TL494 Flyback driver oscillator section

[TofikLupus]

Hello everyone,
I have no idea how old this forum is but i'm new here, so I hope i chose the correct section for this thread. Anyways, to my question.

Recently I was experimentally building the TL494 driving circuit according to this schematic:

I didn't have the Gate driver ICs so I experimentally tried to build a BD139/BD140 totem pole stage, which I used to drive the gates of IRFZ44N MOSFETs, which were then driving the GDT. The issue i got with this circuit is that if i use the pullup resistors, I basically get the inverted signal of what the output should actually be. In the push-pull mode, this results in the two signals overlapping each other whenever the duty cycle is less than 50% (which for the TL494 is always the case, because the dead time is set to be 5% at minimum). That means that the MOSFETs are in short circuit for the amount of time when the overlap happens. I drew a graph of what i mean, It's in the attachment.

So my question is, is it possible to avoid this behavior by using the TL494 as a high side switch and hook the totem pole stage in between the emitters and pulldown resistors?
Also, I ordered a couple of TC4420 drivers, which are the same thing as the MAX ones stated in the schematic, so when they arrive, I would like to use them for this circuit, but it makes me wonder if the same issue will affect the driver ICs. But that would mean that the original schematic is wrong, which I don't think it is, since it was proven to be working by the author itself. So i'd rather ask for some kind of explanation which could clear this all out to me. TC4420 are non-inverting drivers, so in my way of logic it should also suffer from the brief short circuits and should at least heat up significantly.

By the way, here's a photo of the circuit, it's on a breadboard, very messy but it was the initial build, which i tidied afterwards but i don't have any more recent photos.

I ordered all the chips from Aliexpress so I won't be able to try it out immediately but I will get back to you after i did try them.

Thank you all for your replies and suggestions, and have a nice day, I will be around.

[Mads Barnkob]

Welcome to the forum

It is many years since I made this circuit, so I can not recall everything about it, but I did make it from scratch using only the TL494 application notes, not just finding a design on google. This is a good chance for me to find it again and do some measurements.

There must be something in your breadboard setup that is not right, this is not the only design with either pull-up or pull-down resistors, this is practice described from TI.

Could it be that you use a extra couple of transistors in between, since you have TL494 -> BD139/149 -> IRFZ44N -> GDT, that you get the signal inverted.

[TofikLupus]

Hi Mads, thank you. I was mostly looking into datasheets and only used the schematic as a sort of inspiration, mainly the PWM part, but even the TI datasheet displays the output as inverted, using the pullup resistors. I think none of the stages except the pullup resistors should be inverting in my circuit. I will post my schematic, to give an idea of what it looks like, although it's very similar to your schematic.
I used 4k7+25k pot and 1n cap, with this i set my frequency range to 17-100kHz, then I used a single 10k pot as a divider(without the 1k resistor) for PWM regulation, and instead of halfbridge GDT driver i wound the GDT as 15+15 turns primary with center tap to avoid having to use one of the mosfets as a high side switch(this winding configuration causes really bad overshoot and ringing so I wanna try the MOSFET drivers so I can use a single primary). I will experiment with it more, try out more combinations and will tell how it turned out. However it won't be earlier than next week, because I don't have an oscilloscope at home and I borrow one at my school for my measurements.

The first attachment is my original circuit, the second one is the one which I think should work correctly, with non inverted signal outputs, because the output stage of the TL494 is connected in follower mode (pulldown resistors instead of pullup). I also drew the full bridge section, i think there should be no issue with this part.

The output power transistors will be Infineon SPW47N60C3, 650V 47A.

[erlendse]

What will the circuit be used to drive?

Flyback (converter) is a topology, but also a nickname for the line output transformers in TV's.

If you just want square-wave into a transformer, IRF actually offers a full bridge driver that can drive a H-bridge of FET's.

[TofikLupus]

Flyback not as a topology, but as HV transformers from TVs or monitors, or my own homemade HF transformers. I already have a bunch of TL494s from various sources and i don't wanna buy any additional ICs and I kinda like the idea behind GDTs. Today I was trying the totem pole driver to drive the GDT directly without the Mosfets, but got a weird oscillation and unusable waveform on the negative swing, not caused by the Transformer but the driving transistors.. I forgot to take a picture but it might have just been a loose contact on the breadboard... breadboards are a piece of junk. I will try it with the TC4420s when they get shipped and hopefully I will finally get a reasonable output waveform from the GDT and will be able to proceed to building the power section, the bridge.

[futurist]

That breadboard could be the problem, I suggest you solder everything on protoboard as it shouldn't take long
I have few bad experiences with breadboards and I try to avoid them when possible..

[Mads Barnkob]

I found my old TL494 project box, I actually rebuilt the bridge to use some much bigger 88N30W MOSFETs and it would no longer oscillate, so it is not in a working condition right now, so bare with me if it takes a little while before I can reply with some measurements

[erlendse]

See if you have some UC3842's around too.

They should work great for doing limited input current, regulated high voltage supplys.

[Mads Barnkob]

You are absolutely right, I made a mistake and it is actually driving the GDT inverted, which properly explains why the finalized and boxed version never worked, but then I moved on to another project, as the prototype worked fine.

I am wondering if those pull up resistors were even in the prototype then...

[TofikLupus]

Thank you all for your advices and opinions,
@futurist - I tried to solder wires to the transistors, but that didn't help either, i will just abandon the discrete GDT driver and wait till the 4420s ship, i believe it should be by today.

@Mads Barnkob - I'm glad you could diagnose it and find the issue with your finalized driver, so that means I will use TL494 as a high side switch. 88N30W looks like a nice and beefy FET, with low Rds and high current rating, although not too high voltage, I assume you don't plan to feed the bridge from mains. 100V aleady gives impressive results, but my goal is to be able to use the driver without the need of a mains stepdown transformer, so called offline driver. But at the same time, I want it to be able to run on lower voltages. I'm trying to make some kinda universal driver.

