high frequency voltage dividers and measurement / High Voltage / Forums

Forums

4hv.org :: Forums :: High Voltage

« Previous topic | Next topic »

high frequency voltage dividers and measurement

1 2 next

Move Thread

LAN_403

ragnar

Fri Apr 13 2007, 01:43PM

Registered Member #63 Joined: Thu Feb 09 2006, 06:18AM
Location:
Posts: 1425

Hi all,
in designing an off-line class-E coil using 1500V MOSFETs... I find myself suddenly confronted with the issue of measuring the drain-source voltage with my oscilloscope.

My scope may be rated to 400V peak at DC... but the input voltage is derated with frequency. At 10MHz, it should really only be see 10V on the input.

The only high frequency high voltage I'll be measuring is the MOSFET drain-source voltage. My best idea is to permanantly install a 10:1 divider across the drain and source, which can be used as a test point at any time, and since I have a nice 10:1 oscilloscope probe already, the ratio is now 100:1
(About right for 1500V divided down to 15V, with a 10V recommended max input on the oscilloscope at that frequency)

Would some experts here kindly offer advice on the type of divider I should use (I suspect capacitive?)
I don't care about measuring DC, but I'd like to be able to see the harmonics on the DS waveform up to, say, 50MHz.

Considering the value of the shunt capacitor that will be on DS anyway, I'm thinking any capacitive divider should load < 10pF on the circuit.

Any recommendations, especially on capacitor types?

Thanks,
Matt

Sulaiman

Fri Apr 13 2007, 03:06PM

Registered Member #162 Joined: Mon Feb 13 2006, 10:25AM
Location: United Kingdom
Posts: 3140

The 'best' would be a vacuum capacitor something like this ebay item 290101674138
with a LARGE division ratio (e.g. 1000:1)
this would allow low-loss such as NP0 to N750 ceramic or polypropylene at the 'bottom' which will swamp any 'scope/lead capacitance.

Or an NP0 ceramic could be used as the high voltage element, eg NP0 to N750 transmitter-type ceramic.
or a string of cheaper polypropylene capacitors, but beware of low allowable rms voltages at hf.

P.S. I thought the choices available on eBay here in UK were poor (compared to USA for example),
but I just tried to find suitable capacitors using ebay .au - wow - what poor choices you guys have ;)

Marko

Fri Apr 13 2007, 04:24PM

Registered Member #89 Joined: Thu Feb 09 2006, 02:40PM
Location: Zadar, Croatia
Posts: 3145

Hi matt

I don't get it, why capacitive divider?

I would just mount a simple 1:10 resistive divider on board.
720k+80k or etc.

Note that if resistors are too large, your reading will be scrambled because of capacitance of probe cable. (I think.) So it's better to use some fatter resistors and let them dissipate a watt or so.

Aand.. only 10V at 10Mhz??? I never tought that deratings should be that gross. Is that your new scope?

I pumped my labtech with even up to 50V+ at 2Mhz and it was happy. Inputs just say 600V max... Scope is ancient so I don't know much more about it!

Sulaiman

Fri Apr 13 2007, 04:58PM

Registered Member #162 Joined: Mon Feb 13 2006, 10:25AM
Location: United Kingdom
Posts: 3140

Because 10pF @ 10 MHz is <1.6 kOhms

10 pF is good for 'scope/lead capacitance, probably worse

Marko

Fri Apr 13 2007, 05:31PM

Registered Member #89 Joined: Thu Feb 09 2006, 02:40PM
Location: Zadar, Croatia
Posts: 3145

I tought resistive divider would be way more precise..

Most commercial probe dividers are resistive, AFAIK?

Capacitive divider would have to be of a large value compared to probe capacitance to be accurate, and with class E, I fear it might interfere too much with shunt capacitance for high frequencies he is aiming.

If there isn't some other reason.. in that case you help me

Sulaiman

Fri Apr 13 2007, 07:43PM

Registered Member #162 Joined: Mon Feb 13 2006, 10:25AM
Location: United Kingdom
Posts: 3140

Yes, a normal 'scope probe will have a resistive divider to enable d.c. display,
the divider/probe that I described will only show the ac component of the signal
the 0 volt axis will now be the AVERAGE of the input,
so for example a flyback output would have the wave-shape that you'd expect
but with the trace both above and below the 0volt position
the areas above and below the 0volt axis will be equal.

