Royer Oscillator as a DC-DC converter

[adastra]

We have a relatively old (ca. 1970s) design for a DC-DC converter which centers on a saturable core toroid transformer. A center tap feedback winding provides the signals to the bases of two complementary NPN transistors; these transistors nominally switch at around 5 kHz.
At room temp and above, the circuit starts up fine. At cold (below 0°C), the unit sometimes comes up screaming in the hundreds of kHz and stays that way - dissipating lots of power but doing nothing useful.
Total secondary load is around 0.75 watt.
Obviously a driven power supply is the way to go, but our customer still wants this groovy-man era item.
Has anyone had experience with this?

 

[Skogsgurra]

This may be due to the fact that transistor hFE is too low at low temperatures. That prevents the oscillator from starting at the right frequency. Instead it starts at a much higher frequency where the feed-back to the bases is strong enough to keep oscillation going. I would try a few more turns in the base feed-back winding. Or have you done that?
An alternative is to use transistors with a guaranteed hFE.

Gunnar Englund

http://www.gke.org

--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...

 

[adastra]

Thanks Skogsgurra. Transistor hFE definitely is a factor. Vbe increases at cold, making the situation even worse by reducing the base current.
I've also added a series RC network across the B-C of each transistor as an LPF (adding to the Miller capacitance). This helps with starting up, but "rounds" the square waves, reducing efficiency. Frequency response seems to be a double edged sword on this one.

 

[Skogsgurra]

Yes, you are right, Vbe is higher at low temperatures. A few more turns should also help against that.

Gunnar Englund

http://www.gke.org

--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...

 

[Comcokid]

As a hobby, I occasionally rebuild tube-type mobile transceivers built in the 60's. I've noticed the transistor inverter/toroid type supply don't quickly start at cold temperatures, and I've always assumed that this was because the charcteristics of the germanium power transistors used. There's about a 3 second pause before starting when power is applied at cold temperatures. However, I've never scoped it - you hear the switching audibly. Secondary load on this kind of unit is 20 to 100+ watts.

 

[logbook]

I assume the guts is similar to this:

http://www.edn.com/archives/1995/021695/graph/04di1fg1.htm

Clearly this is a purely analog circuit, so nobody now knows how to design them!

I suspect a common-mode feedback path is causing the problem (in addition to the low transistor gain at low temperatures). You may be able to suppress this mode with a common-mode choke supplying both transistor bases.

Failing that maybe a base - ground RC snubber on one or both sides will reduce the gain to the common-mode oscillation sufficiently.

 

[bogeyman]

The royer convertor is still in use in some modern convertors, it is simple, easy, cheap and reliable.
The bias network should have a forward biased diode in the lower end to compensate for the transistors Vbe change. This is often replaced with a resistor to cheapen the price, not fitting the diode can make starting a bit hit and miss.

 

[adastra]

Thanks, all. One issue with the circuit is its symmetry. A simple circuit analysis is useless in that, with ideal components, the base current splits evenly (like balancing a ball on the end of your finger). In reality of course the parameters aren't exactly the same and the base current isn't split identically.
That said, a startup circuit to deliberately bias one transistor differently than the other seems to help. The challenge is to make that part of the circuit "invisible" once it's up and running.