@erlendse - I might have some, I don't know, but if I remember correctly, UC3842 isn't a push pull driver, it only has a single output. I will check it later. Speaking of current limitting, I will probably do it using a current transformer in series with the main transformer's primary.

[erlendse]

Well, the drive topology depends on what you want to archive.

flyback is good at building up major voltage.

The push-pull ones give more current, but is limited in voltage.

Since you haven't told your goal, it's somewhat hard to suggest.

[TofikLupus]

Well I said I wanted to build a push-pull oscillator, hence all the talk about TL494. I don't want to build a flyback one(yet, I will get to that too). Anyways I'm home and my 4420s are here so I will try them out asap. But no oscilloscope shots sadly, I don't have one, so next week I'll borrow one at school and share the results. I got a green unlabeled ferrite toroid(couple of them) so I hope it's not garbage. I already would the windings, all of them with a UTP wire, all 5 twisted together.

[Mads Barnkob]

Unless you bought that green ferrite ring core as a high permability core in the range of 4000-6000, it is most likely just a regular iron powder core from a old power supply, those could be made from N3 material which has a permability down in the range of a few hundreds, absolutely not useful for a GDT in hundreds of kHz region.

[TofikLupus]

Hi, I'm back.

@Mads Barnkob - No I didn't buy the core, it is most likely what you say it is. Not sure if it is iron powder, but I could tell the relative permeability is pretty low. I found one more ferrite core, which had no color coating, and it was even worse. So I decided to temporarily stay with the green one, until I buy a suitable core.

So today I moved on and tried out the TC4420s in the circuit. And the result is - A lot better than I could get with the discrete driver.
I also took several photos of it, including the waveforms on the scope. Sorry for the glare, the scope doesn't support screenshots so I had to take photos of the screen.

I also tried to drive a Mosfet (IRFP250) from the GDT, it worked flawlessly, had a little bit of noise in the switching transitions while the power to the flyback primary was on but that wasn't too much of an issue and it will probably fix after I use all the zener diodes and damping resistors.
I couldn't drive a half bridge with the GDT yet (no supply), so I just drove one MOSFET in a flyback mode, and it worked pretty nice, but before I managed to take a photo of the arcs, I killed a protection diode across the mosfet, BYW29E-150, so I desoldered the diode and ran the MOSFET alone, with no protection. Of course for the sake of keeping the MOSFET alive, I ran it at 12V and 20V for a very short while, just so I could take a photo of it. With the diode I could run it up to 35V volts and the arcs were hot, about 7cm. It drew 5-8A from the supply.

http://imgur.com/a/JPpao Here is an Imgur album, it contains the rest of the photos I took, as well as descriptions of what's going on at each photo.

[Mads Barnkob]

Great to see it work and looking forward to the future improvements.

I have to do some more debugging on my own once I get it out again, after rebuilding the TL494 output setup, the waveform on the GDT was strange, maybe there are some more issues not yet found on the final PCB that I made for it.

[TofikLupus]

So I'm back, 2 years later, and decided to resurrect this project.

The first thing I've done is I sent a request for a PCB to a chinese PCB manufacturer, now it's finished and making its way to my home. Everything will be attached once I get to my computer and render the schematics. For now, I only have the render of the PCB top side. Anyways, everything will be in SMD as an experiment of how much power it can handle, especially SMD version of the 4420 gate drivers.

Now, I bought a large ferrite core, it's an EPCOS MnZn ferrite ring made of T35 material, a thin shielded 4 conductor cable as the winding where the shield is used as the primary and the 4 wires as secondaries, sorry for the bad quality photos here.

Then I asked a friend of mine to machine a multisection bobbin for me out of nylon stick. I drew the 3D model and cross-section plan for him in Sketchup. It is a 26 section bobbin with 2 sections just for termination of the windings so there's actually 24 usable sections. See attachments for more details. Yesterday i finally wound the secondary with my newly built coil winder, it has 3610 turns ±10 turns of 0.118mm wire. It was supposed to handle 168 turns per section but my calculations turned out not so accurate since i didn't count with that the bobbin cannot be made that precisely with just a lathe and the fact that i just couldn't wind the windings exactly one turn next to another without a slight overlapping in just a 1mm wide groove. That's some ant work there. So some sections carry 130-150 turns and only a few of them have full 168 turns on them. That shouldn't matter too much though as far as i'm concerned.
As for the bobbin construction, I inspired myself on youtube channel "How To Do" and also on this photo of flyback crossection that I made 2 years ago. See attachment.

Then I made a potting shell from a 1mm PET-G clear sheet. I plan to pot it with a clear epoxy resin, to see what's happening inside if there are gonna be any failures, for future improvement.
The power stage will be a full bridge of IRFP250s, bought locally because i don't want to rely on chinese fakes to explode in my face.
The power supply for the bridge will be an industrial 24V 20A SMPS until i decide to switch to the full 320V supply, for that i will have to buy higher voltage mosfets. I'm thinking of something 20A+ because of lower Rds on ratings and thus less heat dissipation, but these are quite expensive. Most likely I'll stop at IRPF450s anyways

Now I'm waiting for the rest to arrive from China, that is most of the components for the PCB and the PCB itself.
Just as a note, the PCB doesn't count with any external shutdown or current limiting since it's still an experimental build and i need to first see whether it works or not.

I think that's it so far. Sorry for disappearing for so long. And also merry christmas and happy new year to everyone.

[setkiki]

Thank you for the information provided, which gives me a new understanding of the transformer, thank you
https://www.allicdata.com