To save confusion you could have a resistive divider to keep the 0volt axis where you'd expect it
and a capacitive divider for high-frequency waveform accuracy.

Have a look at this thread for more info.

ragnar

Sat Apr 14 2007, 04:41AM

Registered Member #63 Joined: Thu Feb 09 2006, 06:18AM
Location:
Posts: 1425

Hi Firkragg,
I got the general impression that resistive dividers were a big no-no at high frequencies from the stray capacitance figures given here:
http://www.repairfaq.org/REPAIR/F_hvprobe.html
Read sections #8 and #9
(Otherwise I'd just make a divider with a handful of resistors -- and then stray inductance becomes a problem, too)

The probe/divider won't be "interfering" with the shunt capacitor as such, since it's going to be permanantly mounted on the board between drain/source, allowing me to just read a tap or test point.

I already have some 10kV 250pF ceramic caps, and I'm thinking about how one could be used. (In series with a 50V NP0? hmm)

WaveRider

Sat Apr 14 2007, 08:19AM

Registered Member #29 Joined: Fri Feb 03 2006, 09:00AM
Location: Hasselt, Belgium
Posts: 500

You can buy scope probes good to about 250MHz relatively reasonably (about EUR 50 ea.). Barring that, here's my rough wide-band RF measuring setup:

1. high value resistor (4.7 k or so is about the limit at 50 MHz, sized depending on voltage to be measured.. a capacitor can also be used, but you will need to calibrate wrt frequency...best if you have a digital scope)
2. length of 50 ohm coax with BNC connector on one end
3. Suitable BNC T
4. 2W BNC 50 ohm load

Here is a rough drawing of the setup for use with scopes with high impedance inputs.

Of course, its even better if your scope has a proper 50 ohm input setting (as many scopes made for RF work do) in addition to its standard 1 Mohm/10pF input...

Of course, the performance of this setup degrades as the frequency rises such that the parallel input capacitance loads the termination significantly and produces significant reflection. Up to 50 MHz, it should be good, tho... Test it using a signal generator first to calibrate... Of course, make sure that the specs of your scope, the 50 ohm termination are not exceeded. Also, use low-inductance resistor with suitable power rating...keep all leads as short as practically possible.

The highest voltage that I have measured with this is about 300V. Measuring high RF voltages is always a challenge. If waveforms are what you are after and not accurate voltage measurements, try replacing the resistor probe with a two-loop 'tickler' coil and bring it near the RF lead you want to measure (you will see the current waveform this way).

For the best measurements, you will need to incorporate test circuits on your board (directional couplers, possibly). Hope this helps

Cheers,
Bill

Steve Conner

Sun Apr 15 2007, 09:44AM

Registered Member #30 Joined: Fri Feb 03 2006, 10:52AM
Location: Glasgow, Scotland
Posts: 6706

I have used WaveRider's method on high-speed ECL logic with some success. I used a 1k resistor at the probe end, to get a 21:1 probe, and a 400MHz Philips scope with a built-in 50 ohm termination.

The main drawback, as I see it, is the stray circuit constants associated with the resistor as you make it bigger (electrically and physically) to increase the division ratio and deal with higher voltages.

WaveRider

Sun Apr 15 2007, 11:12AM

Registered Member #29 Joined: Fri Feb 03 2006, 09:00AM
Location: Hasselt, Belgium
Posts: 500

As Steve says, there are limitations with this method. As the value of your resistor increases, parasitic capacitance can spoil your measurement. This is why "high voltage" measurements are difficult using this technique. At this point, you need to design clever compensation networks along the lines of what has has been described (IIRC) by EasternVoltResearch and possibly others.

Also, if you are using a digital sampling scope, you can "compensate" in software (within limits) using calibration results from a sig gen with known frequency and output level.

However, by the time you've done all this, you could have gone down to your local electronic equipment retailer and bought a properly compensated 100:1 probe....

Let us know how it goes!!!

1 2 next

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