About Current Probes and Transformers

#140 In this video I look at my old current probe in a bit more detail to see just why I wasn't getting the intended current to voltage ratio. As it turns out, the reason is not that simple - its a combination of measurement error and overlooking some of the parasitic aspects of how the transformer works...
Late edit: at 19:52 on the right side its Cprim
More on the topic:
Crystal Oscillators Drive level: kzbin.info/www/bejne/eoKUeKacrJqhoZI
Transformer LTspice: kzbin.info/www/bejne/o6a4qJhujZpkpbs
Documentation:
www.arbenelux.com/wp-content/uploads/2020/02/Current-Probe-Calibration-Application-Note.pdf
www.pulseelectronics.com/wp-content/uploads/2020/12/Introduction-Transformer-Magnetics.pdf
Special Thanks to all my supporters on Patreon! Especially @afiskon, Ralf B., Paul Pr. Richard and Jonathan Alvarado!
If you liked this video be sure to check out my other videos and you can also subscribe to be up to date with all the new ones!
If you want to support the creation of more and better videos please consider checking out: www.patreon.com/feszelectronics

[@bgable7707]

WOW, that was excellent and a LOT to understand!

[@Samuel-km5yf]

Another fantastic, educational video from FesZ. Great stuff! Thanks!

[@iwbnwif]

Thanks for the clear explanation of transformer parasitics 🙂

[@anoimo9013]

very good explanation of concepts and practic issues

[@p_mouse8676]

Great video! Lots of these variables and non-linearities are also applicable to low frequency transformers btw. 👍🏻

[@FesZElectronics]

Of course! I was not trying to talk about high frequency transformers, but rather transformers in general. You get the same effects with a 5 turn RF trafo as you get with a 1000 turn audio trafo. To some extent things like parasitic series resistance and parallel capacitance are turns dependent, more turn, more parasitics.

[@mr1enrollment]

excellent again. It would be interesting in the end to also empirically determine the probe response by doing a sweep, observing the response at the frequency of interest, and then applying a correction factor for your first measurement. This could include multiple sweeps at various current levels as well as the frequency. Your analysis shows clearly how models need careful construction in order to represent the real world, and how greater understanding of all the parasitic and non linear responses is needed to refine a measurement. AND The consequence of a complex model will rapidly slow down a circuit model containing more and more representative component models.

[@qno-oj3py]

Thanks for your work. I like it.

[@windward2818]

PCB current measurement by inductive means requires access to the current in the trace, which is difficult because you cannot put a current probe around a PCB trace without using a wire to bring out the current, and as mentioned in the video, this length of wire will add inductance to the circuit. So how to measure PCB current? You have to design not just a circuit that works but also a means to test the circuit. In PCB design you must design for test. If you look at power distribution networks many times there are key points in the circuit where you would like to measure current, not necessarily for stability of a voltage regulator but for design considerations with specific components. Like capacitor RMS current and transistor peak currents. To make relatively low frequency measurements (BW 2.5MHz and below) you can use an accurate non-inductive sense resistor (e.g. 10 Milli-Ohm SMT 2512, +/-0.5% tolerance, not cheap) as part of the design to make the measurement. You can choose to leave the sense resistor in the design (it may need a little damping resistance anyway) which means the circuit will behave as tested or replace the sense resistor by a zero ohm jumper if cost is critical. You can always bring out a loop to do the testing if it can accommodate the small increase in inductance during the test. I think you will find the sense resistor method as having a very low noise floor as compared to the inductive current probe. To measure the current you have to make a floating differential amplifier which usually has an output to a scope voltage differential probe, which most test benches already have. On my designs I like to use an off-the-shelf differential amp and cable driver (current mode amplifier into 50 Oms as an option) with the end of the probe PCB very small and pointy so I can solder the leads directly at the sense resistor with minimal length (10mm or less). You can run the probe off of batteries to keep everything low noise and floating. For current probe gain =100 with 10 milli-ohm sense you get 1 volt per 1 amp, which is handy, or you can add more gain stages, like 10 or 2 usually increasing the noise and offset errors. My probe design gets updated from time-to-time because the semiconductor companies continually improve the performance of their instrumentation amplifiers, and I must say, even though expensive, the performance is getting really very impressive.

[@hedleyfurio]

👍👍

[@susceptibility_]

I think the second part of equation at 19:52 should have Cprim instead of Csec, because 1.) those expressions are not equal (trivial) and 2.) the Lprim/Lsec ratio multiplied by the Rload gives the load seen on the primary side, so then you have to multiply it with the Cprim to obtain the cutoff frequency. And as far as I can tell this formula is only valid if we consider capacitance on only one side.

[@FesZElectronics]

Hello Laszlo. You are right, I did add a card to mention this issue and I also added a "late edit" in the description. Regarding the second thing you mentioned, that is also correct, this formula only refers to the case with a single sided capacitance - it should contain both terms to account for both primary and secondary capacitance at the same time.

[@marcinwitkowski2981]

just use Log detector like AD8307 = no Cload, and U have DC V like 25mv/dB

[@colindejager1250]

Hi Fez - Am I mistaking you for another guy who worked with me in a certain Communications exchange building? Love your work anyway!

[@FesZElectronics]

I don't remember doing that... Probably it was someone else..