Okay, heres the thing. I have made a program on my atmega32 that measures RC servo pulses width, and makes a PWM 0-100% from 1-2ms pulses from the receiver. OK.
Then I need to amplify the PWM signal to my (roughly) 3-6 volts DC motor with a transistor. For the amplification I use an external power supply, not the uC ps from the dev. board (which is connected to mains power).

The problem really is that when PWM duty cycle increases, motor speed decreases. I use a NPN BDX53A transistor. I can post a schematic if you like. I know PWM signal is not inverted because I see it on my scope. I have also looked at transistor emitter which is just an amplification of the PWM signal from atmega32. Yet the motor speed increases with PWM decrease and decreases with PWM increase. What fundamental thing have I gotten wrong here?

edit:
Made a schematic anyway

 

Hmm, tried another NPN transistor now, and it seems to work fine. Does this have something to do with the BDX53A beeing a darlington pair transistor? A PNP switching a NPN or something?

 

I don't see Darlington mentioned in one spec sheet but I do in another. What I do see are some really poor curves regarding gain vs Ic etc. The gain is 100 at Ic = 1A but only a gain of 1 at Ic = 10A

 

So it's gain is restricted to 10A, sort of speak? Doesn't really matter for me, the motor uses not more than 0.3 amps when stalled. Thanks for the info

 

I think problem is with your transistor BDX53A , the inductance of your motor coil connected in series with the transistor diverts it from the normal characteristics.

 

Something about that transistor stinks, it's either poorly designed or made that way for an unknown purpose. Just use something different, it's not a critical circuit.

 

The reason the motor slows down with an increase in duty cycle, is because a NPN transistor functions as an inverter. You are basically seeing the inverse of the applied duty cycle. A change in software will fix it.

 

Without seeing more of a schematic it is almost impossible to say what happened. I suspect you had a drive problem to the transistor. The Darlington takes a minimum of 1.2 to 1.4V on the Base Emitter before it can conduct current. A standard transistor takes 0.6 to 0.7V to meet the same criteria.

While a common emitter is an inverter, this is only in reference to Collector ground. When the Base sees a plus voltage (creating current through the Base Emitter) the transistor will conduct, putting a ground on the collector. So for the load on the collector it isn't really an inverter.

I have a work in progress for the AAC experiments I really need to get off my duff and finish...

555 PWM Oscillator

 

Okay, I might need some further assistance on the transistor theory then. I will read the AAC ebook about transistors, but I will ask one more question first.

The BDX transistor was the only power transistor I had. Some of my other transistor works in this setup, but some of them won't switch on with the 5volts peak PWM signal from microcontroller (tried the whole range from 0-100%). A couple of them did, but the motor turned so slow... When using a switch in series with a resistor from the external power supply (also 5v) the motor turned as it should.

 

Best guess, you had the collector/emitters swapped, or they were the opposite polarity of transistors.

 

Checked both, it is wired correctly, unfortunately(?). Can it have something to do with the to voltages coming from two different sources?