decent 20W isolated power supply / General Science and Electronics / Forums

Forums

4hv.org :: Forums :: General Science and Electronics

« Previous topic | Next topic »

decent 20W isolated power supply

1 2 next

Move Thread

LAN_403

Ben Solon

Sun May 13 2012, 04:42PM

Registered Member #3900 Joined: Thu May 19 2011, 08:28PM
Location:
Posts: 600

i have been working on a power supply for this-

1336925840 3900 FT0 Iso Gate Drive Uvlo Ovco

-circuit i have come up with. the gate drive circuit itself has under-voltage lockout and over-voltage cutoff(because i am trying to use an unregulated power supply). i plan to drive two of them, one for high side and one for low side. the low side is just because i am overly paranoid of extreme failure resulting in 320v through my low voltage circuit. i may add an optocoupler in place of the gdt, but thats for experimentation to decide.

but back to my problem. i want to design a smps to drive 20W into each driver board, and i am becoming very frustrated. mains flybacks tend to burn, push-pull's have too low an efficiency, and even when i drove my core with the mazzili driver, the secondary circuit could only pull a bit over one watt before the primary was drawing over 5 amps on a 12v circuit.

i could just be awful at building these things, but i would like some help with this... i would like to use push pull, so what is the best way to do this? more accurately how should i drive the switches? tl494? 555? the primary power is coming from either 12 or 24v, and the core is a crt flyback of 4uH/t^2.

thanks

radiotech

Sun May 13 2012, 04:59PM

Registered Member #2463 Joined: Wed Nov 11 2009, 03:49AM
Location:
Posts: 1546

Just be sure your crowbar doesn't take out the transformer, diode or transistor.

Ben Solon

Sun May 13 2012, 05:25PM

Registered Member #3900 Joined: Thu May 19 2011, 08:28PM
Location:
Posts: 600

radiotech wrote ...

Just be sure your crowbar doesn't take out the transformer, diode or transistor.

Yep, it's designed to drain the filter cap, so most of the strain will be on the transistor. And it's only on untill the over voltage falls below the threshold anyways. And the smps will designed to provide voltage below that threshold. So i am not too worried about it.

Wolfram

Sun May 13 2012, 06:01PM

Registered Member #33 Joined: Sat Feb 04 2006, 01:31PM
Location: Norway
Posts: 971

Your overvoltage shutdown senses the voltage that it's controlling, so it will probably work more like a shunt regulator. Or it might possibly oscillate. The undervoltage shutdown has a similar potential problem. When it trips, the UCC will stop drawing current which will make the supply rails rise a bit, which makes it turn on again. Some hysteresis would probably fix this.

I see the output terminals are marked base and emitter. I hope you're not planning on using this circuit for driving a BJT. The UCC is not made for driving BJTs, and without a series resistor you're sure to kill something. It might possibly work with a series resistor, but not very well.

Also, optocouplers are quite slow so keep this in mind if you want to use them.

Ben Solon

Sun May 13 2012, 06:59PM

Registered Member #3900 Joined: Thu May 19 2011, 08:28PM
Location:
Posts: 600

Its actualy meant to act as a shunt regulator, and I don't care if the ovco oscilates. Its only meant to clip peaks, as the tranformer wont normaly put out more than 30v. How do I add hysteresis to this however? On the uvlo I don't see how I can add positive feedback as this is not a comparator. The tl431's use comparators, but you don't have access to the output.

And as for the base, it was a long day, and they are igbts. Close enough. And for optocouplers, it doesn't need to be too fast, and the maximum frequency will be 60khz. And I realize that this is very high for optocouplers, but there are workarounds. But can anyone say anything about the power supply itself? The circuit provided is just to show what it needs to be capable of driving.

Edit: I would like to use the tl494/ka7500C controller, but the noise on that thing is dreadful when prototyping on a breadboard. The output, rather than alternating pulses between the two output transistors, will "skip a beat" randomly and often. And I don't know where it would have to be decoupled to fix this problem. Throwing .1uF caps onto every pin isn't a very clean solution to a rather messy circuit to begin with.

Sulaiman

Mon May 14 2012, 05:06AM

Registered Member #162 Joined: Mon Feb 13 2006, 10:25AM
Location: United Kingdom
Posts: 3140

If you only want to clip the peaks why not just use a TVS ?
They can handle large current 'spikes'

Ben Solon

Mon May 14 2012, 11:47AM

Registered Member #3900 Joined: Thu May 19 2011, 08:28PM
Location:
Posts: 600

Good call. Mainly because I have too many 431's and 139's but it would greatly simplify things. Thanks.

Neet Studio

Thu May 17 2012, 06:04PM

Registered Member #4037 Joined: Fri Jul 29 2011, 03:13PM
Location:
Posts: 86

You are driving a CRT type of flyback which probably designed for "low" tens of kHz flyback frequency. May be something like the good old LT1070 (or variants) which runs at 40kHz typ. would do the job.

Added bonus:
- All in one rugged chip with built-in power transistor and you can always simulate the circuit with LTSpice.
- There is an optional â€œflyback regulation modeâ€ built into the LT1070. By regulating the value of the flyback pulse, the output voltage can be regulated with no direct connection between input and output.

Product page:

AN19 - LT1070 Design Manual:

On their graph in the product page, 12V would just be around 20W for a flyback type.

IMHO, I would try to use a properly designed (current mode) PWM chip instead of crappy 555 circuits. The coarse regulation from flyback pulse would get rid of all your regulation circuits and improve on your efficiency. You are not throw power away running in open loop and shunting/dropping it with linear regulators.

You could always use a proper H bridge gate driver chip for with all the high voltage level shifting circuits built-in.

Ben Solon

Thu May 17 2012, 08:02PM

Registered Member #3900 Joined: Thu May 19 2011, 08:28PM
Location:
Posts: 600

thanks, i have looked at a few ic's that do this, mostly from fairchild though, and they don't carry many of chips like these in through hole packages. i will have to try out this one. my main issue is that i want to keep it as basic as possible. the current sensing is nice, but it can also bite you if you don't have many low ohm power resistors. it just doesn't work with anything over 3 very well.

Neet Studio

Thu May 17 2012, 09:51PM

Registered Member #4037 Joined: Fri Jul 29 2011, 03:13PM
Location:
Posts: 86

Gut feel tells me that the 20W is a peak power, not an average power. (Unless you are driving multiple secondaries from the same primary...)

Your driver is rated for 3W max. Just using the datasheet's 15 ohms typ. value for your driver output resistance high means that at most the average current you could be delivering is no more than 0.4A.

Without knowing what you are driving and what frequency, I can only guess at an order of magnitude here:
P = C*V^2*f assuming you are driving 10nF gate, at 15V at 500kHz
P = 10e-9 * 225 * 500e+3 = 1.125W

If you are driving capacitive load like a MOSFET or IGBT, then essentially all you would need to do to supply that gate charge is dumping charge from a capacitor (say a low ESR ceramic at 10x capacitance of the gate) into the gate.

At about 1W of average power, there are tons of 1-2W isolated power supplies (with 1000V isolation) you could buy for a few dollars.

1 2 next

Moderator(s): Chris Russell, Noelle, Alex, Tesladownunder, Dave Marshall, Dave Billington, Bjørn, Steve Conner, Wolfram, Kizmo, Mads Barnkob