Opto feedback for SMPS...common collector or common emitter connection?

[treez]

Why is the common collector connection of the opto-transistor so much easier to make stable than when in the common emitter configuration?

You can see this in the attached LTspice simulations of two flybacks (.pdf schematics also attached)
These flybacks have the same vin, vout, pout, efficiency, Fsw, etc, but have the opto transistor connected differently (one in common collector, and one in common emitter)

Stabilising the one with the common emitter connection of the opto transistor is far more difficult.
(is this due to Miller effect in the CE version?)

 

[betwixt]

I'm not sure it makes much difference. One is configured to invert the signal, the other isn't and they feed different pins on the IC.

Brian.

 

[treez]

From reading the Basso book "switch mode power supplys" (pgs 288-308) I am inclined to agree with you.

However, on the simulator, the common collector connection is far easier to make stable than the common emitter connection.

I have read the sections of the Basso book “switch mode power supplys” concerning opto feedback, and Basso directly contrasts the common collector and common emitter opto feedback connections.
At no point does Basso state that the common collector feedback configuration facilitates higher bandwidth feedback loop than the common emitter connection of the opto transistor.
At no point does Basso state that the common collector connection of the opto more easily facilitates achievement of stability than common emitter connection.
Therefore it must be concluded that the common collector connection is no better than the common emitter (CE) connection.
(I am referring to pages 288, 290, 305, 306, 308 of Basso’s book).

My intuition, however, tells me that the common collector (CC) configuration is superior than the CE connection, because with CC connection, the collector is directly connected to the supply, and thus there is ultra low resistance in series with the highly capacitive base-collector junction of the opto-transistor. –this means a higher pole frequency, and less feedback stability problems?

Do you think that this correct?

 

[betwixt]

I do not have a copy of Basso's book but what you say makes sense. It would be interesting to see the effect of adding a resistor between the transistor's base pin and emitter pin makes a difference. I very rarely deal with SMPS feedback circuits but in high speed data the resistor can dramatically speed up the response time, presumably because it reduces the time constant of the B-C capacitance. It does also reduce the devices sensitivity although that may not be a problem in a PSU. If your simulation allows it, see what happens if you add a 10K resistor. I suppose the opposite effect would be achieved by adding a small capacitance from base to emitter instead.

Brian.

 

[treez]

Thanks, though I think that adding the base resistor possibly affects the CTR(?) ....in any case, ill try it and see, but I can do that with either "common collector" or "common emitter" connection methods.
What is of interest here, is whether "common emitter" or "common collector" connection method is more conducive to achieving higher feedback loop bandwidth whilst maintaining stability?

The following LTspice simulation shows both connection methods in two SMPS's in the same simulation.
The "common collector" connection method is far, far easier to make stable.
The "common emitter" connection version has an unfortunate 10% overshoot on start up, wheras the "common collector" version has no overshoot on start-up. (schematic also attached)

 

[betwixt]

I don't have the time to install LTspice and set it up at the moment but I would guess the overshoot has little to do with the opto-coupler connection and more to do with the compensation network on the LT1243. The reaction time from current change in the opto LED to it's output will be identical in both configurations, the opto-coupler itself will be unaware of the current source and it's load. You would have to simulate it with an additional inverter stage to confirm which was responsible.

Yes, the resistor does reduce CTR, that's what I meant by reducing it's sensitivity. Access to the base pin is very useful when high speed operation is needed though.

Brian.

 

[treez]

I would guess the overshoot has little to do with the opto-coupler connection and more to do with the compensation network on the LT1243

the compensation network is different for the "common emitter" version because that's the only way it could be made stable.
In "common collector" connection, the collector is connected directly to the Supply Rail, and thus there is low impedance seen in series with the base-collector capacitance....thus it has a faster response........in common emitter connection, the collector sees a higher impedance to the supply rail (the emitter resistor, which in the schematic is an "effective" resistor, inside the LT1243)...thus the RC time constant is longer in "common emitter" connection, thus signal transfer is slower, at least this is the theory, its something to do with the miller effect I think.

LT1243 is pin for pin the same as UC3843: LT1243 datasheet...
http://cds.linear.com/docs/en/datasheet/1241fa.pdf

page 17 of this states how "The common-collector configuration eliminates the miller
effect of the output transistor’s collector-to-base capacitance"
and generally increases achievable loop bandwidth.
http://cds.linear.com/docs/en/datasheet/4430fc.pdf