Author Topic: Agilent N2772A (Fluke DP120) differential probe repair attempt + teardown  (Read 7790 times)

0 Members and 1 Guest are viewing this topic.

Offline 6151kokodef

  • Newbie
  • Posts: 1
Hi, I'm new to the forum and this is my first post.

I recently bought this Agilent N2772A differential probe for measuring mains. However the probe inputs measures open circuit and there's no output at the BNC.

I then took the probe case apart, removed the red&black cables and applied signal directly to the gold connectors on the PCB. There is output at the BNC, but it is the input multiplied by ~6.72 rather than divided by 20 or 200. So I suspect the red and black cables have internal resistance and therefore stripped them. They turned out to be shielded coaxial with a super thin core wire. However I just couldn't measure any continuity through either cables.

The two PCB inputs measures only 85kohm in between and 45kohm to ground. I tried to put 5.1MEG resistors in series with the input to fake the stated 5MEG input impedance. But this produced a really poor square wave response with ~12us rise time (c.f. spec states 17.5ns).

What could be done to make the probe functional again?

Here are the instruction and service manuals for the probe:
http://media.fluke.com/documents/dp120___iseng0600.pdf
http://media.fluke.com/documents/DP120___sieng0100.pdf

I also noticed nobody has posted internal photos of the N2772A or Fluke DP120 before - so I've attached some here.

Thanks!
 
The following users thanked this post: thm_w

Offline TheRuler8510

  • Regular Contributor
  • *
  • Posts: 176
  • Country: us
  • RF Engineer
6151kokodef,

Thanks for the pics.

Did you ever get it to work?

Thanks,
"There are no facts, only interpretations."
--Friederich Nietzsche
 

Online Mechatrommer

  • Super Contributor
  • ***
  • Posts: 9639
  • Country: my
  • reassessing directives...
how much price did you get this? last time i got 1141 diff probe at cheaper stated as broken. all i have to do is resolder the unsoldered coax output cable and all are working 100% again. your unit probably just a broken cable, i believe the board is still 100% functional. be carefull not to change those trimpots and variable capacitor they should be already calibrated from factory (but later you are going to recalibrate anyway so who cares? just dont further damage the undamaged. see below). what type 5Mohm resistor are you using. carbon? wire wound? use carbon is better, dont use wire wound thats a sin.

freq respond can be compensated by adding parallel (or to ground) C to 5Mohm, or several "parallel RC" in series to simulate distributed/characteristic impedance of the original coax cable, time to fire up spice simulator now?. also looking at how a normal dso probe is compensated by RC network is a good learning experience, if you need to, buy a cheap china dso probe and destruct it to see the internal to get the feeling.

after you sorted it out you need to recalibrate the device again anyway for both input matching, lets hope the designer have put that in the board among those trimpots/varcaps and in manual on how to recalibrate. this is only way if you cant find spare part of the original cables. hope that help. glad to see the internal of it thanks, just cant figure out the ICs used and dificult to trace since its using more than 2 layer board.
if something can select, how cant it be intelligent? if something is intelligent, how cant it exist?
 

Offline TheRuler8510

  • Regular Contributor
  • *
  • Posts: 176
  • Country: us
  • RF Engineer
how much price did you get this? last time i got 1141 diff probe at cheaper stated as broken. all i have to do is resolder the unsoldered coax output cable and all are working 100% again. your unit probably just a broken cable, i believe the board is still 100% functional. be carefull not to change those trimpots and variable capacitor they should be already calibrated from factory (but later you are going to recalibrate anyway so who cares? just dont further damage the undamaged. see below). what type 5Mohm resistor are you using. carbon? wire wound? use carbon is better, dont use wire wound thats a sin.

freq respond can be compensated by adding parallel (or to ground) C to 5Mohm, or several "parallel RC" in series to simulate distributed/characteristic impedance of the original coax cable, time to fire up spice simulator now?. also looking at how a normal dso probe is compensated by RC network is a good learning experience, if you need to, buy a cheap china dso probe and destruct it to see the internal to get the feeling.

after you sorted it out you need to recalibrate the device again anyway for both input matching, lets hope the designer have put that in the board among those trimpots/varcaps and in manual on how to recalibrate. this is only way if you cant find spare part of the original cables. hope that help. glad to see the internal of it thanks, just cant figure out the ICs used and dificult to trace since its using more than 2 layer board.

