Blog 2: Switching MOSFET's at +100kHz frequencies in high voltage environments - it's a precursor to getting Flyback Transformers to work properly.

RE: Blog 2: Switching MOSFET's at +100kHz frequencies in high voltage environments - it's a precursor to getting Flyback Transformers to work properly.

shabaz — Mon, 22 Jan 2024 19:54:58 GMT

Hi Colin,

As another project if you fancy it one day: it's possible to make the Pi Pico into a convenient test tool, using [mention:acaf6a9338de4eef8f6717d5561ed01d:e9ed411860ed4f2ba0265705b8793d05] Pico SCPI labTool (PST). I didn't realize but now it's got a PST documentation wiki. It has PWM capability, although I believe we hard-coded the PWM frequency for now.

The PWM can serve at least two purposes on PST; one is for an isolated DAC board (which will take that PWM and convert it to an analog output, hopefully on a couple of channels), but the DAC board still needs to be worked on (a schematic for a test board is done), and the other is to use it simply as a frequency or PWM output for general circuit stimulus!

I wanted to add a display to it, which would be used for monitoring purposes (not for controlling, because that's too fiddly, it is better to control with SCPI, and see the current live measurements on the display), but then later realized that with PWM, there could be a need for directly controlling from buttons, since there could be occasions where it is desired to detatch from the PC, especially when working with high-voltage projects!

Unrelated, I was working on a pico-sized GUI and button library in C++, to make pico-sized appliances (they don't need to be pico-sized! but I liked that Waveshare display/buttons board, it was a quick way of knocking up microcontroller based projects that need a bit of info display/control) , and at the time I tested it with PWM (using the hardware/pwm capability, not PIO). Very basic, the up/down controls on the mini joystick control the frequency, and left/right control the duty cycle, no acceleration etc for now.

I've not done much with it recently, the code is very ropey, but if anyone's interested and wishes to extend it, I'd be happy to stick the code somewhere. It's nothing special, but those rectangles on the display are all individual C++ text window objects! Each box is a mini text window, which automatically scrolls text if it exceeds the height of each box (e.g. you could have a multi-line box).

You can see small bugs, like the white box isn't perfectly rectangular, there is an artifact just below where it says 56%.

To keep the library simple, it only supports 3 or 4 font sizes.

RE: Blog 2: Switching MOSFET's at +100kHz frequencies in high voltage environments - it's a precursor to getting Flyback Transformers to work properly.

jc2048 — Mon, 22 Jan 2024 11:11:02 GMT

Interesting blog. Maybe with all this good blog activity, Randel might let the challenge run on a bit longer.

"Does this match the specification? I’ll leave it up to you to decide."

Whilst the datasheet specification looks very impressive, switching characteristics are a secondary attribute: they depend very much on the test used to measure them. That's because the thing that slows is capacitance, and you partially get round that by throwing more current at the capacitance so that the voltage changes quicker.

The manufacturer wants the part to look good in a comparison, so the test is designed to show the part at it's best within the limits the part is rated for. In the case of the NTD5867NL, they have the device switching 20A to 48V in the output circuit, and they're initially throwing several amps at the gate (4A in this case, if the gate is metal rather than heavily doped silicon). In your case, the output is very lightly loaded and the gate current is probably initially something like 100mA - the output pin of the Pico is driven high by an internal MOSFET with an on resistance of maybe something like 20-25 ohms. So you won't see the figures they are showing you on the datasheet.

One thing that does help you is that the microcontroller pin driver is fast, so you aren't being slowed by the performance of the driver in the way Javagoza was with his discrete bipolar circuit. The limitation, though, is that you can only get the gate to 3.3V, so fine for switching lowish currents with a logic-level MOSFET, but may struggle switching higher currents.

Implementing a digital control loop is quite difficult (but rewarding to look at because there are things that can be done digitally that would be impractical with an analogue loop). You might find it more straightforward to implement an analogue control system first (in a chip so that the difficult part has been done for you), get the converter working, and then move on to experimenting with digital control algorithms knowing you weren't fighting two things at once. Having the analogue system would also let you see how the controller deals with different scenarios, like the way the controller has to back off after quickly charging the output capacitor, if a light load isn't then taking the energy being provided, and would give you a benchmark for comparison with your own algorithm.

RE: Blog 2: Switching MOSFET's at +100kHz frequencies in high voltage environments - it's a precursor to getting Flyback Transformers to work properly.

DAB — Sun, 21 Jan 2024 20:27:42 GMT

Very good update.

RE: Blog 2: Switching MOSFET's at +100kHz frequencies in high voltage environments - it's a precursor to getting Flyback Transformers to work properly.

dougw — Sun, 21 Jan 2024 00:35:15 GMT

Nicely done with the Pico-Arduino PWM