솔루션 찾기

Top256YN bad cross regulation on multiple output design

Posted by: thebrakshow on 09/07/2021


I have attempted to create a flyback design with multiple outputs each with a unique ground. One of the main issues I'm facing is when there are strong step loads on either secondary winding and then removed, I can read as much as 33V on my 24V rail. The 12V rail stays nicely regulated. Is this to be expected? I figured there would be some amount of voltage fluctuation on the non regulated rail but not this much. I didn't see an options PI Expert about some sort of cross regulation scheme. AN-43 has a quick blurb about cross regulation with a couple extra resistors and a zener however it doesn't go into very much detail. Am I going down the right path in thinking I need some sort of cross regulation, or could there be something else throwing off my design?

As for the basics, the design boots up just fine with full load on low line (85V - Actually the supply starts as low as 70VAC fully loaded) and all the way through max voltage (265V). The outputs have less than 75mV ripple ( 20Mhz BW limit, using tip and barrel method with a 0.1uF capacitor solder directly across the probing point). There is significant HF switching noise on the outputs ~ 1Vpp in some cases. The current through the drain pin has perfectly linear ramps. The initial current spikes are faster than 220ns so the lead edge blank time takes care of the low line spikes. The drain pin never sees more than 550V with max load/high line with repetitive cold start. The primary inductance was measured at 444uH with a DER-5000 set to 100KHz, and double checked measuring 25%/75% di ratio and average bus voltage across the input capacitor. Leakage inductance was measured around 8uH.

I will say one thing I do find odd about the design is that it only ever operates at or near the 132KHz range when the input is at low line (~85VAC) with a fully loaded output. From there, the frequency just reduces as the input voltage goes up or if the load goes down. The odd part is that the switching freg can go lower than 66KHz.. For reference I have the F pin directly connected to the Source pin. 


I am by no means an expert at flyback designs so bear with me :). I am happy to provide any scope capture or picture of the setup as requested. 

Thanks so much for any help in advance!



Submitted by PI-Wrench on 09/08/2021

Unfortunately, this is just about the worst-case scenario for cross regulation - two outputs with separate ground systems. Since all the regulation is referred to the 12V output, there is no way for the control loop to know what is going on with the 24V output except by the indirect mechanism of coupling between the two output windings. This does not work all that well even for stacked outputs that share a common ground. If the voltage isolation requirement between the two output systems is not too severe, you might be able to improve cross-regulation somewhat by winding the 24V secondary directly on top of the 12V output winding. This might improve cross-regulation, but the 24V output will probably still fluctuate, especially in response to 12V loading.

Also, i would expect to see more high frequency "grass" on the 24V output, as there is no return path for the high frequency edges that get coupled to the secondaries from the primary via the transformer interwinding capacitance. A 2nd Y capacitor from the 24V return to the primary B+ might help with this, if it is an issue. 

What is your 24V regulation requirement? If you need extremely tight regulation on the 24V output, you may need to add a pre-load or a post regulator. There are circuits for an active preload that may help in your situation. A lot of this will depend on the load requirements for both outputs. It would help if you can describe the required loading for both outputs.


The behavior you describe in the last paragraph is normal for the TOP-HX, and is described in the data sheet.

Thanks for the response!

Yeah unfortunately, the 24V supply's ground is going to be referenced to a high voltage AC capacitor doubler which creates VBUS for a AC induction motor. The 12V supply's ground will be ultimately connected to Earth Ground. So the isolation between the two needs to stay. 

The 24V supply mainly powers a switcher (to 15V) which then goes off in to multiple other rails. The initial switcher is a 36V tolerant part so it should be just fine as long as we can keep the 24V rail below 30V (for margin). There are a couple of other devices off of 24V and would be fine if the voltage swings +/-1, +/1.5V, More than that and it becomes an issue. Total current for the devices directly off of 24V is ~100-150mA (not including 15V switcher). I don't believe simply adding preload will work as it takes ~200mA of load to stop the rail from floating past 28-29V. So i guess a post regulator is my only option. I supposed a floating reference LDO could work or a BJT. Are there any other techniques I can use to improve the windings/construction of the flyback transformer?

