Keysight N2790A 100 MHz differential probe repair and calibration

[TheSteve]

I recently bought a Keysight/Agilent N2790A 100 MHz 50:1/500:1 differential probe with autoprobe interface from everyone's favorite auction site. It was sold as parts so I figured it would need a repair, and it did.

There are no service docs for this probe and I'm not sure who actually made it(It was made in Japan so very unlikely to be Sapphire).

When I received it I noted that someone had already attempted a repair, and whoever it was clearly had no repair skills whats-so-ever.

A quick test revealed the red input lead was open. So just swap the lead and it is done right? Not so fast.
Testing the black lead revealed 68 ohms of series resistance. I cut open the bad red lead and found a blown inline resistor.
I then cut open the black lead to verify the exact resistor. I was able to verify each lead should have a series 68 ohm resistor internal to the shrouded male banana plug.
A quick check with Digikey and I had replacement Pomona shrouded male banana plugs in the shopping cart. I then looked into the 68 ohm resistors and compared many until I found what I believe is the exact original. It is a slightly special resistor made by Vishay BC Components, NFR25H series. These are fusible, non-flammable metal film resistors. They are covered in a grey epoxy which perfectly match the originals. The part # is NFR25H0006809JR500. I don't know if they alter the performance of the probe at all or are purely a safely feature. Either way I am glad I have the exact resistor installed.

With the new leads and resistors installed the probe worked but the readings were less them perfect - it read 10% low and the voltage offset between ranges was off by a decent margin. I had a look at all of the adjustment pots and it was clear whoever previously had a go at a "repair" decided the solution was to turn every pot fully counter-clockwise.
As I mentioned previously there are no service docs for this probe so I had a look at the typical calibration procedure for some other high voltage differential probes - how different can they be?
After a little experimentation I believe I have the probe very nicely dialed in.
For anyone who needs to also do this here is the procedure I recommend:

Warm the probe up at least 10 minutes before starting. I would also set the offset adjust in the autoprobe head to the midpoint before starting.
Btw, opening the probe requires lifting the front label - there are screws at either end. Use a heatgun/hairdryer to warm up the label so it can be lifted at the ends without damage.

#1 - Range offset adjust
This is so that the zero point remains the same for both the 50:1 and 500:1 ranges.
Connect(short) the probes together. Use a DMM set to DC volts to measure the voltage between the points labeled "ground" and "output" on the PCB.
Set the probe to the 50:1 range and adjust RV5 to get 0 volts DC.
Then switch to the 500:1 range and adjust RV2 to get 0 volts DC
You will need to switch between ranges a few times and perform little tweaks until both ranges show 0 volts, they interact so some trial and error is needed but you can get it so both ranges will only show a millivolt or two.

#2 - Square adjust
This adjustment is used to get a clean square wave with no over or under shoot.
Select the 50:1 range.
Feed in a 10-20 Vpp 10 kHz square wave. Connect positive of your waveform gen to the red input and the negative to the black input.
Adjust CV1/CV3 for the cleanest square wave on your scope.
Swap the input leads and repeat this time adjusting CV2/CV4 for the cleanest square wave.

#3 - Low frequency CMRR adjustment
This is to adjust the low frequency common mode rejection ratio.
Select the 50:1 range.
Connect(short) both probes together and connect them to the hot side of the AC line.
Adjust RV1 for the smallest signal. You should be able to get the sine wave reduced to a millivolt or two of noise.

#4 - High frequency CMRR adjustment - ***I was not able to complete this adjustment as I don't have a suitable source***
This is to adjust the high frequency common mode rejection ratio.
Select the 50:1 range.
Connect a 150 volt 20 kHz square wave to both probes(probes must be shorted together)
Adjust CV1/CV3 for minimum signal.

#5 - Accuracy adjustment
This is to adjust the overall amplitude accuracy of both the 50:1 and 500:1 ranges
Feed 150 volts DC into the probe. Measure the voltage between the "ground" and "output" test points on the PCB.
Set the range to 50:1 and adjust RV3 for 3 volts DC on the DMM.
Set the range to 500:1 and adjust RV4 for 0.3 volts DC on the DMM.
Repeat both steps until both ranges are correct, as with the zero offset adjust there is a little interaction between the adjustments.

