The rough schematic of the circuit is presented in http://wintermute.csbnet.se/~zilog/smps_balancer_rough.jpg , the aim of the circuit is to keep the two center-tapped voltages (VH-VC) and (VC-VL) locked to each other, that is to keep them tracking even when an external voltage source is attached to one of them.
This is done by switching over the transformer T1, where both halves of the half-bridge cause magnetic flux in the same direction in the transformer. To allow the transformer to rest/reset from this, the transistors are only on for 25% of the time each.
The transformers:
T1 is wound on a EFD30-core with Ae = 69mm^2, using 31 turns per winding which should yield dB = 2100G when the circuit is used with VH = +45V, Vc = 0V and VL = -45V, as are the operating parameters designed for.
T2 is trifilar wound on a TN13/7.5/5 toroid of 3F3 material, using 20 turns per winding which yields dB = 4400G (Bmax = [-2200G, +2200G] since enough reset-time is given)
I have observed the gate drive-waveforms to be sane for all operating voltages, so I suspect that transformer to behave as it should, the problem lies with T1, as this transformer seems to be pushed into saturation.
The circuit is being fed with +-25V in the picture below:
The waveforms in this image are:
[top: 50V/div, "CH1" in the schematic, zero is 2 divs from the top]
[center: "CH4" in the schematic]
[bottom: 50V/div, "CH2" in the schematic, zero is 2 divs from the bottom]
http://wintermute.csbnet.se/~zilog/assymetrical_voltage_balancer_waveforms.jpg
From the traces it is visible that as one of the transistors switches off, the transformer resets itself much more violently than when the other switches off, causing the 200V-rated mosfets to reach avalanche with as little as 25-30V input voltage. The transformer also starts to screech when I turn up the voltage, indicating too much flux. I have checked and rechecked that both windings on T1 are equal.
When compared with the centre trace which tells about when the gates are turned on, you can see that the transformer resonates, possible that this ringing causes flux-walk in the core? I have tried alot of different snubbers (the traces are taken with no snubbing), the traces look similar, and the assymetry at turn-off still remains no matter what values I use, just results in more energy being burned in the resistors.
What is the cause of these problems? I feel that I need some input to be able to continue.
/Daniel
This is done by switching over the transformer T1, where both halves of the half-bridge cause magnetic flux in the same direction in the transformer. To allow the transformer to rest/reset from this, the transistors are only on for 25% of the time each.
The transformers:
T1 is wound on a EFD30-core with Ae = 69mm^2, using 31 turns per winding which should yield dB = 2100G when the circuit is used with VH = +45V, Vc = 0V and VL = -45V, as are the operating parameters designed for.
T2 is trifilar wound on a TN13/7.5/5 toroid of 3F3 material, using 20 turns per winding which yields dB = 4400G (Bmax = [-2200G, +2200G] since enough reset-time is given)
I have observed the gate drive-waveforms to be sane for all operating voltages, so I suspect that transformer to behave as it should, the problem lies with T1, as this transformer seems to be pushed into saturation.
The circuit is being fed with +-25V in the picture below:
The waveforms in this image are:
[top: 50V/div, "CH1" in the schematic, zero is 2 divs from the top]
[center: "CH4" in the schematic]
[bottom: 50V/div, "CH2" in the schematic, zero is 2 divs from the bottom]
http://wintermute.csbnet.se/~zilog/assymetrical_voltage_balancer_waveforms.jpg
From the traces it is visible that as one of the transistors switches off, the transformer resets itself much more violently than when the other switches off, causing the 200V-rated mosfets to reach avalanche with as little as 25-30V input voltage. The transformer also starts to screech when I turn up the voltage, indicating too much flux. I have checked and rechecked that both windings on T1 are equal.
When compared with the centre trace which tells about when the gates are turned on, you can see that the transformer resonates, possible that this ringing causes flux-walk in the core? I have tried alot of different snubbers (the traces are taken with no snubbing), the traces look similar, and the assymetry at turn-off still remains no matter what values I use, just results in more energy being burned in the resistors.
What is the cause of these problems? I feel that I need some input to be able to continue.
/Daniel
I have now found the problem.
I had a bootstrap-winding for taking over the gate-drive after power-on, apparently this caused flux-walk in one direction. I had completely forgotten that this was still connected when I did the tests. 3.3n / 27ohm for snubbers seem to keep avalanche away aswell.
I had a bootstrap-winding for taking over the gate-drive after power-on, apparently this caused flux-walk in one direction. I had completely forgotten that this was still connected when I did the tests. 3.3n / 27ohm for snubbers seem to keep avalanche away aswell.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.