Author Topic: Tektronix TCP202 current probe repair - Schematic and suggestions needed  (Read 18494 times)

0 Members and 1 Guest are viewing this topic.

Offline Tantratron

  • Frequent Contributor
  • **
  • Posts: 393
  • Country: fr
  • Radio DSP Plasma
    • Tantratron
Also sounds like you're not getting any signal out of it.  Can you see any deflection with just a test DC current of 1A?

Just connected a DC load from then going up to 1A while the TCP202 clamped... no deflection at all on my TDS540C while pushing at different times the force trig
 

Offline MarkL

  • Supporter
  • ****
  • Posts: 2120
  • Country: us
If it fails on both your TDS504 and the 1103, it sounds like it's the probe.  You could also do one final check and do the 1A on/off test while measuring the output of the probe with a DMM.  Also check the power supply pads on the 1103 TekProbe connector.  Pinout is here:

  http://w140.com/tekwiki/wiki/1103

Do either of the offset adjustments on the probe (not on the 1103) move the trace?  Or change the DMM reading?

There are two connectors inside the probe housing, and one inside the compensation box.  Since it was presumably working before shipping, you could check to make sure that everything is seated properly.

Be careful disassembling the probe head.  There's a tiny ball bearing which can (and will) roll away unexpectedly.  If you lose it, the jaws will not "click" closed.

If you purchased the probe as "working", I would check with the seller before opening anything.  They might not accept a return if you break any seals.  And if it was supposed to be working and they don't want you to mess with it, I would return it.
 

Offline Tantratron

  • Frequent Contributor
  • **
  • Posts: 393
  • Country: fr
  • Radio DSP Plasma
    • Tantratron
Thanks Mark for helping me.

Could it be a situation that the trip from Austria to France would have created an internal connector loose via chock or vibration ?

In the meantime, i've written to the eBay seller whom I now discover is a professional seller plus he works as a calibration technician to big company called TRESCAL Group (found his name on LinkedIn). What is wierd, this man is in the know of all these advanced and legacy testing gears plus he produced a calibration certificate on behalf my company.

However one thing which annoyed me was the fact he said the probe to be in excellent condition but he never showed a picture where the internal jaw is partially cracked (see attached picture). Just hoping this will end OK because I've paid 500€ + 18€ shipment...

Attached pictures I've sent him, you'll see my test using an induction lamp, my good old P6021 and P5210 then the flat output TCP202.

Albert
 

Offline MarkL

  • Supporter
  • ****
  • Posts: 2120
  • Country: us
Yes, I was suggesting that maybe a connector popped out during shipping, assuming the seller was being forthright.  It happens.

But metal parts around the core exposed, it looks like?  If so, that piece of metal shielding is connected to ground when the jaw is locked.  The broken insulator is supposed to protect you and your scope when measuring current through uninsulated wires.

Besides the trouble with the functionality, that's a safety issue.  Considering what you paid, I would return it as defective immediately.
 

Offline Tantratron

  • Frequent Contributor
  • **
  • Posts: 393
  • Country: fr
  • Radio DSP Plasma
    • Tantratron
The seller did not respond to my query sent yesterday so this morning I declared the issue to the eBay protecting system. He now accepted to refund me so I will ship back the probe. He claims to have tested the probe, that it worked prior shipping so my oscilloscope to be the problem. Anyway he did not comment on the other issue you have raised regarding the lack of isolation inside the jaw having in mind his eBay add said excellent condition.

One last question, my TDS540C does self-switch to 50 ohms, DC and displays A (Amperes) instead of V (Volts) once the TCP202 is inserted then locked into of of its Channel input. Do you know if the electrical self-recognize this TCP202 probe by the TDS540C is just passive coded electrical connector or is managed by some active electronics in the probe internal PCB's.

What is certain, no signal comes from this probe wether connected to a valid 1103, a valid TDS540C (I've tried all 4 channels) and there is a safety issue inside the jaw system. If some of you have such probe to sell and in good working condition, let me know since I need this probe along my actual tek P5210 to perform testing on behalf the University of Toulouse.
 

Offline Weston

  • Regular Contributor
  • *
  • Posts: 217
  • Country: us
That sucks that the probe you bought was a dud, best of luck in getting that resolved.

