Two switch forward based on UC3845

Silvio

Active member
Its not a question of re-doing the pcb but getting a good Strong waveform. It is clearly seen that the GDT is absorbing a lot of power and exhausting the available power from the drive. The waveform clearly shows this and charging the gates is taking forever, hence the slow rise and fall times. Your best option I guess is to make another totem pole after the drive to add some current to the primary of the GDT. If you like to try something else well considering the GDT primary inductance is only 500uH then you can put a 10 to 20 ohm resistor in series with the primary to limit current. (remove everything else and feed from the drive directly) This may cure the problem. Do not forget that with a 2 switch converter you are driving 2 gates simultaneously. Experimenting with the resistor value may get you a good balance to get enough current and not loading the driver chip more than necessary.
 
Do not forget that with a 2 switch converter you are driving 2 gates simultaneously
Well, it makes sense since peak currents are basically double that of a half bridge where transistors are driven one at a time. But I was thinking, a BC546/BC556 totem pole can only sink 100mA continuously, and a UC3845 can do 500mA. Even a TL494 can do 200mA. So how is it helping?
 

Silvio

Active member
Well, it makes sense since peak currents are basically double that of a half bridge where transistors are driven one at a time. But I was thinking, a BC546/BC556 totem pole can only sink 100mA continuously, and a UC3845 can do 500mA. Even a TL494 can do 200mA. So how is it helping?
Well it all seems that the primary inductance is low for this purpose and loading the chip. I myself made GDT and lower inductances than 1mH and using 15v did not gave me good results. I also learnt that symmetry is also important in GDT and you get best results using toroidial cores. If you try a higher inductance I guess you will get better results. All I know is that when loading the driver the waveform gets deformed. The series resistor I mentioned I seen it used in a pro amp having a GDT of 470uH. This was driving a couple of IGBTs. To tell you the truth I monitored the output waveform from this GDT and the wave was not that square either. Regarding the totem pole well in that case use something more beefier like BD139/140.
 
Forgot to mention, this is the GDT

 

Silvio

Active member
You had an EE16 in your schematic and that was what I was imagining you had as a GDT. The ring core is a good choice. The one I have also has more or less the same inductance. well you can drive that directly without any series capacitor. Just put a 1k load resistor in its outputs and check waveform before hooking up any mosfets. If you are happy with the result just put a series resistor to the gates and all should be well.
 
I'm revisiting this since I've come across the need to use GDTs again. It seems that the BC546/BC556 isn't really enough to drive a GDT. I've built a similar SMPS for another project and while the drive is good enough for the main transistors, the poor totem pole pair driving the GDT blows up after a while. So is it the peak currents that blow them up and how is a chip like the UC384x able to drive a transistor, being only able to drive about 200mA?
 
So out of curiosity I built a single switch forward (with reset winding). All works fine but the output can't sustain even a small load (few hundred miliamps) without dropping (for example 500mA drops from 25V to about 10). The transistor doesn't heat up, the output diodes don't, I tried different inductors, I tried raising the VCC on the IC...nothing. I can't figure out what's wrong.
Feedback is disconnected and the current sense pin is grounded.
 
Last edited:

Silvio

Active member
So out of curiosity I built a single switch forward (with reset winding). All works fine but the output can't sustain even a small load (few hundred miliamps) without dropping (for example 500mA drops from 25V to about 10). The transistor doesn't heat up, the output diodes don't, I tried different inductors, I tried raising the VCC on the IC...nothing. I can't figure out what's wrong.
Feedback is disconnected and the current sense pin is grounded.
Do you have headroom voltage to regulate the output? You cannot sustain the output with load without headroom voltage to spare. It is quite obvious that the output is dropping as you did not use regulation. Your output is showing the maximum voltage the smps can give. Without regulation the pulse width is at maximum :)
You should connect the feedback and regulate for 12v for example and then you will have some kind of regulation. As soon as you load the output the voltage tend to drop then the feedback will give a signal to the chip to open the pulse width to regulate the output.
I hope that helps you with better understanding principle of the smps.
 
Yes that is true, the chip is limited to 50% duty cycle with this topology so without regulation that's the maximum voltage I can get out of it.
 
Top