poor man isolated scope

[Electrons]

hi

This question has come up a lot here but since I work with low voltage and I donot have Earthing in my workshop at all I will ask:

Problem: I have to measure high side gate signals of mosfets in SMPS circuits DC/DC. Inputs/Output voltages doesnot exceed 70v DC . the switching frequency is under 50khz. however mosfet rise/fall times are around 50ns-100ns. Donot know if this is too fast ?

Test equipment i currently have: Two Rigol 1054z

Current Solutions: use 2 channels reference to GND and use math function  ( the waveforms are noisy and full of artifacts)

The perfect solution: would be to use Isolated Scope from TEK or that cleverscope CS448 the gate waveforms are mouth watering. however i cannot afford anything above 1000$

I have 2 solutions to consider :
1. Differential probe like testec and micsig . however i have read on this form about problems with common mode rejection at high frequency. but I am afraid that this will affects the gate drive signals and will not show the issues that you would like to see ?
2. Dedicate one of my scopes to make it a isolated scope by using an isolation transformer and use all bandwidth available to measure the high side and maybe see signals like CS448. since circuit voltages are non-lethal the circuit under test is battery operated. the only prblem i can think of is that i cannot measure the low side and high side together on the same scope. it will be basically a 1 channel scope.

[Fungus]

Current Solutions: use 2 channels reference to GND and use math function  ( the waveforms are noisy and full of artifacts)

Is there 70V to ground?

Are your inputs AC or DC coupled?

Screenshot?

[Electrons]

Current Solutions: use 2 channels reference to GND and use math function  ( the waveforms are noisy and full of artifacts)

Is there 70V to ground?

Are your inputs AC or DC coupled?

Screenshot?

1.let me try to clarify
Normal operation : 60v battery supplies the SMPS directly no no connection to mains or earth whatsoever.
When measuring the NON-Earthed scope is connected to the circuit by connecting the GND clips or (springs) to Battery Negative (Circuit GND).
2.Are your inputs AC or DC coupled? DC coupled .
3.I donot have screen shots right now but will try to get some shots later.

However I have read in another thread someone saying that 2-probe math function is worse that differential probe.
Source (Topic): How to probe high-side Mosfet gate
I qoute filssavi here "The old two probe subtraction trick can work for general electronics, however it is not applicable to high side gate drivers since it has basically no CMRR, this means you will have huge ringing during the commutation transients. that basically render looking at the gate useless (you can see if the transistor works by looking at the output voltage or current."

[Martin72]

Hi,
Like fungus said, a screenshot would be helpful - But also a snip of the schematic.
Other questions:

- You measure the output of the driver, going to the gates of the switching FETs, right ?
- Are the outputs isolated by e.g. a drivertransformer ?
- Level of the switching signal, ON = xxVpeak ?

You see, a part of the schematic and screenshot will be...

Also important: Short ways:

When measuring the NON-Earthed scope is connected to the circuit by connecting the GND clips or (springs) to Battery Negative (Circuit GND).

When you measure against GND as the common potential, there is a difference if you choose the nearest GND point or take someone somewhere which can be reached easily.
Plus the lenght of the GND leads from the probe..

Isolated Scope from TEK

A used one is not expensive, the last THS 710/720 were sold between 280...400€(ebay germany).

[David Hess]

mosfet rise/fall times are around 50ns-100ns. Donot know if this is too fast ?

That is not too fast.

Current Solutions: use 2 channels reference to GND and use math function  ( the waveforms are noisy and full of artifacts)

Verify the performance by connecting both probes to the same fast signal source, like the MOSFET source pin.  The subtracted result should now be flat so trim the probe compensation and gain for the best common mode rejection.  On an analog oscilloscope, the probe gain would be trimmed with the vertical variable control; DSOs do not typically handle this quite as well but it should be possible.