I bought a working TCP202 off ebay in good condition in the beginning of march for $450, its really weird how ebay prices have almost doubled since then, I wonder what causes such price trends.

From what I know, the read out is done via some standardized resistor value on one of the tekprobe pins. On the internet at large / eevblog forum there are tables of what resistor corresponds to what readout.
 

Offline MarkL

  • Supporter
  • ****
  • Posts: 2120
  • Country: us
I'm sorry the purchase didn't work out for you.  I'm sure there's one out there somewhere with your name on it.

The TCP202 has a 24C02 EEPROM in the compensation box which is read by the scope.  I2C is used on the TekProbe II interface.

Probes with readout pins have a resistor that will set the scale multiplier, but I'm not aware of any values that will make Tek scopes change the units or termination to 50R.
 

Offline Tantratron

  • Frequent Contributor
  • **
  • Posts: 393
  • Country: fr
  • Radio DSP Plasma
    • Tantratron
The TCP202 has a 24C02 EEPROM in the compensation box which is read by the scope.  I2C is used on the TekProbe II interface.

Probes with readout pins have a resistor that will set the scale multiplier, but I'm not aware of any values that will make Tek scopes change the units or termination to 50R.

Interesting to know that the TCP202 has a digital I2C serial bus to communicate between the probe part number and the digital scope. I guess same design (24C02 EEPROM with I2C bus) inside my high voltage differential probe P5210.

Would you have a document or datasheet which details the precise signal mapping of the TekProbe output pins, there 4 pins on the left and 3 pins on the right then the BNC connector (GND and signal) when inserting inside say my TDS540C front panel ?
 

Offline MarkL

  • Supporter
  • ****
  • Posts: 2120
  • Country: us
...
Interesting to know that the TCP202 has a digital I2C serial bus to communicate between the probe part number and the digital scope. I guess same design (24C02 EEPROM with I2C bus) inside my high voltage differential probe P5210.
I have a P5205 and it does not have an EEPROM.  The x50/x500 button changes the resistance on the "data" pin closest to the flange on the BNC collar.  So, it changes the scale on the scope like the old-style probes.  The 50x position is 1126R, but interestingly the probe needs power to set 676R for the 500x position.  There must be a little bit of electronics involved with the readout pin.

Quote
Would you have a document or datasheet which details the precise signal mapping of the TekProbe output pins, there 4 pins on the left and 3 pins on the right then the BNC connector (GND and signal) when inserting inside say my TDS540C front panel ?
It's an undocumented interface, but there's some info here:

  http://w140.com/tekwiki/wiki/Tekprobe_BNC_connector

A better resolution version of the pinout is here, along with some info on the evolution of Tek connectors here:

  https://download.tek.com/document/TekVPI.pdf

And someone has an adapter board with labeled pad functions:

  https://www.eevblog.com/forum/circuit-studio/example-project-tektronix-tekprobe-adapterbreakout/msg1201386/#msg1201386

If you search around the net there's various reverse engineering projects.


I like TekProbeII probes because now that Tek's patent on it has expired (US4708661A), other major manufacturers are offering adapters to use Tek probes on their scopes.  There's no need to re-buy all your expensive active Tek probes if you move to another scope manufacturer (well, minus the cost of the adapters).
 

Offline Weston

  • Regular Contributor
  • *
  • Posts: 217
  • Country: us
I don't think I ever explicitly check the square wave response on my TCP202 when I got it, so I am unsure if this is a new issue or an old issue, but I have realized the square current step response is quite bad on my probe. It almost seems like a HF/LF compensation issue, but there is no way to adjust that. Here is a waveform where I compare against at CT-2 current probe.

1016128-0

This has to be a defect, right? No way this is acceptable behavior for a commercial probe. Given that the probe works down to DC the hall effect sensor is still good.

Right after degaussing before it jumps to the correct DC level the response looks more compensated. Could something have damaged the JFET that isolates the degaussing circuity from the amplifier? I have inspected the PCB and everything seems visually fine. I am pretty close to just replacing every IC on it that I can identify.

