Hay,
I know this might be a self-explanatory question, but none the less I would like to know if you have experience with using different current sense amplifiers with different bandwidth.
I see there are some new current sensors, with 1Mhz bandwidth, already out from TI. eg. INA241A
What´s your take on current sense bandwidth vs. MCU performance and overall system performance?
Hey @Juan-Antonio_Soren_E !
Been missing you here man! Where you been?
I’d say the frequency you’re measuring is the frequency of the commutation signals, the (ideally) sinusoidal waveform of the currents. This frequency will be proportional to speed of rotation and motor PP, so something like 1000 RPM is about 185 electrical rotations per second on a 11PP motor.
So this isn’t the kind of frequency that will come anywhere near the bandwidth limits of the sensors, if I understand the situation correctly.
185 is so much lower than 1MHz its basically like DC, and the gain attenuation should be barely noticeable.
Hi @runger
Ya, I was kinda stuck in some old thought patterns, so had to go through this process of letting go. Like they say, In order to go somewhere, you gotta leave something behind.
Now I am back to the original idea, when I first started exploring these electromagnets and their influence on permanent magnets for moving stuff, using the Teensy 4.0 as the MCU part, doing the design as modular as possible. Not so much concerned about price, but have my focus on how to aid my bike as efficient as possible.
Has started working on 100v mosfets (not stuck @24v no more), giving room for some spikes using like 36 or 48 volt power source (battery).
For future improvement, I will try to implement a movement sensor into the controller itself. Since the T4 is a real improvement from 120Mhz MCU to 600Mhz, there should be some overhead for these space readings, acceleration and so forth. Naturally a wireless module add on for iOS integration.
Regarding the Bandwidth of the current sense amplifier I believe the speed of the ADC (Analog to digital converter) is related to how fast we are able to measure current flow or change in current flow. Having higher bandwidth is kinda like having higher refresh rate on a monitor, right ?
Now, regarding the switching freq. When switching the FET´s, surely it happens faster then the electrical revolutions and in order to implement a smooth switching paradigm, we must have a way to perceive the changes in current flow we make by switching. Since these changes in current flow happens so fast, I will say, higher bandwidth is better, but is limited to the speed of the ADC.
i have used Bumblers’ V0 adc library (search in technical for “adc”) to get 12 bit conversions in about 0.8 us
going 10 or 8 bits speeds it up some but not a lot.
since that is your low end i think you need an external adc.
to get that 0.8us you have to use hardware averaging of 0 and that makes for some fairly noisy data.
data starts to look clean at average of 4 but then your sample time goes to about 1.5 us
I found this article on ADC aliasing from TI. Maybe 1Mhz bandwidth is over the top, even for the teensy 4. It depends on the acquisition not so much the conversion rate. I believe
