EEVblog 1413 - Micsig CP2100B Current Probe TEARDOWN and TEST

[EEVblog]

Teardown and measurements with the Micsig CP2100B 2.5MHz oscilloscope current probe.
Also a demonstration of mains switch mode power supply current and noise measurement, and also some performance tests.

Use the coupon code FLUX for 15%
And paying with crypto gets you a further discount on top of that!
https://www.eevblog.com/product/cp2100b/
Or get the cheaper lower bandwidth 800KHz CP2100A model:
https://s.click.aliexpress.com/e/_A1gq4i
https://s.click.aliexpress.com/e/_AKl2Y2

00:00 - Intro
05:39 - Teardown of base unit
14:25 - Clamp unit teardown
23:40 - Mains switchmode current measurement
34:00 - Measuring the Earth's magnetic field!
34:42 - Noise measurement
35:26 - Low current measurement

[Kleinstein]

Whatching the scope in the last minute of so shows a funny effect: the amplitude is randomly changing quite a lot.  I don't think this is a fault of the proble. It is more like a fault of the scope triggering right (a good scope should warn about that - a fluctuating frequency is kind of a warning though hidden). Averaging over shifted traces reduces the signal. It is a kind of trap for the young players when using the averaging mode.

The shields are nice for a tear down, but maybe not too good a contact for a really sensitive signal.

I would be a bit worried about the magnetic core not being laminated very fine. So higher frequencies may not pass through the core as much. They somehow seem to have a good compensation for that, even without the trimmer caps.
Chances are a rectanglular waveform would not look as nice.

For a little more signal one could wrap some 2 or 5 turns through the sensor if needed.

It looks like they just switch the gain of the instrumentation amplifier - so maybe a trimmer for each range.

Averaging 2 back to back sensors could reduce the offset drift - at least the part that correlates between the 2.

[Rdx]

Wondering about component U4 on the head unit.
It probably is an alternative footprint for U5 I would assume?

What is a usual workflow for such  stuff? Just place two components and wire them in parallel (like apparently they did here) and then manually delete one version each out of the BOM?
Well I guess it depends on your CAD package whether it does have a workflow for these things.

[thm_w]

I would be a bit worried about the magnetic core not being laminated very fine. So higher frequencies may not pass through the core as much. They somehow seem to have a good compensation for that, even without the trimmer caps.
Chances are a rectanglular waveform would not look as nice.

For a little more signal one could wrap some 2 or 5 turns through the sensor if needed.

It looks like they just switch the gain of the instrumentation amplifier - so maybe a trimmer for each range.

Averaging 2 back to back sensors could reduce the offset drift - at least the part that correlates between the 2.

Youtube comment from Yuxian:

Well this is widely used in current sensing for hall sensor in parallel to cancel temperature coefficient. And I work for the company in the video (multidimensional). Max bandwidth we have now is 1-2mhz. And this not yet publicly available.

I’ll double check that. It might. But looking at the design. The ferrite core should not have a bandwidth of 1mhz. The magnetic permeability drops dramatically at high frequency. For those material. The best I known of caps at about 2mhz.

Wondering about component U4 on the head unit.
It probably is an alternative footprint for U5 I would assume?

What is a usual workflow for such  stuff? Just place two components and wire them in parallel (like apparently they did here) and then manually delete one version each out of the BOM?
Well I guess it depends on your CAD package whether it does have a workflow for these things.

Yes, place two.
Another option is place one with a dual footprint in PCBLIB, but they clearly did not do that as you noticed the U4/U5.
In the schematic you can select it as "no BOM" for altium, or manually delete it as you say.

[tchicago]

Curious why these HV probes are not made galvanically isolated. It would feel much safer knowing that there is no direct connection to the HV circuit whatsoever. And that the info about the signal passes via the highly rated isolation optocouples or transformers. Also should be easier to fight the common mode rejection issues.

[Kleinstein]

The scope is usually grounded. For safty reasones the high voltage probe also needs to grounded one side. Some of the high voltage probes come with an extra ground lead, just in case the BNC connector at the scope is not enough.

[2N3055]

Curious why these HV probes are not made galvanically isolated. It would feel much safer knowing that there is no direct connection to the HV circuit whatsoever. And that the info about the signal passes via the highly rated isolation optocouples or transformers. Also should be easier to fight the common mode rejection issues.

Errr... It doesn't work that way.
It is current that kills you and equipment.
And current flows into ground, all microamps of it...
Making DC-100MHz galvanically isolated probes is harder than you think.

Tektronix has some fibre optic isolated probing solutions... How much your house cost?

[thm_w]

Curious why these HV probes are not made galvanically isolated. It would feel much safer knowing that there is no direct connection to the HV circuit whatsoever. And that the info about the signal passes via the highly rated isolation optocouples or transformers. Also should be easier to fight the common mode rejection issues.

As mentioned, its hard and expensive.
Look up optocouplers and what kind of bandwidth you can get out of those. Optocouplers cannot easily transmit analog signal linearly anyway. Special linearized optos exist eg HCNR201 is only 1MHz, for $6.

Dedicated analog optos exist (isolation amplifiers), only 100-200kHz bandwidth, price is over $5.
https://www.digikey.ca/en/products/detail/broadcom-limited/HCPL-7510-000E/825318

Its probably easier to digitize the signal and send it over isolated USB or something instead.

[trophosphere]

Curious why these HV probes are not made galvanically isolated. It would feel much safer knowing that there is no direct connection to the HV circuit whatsoever. And that the info about the signal passes via the highly rated isolation optocouples or transformers. Also should be easier to fight the common mode rejection issues.

As mentioned, its hard and expensive.
Look up optocouplers and what kind of bandwidth you can get out of those. Optocouplers cannot easily transmit analog signal linearly anyway. Special linearized optos exist eg HCNR201 is only 1MHz, for $6.

Dedicated analog optos exist (isolation amplifiers), only 100-200kHz bandwidth, price is over $5.
https://www.digikey.ca/en/products/detail/broadcom-limited/HCPL-7510-000E/825318

Its probably easier to digitize the signal and send it over isolated USB or something instead.

As an example:

[Wolfgang]

Hi,

I bought one and I wanted to see how reliable the specs are, The surprise was that even short runs of open wiring will result in major deviations from the bandwidth and rise/falltime specs.

Details here:
https://electronicprojectsforfun.wordpress.com/testing-a-micsig-cp2100b-current-probe/

[thm_w]

Hi,

I bought one and I wanted to see how reliable the specs are, The surprise was that even short runs of open wiring will result in major deviations from the bandwidth and rise/falltime specs.

corrected link: https://electronicprojectsforfun.wordpress.com/testing-a-micsig-cp2100b-current-probe/