1016132-1
 

Online tautech

  • Super Contributor
  • ***
  • Posts: 28148
  • Country: nz
  • Taupaki Technologies Ltd. Siglent Distributor NZ.
    • Taupaki Technologies Ltd.
 :-//
Why is the trigger level outside the waveform ?
Avid Rabid Hobbyist
Siglent Youtube channel: https://www.youtube.com/@SiglentVideo/videos
 

Offline MarkL

  • Supporter
  • ****
  • Posts: 2120
  • Country: us
I don't think I ever explicitly check the square wave response on my TCP202 when I got it, so I am unsure if this is a new issue or an old issue, but I have realized the square current step response is quite bad on my probe. It almost seems like a HF/LF compensation issue, but there is no way to adjust that. Here is a waveform where I compare against at CT-2 current probe.
...
I have some non-flatness with my TCP202, similar to yours but not as extreme.

Do you have 50R termination on in the scope?  I never used an R&S scope, but I don't see an indication of the termination.  Or maybe using an external terminator?

What do you get with different input frequencies?  Does it eventually flatten out with lower frequency?

If it helps for comparison, below are some screen shots measuring the current from a signal gen output into my scope's Ch1 with internal 50R termination set.  The TCP202 is attached to Ch3 with a Tekprobe II adapter which sets 50R automatically on the scope.  I'd like to tweak the probe to be a little better, but as you point out there are no adjustments, except for the offset and gain. The TCP202 specs provide a very wide berth for "System Aberrations 10% p-p", but I think your probe should be doing better than you show.

The AM503 service manual has some good background on how all these DC current probes work in relation to combining the signals from the Hall sensor and AC transformer.  It may help you understand where to dig.  Perhaps there are some factory selected components in the TCP202 that could be tweaked.
 

Offline Weston

  • Regular Contributor
  • *
  • Posts: 217
  • Country: us
I am using an external 50 ohm load, I have already checked everything in relation to the tekprobe power supply, scope, and load.

I saw the "System Aberrations 10% p-p", so  guess this is technically in spec? I guess I am just surprised its so bad.  Talking about this issue with a friend got them to check their AM502 + A6302 and they were seeing a similar response, although not as bad. Based on that + your measurements I guess its just bad compensation and not any damage to any of the active parts. I wonder if something drifted over time.

It just seems to be some mid-band weirdness, its better at LF or HF. I measured a bode plot today using a 50 ohm shunt.
1016876-0

I will have to read through the AM503 manual and see if I can figure out any ways to tweak the compensation networks, thanks for the advice!
 

Offline Tantratron

  • Frequent Contributor
  • **
  • Posts: 393
  • Country: fr
  • Radio DSP Plasma
    • Tantratron
It just seems to be some mid-band weirdness, its better at LF or HF. I measured a bode plot today using a 50 ohm shunt.
(Attachment Link)
One issue or explanation, this is a hybrid technology probe where they mix or fuse the hall sensor with the current sensor outputs. Normally tektronix design should compensate actively and passively the effect to get it much less that 10% otherwise what is the point to pay so much.

Similar topic happens with Rogowski coil where you cascade or hybrid three types of integrator (two passive and one active)...

So far I've never owned any hall-Transformer probe plus as mentioned before I got unlucky with a purchase on eBay-Austria. I've been refunded thanks to PayPal protection but do you guys confirm acquiring TCP202 is still good investment for laboratory use now we find this non-flat bump response ?

For years I've been using P6021 and P6022 but these are only AC probes with incredible performance... the DC-AC philosopher stone is not easy to design or find !

 

Offline Weston

  • Regular Contributor
  • *
  • Posts: 217
  • Country: us
I spent what is probably too much time this weekend fiddling with the TCP202. I reverse engineered a good portion of the schematic and took a lot of measurements (as a side comment, this board is really weirdly routed, did they use an autorouter or something??). The pulse aberrations are due to the nulling amplifier, nothing in the passive output network of the TCP202 seems to have relevant time constants.

As previous comments have said, the hall effect amplifier is amplified by a OP27 and then further amplified and buffered by a MC1458 + PNP/NPN stage. The schematics for the AM503 and the AM503A have a similar configuration. One thing I noticed was that those units had adjustable feedback on the nulling amplifier output stage and had a R + C feedback network on the hall effect amplifier stage. The TCP202 actually has two sets of unpopulated pads for what I assume is a R + C feedback stage.

I took some interesting waveforms from the TCP202, it is clear that the nulling amplifier is under damped. Here is a capture of the input current (green) probe output (blue) output of the nulling amplifier (orange) and output from the hall effect sensor ( yellow) (orange and yellow  should be inverted).