You may want to repeat the process a second time to verify everything, then test it and close it up.
In my case while several of the pots were clearly messed with the variable capacitors that change the square and CMRR adjustments didn't appear to have been messed with. If I do get a suitable high voltage source for the square wave I will further verify the CMRR at higher frequencies. For now it seems to be working quite well though. I also think that perhaps with the variable capacitors two are designed to be adjusted for the cleanest square while the others are designed to reduce high frequency CMRR. There is likely interaction between them as well. These are best left unless adjustment is absolutely required.

Any questions let me know.

Here are some pictures of the probe:

[HighVoltage]

Nice repair job, congratulations.

So, the two 68 Ohm resistors were molded in to the input cables?
You have pictures?
I would not have expected that?
Why not having them on the PCB?

Where did you place the new resistors, after replacing the cables?

Reminds me of repairing many Agilent / Keysight current probes by now.
Some previous owners also though it would be a great idea to turn all pots and variable capacitors to their limit. 

Also too bad, that we can not get schematics / service manuals, even for older Agilent current and differential probes.

[TheSteve]

Yes, the resistors are molded into the banana plug end. They were soldered to the plug and then the wire soldered to the resistor just as you would expect. They then mold the plastic around it so they have pretty good strain relief. I added some extra heatshrink internally when I assembled the Pomona connectors so the wire is pinched very tightly in the plastic to provide strain relief and not break the connection to the resistor.
I have no pictures of the internals of the original ones as I had to cut them apart, so only little bits were left.

The resistors could be a safety feature but perhaps the real purpose is impedance matching for higher frequency performance. 100 MHz is pretty crazy for a high voltage differential probe.

Did some good playing around with the probe last night and it really is working great, I am very pleased.

[EEVblog]

The resistors could be a safety feature but perhaps the real purpose is impedance matching for higher frequency performance. 100 MHz is pretty crazy for a high voltage differential probe.

Its would be an impedance matching thing.

[EE-digger]

Nice job and great images.  Always fun to look inside a new toy.

Like Dave says, the resistors are for impedance, probably similar to damped ground leads.  You do have 100MHz BW so these would cut down ringing on edges.

Do you know if the scope end box has a compensation adjustment?  I just went thru this last week on my "new" 1153A succesfully.  I think the documentation is in the 1141A for that family.

I just received an ebay N2796A yesterday and was playing with it today    .  Very nice, if not as compact as Tek probes but still nice.

Totally bummed out this evening as a square wave at low frequency has significant peaking and the seller has no returns (well, unless it doesn't work). I'd hate to return it though as it's like new.

I'm hoping there's a control under the label.

Why do we have to remove labels   

[TheSteve]

The probe head end only has a zero volt adjust(which is user accessible). I forgot to take a picture when I had it open.

I have an N2795A (1 GHz version of your N2796A). If you want to start a new thread with some sample screen shots and pics of exactly what you're probing I can repeat the measurements with my probe to compare.

[EE-digger]

Thanks for the offer.  I will do that.  I probed the waveform with my 1152A (other scope) and 1153a (same scope) at the same time.  Waveform was perfectly flat topped.

I need to correct my comment on the controls for the 1153A.  The head end box only had the offset adjust UNDER the label and another control left alone as minor adjustment did not have visible effect.

[EE-digger]

No photo but the label came off the probe head box 

There are two controls in this model.  One is offset (hole closest to scope) which responded nicely to zero the probe.  The other (closest to cable) has no visible effect.  So, unless there is some form of compensation in the box, this probe may have to go back 

Your N2795A probably has the same controls.

[TheSteve]

Bummer, my N2795A has always been nice and clean. My 1152A's out perform it, but they don't have headlights!

[EE-digger]

** moved to new thread **

[TheSteve]

Recently picked up a case for this probe. It has deserved one for a long time(2 years it seems). I often go for Pelican cases but there are no suitable/reasonable cases that would fit such a probe.
I selected a Flambeau 6785TE from Digikey. Not bad at $22 USD and they have pick and pluck style foam on the bottom half.