Author Topic: Shielded cable for sensing 900V, 15 kHz PWM signal over 10 meters  (Read 1948 times)

0 Members and 1 Guest are viewing this topic.

Offline matbobTopic starter

  • Contributor
  • Posts: 23
  • Country: ca
Hello All,
   
    In one of my projects, I want to sense a three phase switching voltage coming from a PWM inverter connected to a motor. The DC-link voltage is about 900V, so the switching happens between 0V and 900V at about 15 kHz. The voltage measurement has to be done at the motor terminal. The measuring instrument (maybe an oscilloscope or power analyzer) is about 10 meters far from the motor terminal. I would like to run a cable over this 10 meter carrying these switching signals. I need a bandwidth of about 20 MHz for preserving the sharp edges in the signal. The signal being large amplitude will not be affected by noise, but I want to avoid capacitive coupling from this cable to nearby equipment, so I need shielding for this cable. The shield will be connected to ground. At the end of this cable, I will connect a differential probe for connecting to an oscilloscope.

    I tired searching in websites of oscilloscope probe manufactures, but I could not find anything beyond 2-3 meters. Can anyone suggest a cable assembly for this purpose?

Thank you.
 

Offline jmelson

  • Super Contributor
  • ***
  • Posts: 2851
  • Country: us
Re: Shielded cable for sensing 900V, 15 kHz PWM signal over 10 meters
« Reply #1 on: January 29, 2019, 07:50:45 pm »
Make your own.  Get 50 Ohm coax cable, and build an attenuator that will be at the motor end.  You sure don't have to worry over capacitive loading when measuring this system!  So, a 100:1 probe would need a 4950 Ohm resistor, shunted by a capacitor that is 1/100th the cable capacitance.  You can look up the capacitance of various cables per foot (or m) online.  You must be VERY careful to make sure the 50 Ohm termination on the scope is always available, so I would recommend using an external 50 Ohm terminator at the scope end.  Actually, this will deliver 9 V to the scope end, so most internal terminators would not handle that, anyway.  The 4950 Ohm resistor would need to have a rating of 163 Watts!  OOPS, looks like a problem!!!  Maybe a 1000:1 probe makes more sense, that would give you 900 mV when the input is at 900 V.  The resistor would need to be 99950 Ohms, and power rating of 8 W.  That might be doable with a bank of 16 paralleled 1/2 W resistors.

Jon
 
The following users thanked this post: matbob

Offline djacobow

  • Super Contributor
  • ***
  • Posts: 1170
  • Country: us
  • takin' it apart since the 70's
Re: Shielded cable for sensing 900V, 15 kHz PWM signal over 10 meters
« Reply #2 on: January 29, 2019, 08:58:27 pm »
I'd like to understand this type of thing better, as I have some similar needs.

Why not 3M3 ohm and 50 ohm resistor voltage divider at the source connected to 50 ohm coax and then attached to scope input through feed-through terminator or a tee with a 50 ohm terminator on it?

The divider's output impedance is 3300000 || 50 ~= 50, power dissipated in the divider is < 250 mW at 900V. You do get a voltage ratio of 66000:1, but is that tragic for this app? You can use a lesser ratio at the cost of needing a resistor with some power handling capability.

Does this approach without capacitors seriously curtail the bandwidth? Naively I'm thinking that the scope input capacitance is going to be a large impedance relative to the source, and so it can be ignored for a 20 MHz signal. (Like, 20 pF * 50 ohm = 1ns)
« Last Edit: January 29, 2019, 09:04:53 pm by djacobow »
 
The following users thanked this post: matbob

Offline Amper

  • Frequent Contributor
  • **
  • Posts: 290
  • Country: 00
Re: Shielded cable for sensing 900V, 15 kHz PWM signal over 10 meters
« Reply #3 on: January 29, 2019, 09:01:23 pm »
Couldnt you remote operate the oscilloscope? Put it next to the motor, connect as usual and then run a gpib or ethernet cable to a laptop.
 
The following users thanked this post: matbob

Online jbb

  • Super Contributor
  • ***
  • Posts: 1265
  • Country: nz
Re: Shielded cable for sensing 900V, 15 kHz PWM signal over 10 meters
« Reply #4 on: January 29, 2019, 11:29:18 pm »
The ‘make your own’ probe is fine, but will be earth referenced.  If you need to measure phase to phase, you’ll need to do 2 channels & subtract in ‘scope. Also watch out for ground loops.

I recommend starting with a differential probe with coax output.
 
The following users thanked this post: matbob

Online David Hess

  • Super Contributor
  • ***
  • Posts: 17427
  • Country: us
  • DavidH
Re: Shielded cable for sensing 900V, 15 kHz PWM signal over 10 meters
« Reply #5 on: January 30, 2019, 02:33:01 am »
The ‘make your own’ probe is fine, but will be earth referenced.  If you need to measure phase to phase, you’ll need to do 2 channels & subtract in ‘scope. Also watch out for ground loops.