How much $$ did you pay for yours? Is $300 a good price for a new one?

"There are no facts, only interpretations."
--Friederich Nietzsche
 

Online Mechatrommer

  • Super Contributor
  • ***
  • Posts: 9639
  • Country: my
  • reassessing directives...
How much $$ did you pay for yours? Is $300 a good price for a new one?
$50 (incl the 1142 power module). but i have to pay extra $60 for shipping. i dont think its available new in ebay. the best is "like new" (fully functional) in box that cost $500-$1000 excl shipping. i believe this agilent 2772 is also around that range if not cheaper (used) no i just checked its actually more expensive in ebay than 1141 unit. the one-hung-low brand new is around $300 but goes to ~20MHz BW only.
if something can select, how cant it be intelligent? if something is intelligent, how cant it exist?
 

Offline EEVblog

  • Administrator
  • *****
  • Posts: 31257
  • Country: au
    • EEVblog
Thanks for sharing.
Haven't seen that one before.
Not the usual high voltage resistor ladder input you'd expect to see on the input?
 

Online Mechatrommer

  • Super Contributor
  • ***
  • Posts: 9639
  • Country: my
  • reassessing directives...
it has peculiar arrangement "circled" in the picture. why they want to short those passives (C or L or both) is a mystery to me. i suspect this is why this kind device is overly expensive (read black magic)
if something can select, how cant it be intelligent? if something is intelligent, how cant it exist?
 

Online David Hess

  • Super Contributor
  • ***
  • Posts: 10946
  • Country: us
  • DavidH
They turned out to be shielded coaxial with a super thin core wire. However I just couldn't measure any continuity through either cables.

The center conductor is resistance wire used as part of the probe compensation and usually very difficult to solder to.

Quote
The two PCB inputs measures only 85kohm in between and 45kohm to ground. I tried to put 5.1MEG resistors in series with the input to fake the stated 5MEG input impedance. But this produced a really poor square wave response with ~12us rise time (c.f. spec states 17.5ns).

What could be done to make the probe functional again?

I would try replacing the probes with a pair of cheap x10 oscilloscope probes that include their compensation at the probe end instead of the BNC end.  The attenuation factor will be screwed up but maybe you can fix that either in the differential amplifier or at its output.

The AC and DC common mode rejection ratio would need to be recalibrated which may not be possible.

The bandwidth is low enough that it a pair of attenuating probes made from RG-174 with 5 MOhm resistors on the end could work if you can get the compensation right.
 

Offline MarkL

  • Supporter
  • ****
  • Posts: 1631
  • Country: us
it has peculiar arrangement "circled" in the picture. why they want to short those passives (C or L or both) is a mystery to me. i suspect this is why this kind device is overly expensive (read black magic)
Looks like two passives in parallel to me (left and right), and there are two sets of them (top and bottom) in each of your circles.  Maybe not so magical?
 

Online Mechatrommer

  • Super Contributor
  • ***
  • Posts: 9639
  • Country: my
  • reassessing directives...
Looks like two passives in parallel series? to me (left and right), and there are two sets of them (top and bottom) in each of your circles.  Maybe not so magical?
look carefully the trace in the middle (longest trace in circled area). 2 sets of C in series, and then they are paralled. hence we have 4 passive in series and parallel combo, thats not magic, whats magic is the solid copper trace in the middle shorting those parallel making them useless to my eye. edit: ok i got it now you are right, parallel C and then seried, i need sometime on drawing board to figure it out. i still find its funny though.
« Last Edit: July 22, 2014, 09:14:56 pm by Mechatrommer »
if something can select, how cant it be intelligent? if something is intelligent, how cant it exist?
 

Offline TheRuler8510

  • Regular Contributor
  • *
  • Posts: 176
  • Country: us
  • RF Engineer
To anyone interested:

Techni-Tool has these units in their clearance category for $298.  I just ordered one...seems like a bargain.

