Two-switch forward current source is heating up when idle

[Artlav]

I made a constant current power supply for an arc lamp, based on a welding inverter design.
24V input, 8A at about 16V output.
Transformer is a 40x25x11 ferrite ring.

Problem is, it's heating up at no load, and is not heating up when loaded.
Unloaded, it consumes about 1A, with the heat going out of the bridge's diodes and fets.
As the thing heats up, it goes closer to 2A.
Does not look like an overvoltage - nothing changes with the secondary disconnected from the rectifier, and there are no overshoots or problems in the bridge itself.

So, is that a problem (it works fine and stays cool under load), and if so, what's causing it?

[T3sl4co1l]

Why use transistors 50 times larger than necessary?  The gate charge is embarrassingly high, even for 50kHz operation.  Without a voltage limit, I would guess it's running at full duty cycle and circulating current through the primary and burning losses in hard switching.  This would be obvious on the oscilloscope.

A current source is supposed to be shorted when not in use.

Tim

[Artlav]

Why use transistors 50 times larger than necessary?

It's the closest thing i had on hands...
The driver can handle them, however - i see 200ns rise/fall times.

I would guess it's running at full duty cycle and circulating current through the primary and burning losses in hard switching.  This would be obvious on the oscilloscope.

What should i be looking for?
The controller can only give 50% DC.

A current source is supposed to be shorted when not in use.

There is a time between it's turning on and the lamp getting ignited, where it should run into no load.
Not a long time, but still.

[Richard Head]

Have you confirmed that the magnetising current is a reasonable value?
If you have a current probe clip it over the transformer primary and check the primary current waveform to confirm the transformer isn't saturating at full duty cycle.You could also just measure the CT secondary voltage I suppose.
Also, the primary Shottkies for steering the magnetising current back to the bus can have a huge reverse capacitance that could be causing problems. You would definately need a current probe for that measurement as the current would be a shoot through current that the CT won't see.

[DanielS]

I would guess it's running at full duty cycle and circulating current through the primary and burning losses in hard switching.  This would be obvious on the oscilloscope.

What should i be looking for?
The controller can only give 50% DC.

What to look for? Nothing, the circuit will look like it is operating fine except for the high idle power draw and dissipation in the core. The circuit uses current feedback and drives the primary however hard it needs to in order to reach the feedback current.

If you want to reduce that wasted power, you need to add open-circuit detection that either resets the PWM into soft-start mode until a load is detected or lowers the current set-point/duty-factor until then.