Push-Pull SMPS overshoot problem

[circuit]

Hello from Lithuania 😉

I'm designing a small SMPS for my huge graphic VFD from Noritake. This display needs DC 60V and AC 6V 100kHz. For even illumination, that AC supply signal must be symmetric. Also, PWM is needed.
This VFD is going to be used in my power amplifier, so supply should be really well built. It will be supplied from DC 5V and total power is about 5 Watts, so it is 1A.

Problem #1:
I can't find a SMPS controller chip, that could operate on 4.5V-5.5V. Something similar to TL494 would be OK. Only 1A will be drawn through power switches, so, in best cause, they could be integrated into that chip.
I have temporarily implemented it on small microcontroller ATtuny15.

Problem #2:
Simplified schematic looks somewhat like this:

An externally hosted image should be here but it was not working when we last tested it.

Notice signal on transformer's secondary. Well, this is what should be theoretically, but I get something a lot worse:

An externally hosted image should be here but it was not working when we last tested it.

I have marked pulses that shouldn't be there. If those diodes are removed, pulses go few times larger than supply voltage and fries mosfets.

This is caused because of high parasitic inductance in transformer's primary wingding. I have seen a lot of schematics and working power supplies of this type, but none of them had such problem.
What am I doing wrong? Does here matter ferrite core material?

Thanks in advance. Oh, and sorry for my English 😉

 

[circuit]

Some more info.
Here is an example of transformer current:

An externally hosted image should be here but it was not working when we last tested it.

Mine is running at 100kHz, that is, a maximum pulse width is 5µs. According to picture above, it goes to peak of 300mA. Is it OK?

I'm using EPCO's transformer ER 11/5 with N87 core. Also tried N49 and T38 - result the same, only with T38 that pulse is slightly smaller.


Anyway, I'm looking for a way to fix this.

 

[HVYMETL]

The waveform that you're seeing is due to the magnetizing current (I-m) of your transformer. This current flows into the transistor which is ON, then when that transistor is turned off, I-m has to flow somewhere so it forces the drain high, and the opposite drain low until the reverse diode conducts. The integral of volt-seconds while the switch is ON has to equal the volt-seconds when the switch is OFF. That accounts for the part of the waveform which you circled. After this is finished, the resonant voltage you see is just the ringing of the natural parasitics of the transformer and capacitances.

 

[megajocke]

This output is for the VFD filament and the waveforms are taken unloaded, right? When connected to the filament the flyback pulses you see will go away or at least be attenuated.

 

[circuit]

This output is for the VFD filament and the waveforms are taken unloaded, right? When connected to the filament the flyback pulses you see will go away or at least be attenuated.

Wow, thanks. Somehow I didn't realized that. Well, everything works fine now.
Will post some photos when my 256x64 grapchic VFD lights up 😉
Thanks again.

 

[jamesrnz]

i have over shoot too

on my 12 to 24 volt full bridge i have "overshoot" too

it seems i have always had it when the transistor turns off i get a 90 to 100 volt pulse.

i wonder if i am measuring something wrong

😕

 

[megajocke]

Full bridge or centertapped push-pull? Where are the 100V pulses? Primary?

 

[jamesrnz]

regular full bridge

on the primary


which i thought that they couldnt go much above the supply voltage in a full bridge


nothing special about 2 t of foil on an etd49 at 100khz with a series cap of 25uf

 

[megajocke]

I'd guess it has something to do with the layout then, if the inductance is a bit high in the supply leads this could happen. How far from the transistors are the power supply caps?

 

[jamesrnz]

mmmmm.
about 20 inches in the dc power supply that i am using to power the board.

should i have bulk caps near the transistor inputs?

thanks for your time

 

[megajocke]

Yes, that's probably it. The bulk caps need to be sized to take the input ripple current too.

 

[jamesrnz]

thanks, i'll try that when i get back home in acouple of weeks, since i am a truck driver

 

[luka]

Hi

that 25uF is way (too?) big for 100kHz, coz you don't have that big current in primary that you would need that big, 2.2u should be good enough, I have it @ 50kHz

@circuit

is your supply unregulated? Ah you are using uCPU...your dead time is big, probably you leackage inductance too... post more pic

 

[megajocke]

Not at 12V input voltage. 25uF seems a bit small, but it depends on power of course. I'd chosen push-pull instead though.

 

[luka]

Hi

Building half bridge for 12v input is pointless, push-pull or fullbridge is the way to go, more push-pull... and those 2 don't need series cap

 

[jamesrnz]

ok here is my schem, it converts badly

 

[jamesrnz]

this is a representation of the signal i have at the source drain of the upper and lower mosfets on either side of the transformer.

i dont have any bulk caps at the input

i have it hooked to a 10 ohm load at 18 volts output at 1.8 amps.

the input is 12 volts at about 3 amps

the transformer is an etd49 with 2 turns 1" wide foil .040, and the secondary is 5 turns foil.

i am running at 100khz transformer (200 controller)

the current mode control 6,7 are shorted to disable. and the current sense pin 9 is grounded. the voltage divider is a little different but feeding back fairly stable at 18 volts( i have a small pot in there)

it is on a circuit board i made that is fairly balanced in its layout. i have quieted it down a lot.

i may have some ringing on the rectifiers that i have to deal with.

fixed the phasing on the gate drive transformer so i dont create plasma

uhmmmmmm i am running it at low current while i sort out the problems before i load it heavier

 

[megajocke]

Maybe I misunderstood the 12V to 24V part. Is this the range of output voltage or the input and output voltage?

 

[jamesrnz]

12 v ish in 12~24 out @ 50 amps sorry

 

[jamesrnz]

this is a representation of the signal i have at the source drain of the upper and lower mosfets on either side of the transformer.

i dont have any bulk caps at the input

i have it hooked to a 10 ohm load at 18 volts output at 1.8 amps.

the input is 12 volts at about 3 amps

the transformer is an etd49 with 2 turns 1" wide foil .040, and the secondary is 5 turns foil.

i am running at 100khz transformer (200 controller)

the current mode control 6,7 are shorted to disable. and the current sense pin 9 is grounded. the voltage divider is a little different but feeding back fairly stable at 18 volts( i have a small pot in there)

it is on a circuit board i made that is fairly balanced in its layout. i have quieted it down a lot.

i may have some ringing on the rectifiers that i have to deal with.

fixed the phasing on the gate drive transformer so i dont create plasma

uhmmmmmm i am running it at low current while i sort out the problems before i load it heavier

any comments on this scope pattern or is this normal at this point of measurement when the transistors turn off?