The voltage on the output pins of the device that go to the custom amp is only 0.3 volts, instead of 0.6!
So it's possible that the sensor itself is faulty in some way or maybe I can just increase the current a little bit, hmmm... I don't want to kill the poor sensor. What do you guys say?
UPDATE: Increasing R14 to bias the output at 0.6v still doesn't fix the problem. This is to be expected when dealing with a differential signal I guess. I just wondered maybe I was under the working point of some transistor inside the IC and that was causing the problem. I will try measuring the Hall sensor voltage directly.

[Alex Eisenhut]

Found some time to get back to this probe...
Setting the high frequency correctly is fairly easy. The final trim pots can modify the signal by orders of magnitude. When getting closer to DC this final stage does nothing, so I went back to the Hall sensor and the first stages.
Now, I feel a little stupid that I didn't notice it before, but the schematic does give some voltage test point levels around the custom differential amp. Why couldn't they put the letter "V" following the number? Anyway, the outputs should float on 8 volts, which they do, and the input leads (pins 2 and 10) should be at 0.6 volts. There are even wires going to some nice test socket (pins 1 and 3).
According to the manual, the current flowing through the Hall sensor should be about 20mA and is right about there.
The voltage on the output pins of the device that go to the custom amp is only 0.3 volts, instead of 0.6!
So it's possible that the sensor itself is faulty in some way or maybe I can just increase the current a little bit, hmmm... I don't want to kill the poor sensor. What do you guys say?
UPDATE: Increasing R14 to bias the output at 0.6v still doesn't fix the problem. This is to be expected when dealing with a differential signal I guess. I just wondered maybe I was under the working point of some transistor inside the IC and that was causing the problem. I will try measuring the Hall sensor voltage directly.

Sensor's dead. Sorry. Maybe you can just re-purpose it as an AC probe.
I never found out exactly how the hall sensor works. It seems to be a thin-film plated right on the ferrite core. Contacts are somehow made and the two core halves glued together with something that even epoxy remover can't touch. I don't know why a shock breaks it.
It's quite baffling.