Hi to you all.
I am having this issue which suppose to be easy because it is just resistive attenuator but trust me it did kept me awake at night and kept me busy.
I need to measure a high voltage ( round 1000V ac mainly transient and 600V DC) thus I need an attenuator factor of 25. I am using a T type attenuator with a series resistance of 9M ohms and a shunt resistance of 801K. The input or the output impedance is 10M. This circuit is a conditioning circuit for my measurement circuit which has an input resistance of 10M.
I am having two issue:
1- The attenuation for DC voltage differ from AC voltage ( the difference start even at frequency of 10Hz).
2- The attenuation bandwidth is low ( less than 10Khz at -3db).
I checked the internet for attenuation circuit and the popular one are T type and PI type but most of the explanation is done for low input impedance and I guess for also a low voltage.
Do these T type or PI type attenuator also work for high voltage with high impedance ( the high impedance feature is needed because the attenuator will be connected to a safety critical circuit)?
I need a bandwidth of 30Khz at -3dB which suppose to be easy to achieve with a passive attenuator.
Does anyone have a solution or this problem?
Thank you in advance.

 

'Scope probes use basically the same values as you are. The bandwidth is reduced by the stray capacitance on the output. They compensate for this by having a small capacitor across the 9MΩ resistor. This capacitor is a trimmer so the compensation can be adjusted but in your case it would probably be best to use a trimmer on the divider output to avoid needing a high voltage trimmer.
http://www.syscompdesign.com/assets/images/appnotes/probes.pdf

 

rabmerab, do I understand correctly that the source that you are measuring has 10 MOhm impedance?

 

'Scope probes use basically the same values as you are. The bandwidth is reduced by the stray capacitance on the output. They compensate for this by having a small capacitor across the 9MΩ resistor. This capacitor is a trimmer so the compensation can be adjusted but in your case it would probably be best to use a trimmer on the divider output to avoid needing a high voltage trimmer.
http://www.syscompdesign.com/assets/images/appnotes/probes.pdf
View attachment 126774

Yes, that was the first configuration I tried. The frequency response wasn't what I want. I think it was round 15Khz a -3dB.
Is T type or PI type attenuator not suitable for high voltages?

 

rabmerab, do I understand correctly that the source that you are measuring has 10 MOhm impedance?

The voltage sources is a 600V DC that we want to monitor. We want the attenuator input impedance to be as high as possible (10Mohms) so not to affect the measurement.

 

More oscilloscope probes: https://www.google.com/url?sa=t&rct...lyt627&usg=AFQjCNFVjc598oK4W5WW16-UugMwaec0Aw

 

Yes, that was the first configuration I tried. The frequency response wasn't what I want. I think it was round 15Khz a -3dB.
Is T type or PI type attenuator not suitable for high voltages?

I don't think T or PI would give you any benefit. They are generally used to match impedances in RF circuits.

Do you know the input capacitance of whatever is connected to the output of the divider including any cables?

 

Thank you for the reply.
I don't have that information with me now. I will check tomorrow and I will post it.

 

Do you know the input capacitance of whatever is connected to the output of the divider including any cables?

To determine the frequency response I am using the HP Agilent 33120A function generator ( generate sine wave with different frequecies) and scope Fluke 199C (http://www.farnell.com/datasheets/68138.pdf) which has an input impedance of 1MOhms and input capacitance of 15pf.
The final use is output of the attenuation circuit will be connected to a SCMVAS-M200 ( https://www.dataforth.com/catalog/pdf/scmvas.pdf) which is itself a voltage divider by 200 with an input impedance of 10MOhms. My attenuation circuit and the SCMVAS-M200 module are connected via two wires so it is hard to say what is the input capacitance. I open the SCMVAS-M200 module and it is just a restive voltage divider.

 

The ratio of the two capacitors should be the inverse of the ratio of the two resistors.

If you change what the divider is connected to (that is if the load resistance or capacitance changes) then one or other of the capacitors will have to changed to match.

Note that connecting a 'scope probe constitutes a change in the load resistance and capacitance so the response you get will not be the same as with probe disconnected.

 

The ratio of the two capacitors should be the inverse of the ratio of the two resistors.

If you change what the divider is connected to (that is if the load resistance or capacitance changes) then one or other of the capacitors will have to changed to match.

Note that connecting a 'scope probe constitutes a change in the load resistance and capacitance so the response you get will not be the same as with probe disconnected.

The SCMVAS-M200 high voltage attenuator in the attached picture doesn't use any capacitor ( I opened it to see; one through hole 10 MOhms resistor and the rest is SMD resistor) and according to the manufacture it is -+ 0.03%. I tried to reproduce the circuit with different attenuation factor for my application but no luck. I have to say the 10M resistor was expensive about £8. In my voltage divider I used SMD resistor with 1% tolerance.

For the attenuator with capacitor I used 4 x 2.2M to get near 9M and I used 1M resistor . Parallel to 1M, I used 41 pf. I used a variable (2pf to 8pf) capacitor parallel to 9M resistor. From the spreadsheet result it is clear that there is more than 3dB between attenuation at DC current and attenuation at 100 Hz. The resistor are thick film technology.
I used 1Khz 5V to adjust the capacitor until the output voltage follow the same waveform as the input.

 

Have a read here: http://www.repairfaq.org/sam/hvprobe.htm