1018216-0

I guess if I took bode plots of a lot of things I could calculate more optimal feedback network values, but I already sunk more time into this than I wanted to and was getting tired of it. So instead I just iterated through possible tweaks of the two feedback networks. I tweaked the feedback resistor on the output stage, lowering it from 10k to 4.3k and that seemed to make the biggest improvement. Installing a 140p + 83k network on the unpopulated pads seemed to contribute a small additional improvement, lots of changes there just made the waveform worse, including RC corner frequencies used in the AM503 and AM503B.

I changed the current level, so the scales do not exactly align, but you can see the waveform is now greatly improved!

1018204-1

I assume manufacturing tolerances on the hall effect sensor lead to a range of gain values, so it seems to make sense that by changing the loop gain elsewhere I could improve the step response. The hall effect resistance in my probe seems to be lower than typical, I after the 220 resistors that drop the voltage from 5V to the hall effect sensor I was only reading ~+-1.8V instead of the +-3V others were mentioning. I think this may be related to why changing the loop gain was beneficial in my case.

« Last Edit: July 06, 2020, 09:59:37 pm by Weston »
 

Offline MarkL

  • Supporter
  • ****
  • Posts: 2120
  • Country: us
...
I saw the "System Aberrations 10% p-p", so  guess this is technically in spec? I guess I am just surprised its so bad.  Talking about this issue with a friend got them to check their AM502 + A6302 and they were seeing a similar response, although not as bad. Based on that + your measurements I guess its just bad compensation and not any damage to any of the active parts. I wonder if something drifted over time.
...

I dragged out my old AM503 + A6302 and looked at that too.  I think it has a bad core or a bad mating between the transformer surfaces.  Squeezing the surfaces improves a 10% overshoot problem by several percent.  So, mechanical problems can cause bad waveforms too.


...
So far I've never owned any hall-Transformer probe plus as mentioned before I got unlucky with a purchase on eBay-Austria. I've been refunded thanks to PayPal protection but do you guys confirm acquiring TCP202 is still good investment for laboratory use now we find this non-flat bump response ?

The small level of non-flatness I'm seeing in my TCP202 was surprising.  I guess I'm going to have to temper my opinion of this probe a bit.

However, I'll also say that in the 20 years I've been using it I never noticed it.  The resulting waveform was always close enough to what I expected to see, and I never required a high level of precision.

The aberrations may be a good question for the TekScopes mailing list on groups.io.  I also don't know if the behavior is typical of other manufacturers' DC current probes.

 

Offline Weston

  • Regular Contributor
  • *
  • Posts: 217
  • Country: us
I always considered these probes the gold standard in current sensing so it is a bit discerning to see this behavior. I am still trying to figure out if my probe is within specs or is defective. Based on what others are seeing, if this is the case I am surprised its not a well known thing. I guess, similar to how I have not noticed until now, most people do not probe square currents where this behavior is most evident?

In favor of the argument that these probes are all that bad, this post shows the weird crossover behavior in simulation; https://www.eevblog.com/forum/testgear/p6042a-tek-current-probe-replacement-circuits-an-61-lineartechnology/msg3085619/#msg3085619

On the other hand, upon taking apart my probe and making sure that the surfaces were clean and in contact with maximal force, I was able to reduce my LF overshoot by a just barely perceivable amount. I have heard that the cores are fragile, it may be possible that we all have at least slightly damaged transformers with these ebay acquired probes.

The research group I work in has some more modern current probes based on the same architecture. Next time I am in lab I will have to test and see if I can observe similar behavior.
 

Offline MarkL

  • Supporter
  • ****
  • Posts: 2120
  • Country: us
I think you're right that most people are not usually looking at perfectly square current waveforms.  I know I'm not, which is why I never noticed.  Nevertheless, I still think it's a good test of the probe's signal integrity.

My TCP202 was purchased new by me in 2000 and has never been mistreated.

The A6302 was an ebay purchase, along with two AM503's.  The A6302 has the same behavior on both AM503's, so I'm going to say the issue is with the probe.  The AM503 has four compensation adjustments, but none can bring the output flat.  Or even affect the waveform significantly, for that matter.  I'm going to have to go through the official cal procedure for both of them.

I have a second TCP202 which I will retrieve and test.  It was purchased from another user on this forum in the Buy/Sell section a couple of years ago.  It hasn't seen as much use, but again I never noticed anything because I wasn't testing square current waveforms with it.  It's a newer version with a black cable instead of gray, and the serial number is on a tag on the cable instead of on the probe.  But it's still labeled as a TCP202.

I may also try swapping the comp boxes on the two TCP202's to see if the issue moves.

It's also interesting to note there's no response flatness in the verification procedures or specifications for either the TCP202 or A6302/AM503.  Only BW, rise time, and DC accuracy.  You could have a completely bizarre waveform and still meet the verification (and still be within 10% "aberrations").

The Agilent/Keysight N2893A (15A/100MHz) is similar in ratings to the TCP202.  The specifications are even sparser.  Accuracy is +/-1% of reading +/-1mV (DC or 45 Hz to 66 Hz), and rise time 3.5ns.  They include a Bode plot showing (typical) flat response out to 10MHz and then a rolloff to -3dB at 100MHz.  The amplitude scale is in 10dB increments, which is hardly enough precision to show anything interesting.


If anyone else has a DC current probe from Tek or another manufacturer, it would be great if you could take a few minutes to test it and post a screen shot or two of a square current waveform.  It could help answer the question if non-flatness is par for the course, or if all of us so far have messed up probes.
 

Offline MarkL

  • Supporter
  • ****
  • Posts: 2120
  • Country: us
Ok, here's some screen shots.

Since I mentioned it, first up is the A6302 overshoot.  Screenshots included of normal pressure and then significantly more pressure applied to the transformer halves.  The surfaces are absolutely clean, and under a microscope everything looks perfect.  Maybe there's a tiny crack in the transformer core where it's potted that I can't see.

Second up is some screenshots with the newer TCP202 that I purchased used two years ago compared to the older TCP202.  At a low frequency it's absolutely flat.  I couldn't ask for much better than that.  But a closer look at the edge reveals a ringing problem, and by the looks of it at *two* different frequencies.  Great.  It's still within +/-10% aberration, but barely.

All three current probes are clamped onto the same wire.  The order doesn't affect the waveform shapes.  All the differences appear to be introduced by the probes, and confirmed by the shot that also includes Vin (scope measuring across the 50R load which is creating the current draw).  I must admit it's a bit disappointing.
 

Offline MarkL

  • Supporter
  • ****
  • Posts: 2120
  • Country: us
Here's some more screen shots of a couple other DC current probes, again clamped onto the same wire as the TCP202/AM6302 tests above.

The first is an A6303 which is rated for 15MHz/100A and is also used with the AM503 amplifier.  It has three tweakable compensation adjustments on the probe body and it's possible to get the response very accurate to the actual current.  The rise time is showing a BW of only 13.5Mhz so a little more tweaking is in order.  Perhaps an examination of what these pots do can be applied to the the TCP202 and/or the A6302.

I'm not for a moment suggesting anyone use the A6303 for small circuit probing.  The thing is huge.  I was interested in seeing its response since it uses the same Hall/transformer design.

The second is an Aim-TTi iProber 520 which is rated for 10A but is only 5Mhz.  It's normally used open-ended to measure current in PCB tracks, but also comes with a horseshoe shaped attachment (presumably ferrite) to capture the flux around a wire.  It uses a fluxgate magnetometer instead of a transformer/Hall device.  It has a completely flat response apart from some settling after the edge.

That's it for me.  I'm out of DC current probes.  Screen shots from other DC current probes from Tek and other manufacturers welcome for comparison.

EDIT: Oops - had the iProber set on 500Hz BW limit.  Screenshot fixed.
« Last Edit: July 08, 2020, 12:53:50 am by MarkL »
 

Offline esc

  • Newbie
  • Posts: 5
  • Country: us
Re: Tektronix TCP202 current probe repair - Schematic and suggestions needed
« Reply #45 on: August 28, 2020, 08:16:14 am »
Glad I found this thread.

I also have acquired a TCP202 in good condition. The probe "operates" but it doesn't "work".

I did test it with DC 1A and the display seems accurate. Also tested with DC 50mA and it seems accurate, but I need to degauss to go back to the original 0 value - after the DC 50mA test the trace is about 10mA higher then 0 every time. This may be normal, not sure.

