I am confused with your schematic, since the dot side of the transformers indicates the design is more like a forward. But based on your waveform, the circuit is a flyback if L5 is shorted. L5 must be shorted for a flyback design.

Your output is controlled by the bias winding, normally the Voltage across C54 is 6.3V plus the voltage rating of D14, and the output of power supply should be the voltage across C54 times the turn ratio between the secondary winding and the bias winding.

Let's asume turn ratio between seconday and bias winding is N, then the rating of the zenner should be (24/N-6.3)V. Try the zenner with calculated voltage rating first, and Let's see what may happen. Actually even the output is 24V, you have to spend a lot of time on loop stability. So why not use  feedback from the optocopler. With an optocoupler, then youo not have to worry about the 24V setpoin, and the loop is a lot easier. You can start from any DER or RDK board for the loop.

Regards

Hello PI-Cochrane,

Schematic error correction was stated in my previous post.  Primary dot is on pin 1 and L5 is shorted.  This is a flyback design.

Got 24V output with my last xfmr design iteration (I used PI Expert 9 - primary feedback) but only with low Vin (<220Vdc) and will not regulate under load.  Note Vin was specified from 200 - 375V.

Changed to opto/zener feedback.  Unable to close loop (auto re-start) with V0 = 24V using 22V zener;  but loop will close with Vo = 22V using a 20V zener.  However, circuit will not regulate under 15 ohm load and goes into auto re-start.  Upon removal of load, output voltage wouldn't reach regulation point and stays in auto re-start.

I need urgent help as we're running out of time and may need to abandon this design.

Thanks for your support,

Hai 

The start up issue should be caused by the capcitor charge of the output capaciotr C12 and C13. I believe thepower supply can start up sucessfully  if only one 470uF were used. Could you double check this?

And schottky diode is very important for start up issue too, since the flyback always operates at DCM mode for early stage of start up.And reverse recovery can cause the higher initial current spike on the primary side. But definitely your circuit is ok since schottky diode is used in your circuit.

Regards

Yes, the start up issue with Vo = 22V (Vz = 20V)  at various input voltages is resolved by reducing the output capacitance as you suggested.  Thank you!

We still have problem trying to reach 24V output (Vz = 22V).  I assume we need to readjust our transformer ratio?  If yes, what is your suggestion?  Our current ratio is 3.7 with input inductance of 1.23 mH.  BTW, our bias ratio is a low 3.44 since it was designed for primary feedback.  I assume it's still ok because the feedback current is controlled by the opto.

Our second problem is the output loose regulation under load.  Please see image attached.

Thanks,

Hai 

Further reducing the output capacitance from 377 uF to 47 uF enable the output to start up at 25V with full input voltage range from 190 Vdc to 310 Vdc.  However, loss of regulation still occurs under 15 Ohm load (pics attached).

Please help as this is my last chance to save this design.  I'm running out of time!

Thanks,

Hai 

I suggest you using optocoupler, not the third winding fors the volateg sense and feedback. If optocoupler is used, then the volateg rating of the zenner should be 24V-1V. 1V is caused by the led of optocopler.

If the setpoint and regulation can not meet your SPEC, then you have to use the TLV431 to regulate the output. With TLV431, the output is controlled by the reference and voltage divider.

Regards

As stated on 3/27, I've switched to opto/zener feedback using LTV817A opto, 23V zener, and 499 Ohm resistor.

Struggled to get circuit to startup at 24V until reducing output capacitance from 660 uF to 47 uF.  Now circuit will startup but will not regulate with 1.7A peak load.  Please see waveforms attached.

Note that primary current (green) increases under load and goes from DCM to CM but output voltage started to drop out in less than 1 msec.

Please help!

Thanks,

Hai 

Hi KIwibui,

Could you upload your latest schematic? The schematic will help me a lot. By the way, I can not open the files you uploaded the last two times.

