HV 50MHz Differential probe [DIY PROYECT]

[hazard]

Hi there
For quite some time, I managed to measure differential voltages without relying on a differential probe,  but now I needed one. Since they are extremely expensive they are in Argentina, I decided to design my own, based on the MICSIG DP10007 (although with a bit of my own personal touch). The goal was for it to have a bandwidth of 50MHz, tolerate input voltages up to 700Vpk, have x50 and x100 scales, and be powered by a 9V battery. I was able to achieve almost everything I set out to do, except for the bandwidth; it barely reached 15MHz at the -3dB point. I've already looked for different solutions to improve the bandwidth, without significant changes. I'd like to read any recommendations. Here are some images.

In the bode plote can be seen both the x100 and x50 frequency responses

[moffy]

Just a quick observation, you probably don't want that multiturn Bournes pot, they have parasitics, better off with one of those cheapish mini trimmers like this:
https://www.digikey.com/en/products/detail/bourns-inc/TC33X-2-104E/612859
much better HF performance, but you do loose precision, in such a case go for 0.1% resistors and a smaller trim value.
Replace the 2k2 resistors for 1k for the LM7171, every little bit helps, and you might try increasing the 10p cap values to say 12p or 15p and see if that gives better BW.
Overall, I really like the design.

P.S. You might consider two levels of capacitance trim, a coarse/largish and fine/smallish.

[gbaddeley]

C1 etc should be in parallel with each of their partner resistors?

[hazard]

Hello everyone, thanks for replying.
After looking around the input and output stages, I realise that the poor bandwidth response seems to be around the output stage...
I tried to down the R18, R19, R20, R21, R22 and R23 to 510 but with similar results... in your opinion, and after taking a look inside the DP10007 teardown and its scheme, how much better would it be to use the THS3091 instead of LM7171 as differential output stage?
On the other hand, I used an RF-grade microrelay to switch the gain resistors in the output stage, which has a shielding can. In your opinion, could its capacitance interact to lower the bandwidth?

[moffy]

Hello everyone, thanks for replying.
After looking around the input and output stages, I realise that the poor bandwidth response seems to be around the output stage...
I tried to down the R18, R19, R20, R21, R22 and R23 to 510 but with similar results... in your opinion, and after taking a look inside the DP10007 teardown and its scheme, how much better would it be to use the THS3091 instead of LM7171 as differential output stage?
On the other hand, I used an RF-grade microrelay to switch the gain resistors in the output stage, which has a shielding can. In your opinion, could its capacitance interact to lower the bandwidth?

The THS3091 is a current mode opamp, the +/- inputs are different in more than just phase, so I am not sure that you would get the desired CMRR with the current resistor configuration.

[hazard]

Hello everyone, thanks for replying.
After looking around the input and output stages, I realise that the poor bandwidth response seems to be around the output stage...
I tried to down the R18, R19, R20, R21, R22 and R23 to 510 but with similar results... in your opinion, and after taking a look inside the DP10007 teardown and its scheme, how much better would it be to use the THS3091 instead of LM7171 as differential output stage?
On the other hand, I used an RF-grade microrelay to switch the gain resistors in the output stage, which has a shielding can. In your opinion, could its capacitance interact to lower the bandwidth?

The THS3091 is a current mode opamp, the +/- inputs are different in more than just phase, so I am not sure that you would get the desired CMRR with the current resistor configuration.

I agree, but after checking Dave's hand-drawn schematic of the DP10007 and looking closely at the PCB photos, it looks like micsig used the THS3091 as a VFA.

[moffy]

Hello everyone, thanks for replying.
After looking around the input and output stages, I realise that the poor bandwidth response seems to be around the output stage...
I tried to down the R18, R19, R20, R21, R22 and R23 to 510 but with similar results... in your opinion, and after taking a look inside the DP10007 teardown and its scheme, how much better would it be to use the THS3091 instead of LM7171 as differential output stage?
On the other hand, I used an RF-grade microrelay to switch the gain resistors in the output stage, which has a shielding can. In your opinion, could its capacitance interact to lower the bandwidth?

The THS3091 is a current mode opamp, the +/- inputs are different in more than just phase, so I am not sure that you would get the desired CMRR with the current resistor configuration.

I agree, but after checking Dave's hand-drawn schematic of the DP10007 and looking closely at the PCB photos, it looks like micsig used the THS3091 as a VFA.

That is interesting, the current mode opamps maintain BW with modest gain changes, but they also tend to consume a bit more current. You might want to clear any ground plane directly beneath the chip and its feedback components, they can be quite sensitive to small capacitances in the wrong place. Would love to know how it goes if you try it.