Forward converter 230Vac to 14Vac/20A

[Svele]

I have a problem to make forward converter - it works good with lamps load:four halogen bulbs 12V/50W and one car bulb 12V/20W connected in parallel, and I adjust the duty cycle with trimmer and everything works fine (power consumption from mains is about 0.75A).But problem is that converter should work like accu charger for car batteries 12V and load of that converter is changing through time, so I should have some regulation of voltage/current for that converter.In attachments there is schematic of my converter, way of manual regulation of duty cycle and calculations of ferrite transformer.I must say that I do not know any of specifications of ferrite core, so I calculate windings for low frequency (about 25.5kHz), because I suppose it is core older than 20 years and works on low SMPS frequency.
In fact, it seems that I have two problems.
First, current sense won't work good because it won't switch off IGBT on high current - never increase to 200mV, always stays on 185mV never get higher, but when I connect these 200mV from external supply SG3524 switches off the IGBT.
Second, I make voltage control with CNY17-3 and TL431 but when I change threshold of TL431 with trimmer connected to reference pin of TL431 suddenly converter turns off (SG3524).I thought that when I change voltage on reference pin with trimmer or change the input voltage of a converter (with manually regulated autotransformer), duty cycle should increase or decrease, in a manner to maintain constant output voltage, but it never happens.Only, when reference voltage on the other side of opto coupler comes closely to a reference of 2.5V converter is working with buzzing sound, on connected scope to the gate of IGBT, oscillogramme is not stable and after that when the reference is achieved, converter is turned off.
My question is how to make good voltage/current stabilisation and to hold egzisting configuration.Please help me!

 

[Redwire]

The circuit looks good, maybe I can give some things to look at.
I think you need an output filter capacitor, or there is all ripple on the output and feedback circuit.
You need a resistor (~1k) across the opto-LED so TL431 gets 1mA bias power. This keeps the TL431 running without the LED being slightly on.
Ensure pin 5 has a good path to the 0.1R resistor. I use 10uF cap on VREF to keep it quiet.
Are you really using a 100uF compensation cap on pin 9? That seems big and slow.

 

[MicrosiM]

Circuit looks OK to me, Except output capacitor as mentioned by RedWire.

Some oscilloscope measurements would help us understand more

 

[Svele]

Sorry for badly writing of values in schematic!Compensation cap is 100nF.Please explain where to put a resistor across the opto-LED?Is it in parallel with opto-LED or somewhere else?And I will try tomorrow to put some pictures of oscilloscope measurements...

 

[Redwire]

The resistor is in parallel with the opto-LED, because the TL431 draws 1mA which can turn on the opto a bit 5.1V-2.5V/1mA is 2k6 max but I use 1k. Try 47uF across the 5V1 zener to give smooth power to the TL431.

 

[Svele]

Thank You once again for Your replies,voltage loop of error amplifier works good, but I have no equal voltages of both sides of secondary windings, probably because of non simetrical conducting wires on board, everything is now seems to be okay, now I will spend my time with family for holidays,maybe I will try to make 3kW push pull 14Vdc SMPS next year for car/truck starter.Goodbye!

 

[Svele]

Further questions about my converter

Hi again!
I must say that I still have a problem with current /voltage stabilization in a way that only when I try manually to set duty cycle with trimmer 5k and diode 1N4148 between pin 9 of SG3524 and trimer ends of +5V and GND, converter works good with different loads (duty cycle is manually presetted with trimer acoording to load) but when I try to connect converter with regulated autotransfomer in series with main power supply I always start with 50% duty cycle which is normal because voltage stabiliziation is still not work when input AC voltage is around 50 to 100V.Shall I try to connect external +5V to supply optocoupler and try to get voltage stabilization when I increase input AC voltage with autotransformer.Load on ouput is few halogen lamp bulbs with power around 250W, so duty cycle should decrease when I increase input AC voltage?
And second thing happen with shunt resistor when converter is connected to mains: with maximum load of above mentioned 250W square input on IGBT gate pin is completely chopped on little pieces so I should decrease of value of that resistor?When I put pin 4 of SG3524 to ground, gate pin is okay, square impuls is looking good.
And finally, when I connect converter to AC input which voltage is slowly increased with regulated autotransfomer, with input AC is around 60 to 80 Vac, current to converter quickly increases over 2 amps and soonly after that switching IGBT is dead!With scope, voltage shape of gate lokks good but on Collector on IGBT I have very big peak (about 500V with input AC voltage around 80V) and after that peak I have pretty normal square impuls, and furtherly increasing input AC voltage Peak on IGBT not increased any more but current is growing further...And finally, around 100Vac IGBT is dead.
So What should I do to put converter in normal operating condition with working voltage/current stabilisation?