I just want to be sure I understand you comment about the 24V Y capacitor, can you have a look at my linked file and confirm this is what you are recommending? On the topic of Y capacitors, is there an application note on how to select these? I don't really see much about it in the HX datasheet or AN-43. lastly, should the bias winding also have a Y safety cap?

Thanks so much for your help!

첨부 파일 파일 크기
24V_SafteyCap.png 223.64 KB
Submitted by PI-Wrench on 09/08/2021

I would still recommend winding the 24V output secondary on top of the 12V for improved cross regulation. Since you have the 24V output referenced to AC potential, it would be necessary to use something like triple insulated wire for the 24V output for enhanced isolation to the 12V. Winding both outputs using triple insulated wire would be a belt-and-braces approach.  This measure might take care of some of the voltage rise you are seeing on the 24V.

You will also need to supply some safety physical creepage distance between the 12V and 24V supplies by physically separating the 2 outputs. Pulling intervening pins on the output side of the bobbin might work, or having one of the outputs exit on flying leads.

Absolute worst case would be the 12V loaded to maximum, with a minimum load on the 24V output.  An active preload using a TL431 and a PNP Darlington transistor might be one solution for your issues once you have the 24V output excursion under better control. The active load would only sink significant current if the 24V output rises above its maximum limit. You can set this limit as you please with the voltage divider driving the TL431 REF pin. Of course, with an actual post regulator, you can absolutely nail the output overvoltage problem.

The scheme you show for the extra Y cap was the one I had in mind. This should only be necessary if the HF  "grass" on the 24V output is an issue. The bias winding doesn't need a Y cap, as it is primary-referred and will not have issues with HF noise. In many instances, the bias winding is wound multifilar and is used as an impromptu shield between one of the primary halves and secondary.


Submitted by thebrakshow on 09/09/2021

Its definitely worth a shot, today I will rewind the transformer to remove the single turn of insulation tape between the 2 secondaries. Both secondaries are wound with TIW. Currently I'm using a 7 x 7 bobbin with everything on the "AC" side and only the 12V secondary on the other. See attached diagram (N2 = Bias, N3 = 12V, N4 = 24V). What is not shown in the diagram is that N2 is bifilar and N3 is trifilar. Lets see what this buys us in terms of voltage fluctuations, before I go down in a rabbit hole of post regulator solutions :)


By any chance, and I can start a new topic/thread for this, do you guys do layout reviews? If so, is there a non-public way to do so?  

As always, thanks for the help!

첨부 파일 파일 크기
Screen Shot 2021-09-09 at 10.22.43 AM.png 121.59 KB
Submitted by thebrakshow on 09/13/2021


I have retested with a rewound transformer which now has the two secondaries wound together without any insulation tape. It didn't seem to buy us much as I still see the 24V rail shoot up to ~30-31V when the load is pulsed on and off. I'll move forward with a post regulator scheme. 

Switching to the topic of "grass" on the outputs:

While we do see an acceptable voltage ripple ~75mV ish (pk-pk), We do see significant HF voltage spikes as the switcher turns on and off. The HF is ~70MHz and can be as much as 400mV pk-pk ( 20Mhz BW limit on scope). Similar noise is observed on both 12V and 24V rails. I have soldered on  100pF,100nF, 1uF MLCCs and 10uF ecap to attempt to reduce this but only succeeded in drop it ~50ish-75mV. Is this amount of noise pk-pk to be expected? I assume some of it has to do with the additional parasitics due to my hand winding. We did see similar amounts of noise with a properly fabricated transformer we had made. 

Again thanks for the input!

Submitted by PI-Wrench on 09/13/2021

It sounds like the switcher is dumping HF noise back into the 24V output. Try isolating it from the 24V output with a pi network.