Hi at switching frequencies give less heat

[Simon]

I did some experimenting yesterday with an Infinion half bridge chip and a couple fans. I started driving at 100 Hz and increased in 100 Hz steps up to 1 kHz as expected the temperature of the chip got hotter as I increased frequency. I went right through to 3 kHz but found that as I was leaving roughly 2.7 kHz behind and going into 3 kHz temperature on the chip actually started to drop back to where it was when I was at 100 Hz. I'm not entirely sure what to make of this. Presumably this has something to do with the switching frequency and inductance of the fans. I was using 2 fans so I don't know if they were interacting with each other or if it is purely the frequency and the fans themselves. When I was up at 6 kHz it was about the same as being at 3 kHz. Going much higher actually meant the fan slow down but up to 6 kHz the power consumption was the same so it was not due to the fact that less power was getting into the fans. At a low PWM frequency it was very evident that current consumption increased down at 100 Hz and dropped off quickly going up to 3 kHz which I suppose is normal as at lower duty more time is spent switching than actually shoving current into the fan.

[d-smes]

That's going to be hard to answer without some corresponding voltage / current waveforms.  I suspect near 3 kHz, your switching times started to match the natural resonant frequency of the parasitic inductance / capacitance of your wiring and load.  In other words, your inverter transitioned from hard-switching to soft-switching (i.e. zero-voltage switching).

[Simon]

Well up to 6KHz gives the same results. It seems that I stay under 1KHz or go to 3KHz or more. Too high and not a lot happens but 6KHz is about the same as 3KHz

[bktemp]

You need to give us more information:
What type of fan?
Name of the Infinion half bridge chip?

[Simon]

Just a couple of old fans I got from a throw out at work. the chip is a BTN8982TA

[bktemp]

The BTN8982TA seems to be fast enough for at least 10kHz, so that shouldn't be the problem.

What type of motor does the fan have? Is it a BLDC motor? Most BLDC fans don't like PWM.

[Simon]

It's a brushed fan or rather they are to brushed fans. Yes the driver chip is pretty good. I'd never drive a brushless fan with power PWM I already have a 2 channel signal controller for those but I am often asked to control a brushed fan because people are stupid and won't see that although a brushed fan is one quarter of the price it will only last one quarter time and then they expected it to be speed controlled which adds more cost so I decided to take money from people who are willingly giving it. The fans together draw between 8 and 9 amp on full power I've been running them up to 97 or 99% so that I have maximum current going through but I am also carrying out switching which will generate the heat that I am trying to investigate for a real-life scenario.

[bktemp]

Ok, no that it is clear you are using normal brushed motors, I think I can explain what you are seeing:
A motor can be seen as a series connection of an inductor (the winding), a resistor (copper resistance of the winding) and a voltage source (back emf).
The inductance and the resistance form a time constant. As long as the PWM frequency is higher than this time constant, the current is more or less constant (there is a bit of ripple, but the current does not go to zero between the PWM pulses). For typical motors this time constant is in the range of 1ms. That's why the PWM frequency of typical motor drivers is in the single digit or lower 2 digit kHz range.
Going higher reduces the ripple, but also increases switching losses and also losses in the iron core of the motor. Sometimes >15kHz are necessary if you need to be outside of the audibility limits.
If you go below 1kHz, the ripple current increases. In your case you have both a low and high side switch in the half bridge, meaning the motor is connected to either the supply voltage or shorted. With a low PWM frequency the current ripple gets so high, the direction of the current actually reverses when the motor is shorted by the low side switch. So the motor accelerates and deccelerates at the PWM frequency. The high peak currents because of the acceleration and decceleration generate a much higher rms current therefore everything gets much warmer than at a couple of kHz.

[Simon]

I see, that makes sense.