Regards

Hello PI-Cochrane,

Please see schematic attached.  I also resend the voltage and current waveforms under transient load of 1.7A peak.  Note the loss of regulation.

Also, why the circuit can only startup with low output capacitance.  Whereas, I've seen design idea (i.e. DI-144) that has output capacitance of 680 uF.

Your prompt reply is much appreciated.

Thanks,

Hai 

BTW, the transformer electrical characteristics are as follow:

Primary Inductance = 1.23 mH

24V Voltage ratio is 3.7: 1 @ 120 KHz

Bias voltage ratio is 3.5:1  @ 120 KHz

Thanks,

Hai 

DI144 has no M pin resistor to line input, and thus the maximum current limit  and power capability is higher than yours.

RCD snubber makes some diffeence too. The RCD snubber used in DI144 is much more efficient than yours. I suggest shorting  M pin to ground and using RCD snubber  for your circuit.

The TLV431 circuit affects the start up too. For 22V zenner, the optocoupler has no current  unless Vo high enough( around  23V). But with TLV431 as the error amplifier, the led current of the optocoupler during start up  should be much higher and earlier than circuit with  22V zenner, since the bias of TLV431 is from the LED of optocoupler. As you know, the lower LED current results in lower current limit/switching frequency, and thus less soft start effect.  For this point of view,   a cap paralleled with the 22V zenner should be helpful for the start up issue.

DI144 is a very good reference. If the issue can not be solved with better RCD snubber and higher current limit (shorting M pin to source pin), then you may have to use same TL431 circuit as DI144.

I donot know why I can not open your waveforms...But anyway,try these three possible solutions first.

Regards

I've shorted M pin to Source pin since last week with no improvement.

I've switched from opto/zener to the TLV431 circuit of DI-144 including the post LC filter.  I've also added the RCD snubber of DI-144. Adding the RCD snubber caused startup problem so I temporarily removed it.  Without it, startup is soft and reliable at full range of input voltages.  Startup will also work with high output stage filter capacitor (680 uF).  However, the converter will not regulate and will not startup even with a 400 mA load.  BTW, currently  the primary and secondary grounds are tied together (no isolation).

I'll try to figure out the waveform posting issue tomorrow so you can see what happen under load.

Thanks,

Hai 

You  do not have to use 680uF cap. use 330uF cap or 470uF cap, and use Post stage LC filter to take care of the ripple problem.

Please help me with the voltage regulation problem.  The converter is unable to startup or maintain regulation even at load current of 400 mA.

Attached are the voltage and current waveforms during startup and under load (400mA and 1.6A).  I don't understand the condition that caused the voltage drop out.  Note Vbias, the control voltage on pin C, is still at 5.8V when loss of regulation occurs.

Thanks,

Hai 

Could you upload your new schematic?

Regards

Schematic is essentially the same as DI-144 (please see file attached).

Transformer electrical characteristics are as follow:

Primary Inductance = 1.23 mH24V Voltage ratio is 3.7: 1 @ 120 KHz; Pri 52T #25 AWG; Sec 14T #19 AWGBias voltage ratio is 3.5:1  @ 120 KHz; Bias 15T #26 AWGCore # DP-43019 gapped for 1.1 mH 
 Help Request: The converter is unable to startup or maintain regulation even at load current of 400 mA.  Startup is ok without a load with Vo = 23.87V. Attached are the voltage and current waveforms during startup under load (400mA and 1.6A).  Note Vbias is the control voltage on pin C. Thanks, Hai 

I suspect that the loop caused the problem. R27 and C35 together with  R10 and  C36 result in two low frequency zero, and thus the power system may not stable.

My suggestion:

1. open R27, and use 100ohm resistor as R26, then check the problem.

2. If 1 not work, then short R10(keep R27 open and 100ohm R26), and check the problem. You can use DER232 as a reference for theloop compensation.

3. once the loop is stable, measured the loop and optimize the loop design.

Regards