Link:
http://www.techni-tool.com/431TE396

Cheers!
"There are no facts, only interpretations."
--Friederich Nietzsche
 

Offline eurofox

  • Supporter
  • ****
  • Posts: 843
  • Country: be
    • Music
I have one like that in "new" condition and pay 120 Euros on ebay France  :-DD
eurofox
 

Online giovannirat

  • Regular Contributor
  • *
  • Posts: 62
  • Country: at
Re: Agilent N2772A (Fluke DP120) differential probe repair attempt + teardown
« Reply #12 on: December 07, 2016, 06:31:37 pm »
I recently bought a Fluke DP120 differential probe, ecaxtly the same as desribed by as6151kokodef in the first post.

Unfortunately I  have the same problem as6151kokodef because I couldn't measure any continuity through the red cable. These cables turned out to be shielded coaxial with a super thin core wire with a 5MOhm resistance.

I asked Fluke for a spare part, but it is no longer available. :--

Is there any chance to get the probe repaired?  I'm not so skilled to find a decent solution...... :-//

Has anyone solved a similiar problem or may be a spare part available? Can the failure solved with another setup (other cables with a resistance in series.....) but still maintaining the original specs (more or less) ?

Any help appreciated ..

Thank you very much

 

Offline Assafl

  • Frequent Contributor
  • **
  • Posts: 582
Re: Agilent N2772A (Fluke DP120) differential probe repair attempt + teardown
« Reply #13 on: December 08, 2016, 09:24:26 am »
Well, the probes are very similar to the (egregiously expensive) probes made for the isolated scopes (like the VSP210 or VPS212 probes). I don't know if the wire is the same but very well may be... They sealed the probe's ground pin with a plug, and changed the sleeve to fit a safety banana plug (very useful). The VPS series are like 200$ a probe so perhaps try to find a broken one ebay...

Another option (if lucky) is that the disconnect happened on the probe body side. In that case just shortening the wires a bit may work. The service sheet shows the configuration of the internal wire connector and splice (it isn't soldered - it is bent on the inner insulator and has a ferrule that is spring held in place).

Yet another option is to try to pry the probe apart. Not sure it isn't molded though....
 

Offline mk_

  • Regular Contributor
  • *
  • Posts: 182
  • Country: at
Re: Agilent N2772A (Fluke DP120) differential probe repair attempt + teardown
« Reply #14 on: December 11, 2016, 08:45:04 pm »

Any help appreciated ..


send me a PM with info about your location... maybe I can help you, I`m livin in Oberösterreich, 49xx

Regards

Michael
 

Offline hanakp

  • Contributor
  • Posts: 20
  • Country: cz
I know this thread is quite old, but my colleague recently asked me to repair a N2772A with broken cable, too. I have access to a bit better test equipment and better contacts than most people, so here are my findings:

-The 5 Mohm resistance is NOT created by resistance wire, like giovannirat suggested. The center conductor is of the high-resistance type, but it has only about 200 ohm/meter.

-I made a few x-ray images of the probe tips and there is high voltage 5 Mohm resistor inside (made from several ordinary MELF resistors in series). So theoretically, the probe can be fixed by cutting away (shortening) the broken part of the cable.

-There are also some other components connected in parallel the resistor. My best guess is they're either compensation capacitors or spark gaps (or other safety-related components).

-Virtually all these cheap high-voltage probes are actually differential transimpedance amplifiers: the resistors in the tips convert the high voltage to (safe) small current, which is then converted back to voltage fed to the oscilloscope. But even low-voltage probes like Agilent N2792A use this design (I x-rayed it, too).

-Unfortunately, the breaking point of the center conductor cannot be located on x-ray, beause the coax shielding obscures the view.

-So I tried to locate the break point by the classical capacitance measurement method. The inner conductor has diameter 0.05 mm and the insulator around it about 1.85 mm. Using this Pasternack calculator yields about 45 pF/meter:

https://www.pasternack.com/t-calculator-coax-cutoff.aspx

-When I connected the cable to Agilent 4294A impedance analyzer, I measured about 47 pF on good one and 42 pF on the broken one. Unfortunately, this means it's broken somewhere near the tip, so shortening it will not fix it in my case.

