Electronics > Projects, Designs, and Technical Stuff
ATX Flyback Transformer
engrguy42:
--- Quote from: T3sl4co1l on June 20, 2020, 07:06:33 pm ---The heck is up with that input? I don't see a Kx L7 L8 y statement. What's D802 for? Does C1 have a .IC set, is that what's powering this thing? Why is L1 so huge? Why include Z1, are you also testing transient immunity..?
Again as mentioned, it's a forward converter transformer; you won't get much power out of it this way, some watts I guess. Easy enough solution: turn around L10 (or set K5 = -1), double up D5 (use a diode from GND to L1), and use a, well, L1 is already there and roughly the right value so I guess that's alright once these changes are made.
Primary side doesn't need to change because the transistor sees the reflected current charging into L1 -- the current ramp, and I suppose LTC3873 has slope compensation too -- which is basically the same story there as for flyback. :-+
Tim
--- End quote ---
Like I said, the input is a reverse engineer of the actual ATX supply. That's why D802 is there. Because, well, it's there. And like I said, there's more circuit to the story that I haven't gotten to yet. I just added some core stuff and tweaked it 'til it worked.
Regarding L1, you ask why it's so huge, then say it's about the right value. I just threw a number in there. And the "K4 L7 L8 1" statement is there, you just can't see it. And I tried reversing the switching transformer previously with a bunch of different ratios and it didn't make much difference in terms of improving response. And Z1 is there because, well, it's there. This is part a documenting of what's there and part a simulation to see how it all works.
Now that I've answered your questions, can you answer mine?
"I guess what I'm grappling with now is the relationship between the turns ratio, the input Vdc, and the duty cycle of the gating PWM. And whatever else comes into play that, for larger loads, causes the output voltage to shut down (or drop to a very low value), while for lighter loads it just gets a small ripple that quickly damps and returns to the flat 12VDC. Or maybe the MOSFET has to be sized correctly to match the controller response?"
T3sl4co1l:
--- Quote from: engrguy42 on June 20, 2020, 07:44:27 pm ---Regarding L1, you ask why it's so huge, then say it's about the right value.
--- End quote ---
Yes. In context, as language is wont to do. :) It's much too large for flyback (ideal value 0, typical value maybe 1-5uH as transformer leakage), and still probably too large for forward but in the right ballpark at least.
And again, a flyback this size (100s watts) will have Lp ~ 100s uH, and Ls that much smaller still, so L1 is obviously off magnitude in that case. With values as shown, it's not a big deal, but it's also only good for a few watts.
--- Quote ---I just threw a number in there. And the "K4 L7 L8 1" statement is there, you just can't see it.
--- End quote ---
Oh, great. I suppose there are parasitics on components too? Which LTspice isn't kind enough to display by default? :rant: :rant: :rant:
--- Quote ---Now that I've answered your questions, can you answer mine?
"I guess what I'm grappling with now is the relationship between the turns ratio, the input Vdc, and the duty cycle of the gating PWM. And whatever else comes into play that, for larger loads, causes the output voltage to shut down (or drop to a very low value), while for lighter loads it just gets a small ripple that quickly damps and returns to the flat 12VDC. Or maybe the MOSFET has to be sized correctly to match the controller response?"
--- End quote ---
I would guess the controller is doing a hiccup mode because it thinks it's overloaded. Depends on shunt resistor (not so much on transistor size, and what you've got there is way more than enough for the currents shown) and how the control works.
Damping depends on compensation; or maybe it's switching ripple, depends on what components and time constants you're looking at.
Play with this a bit,
http://schmidt-walter-schaltnetzteile.de/smps_e/smps_e.html
and by all means, draw the waveforms by hand, they are quite simple to plot -- you should be able to see why, for example, less inductance means more power transfer in a flyback converter, or why duty cycle doesn't mean nearly as much in CCM as it does in DCM.
Tim
engrguy42:
Thanks. So riddle me this...
The actual supply provides about 170VDC to the switching section (120VAC, then the bridge). In my case I dropped it down to around 40VDC and it seems to work.
My confusion is regarding how to change this circuit to allow for 170VDC to presumably appear across the switching MOSFET and the switching transformer. And add to that the inductive switching overvoltages...
And I measured the inductance of the actual main transformer (which I posted previously...10-20 mH). So I'm trying to understand what the actual circuit will look like in order to work for this condition, and also allow the rated 10-20 amps to flow on the 12V output rail.
From what little I know of forward vs flyback there doesn't seem to be a big difference in components or configuration, though that comes from somone who is basically clueless.
I'm just having a hard time imagining being able to use anything similar to the circuit I modelled to work with 170VDC input and 12VDC and 20 amps output, using those MOSFETS and the little 3 winding transformer I showed in my first post. I'm guessing it's a 2-switch forward converter like in the schematic I found, but that doesn't answer it for me.
I've got a lot to learn.
T3sl4co1l:
Only thing you're missing is the phasing of the transformer, and another diode from GND to L1. This catch diode keeps L1's current flowing, from GND rather than from the secondary, while the transformer resets its flux.
The transformer also needs something to clamp said reset flux, usually an RCD clamp snubber. Add a diode from the primary to an R||C to +V. The R and C should be relatively large (ballpark 10k, 33nF). Alternately, a diode into a zener (say UF4007 + P6KE200) which will clamp at a stable voltage rather than a voltage varying with PWM.
In short, the primary looks like a boost converter, and you're just burning its output in a load resistor or zener.
Tim
engrguy42:
Thanks. This thing is driving me up a freakin' wall. |O
Last night I modified it per the attached schematic, though without the R/C/Diode transformer reset circuit on the primary. And today I added that and still I'm getting what I've been getting all along. If you look at the attached, it's fine upon startup and hits 12VDC and stays flat, but when I apply a 2.4A, 30watt load the output drops to about 5VDC and stays there.
I've varied the values until I turn purple and I can't understand what would be affecting its ability to recover from a load switching on. Unless there's something about the controller I'm missing (which is clearly the case).
With minor load additions it recovers and damps fine, but with this it just loses it completely and gives up. Maybe it runs out of duty cycle?
Oh, and the output inductor value and N1/N2 values came from that calculator website you recommended. Though I've also varied those about 150 times with little change....
Navigation
[0] Message Index
[#] Next page
[*] Previous page
Go to full version