I recommend starting with a differential probe with coax output.

Further, a high voltage differential probe can be safely attached to the motor wiring at the motor and its output to the oscilloscope can be extended with standard 50 ohm coaxial cable for practically as long a distance as required.
 
The following users thanked this post: matbob

Offline matbobTopic starter

  • Contributor
  • Posts: 23
  • Country: ca
Re: Shielded cable for sensing 900V, 15 kHz PWM signal over 10 meters
« Reply #6 on: February 04, 2019, 04:52:25 pm »
Hello All,

Thank you all for your replies.

@jmelson: Making a probe myself is good a good idea. As a hobbyist, I like to do this, but my boss needs a ready made solution.

@djacobow: Thanks for the calculations. Attenuation of 66000:1 looks bad because the dynamic range of the resulting signal will be too low. With 900 V, the voltage in the coax would be 13.6 mV (please correct me if I am wrong) and I am concerned about external noise coupling.

@Amper: No, the oscilloscope has to be at a remote location.

@jbb @David Hess: Yes, you are correct. The coax shield has to be connected to ground and also one end of the divider resistors. As you have suggested, the differential probe + coax would be the best idea.

If I connect a 50 ohm coax cable to the differential probe, I will need a differential probe with 50 ohm output impedance. And at the oscilloscope end, I will have to terminate the coax with 50 ohms (if the scope does not have 50 ohm termination) for impedance matching. This will result in my output signal amplitude getting divided by 2. But that is fine.

Anyone knows a differential probe which can drive the 50 ohm load? I could find only differential probes which are to be connected to 1 Meg input impedance.

Thank you.

 

Offline max_torque

  • Super Contributor
  • ***
  • Posts: 1327
  • Country: gb
    • bitdynamics
Re: Shielded cable for sensing 900V, 15 kHz PWM signal over 10 meters
« Reply #7 on: February 04, 2019, 06:15:27 pm »
Doing this properly is, trust me, a whole world of pain!


Consider what happens to your cable should you accidentally short the far (instrumentation end)

Or what happens to the signal edge as it propagates down the length of wire

The consider what effect your spur has on the motor you are measuring


If you are new to the world of Precision Power Analysis on high power PWM motors, then you'll be happy to know you are in for a fun ride......


(BTW i do this for a living! and we use ultra low capacitance, ultra high CMRR probes right at the eMachine.  You don't need a 20 MHz bandwidth in most cases (unless you are looking to characterise the motors phase voltage overshoot) normally a 2Mhz sample rate is sufficient.  You need to capture enough edge information to be able to accurately calculate the RMS voltage (if required) and remember the square term effectively doubles the frequency content....)



So my question to you, is why do you need to make those measurements  ie, For what reason do you need to measure them?  (power analysis? voltage overshoot? ringing frequency? Harmonic analysis ?? etc etc )
 

Offline matbobTopic starter

  • Contributor
  • Posts: 23
  • Country: ca
Re: Shielded cable for sensing 900V, 15 kHz PWM signal over 10 meters
« Reply #8 on: February 04, 2019, 06:54:58 pm »
Hello max_torque,

Thanks for your reply. The measurements are for motor efficiency measurement, harmonic analysis, motor characterization, etc.

If I am using a differential probe at the motor end (giving, say +-10Vmax output) , I think I don't have to be much concerned about what happens at the instrumentation end. Even if the cable shorts, there would be reflection, over-voltage,... and may damage the differential probe in worst-case.

Thank you.
 

Offline PartialDischarge

  • Super Contributor
  • ***
  • Posts: 1625
  • Country: 00
Re: Shielded cable for sensing 900V, 15 kHz PWM signal over 10 meters
« Reply #9 on: February 04, 2019, 08:15:13 pm »
Hi max_torque, have you ever had to work for that same purpose on MV motors? I mean 3.3kV, 6.6kV, 11kV...
 

Online David Hess

  • Super Contributor
  • ***
  • Posts: 17427
  • Country: us
  • DavidH
Re: Shielded cable for sensing 900V, 15 kHz PWM signal over 10 meters
« Reply #10 on: February 05, 2019, 03:51:07 am »
@jbb @David Hess: Yes, you are correct. The coax shield has to be connected to ground and also one end of the divider resistors. As you have suggested, the differential probe + coax would be the best idea.

If I connect a 50 ohm coax cable to the differential probe, I will need a differential probe with 50 ohm output impedance. And at the oscilloscope end, I will have to terminate the coax with 50 ohms (if the scope does not have 50 ohm termination) for impedance matching. This will result in my output signal amplitude getting divided by 2. But that is fine.

Anyone knows a differential probe which can drive the 50 ohm load? I could find only differential probes which are to be connected to 1 Meg input impedance.

Most or all of the probes from Pintek are specified to have a 50 ohm output if you dig into the manuals.  Their manuals indicate that using a double termination decreases sensitivity by 2 as expected.

Pintek seems to OEM most of the differential probes for others.
 
The following users thanked this post: matbob


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf