DCmax Reduction at High Line causing slow startup

Could you please give more details about your problem?  Load?  Is High Line = 265?  Can you please post your transformer design?

Your forum website is slightly confusing.  After previewing my original comment, I hit post comment, but the text was deleted.

I need help with a universal 88-264V input 16W flyback supply with TOP254EN for +300VDC (@6mA), +/-18VDC (@400mA ea).  Schematic is attached.  Overall the design works well (still finalizing the EMI filter), but every so often when power is first applied, the 300V (stacked with 18V winding) does not charge complete during the first on cycle and the feedback loop does not close, so the supply ‘restarts’, and completes the next cycle with ~2sec delay.  I suspect this may be related to the phase angle of the AC mains input at t=0.  The +/-18VDC drives a power LED on the product that flashes momentarily during the restart and can be confusing to the customer.    The problem gets worse at higher input above 132VAC, and from 220V-264VAC there are 3-5 ‘restarts’ taking up to 9sec for the supply to become stable (power LED = solid ON).  I have observed the 300V output charging higher after each restart until the feedback loop (18V winding) eventually closes.  It would appear the reduced duty cycle is affecting the ability to charge the 300V output cap rapidly enough.  I have tried to short V pin to source, but the restart issue remained unchanged.

Using a regulated AC power supply (using the output ON/OFF button for consistent phase angle) I can consistently observe the following: 90-120V starts immediately, 132V = 1 restart, 220V= 3 restarts, 230-264V = 5 restarts.

I can probably accept 1 ‘restart’, but not 3-5 without causing concern for the end consumer.  Do you have any suggestions? 

I don’t have the transformer construction details, I worked with Renco Electronics (sample S010763) to design the transformer because the PIExpert is limited to 100V output.

It is good you connect the V pin to the return node.  This way the duty cycle is not reduced at start up.

I notice Diode D316 in the circuit.  This diode charges capacitor 306.  What I would do is to remove D316 from the circuit and see if the problem is fixed.

Another thing I notice is the value of R312 is only 24 Ohms.  I think you have too much gain in the feedback loop. It  may be affecting your start up.  I think is good you increase R312 to about 400 Ohms and check your start up.

Let me know if this info fixes the problem. Otherwise we may try something else. If the problem does not go away, please post your DRAIN voltage and current waveforms at start up.  I want to see if IDRAIN is going into ILIMIT or duty cycle limit of the TOPSwitch 

Thanks for the response however the diode D316 is not stuffed in the prototype and 24.3ohm for R312 came from PIE9.  See attached file for reference.

I have connected V to source, no change.  R312 = 100 or 390ohm, no real improvement.  Startup at 120VAC input started to have at least one restart where there was none before.  I did find that increasing the DC bulk cap C301 to 66uF did help with some issues at <90VAC, but the larger cap did not seem to have any effect in 230VAC range.

It is difficult to capture drain voltage and current on the same scope without an isolation device, but instead I captured Vo and Idrain for 120VAC and 230VAC input on the same V and time scale.  R312 was 82ohms (after experimenting with change Vin from 75 to 240 on PIE9) for these tests.  120VAC would restart once occasionally.  Ch 1 (yellow) is Vout (300V output), Ch4 (green) is Idrain (1A/div) using AB503B current probe.  Clearly there is some difference in behavior after 25msec. 

One thing I would do is to connect a DC source to cathode of D6 though another diode (cathode of this new diode connected to cathode of D6, Positive terminal of the DC source connected to the Anode and negative to primary return)  The idea is to bias the opto-transistor with a DC source. You can put about 12 VDC.  This condition will eliminate any possible issue with the BIAS circuit.  Turn on the external DC source and then power up the power supply. See if the problems disappears or sill there.

If you still have a problem, then you need to measure the DRAIN current to see if it is as expected.  For some reason the topswitch is going into limit at start up. The limit could be ILIMIT or Duty cycle limit (DC_MAX)

So you need to measure your current waveform and find out which limit are you exceeding.

You have to make sure your transformer is really as the design is calling for.

I tried your suggestion and diode connected 12VDC to the bias circuit (cathode of bias winding rect D305).  The supply powered up immediately at 240VAC input.

I observed the bias winding output during startup and noted that the voltage was <5VDC and is too low to provide external current to the control pin.  See the attached waveforms.  It appears bias winding voltage is being delayed because of the 300V output charging and isn’t of sufficient voltage fast enough.  Is the solution to increase the number of turns in the bias winding?

