Do I really need a current mode control? I didn't see any circuit before that use a current mode control with SG3525. And yes, I've a very little knowledge about that.
It's a very old chip, back in those days it was enough that a switching supply simply work at all, without taking so much space / weight / power / money as a linear supply.
IMHO, any supply that isn't current mode, is either suicide, or hacked enough to be roughly equivalent to a current mode control anyway, in which case you can save a lot of trouble by doing it right from the start.
Anyway, here's the drawing, let me know if it is enough or not.
First I winded the primary, total 8 wires in parallel. it takes only one layer. then brought out 4+8+4 terminal wires as i draw here. then 4/5 layers of insulation tape. then secondary in the same way. this time only 2 wires in parallel. It takes 5/6 layers, then another 2/3 layers of tape
Hmm, so primary is bundled together with itself, that's good. This affects the peak voltage measured at the MOSFET drain, and the required value of the R+C dampers connected to them.
Roughly speaking: your primary wire length seems to be about 60cm, I would guess its leakage inductance to be under 0.2uH. This is a reasonable figure.
I think switching speed will be limited by gate resistance and driver output to about 300ns, which limits how low you can get switching losses, but also prevents voltage overshoot on the drains. As long as you can run switching frequency low enough, overall switching losses can be kept down.
The problem in question is this:
The transformer leakage inductance, combined with drain capacitance, defines a resonant frequency, and impedance.
If the gate drive is fast, that resonance will be excited, leading to high peak voltages.
The drain time constant is,
t_d = pi*sqrt(LL * Coss) / 2
Gate drive risetime should be more than double this, to avoid voltage overshoot. If lower, you will need dampers (R+Cs) or other snubbing methods (e.g., clamp diodes) to deal with overshoot.
So that's alright.
The next problem concerns the leakage inductance between primary and secondary.
For this, the distance between primary and secondary matters. For ~10m of secondary in a single block, expect over 20uH (secondary referred) leakage.
The effect of leakage here, is to cause voltage overshoot at the rectifier, once a filter inductor is installed. An R+C across the rectifier DC side may be necessary, or a clamp snubber, to limit this.
Some related examples here:
https://www.seventransistorlabs.com/Images/Tubescope_Supply2.pngThis isn't really a clear illustration of any concepts I've covered so far, because the rectifiers are mixed (FWB and FWCT types), the filters are choke-input but the one uses a two-winding choke which may be confusing; the current monitoring is primary side, peak mode, uncompensated, and strapped in parallel with the voltage error amp -- if nothing else, a possible example of such a "hack" I mentioned earlier.
Structurally, this is the perfect example of a current-mode circuit, but it is probably still not a good example for present discussion:
https://www.seventransistorlabs.com/Images/Flashlight2_Schematic.pngFirst, understand that the circuits along the top generate PWM, based on the control voltage PWMV. This is equivalent to the circuitry in the SG3525, between the COMP and output pins. IC3B is the error amplifier. IBATT is the current shunt signal, which is exactly in series with the inductor (the output section is here:
https://www.seventransistorlabs.com/Images/Flashlight2Sch.png ). INSP is the inductor current setpoint: a voltage proportional to the desired input current.
At this point, the circuit is a transconductance amplifier, with a voltage input and a current output.
Additionally, IC3A is the outer loop error amplifier. In this case, ILED is sensed, regulating output current; but just as well, the output voltage can be sensed instead, and this is what D1 + R24 do if the LEDs were to become open circuit.
(Indeed, I have hacked one of these boards for use as an inverting Cuk DC-DC converter, with regulated output voltage, and controlled current limiting. It worked first time, and very nicely at that!)
You have 90% of this circuit, in an equivalent form. Fundamentally, you only need to add an inductor and resistor in series with the output rectifier, and an op-amp between the voltage feedback divider and the SG3525 input.
Tim
Home
Help
Search
Login
Register