-So I set out to see if I could replace it with cable from other probe types. Unfortunately, I haven't much luck yet - most high-voltage probes on market use tips with either 4 or 10 Mohm resistance.

-That's because in reality, these probes are made by OEMs and Agilent, Fluke, Yokogawa and others just put sticker with their logos on them. And it's impossible to buy spare parts for them, if your probe breaks during warranty, they simply replace it.

-One of the probe OEMs is Sapphire Instruments in Taiwan:

http://www.sapphire.com.tw/npproducts.htm

-Just compare these two probes for lulz:

https://tmi.yokogawa.com/solutions/products/oscilloscopes/voltage-probes/700924-differential-probe-100-mhz1400-v/

http://www.sapphire.com.tw/nsi9101ds.htm

-Or these two:

https://www.keysight.com/en/pd-1722999-pn-N2792A/200-mhz-101-differential-probe?cc=US&lc=eng
http://www.sapphire.com.tw/nsi200ds.htm

-Unfortunately, I've been unable to determine the real producer of the N2772A probe. It has "made in Holland" sticker, but googling for probe producers in Netherlands came up empty.

-So I'm currently stuck too, please let us know if you have  some relevant information.
 
The following users thanked this post: Mechatrommer, thm_w, p514

Offline hanakp

  • Contributor
  • Posts: 20
  • Country: cz
I have good news - I devised a cheap replacement for "my" broken N2772A probe. It will take me some time to finish it (and write something about it), so in the meantime, have a look at this teardown of the original probe.

At first, I assumed the probe was overmolded plastic, so I tried to dissolve it. But no solvents I had on hand (acetone, dichloromethane, tetrachloroethylene...) made even a dent in it. Then I tried to melt it with a hot air gun, but it was a slow and smelly work, so I abandoned that too. But I noticed the black plastic is not fused with internal white plastic near the tip, which led me to idea it may not be overmolded after all. So I grabbed a Dremel and cut the black cover at its seams. And indeed, the cover is not overmolded, but actually assembled from 3 separate parts, which are glued together (yellow substance in the first photo) near the tip. The metal tip and its white insulator can be easily pulled from the cylindrical metal body, because the internal PCB has a connector for it. This revealed that the cylinder is filled with grey potting compound. On the cable end there is a brass plug that is tightly pressed into the cylinder. I had to put the brass plug in vice and then wrench the cylinder out with pliers. That revealed the PCB and the potting compound over it. The compound is rather tough, but I was lucky (?) there were large air pockets in it, so it soon succumbed to a sharp scalpel. Even so, I later found out that solder joints on some resistors and capacitors cracked during the process. There are 6 mini-MELF resistors on the PCB, all connected in series: 5 pcs of 1M/1% and one 511R/1% at the tip. The latter resistor probably serves to dampen oscillations (signal reflections) at the "entry point" of the probe. There are also 5 capacitors in 1206 package in parallel with each 1M resistor. I measured them on Agilent 4294A with 16034G fixture and it seems they're 10p/5% NP0 types. The coax center conductor is not soldered to the board, but connected to it with a press-fit style rivet.

All this unfortunately means it would be very hard to fix the probe (shorten the cable near tip) without damaging at least some components. Nevertheless, I know there are some really handy people out here, so if anybody manages to do it, please let us know!
« Last Edit: June 06, 2019, 12:01:13 pm by hanakp »
 
The following users thanked this post: Mechatrommer, thm_w, p514

Offline hanakp

  • Contributor
  • Posts: 20
  • Country: cz
Re: Agilent N2772A (Fluke DP120) differential probe repair attempt + teardown
« Reply #17 on: December 16, 2019, 04:35:17 pm »
It took MUCH longer than I hoped, but here is a way how to replace the entire probe tip and cable. Of course, the easiest way to do it was to adapt some existing probe. So I scoured my workplace and evaulated 5 or 6 different probes, mostly oscilloscope ones. Eventually, I decided to use Hantek T3100, because

-it can be disassembled and reassembled easily,
-it's designed to withstand high voltages,
-it's dirt-cheap,
-yet has suprisingly good frequency characteristics.

This is also why I posted T3100 teardown elsewhere on this forum, I recommend checking it out before you attempt the replacement. What particularly surprised me is that its coaxial cable has resistive coating over the center insulator, along with the usual wire shielding. I've never seen anything like it, but I guess it serves to dampen signal reflections at high frequencies. T3100 also has some drawbacks, but more about that later.

In any case, you need to replace the original 99M resistor with 5M one, and it must withstand at least 600 VDC (half of the N2772A rating). At first, I tried free-hanging construction similar to what I originally found inside T3100, but I had to abandon it, because it required high-voltage miniature axial THT capacitors which are hard to get nowadays. So I decided to emulate the SMD design of the original N2772A tips. I'm attaching Eagle 7.x files, so you can easily make your own PCBs. I'm also attaching Gerber files for PCBway, but of course every manufacturer requires a bit different format. The PCB resembles the original one, but employs 0805 components and has larger pads for wires. I originally planned to use 1206 components, but the resulting PCB was too big to fit into the T3100 probe cavity. The compensation capacitors are a bit smaller than in the original tip, I settled for 8p2/5% after a few experiments. I also made a version with 1% capacitors, but there was no discernible benefit. Likewise I tried a few different values (from 220R to 1k) of the tip "anti-ringing" resistor, but again, I saw no improvement. In the end I arbitrarily chose 560R because it was between 511R (in N2772A) and 680R (in T3100). Total assembly length must be under 78 mm, otherwise it won't fit into the probe cavity. The length of the "tip wire" is 43 mm and the "clamp wire" 6 mm in the photos. I used 0.6 mm diameter wire, which is similar to the original T3100 design. Don't forget to put insulation sleeve over the PCB and "tip wire", so it wouldn't short-circuit if it touches the cavity wall. The rest should be obvious from the photos.

Important! Make absolutely sure you use 1M resistors and 8p2 capacitors rated at least to 150 V, because probe safety depends on them! Which also brings me to the main T3100 drawback - the ground shielding is easily accessible around its tip, which is against N2772A safety design. If you need to use the probe at high voltages, it's absolutely imperative you do following:

1. Permanently glue on the retractable cover with hook. This will cover the exposed ground sleeve near the tip. Be careful not to glue the internal retract mechanism. Alternatively, you could machine or 3D print some thick insulating sleeve, perhaps even with banana adapter, but I will leave that to you. In any case, heat-shrink tube will simply not do here, unless you wish to meet the Reaper prematurely (or burn your scope).

2. The grounding cable with alligator clip is not detachable. You must snip it away and then remove all traces of its grounding wire as deep into the probe's body as you can. I drilled it away with a new (sharp) 2 mm drill bit. The wire is not perpendicular to the probe's axis, so I had to carefully drill it away piece by piece. Then you need to fill the resulting cavity with some insulating material. I used Spinner Plast2000 sealing compound I had left over from some other project, but I guess pretty much anything will suffice, as long as it won't fall out over time. Be sure not to leave any air pockets!

3. T3100 comes only in gray color, which is a complication if you need to replace the positive (red) probe. The outer plastic is soft, so any paint probably won't last very long. If I ever need it, my plan is to buy red heat shrink tubing and put it over the probe.

If you've seen some probe that doesn't have these drawbacks, please let me know (we have several N2772As in my workplace and I suspect some of them will need tip replacement soon).
 
The most laborious part is to connect the new coaxial cable into N2772A, because it's thinner than the original one. It took me 3 tries to get everything right, so I've prepared step-by-step instructions:

1. Pull the T3100 BNC pod cable grommet on and then the pod cover. Cut the cable as close to the pod as possible.
2. Slide the cable through hole in the N2772A plastic cap.
3. Slide on the original N2772A cable grommet.
4. Slide on two or three 1 cm long pieces of heat-shrink tubing, 4 or 5 mm diameter. They will serve to increase total diameter of the cable in step 11.
5. Carefully remove about 40 mm of outer black insulation. Be sure to cleanly cut around its circumference, otherwise it will tear during step 10. Try not to cut into the shielding wires underneath.
6. Unbraid the shielding wires. I'm usually using a thick needle for this.
7. Shorten the shielding wires to about 15 mm.
8. Remove the resistive coating from the white center insulator. Leave about 10 mm near where the shielding ends.
9. Slide the original brass ferrule over the resistive coating - it's a tight fit, but it should be possible.
10. Use a hot air gun to soften the outer black insulation and then use small pliers to push the ferrule inside the cable. Don't set too high temperature, the insulation melts easily.
11. Move the shielding wires away and slide the heat-shink tube from step 4 so it would be aligned with the end of the shielding/ferrule. Then shrink it.
12. Bend the shielding wires backwards over the heat-shrink tube, like it was on the original cable.
13. Bend open the aluminium clip and slide it over the wires, then close it tightly with pliers. Try to pull on the cable, it shouldn't slip out from the clip. If it slips, go back to step 11 and apply another piece of tubing.
14. Very carefully strip about 7 mm of the white center insulator. It's rather weak, so you can cut it lightly around the circumference and then tear it away. It's absolutely critical to not even touch the center conductor with knife, otherwise it will almost immediately break.
15. The white insulator is too thin for connectors on N2772A PCB, the cable would fall out. Use about 5 mm long piece of 2 mm diameter heat-shrink tubing to thicken it.
16. Carefully bend the center conductor over the tubing and test if the connector can hold it. If it still has tendency to fall out, it will be probably better to use hot glue than to add another tubing.
« Last Edit: December 16, 2019, 04:37:01 pm by hanakp »
 
The following users thanked this post: giovannirat

Offline hanakp

  • Contributor
  • Posts: 20
  • Country: cz
Re: Agilent N2772A (Fluke DP120) differential probe repair attempt + teardown
« Reply #18 on: December 16, 2019, 04:38:04 pm »
And now for some results.

Of course, you will need to calibrate the N2772A for the new tip first. The process is explained in Fluke DP120 service manual which is linked in the original post. N2772A actually has more trimmers than DP120, but I ignored the "extra ones" and was still able to calibrate it. Just make sure you have both probe cables neatly parallel, with no coiling etc. And if someone has original Agilent service manual for N2772A, please let us know.

I didn't have Fluke 5500 calibrator recommended in the manual, so I used Agilent 81150A generator which has differential outputs. The maximum output voltage of one channel is 10 Vpp, but in differential connection this increases to 20 Vpp. Then I connected the fixed probe to a scope, along with a good one. On the screenshots, yellow trace is the fixed probe and green trace is the good one, the generator is set to 1 kHz square wave. As you can see, the fixed probe has slightly slower edge time, but I found no way how to fix it. It also means it has less ringing, which is good, I guess. There is also a slight delay, but that's because I replaced the output BNC cable for a longer one (the original connector was worn out).

So here you go, it's not perfect, but it works. What's even better, I'm fairly confident it will work for other differential probes with the resistor-in-tip construction, though you will have to find proper compensation capacitors etc. yourself. Please share your results if you do.
 

Offline hanakp

  • Contributor
  • Posts: 20
  • Country: cz
Re: Agilent N2772A (Fluke DP120) differential probe repair attempt + teardown
« Reply #19 on: December 17, 2019, 11:17:44 am »
Today I got curious what the "extra" capacitance trimmers on the PCB do, I marked them in red rectangles in the photo below. They're not pictured in DP120 manual, but it seems they can be used to tune the rise (and fall) time. I touched only the bottom trimmer (near negative tip input) and I was able to match edge time of the good probe, see the screenshots. I assume the top trimmer serves to tune the positive tip input, but it's not clear what setting is optimal or "best".

There is also an "extra" resistance trimmer right under the negative cable, but I don't know what it does. Hopefully, someone will provide N2772A service manual...
 
The following users thanked this post: thm_w


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf