Voltage divider for up to 1500Vdc 5kHz BW / High Voltage / Forums

Forums

4hv.org :: Forums :: High Voltage

« Previous topic | Next topic »

Voltage divider for up to 1500Vdc 5kHz BW

1 2 next

Move Thread

LAN_403

rulmismo

Wed May 20 2015, 01:52PM

Registered Member #4187 Joined: Fri Nov 04 2011, 08:08PM
Location: Spain
Posts: 43

Hi everybody,

I want to make a voltage divider for sensing an input voltage from 500 to 1500Vdc with 5kHzBW, to an output of about 10V peak.

I know that a lot of effort was done by some forum members to design really HV and HBW dividers, but my goal is quite less ambitious.

Tricking a little bit with RS and Farnell readily available componentes I got to these two options:

1) option 1,
coarse film 7.5kV resistors from TE (RS 296-0522)
R1= 6 x 10M, 2W (1,5W dissiped power, too tight?)
R2= 400k, 2W

2) option 2, wirewound with objetive using less components for R1, but with higher L
I couldnÂ´t found any sensible resistors suiting these
R1= 3 x 5M, >13W
R2= 100k

Any comment wil be welcomed.

Regards

Dr. Slack

Wed May 20 2015, 02:50PM

Registered Member #72 Joined: Thu Feb 09 2006, 08:29AM
Location: UK St. Albans
Posts: 1659

A few observations

On your option 1 of 60M input resistance, the total dissipation in that at 1500v input will be 37mW, or 6mW per resistor. There appears to be a mismatch between your specs and your power calculations.

Wouldn't a single 7.5kV 10M resistor be OK to 1500V, with 225mW dissipation?

With 400k shunt resistor, and assuming a scope+cable input capacitance of 100pF (or more, depending on cable length), your 3dB point is only around 4kHz, so you could use a sniff of adjustable C across the 60M input resistor to prop up the frequency response up, even at 5kHz.

Patrick

Thu May 21 2015, 09:46PM

Registered Member #2431 Joined: Tue Oct 13 2009, 09:47PM
Location: Chico, CA. USA
Posts: 5639

what power is available on the main 1500 volt circuit? is it 100uA or 100 amps or what...?

rulmismo

Fri May 22 2015, 03:57PM

Registered Member #4187 Joined: Fri Nov 04 2011, 08:08PM
Location: Spain
Posts: 43

I will review my excel sheet cause probably messed up something with the powers...

I was thinking using a high impedance NI CompactDAQ with 4 differential analog input card from -10/+10V and connect with unipolar cables. I looked for the specs. but didnÂ´t find the inpunt impedance, it is some kind of instrumentation amplifier. It has a 100kOhm impedance from negative to input modulle local ground. It should not affectt if I donÂ´tuse the remaining 3 channels.
How does would match with the compensating cap Dr Slack? I do not see how to predict the BW with such input.

Input side is from a railway catenary power source, so up to several thouusands of amps Patrick. Power is "protected" by a CB at about 1000A.

I have seen some practical designs used for 1500Vdc that applied the following safety measures that I plan to apply also:
- serial quick acting fuse of some mA
- neon bulb or inverse diodes in parallel to the low side resistor. I thik the diodes are
planned to shortcircuit and blow the fuse up and the neon bulb just to limit the voltage to the firing voltage. What would be best option as protective device?
- dual redundant earth connection to return path.

Regards

Dr. Slack

Fri May 22 2015, 06:44PM

Registered Member #72 Joined: Thu Feb 09 2006, 08:29AM
Location: UK St. Albans
Posts: 1659

Rather than draw any diagrams myself, here is one source from picotech

There isn't a bandwidth to predict with a correctly compensated divider, it's flat from DC to light, well, from DC to well above your 5kHz requirement until other non-ideal things bite.

It's not the one you are using as it's +/-5v input, but the NI 9234 four channel diff input module is 305kohm differential input impedance. I'm not going to flog through all the NICompactDAQs seeing which one is 4 channel 10v, it's usually polite if you have a piece of kit in mind to give the part number, rather than set a puzzle for responders.

Patrick

Sat May 23 2015, 04:36AM

Registered Member #2431 Joined: Tue Oct 13 2009, 09:47PM
Location: Chico, CA. USA
Posts: 5639

yes dr. slacks link is as good as it gets.
you can look at some of my previous work, but it may be more than you need. though informative.

Ive acquired a HP 34119A 5kV probe, so im really happy ! but i cant find a pdf for it yet.

rulmismo

Sat May 23 2015, 07:59PM

Registered Member #4187 Joined: Fri Nov 04 2011, 08:08PM
Location: Spain
Posts: 43

Hi, the input module is NI9215 with 4 BNC input,, 420kHz BW. I don't know how to interpretate the input impedance figures for BNC module, the datasheet includes the impedance between two AI- of different channels (200k) , but that"s of no use here. I would need the impedance of the AI+, would"nt I?
So I don't find the equivalent to the 1Gohm scope input.

http://www.ni.com/pdf/manuals/373779f.pdf

I own a picoscope btw, quite happy with it.

Patrick

Mon May 25 2015, 02:51AM

Registered Member #2431 Joined: Tue Oct 13 2009, 09:47PM
Location: Chico, CA. USA
Posts: 5639

are you wanting a 1Gohm a probe? i dont fully understand.

those are always hard. they are so small in current that leakage, stray fields, and radio waves all start creeping in.

rulmismo

Tue May 26 2015, 12:22PM

Registered Member #4187 Joined: Fri Nov 04 2011, 08:08PM
Location: Spain
Posts: 43

Sorry, I meant that I didnÂ´t fully understand how to know the input impedance of the NI9215, in the same sense of the 1Mohm input of an scope.

I corrected the wattage figures and updated the resistor values.

I think that from datasheet 1Gohm is the input impedance figure (see attached schematic), I added an RC network to simulate the 450kHz BW input, the resistive divider and some stray impedances for lead cables.

The simulation gives a flat response up to 10kHz more or less (seem clipboard figure attached).

Any additional suggestion?
Regards

Patrick

Tue May 26 2015, 10:35PM

Registered Member #2431 Joined: Tue Oct 13 2009, 09:47PM
Location: Chico, CA. USA
Posts: 5639

i tried to do this with the same program. but the assumptions you make about stuff you cant really measure, is the real problem. It looks like youve got a boat load of them too.

look at some of my work. you might even be better off with a purely resistive probe just layed out well.

1 2 next

Moderator(s): Chris Russell, Noelle, Alex, Tesladownunder, Dave Marshall, Dave Billington, Bjørn, Steve Conner, Wolfram, Kizmo, Mads Barnkob