VVV
Advanced Member level 5
R34 affects the overall gain, by affecting the "optocoupler" gain. The optocoupler itself has a gain of 1, but the ratio of R34 to R32 is also a gain factor.
A typical value for R34 is 2.7k. However, since all calculations have been done using 1.8k, leave it at 1.8k. You can change it to 2.7k, without too much trouble. The gain increase of 3.5dB would affect the crossover frequency only marginally. The fact that it runs at 5V has no effect, since it is the ratio of the two resistors that counts.
The voltage at pin 1 should be DC. If it is not, the P/S is oscillating or there is noise pickup. This could come from the primary side or the secondary side, coupled through the optocoupler. I can tell you how to determine where it comes from, but I am not convinced this is the problem. Try attaching pictures, they may provide clues. Don't forget, use short GND lead for the scope probe.
Traditionally, transient testing is done from 10% to 100% load. This has its roots in those converters, in which the inductors were selected to go discontinuous at 10% load. Since an inductor in DCM could create problems, the "normal" practice was to avoid that region and test in CCM ("inductor" can mean the transformer of a flyback designed to work in CCM).
By extension, the method has been used in all P/S. There are situations where testing down to 0% is necessary, but you really have to ask yourself how important this is. Is such a situation likely to occur in practice? If so, is it important that the output voltage remain within tight limits? I mean, if you simply apply voltage to a circuit, does it matter that the voltage first drops to nearly zero and then recovers? After all, just prior to that the circuit was unpowered, so this should not matter. Of course, there may be other circuits powered from the same voltage, (drawing negligible current) which may reset when the heavy load is switched in. Under these conditions, testing down to 0% is necessary.
But you can begin by testing from say 0.3 to 3A, just to see if the behavior is different. Perhaps we can draw some conclusions from that.
Basically, the transient response means just the output voltage. Look at the picture. This is what the output voltage should look like on a well-behaved P/S. The difference between the inner horizontal dashed lines represents the effect of the load regulation. The difference between the peak or valley and the flats represents the transient response. Usually, this should be less than 4-5% of the output. Measure also the recovery time. Typical values are from a few hundred µs to about 1ms.
Usually, the transient response is measured with the scope set to AC, and the sensitivity set in such a way as to allow easy measurement of the waveform characteristics. The load switching frequency must be low enough for you to observe the flat portions, i.e. low enough to allow the P/S to recover after the transient.
A typical value for R34 is 2.7k. However, since all calculations have been done using 1.8k, leave it at 1.8k. You can change it to 2.7k, without too much trouble. The gain increase of 3.5dB would affect the crossover frequency only marginally. The fact that it runs at 5V has no effect, since it is the ratio of the two resistors that counts.
The voltage at pin 1 should be DC. If it is not, the P/S is oscillating or there is noise pickup. This could come from the primary side or the secondary side, coupled through the optocoupler. I can tell you how to determine where it comes from, but I am not convinced this is the problem. Try attaching pictures, they may provide clues. Don't forget, use short GND lead for the scope probe.
Traditionally, transient testing is done from 10% to 100% load. This has its roots in those converters, in which the inductors were selected to go discontinuous at 10% load. Since an inductor in DCM could create problems, the "normal" practice was to avoid that region and test in CCM ("inductor" can mean the transformer of a flyback designed to work in CCM).
By extension, the method has been used in all P/S. There are situations where testing down to 0% is necessary, but you really have to ask yourself how important this is. Is such a situation likely to occur in practice? If so, is it important that the output voltage remain within tight limits? I mean, if you simply apply voltage to a circuit, does it matter that the voltage first drops to nearly zero and then recovers? After all, just prior to that the circuit was unpowered, so this should not matter. Of course, there may be other circuits powered from the same voltage, (drawing negligible current) which may reset when the heavy load is switched in. Under these conditions, testing down to 0% is necessary.
But you can begin by testing from say 0.3 to 3A, just to see if the behavior is different. Perhaps we can draw some conclusions from that.
Basically, the transient response means just the output voltage. Look at the picture. This is what the output voltage should look like on a well-behaved P/S. The difference between the inner horizontal dashed lines represents the effect of the load regulation. The difference between the peak or valley and the flats represents the transient response. Usually, this should be less than 4-5% of the output. Measure also the recovery time. Typical values are from a few hundred µs to about 1ms.
Usually, the transient response is measured with the scope set to AC, and the sensitivity set in such a way as to allow easy measurement of the waveform characteristics. The load switching frequency must be low enough for you to observe the flat portions, i.e. low enough to allow the P/S to recover after the transient.