 

[Svele]

And few more informations about converter...
I noticed that both halves of secondary windings had not have the same voltage - when input AC voltage is around 60 V, on secondary with load of 250W/12V, voltage is on one half about 2,3V and on the other less than 1V.Both halves of secondary windings have the same measured inductance - 34uH, where is the problem.
For current sense resistor and for snubber I use wire wounded resistors on ceramic body (I guess that is not problem for relative low frequency of 20,5 kHz) and I use silicon rubber insulators for IGBT aluminium cooler.
Here is scope pictures of voltage shapes on gate (Pic1) with high value of shunt resistor (posted before), where we see chopped square impulses, and according to this, on Pic2 is collector voltage oscillogram on IGBT (when pin 4 of SG3524 is grounded I have only one big peak and good square impuls after that peak), and on Pic3 is the shape of impulse on shunt resistor where it becomes more aslopely when current through IGBT transistor is increasing.

 

[Svele]

On Pic4 is the impuls on gate of IGBT when pin4(SG3524) is grounded.

 

[Svele]

On Pic's where shows the gate voltage it is 5V/div, and where is Collector on IGBT it is 50V/div, frequency is 20,5kHz, on shunt resistor is 200mV/div.

 

[Redwire]

The 3524 has a bug in its 1976 design...lol

I'm not sure but at first look, the 3524 seems to be retriggering its pwm in the middle of a pwm pulse. This would probably be from noise on pin 4 signal. I would like to see Pic3 but from pin 4- if spikes are still there increase 1nF on Pin 4 to 2.2nF or more. Or the ground on 0.1R sense resistor or 1nF is a bad location and noise comes in on ground.

There is a "bug" in the 3524 "... a common problem ... was that any noise or ringing on the output of the error amplifier would effect multiple crossings of the oscillator ramp signal resulting in multiple pulsing at the comparator's output. The [newer] SG3525's latch terminates the output pulse with the first signal from the comparator, insuring that there can be only a single pulse per period..." SGS-Thomson App Note AN250/1188

 

[Svele]

Further questions...

Hello again!I connected 4.7nF ceramic cap instead of 1nF and soldered very close to pin 4 of SG3524 and now you can see the pictures of scope on gate of IGBT,drain of IGBT,Source og IGBT and pin4 of SG3524.I think now the major problem is very big spike on drain which destroyes IGBT when it connected on voltage supply larger of 100Vac.Scope images is on 45Vdc rectified (voltage is taken from lab power supply),current consumption is 0.45A,pic of drain is 5V/div,on drain is 50V/div, on source and pin4 is 0.1V/div.I tried to connect one fast diode UF5408 across the drain (cathode on drain, anode to source) to decrease the spike but nothing happens...What to do next?

 

[Redwire]

I'm a bit puzzled too
What is the timescale on the scope? Same for all pics? They look fine- except for the big spike 600V?! on the IGBT (collector) at 3/4 time through a pwm pulse. It is strange, the gate pulse and emitter current do not agree with what is happening- that the IGBT turns on, then off and then on again! Some things to try:

- SG3524 does not have push-pull output circuit. So it cannot turn on your PNP driver (BC327). Add resistors to ground on pin 11 and 14, say 4k7 to sink current.
- Does the spike happen if you disconnect the opto-coupler (feedback) ?
- Check transformer winding (with UF5408) is wound reverse-phase with primary (dot at bottom)

 

[Svele]

Yes,the timescale is the same for all pics.Yesterday I tried to increase 3x4.7nF1kV ceramic caps in snubber,and peak is decreased very much.And when I put 120nF instead of 3x4.7nF,peak is lost I just have little distorted square impulse on the beginning (little "hill" in the beginning of the pulse").In the evening I will try to implement Your suggestions and see what will happen...

 

[Rogelio]

It might be better if you place a dc feedback voltage to supply pin 15. Just in case ac supply voltage have lowered. The dc feedback supply still stable which give sg3524 better voltage regulation. But you need to a secondary feedback winding and half wave rectifier and filter capacitor for the dc feedback supply voltage to pin 15 of sg3524.