High voltage Attenuator with 60 KHz bandwidth

[Amr Wael]

Hello , I would like to design a high voltage attenuator that attenuates a signal from 1000Vpeak to full scale of 250mVpeak in order to be suitable for an Isolation amplifier IC
The input capacitance of the IC is around 4 pF therefore , if I used only a resistive divider (Clearly with a high resistance value for R1), it will form a low pass filter with the 4 pF capacitor and if I add a capacitive divider with the resistive divider , the voltage becomes dependant on the frequency and introduces an error of 9.8% at around 50 KHz.
is there an easy solution to attenuate 1000V with such attenuation ratio while maintaining the -3dB bandwidth to be DC to 60 KHz ?

Thank you very much in advance.
BR
Amr Wael

 

[KlausST]

Hi,

60kHz isn´t really high frequency. Thus I´d also go with a resistive divider.

For sure you need to use suitable resistors (type) and values.
And you may compensate for the input capacitance.

Sadly we don´t see what you did, so we can not recommend how to improve it.

****
I did something similar, but used an inverting Opamp circuit. So the high ohmic resistor goes the the inverting OPAMP input and the node is called "virtual GND". Due to "no voltage change" on this node ... stray capacitances (to GND) have no influence on frequency response. (Problems come with much higher frequencies).

With the inverting OPAMP you can do much better (simpler, not affecting the measurement value....) input protection with two diodes.

Klaus

 

[danadakk]

Not sure if this is workable in terms of tolerances, but use 400 V Zeners(2) and a
90V, all in series, driving a load :

Example done for 90Vz, could not find a spice model for 400V part -

Of course Zener C is a strong f(V).....but in a high Vz would that matter, investigation necessary.


Regards, Dana.

PS : I like the idea of sticking with Rs for precision reasons.

 

[FvM]

You forgot to tell the intended divider input resistance. Your calculated 10 % error at 50 kHz seems to refer to an unsuitable high resistance in the Gohm range. I would consider a few Mohms as useful value.

I also don't understand the calculation. -3 dB bandwidth refers to about 30% voltage error, how are you complaining about 10 % error.

According to my experience, it's well possible to get 100 to 200 kHz (- 3dB) bandwidth with uncompensated voltage dividers at several Mohm input resistance. For higher bandwidth, compensated dividers are required, possibly with adjustment.

 

[danadakk]

10:1 scope probe, 10 M and 1 M ohm, 150 Mhz bandwidth.

https://www.grainger.com/product/B-K-PRECISION-Oscilloscope-Probe-150-MHz-20FP82

100:1 100 Mhz -

Oscilloscope Probe 100Mhz High Voltage 2000V 100:1 for HP Tektronix, etc. (Black) | AST Labs

Specifications Bandwidth: 100MHz Rise time: 3.5ns Attenuation Ratio: 100:1 Input Resistance: 100MΩ±2% Input Capacitance: 6 pF Maximum Input: 2000V Working Voltage (Vp-p) Compensation Range: 10pF ~ 35pF Operating Environment: 0C ~ +50C / 0 ~ 80% Relative Humidity Pack Contents Retractable Hook...

ast-labs.com

Regards, Dana.