I demonstrated that the delay happens at 100VAC input with the reduced loop gain during this experimentation.  Because of the very slow discharge rate, I think I was fooled, but now if I discharge the bias cap before each power on, I get the startup delay shown consistently.

did you remobed Diode D316  from the circuit?

Yes, diode D316 is not stuffed in prototype.

try to increase the soft finish time by increasing the value of C310.   Also I think is good idea connect this soft finish capacitor as shown in schematic of below link 

http://www.powerint.com/sites/default/files/PDFFiles/der276.pdf

If increasing the value of C310 does not fix the problem, then  disconnect R311 and D309 and R308. Try to start yur power supply and see if the problem has gone. 

Hi.  Yes, I had already attempted power on with the 300V output disconnected, and it powers up immediately.

From what I can tell, the time for the bias winding to charge up is longer than the 17ms and the feedback loop does not close and the auto-restart occurs (for 16cycles).  I can observe this.  I don’t see any evidence of current limit during this time, but it is difficult to capture any individual events that might occur during that time with the time base slow enough to see the entire 17ms.

I have been experimenting with reducing the value of the output cap C314 and this is helping significantly.  Apparently the time constant to charge this 33uF cap far enough within 17ms at start up seems to be the issue.  A smaller cap of 10uF solves the start up delay for 88-132VAC input, but still 1 or 2 delays at 198-240VAC.  I've tried reducing the cap value, but I'm down to 1.5uF and it still has at least one delay in that higher range.  I don’t have low uF value caps in 350V so it is getting tricky to keep going lower in value.  I’m concerned about having that small of an output filter capacitor.

What are the design calculations for selection of the output cap?  PIExpert is limited to 100V, so there is 1/3X the voltage but 3X the current in the design (software) compared with actual circuit.

did you try to increase the value of C310  as I asked you in my prevous message?

Yes, increased to 100uF.  No change in performance.

Removing C310 makes start up worse.  It is definitely needed here.

Because your power supply starts up and reaches regulations at low input voltage but does not regulate at high line, then the problem does not sound as a power capability of the controller.  I can think about a condition when the transformer goes into saturation at high line. Then the DRAIN current will reach ILIMIT prematurely and prevent the power supply to go into regulation. So it is important you be sure your transformer is properly built.  Also check the DRAIN current waveform at start up and look for any saturation symptoms

I recommend you change the value of R312 to about 399 ohms.  My experience tells me that 24 ohms is too low for 15 VOUTPUT.

If your transformer is properly designed then you can try to increase the value of C303. This change will increase the discharging time of the control pin and allows the output voltage to go higher at start up before the controller goes into auto-restart

Another thing you can do,  if you do not exceed the breakdown voltage of U301, is to add more turns to your bias winding. This way the controller will have more bias voltage available and will not go into auto-restart mode because the lack of bias voltage

Please see the following waveforms captured during startup.  This is for 230VAC input, with R312=82ohms, and C314=10uF.  It looks as if current limit is being reached for 254EN specified as 1.2-1.39A.  The drain looks like it is attempting to close again shortly after, but the operation is cancelled by the controller?  The current waveform at that point is clearly violating the leading edge blanking, but I’m not sure why.  The peak current detect behavior is not well detailed in the datasheet.  This might be normal and maybe the only solution then is to change to the next larger part TOP255EN (I limit = 1.58-1.82A).  Even with C314 = .1uF, the supply still had 1 delay at 240VAC input, with similar waveforms as shown here.  For comparison, I captured the waveforms (same scale and time base) at 132VAC, which starts up immediately.  The current waveforms look continuous during start up, but after the current waveform is discontinuous.  The trigger point (Vbias) was the same for all captures shown (except 132VAC after start up which was a random stop).

I contacted the transformer designer about the possibility of saturation and the response was that it was not likely the case.

The actual supply output is +/-18V but those feed 7815/7915 regulators to supply +/-15V for the application. I've tried the higher value of R312 @ 390ohm and it has only slowed down the supply startup.

I think the controller does not have enough juice to regulate at starting up.  One solution, as you said, is to increase the size of the controller.  However, if you do that, you need to make sure your controller and transformer are properly matched and you do not saturate your transformer with the larger ILIMIT of the larger device.

Talking about blanking time, every time the controller turn ON, it is discharging C326. This discharge contributes to the large blanking time. Did you ever tested the power supply without C326?

Di you try to increase C304 to have more time for the controller to charge the capacitance of the high voltage output?