If the transformers primaries are in parallel, and the turns ratios are identical, the secondary voltages will be the same.
We have the Coilcraft flyback transformers in stock, so there's no incentive to wind a custom transformer with multiple secondaries.
-- John Larkin Highland Technology, Inc Science teaches us to doubt. Claude Bernard
m
mA
like
rm,
.no
ion. Since you would need 18 pins on a bobbin, it might be more practical to use two independent flyback converters, each with 4 outputs. There migh t actually be an off-the-shelf transformer that works for this.
ing FET, but it seems like the outputs could vary by a lot. ... ok you pi qued my curiosity, I scribbled out the equations, using V = L dI/dt , and E = 1/2 L I^2, if you vary L by say 20% it looks like the transferred en ergy will also vary by 20% (if I did that right). So for your idea it migh t just work. It might be a good idea to add a DC load at the outputs to so ak up any energy from the transformers' leakage inductance.
I understand you're trying to use parts in stock, I get that. For a flybac k though you have to keep in mind it doesn't follow the transformer equatio n; it transfers energy stored in its magnetizing inductance. E = 1/2 Lma g I^2, and the FET's ON time is adjusted so that Energy in = Energy out. For a given lot of flyback inductors, you can expect Lm to vary by maybe 2
0%. I still think your idea has a good chance of working, especially if yo u add a DC load to each output.
Hmm, interesting. ISTM that only works with sufficiently tight coupling--in the limit of loose coupling the secondary voltages can be anything depending on the loading on each one. Might work fine for this purpose though.
Cheers
Phil Hobbs
-- Dr Philip C D Hobbs Principal Consultant ElectroOptical Innovations LLC / Hobbs ElectroOptics Optics, Electro-optics, Photonics, Analog Electronics Briarcliff Manor NY 10510 http://electrooptical.net http://hobbs-eo.com
Leakage inductances won't affect rectified voltage much, and are a fraction of magnetizing inductance, and the transformers are identical.
-- John Larkin Highland Technology, Inc Science teaches us to doubt. Claude Bernard
The flyback primary inductance is the inductance of all 8 primary windings *in parallel*. They are not independent.
I don't expect leakage inductances to have much effect on output voltage differences.
-- John Larkin Highland Technology, Inc Science teaches us to doubt. Claude Bernard
Yeah, but flyback transformers have pretty loose coupling, 0.8 or thereabouts IIRC. Probably still fine at the +-15% level.
Cheers
Phil Hobbs
-- Dr Philip C D Hobbs Principal Consultant ElectroOptical Innovations LLC / Hobbs ElectroOptics Optics, Electro-optics, Photonics, Analog Electronics Briarcliff Manor NY 10510 http://electrooptical.net http://hobbs-eo.com
25
The coil former has 24 pins, which is enough to accommodate your 16 isolate d outputs and the five or six pins you'd need to drive a centre-tapped prim ary and pull off the base/gate drives.
It's a fairly big core, but coil formers with lots of pins are thin on the ground.
-- Bill Sloman, Sydney
For the Coilcraft Q4436-BL, leakage inductance is spec'd as about 2% of the primary inductance. And I wouldn't expect that to vary much between units.
-- John Larkin Highland Technology, Inc Science teaches us to doubt. Claude Bernard
It will depend a bit on whether your circuit is continuous-incomplete energy transfer, or not.
Pretty simple to spice it out for a first order approximation. A bit of copying/pasting, but don't expect much speed in the sim. Multiple parallel inductors give spice the heartburn.
RL
This is pretty close:
I essentially paralleled all 8 isolated channels. I don't think leakage inductance will matter much. Spice runs better without it.
Loop dynamics depends on a lot of things, so might be tweaked a little more. Or left alone.
-- John Larkin Highland Technology, Inc Science teaches us to doubt. Claude Bernard
You don't need an MCU. Just connect SER IN to VDD, SER CLK to G and have fun.
8*150V*1e-3A=1.2W *Big* fet?Use both channels of IR21531 to drive two out of phase flybacks.
Best regards, Piotr
The inductor used in a flyback is not a transformer. Ideally, both cycles don't overlap. In reality, they do --it is due to the rectifier's finite reverse recovery time and this effect contributes to the losses.
Regulate a 150mW channel? Why would you want to do that? Make it 300mW and burn the surplus power.
Best regards, Piotr
Mostly I used CCFL inverter transformers for that. Cheap, fairly small. Even complete modules can be had for a few dollars:
Or course, with everything going LED and OLED these days it's only a matter of years until the available selection starts to thin out.
-- Regards, Joerg http://www.analogconsultants.com/
Bigger than SOT-23.
We have LTC3803 in stock and they work great.
Here's my circuit so far:
It has ground-side flyback voltage feedback (D3 etc) to make the isolated voltages about 200, so I can regulate them to 150.
Since the Coilcraft transformer has dual secondaries, I guess I could have one transformer power two channels, with a voltage doubler in each. That would be fun. The caps in the doubler could be current limiters, so I could shunt regulate.
-- John Larkin Highland Technology, Inc Science teaches us to doubt. Claude Bernard
$3.33 with free shipping from China. We pay for that.
I won't use ebay or Amazon stuff in my gear. Some of my customers care about part traceability and anti-Chinese stuff.
Some of those things, ebay and Amazon, are OK for breadboards and lab cables and such.
One customer recently elected to buy a bunch of $40 cables from us, when Amazon has them for $6.
-- John Larkin Highland Technology, Inc Science teaches us to doubt. Claude Bernard
I played around with your sim a little bit (the one you posted earlier this morning). Since the datasheet for Q4436-BL states inductance as 22uH +-15 %, I drew all 8 transformers, with 4 of them at 22uH, 2 at 15% low (18.7uH) and 2 at 15% high (25.3uH), and everything else scaled accordingly. The o utputs are dead on until you reduce the coupling factor K; I set it at 0.9 and all the outputs start to drift with an 8M ohm load on each output, up t o 270V. I changed the loads to 1M on each output, and it looks nice; 180V for the lower inductances and 220V for the high inductances. If you want b etter matching, obviously for the one-off unit you can just measure the ind uctances first.
Anyways, that was my Saturday morning exercise. Now gonna write some C cod e (yuck).
This version
uses half as many transformers. It uses voltage doublers in the isolated channels and a doubler in the flyback feedback loop.
The initial swing of the flyback, when the fet turns on, is negative
24 volts, and there's no point in wasting that.Faking in a little leakage inductance doesn't seem to change things.
The fet ON times are awfully short. I need to verify this by breadboard.
I never learned c, which is just as well, since Python is the rage now.
-- John Larkin Highland Technology, Inc Science teaches us to doubt. Claude Bernard
This "efficiency" is so that you can keep R4 warm?
Leakage inductance is mostly to save money I think. if you can store enough enegy in your chosen transformer then you can build a flyback. A gap adds energy storage but also reduces coupling.
Output regulation seems poor, that's possibly the impedance added by the doubler.
-- Jasen.
The entire rackmount beast will use about two watts, and I'm not going to pay the electric bill. It's the voltage swing I didn't want to waste.
That's the reason to have the 150 volt regulators on the highsides.
200, even 300 volts unregulated would be fine.-- John Larkin Highland Technology, Inc Science teaches us to doubt. Claude Bernard
Yup, they'll make the next aircraft carrier :-(
It was just meant as an example. On most projects where I needed a few hundred volts I used the bare CCFL transformers and they cost just a few bucks. Even from US sources they are often produced abroad, sometimes in China.
-- Regards, Joerg http://www.analogconsultants.com/
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.