The degauss works and so does the coarse zero (on the back of the probe head) and the fine 0 on the side of the plastic box at the BNC.

The sine wave in AC looks ok, but it is smaller than it should be. At 10KHz is about 15% smaller, at 20KHz is the worst, 60% smaller and it recovers at 40KHz to 15% again, staying around there to the MHz range. The square wave response is very ugly.

I measured the transformer in the head and I got the following values (compared to the previously posted measurements):
+3V - -3V = 276 ohm - 176 ohm
+3V - HALL1 = 183 ohm - 93 ohm
+3V - HALL2 = 179 ohm - 92 ohm
-3V - HALL1 = 189 ohm - 141 ohm
-3V - HALL2 = 182 ohm - 141 ohm
HALL1 - HALL2 = 110 ohm - 72 ohm
COIL - COIL = 6.5 ohm - 6.5 ohm

These are with the small PCB in the head disconnected. Are these still in range?

Also, the electronics in the plastic box looks good, no visibly fried parts, and no sign of any repair.

Attached some pictures. The load is a 100 ohm metal film resistor.
Any help would be greatly appreciated. Thanks.



 
 

Offline esc

  • Newbie
  • Posts: 5
  • Country: us
Re: Tektronix TCP202 current probe repair - Schematic and suggestions needed
« Reply #46 on: August 30, 2020, 01:45:05 am »
Sine wave picture at 20KHz.
The two traces are supposed to be identical for the 100 ohm load (20mA/Div and 2V/Div).
« Last Edit: August 30, 2020, 01:46:40 am by esc »
 

Offline esc

  • Newbie
  • Posts: 5
  • Country: us
Re: Tektronix TCP202 current probe repair - Schematic and suggestions needed
« Reply #47 on: August 30, 2020, 01:46:03 am »
One more at 10KHz.
 

Offline Tantratron

  • Frequent Contributor
  • **
  • Posts: 393
  • Country: fr
  • Radio DSP Plasma
    • Tantratron
Re: Tektronix TCP202 current probe repair - Schematic and suggestions needed
« Reply #48 on: October 03, 2020, 09:54:26 am »
I don't think I ever explicitly check the square wave response on my TCP202 when I got it, so I am unsure if this is a new issue or an old issue, but I have realized the square current step response is quite bad on my probe. It almost seems like a HF/LF compensation issue, but there is no way to adjust that. Here is a waveform where I compare against at CT-2 current probe.
...
I have some non-flatness with my TCP202, similar to yours but not as extreme.

Do you have 50R termination on in the scope?  I never used an R&S scope, but I don't see an indication of the termination.  Or maybe using an external terminator?

You might remember that I did purchase few months ago a TCP202 to a eBay-Austrian but it was not working plus the jaw was partially broken so got lucky PayPal reimbursed me. This time, I finally acquired to Belgium a TCP202 along with a 1103 and P6243 where all work except the TCP202 would fail the compensation test on my TDS540C.

I've carefully balanced the DC offset after de the degauss, the TDS540C and the probe been running at least 20min. However there seems to be a response flatness issue, please see attached two pictures taken this morning.

What do you think and what would be the procedure to have a better flatness when 1KHz square current loop is sensed ?

I'm still on my learning curve of the TDS540C use and many options. Do you know if the probe compensation output of the TDS540C and the probe calibrator are standard or precision 500mV and 500 mA references ?

Thanks, Albert
 

Offline Tantratron

  • Frequent Contributor
  • **
  • Posts: 393
  • Country: fr
  • Radio DSP Plasma
    • Tantratron
Re: Tektronix TCP202 current probe repair - Schematic and suggestions needed
« Reply #49 on: October 04, 2020, 07:20:39 am »
Earlier this morning, I thought to try leaving into open the probe jaw (see picture) and noticed then upon not easy or stable thumb pressure or slight movement the probe would display a flat square response. It is very unstable mechanically where the slightest movement would loose the correct response, then I did run the TDS540C probe compensation which this times shows a PASS status (see picture).

When I inspect visually the jaws, the external probe it seems quite neat so what failure or mechanical issue causes the probe to not work correctly when the jaws are CLOSED but the current-transformer with Hall sensor would tricky work when OPEN ?
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf