MC34063 vs TL494 (as buck converters)

[AL3xGR]

Hello

These days I'm planning to make an adjustable switching PSU.(20V max - 20-25A)
It will use a step-down transformer and then a buck converter.The switching controller will be a TL494 or a MC34063,both with external FET.

Which of these ICs do you suggest? I read somewhere that bigger FETs have higher gate capacitance and MC34063 cannot drive them ,is this true? In this case,a FET driver wouldn't solve the problem? (I prefer MC34063 since it's easier to use and datasheet has all required formulas for part calculation.) Also,it is better to use a P-Channel MOSFET or N-channel ??

[c4757p]

1) MC34063 is easier to use because it's voltage mode. The structure of the TL494 suggests it could be easily used this way, even if it's often not. (I don't really care for current-mode SMPS generally - too much work to stabilize and not really worth it for the advantages IMHO, unless you're making an actual product and not just a one-off.)

2) You're right, MC34063 won't be happy driving a huge FET. You can, of course, build an external FET driver with just a few discrete components. No real need to use a commercial one (rather pricey in low quantities IIRC).

3) Most of the equations in the MC34063 datasheet are applicable to the general converter topologies, not just MC34063.

FWIW, MC34063 can be replaced with an LM393 and a handful of external parts. Which controller you use isn't really a big deal, they all do very simple things.

I'd use TL494.

[AL3xGR]

Ok.I will possibly use MC34063 with a driver circuit.LM393 needs more compoments (oscillator,etc.) so I wouldn't use it.
Thanks for helping.

[c4757p]

LM393 needs more compoments (oscillator,etc.) so I wouldn't use it.

I wasn't seriously suggesting that, just pointing out how simple these are.

And you'd use one half of the 393 as the oscillator, the other half as the reference comparator. I've done it - I needed a negative in, negative out buck converter, which most controllers out there are obviously not suited for.

I've also done it to save money when I needed a bunch of separate converters. They can all run off the same oscillator, and when all you need is one external voltage reference, half an LM393 for the whole system plus half an LM393 for each converter, you do end up with a lower overall cost in small quantity.

[AL3xGR]

I was reading about the oscillator part in the datesheet right now.For a part that is not easy to find it seems ideal.

I should use N-Channel MOSFET as switch? Or P-Channel?

[c4757p]

For a buck? P-channel, unless you're doing something unusual. N-channel is cheaper, but you'll need to drive the gate well above the source, which means a gate driver with integrated charge pump. Not usually the most cost-effective solution unless you're making a shitload of these and get them cheap in quantity.

[AL3xGR]

P-Channel... It will be only for 3-4 units.

[free_electron]

neither one of these.

why do people stay frozen in time with 30 year old crappy ic's that have no intelligence nor protection ? the 34063 and tl494 are dinosaurs that should have died years ago. in terms of capability and performance they are the snot of the ic world.

take  look at a modern smps contrller from maxim or linear. these things measure current through the coil , protect themsleves, have synchronous rectification, switch much faster  and have better performance. they adapt switch frequency , can do pulse skipping mode for light loads and mouch much more. the tl494 is a glorified opamp and comparator and flipflop in 1 package and the 34063 is roughly the same.

please widlarize that tl494 or mc34063 and bury its powdered remains quitly in your garden... its 2013 .time to get with the flow...

[c4757p]

the tl494 is a glorified opamp and comparator and flipflop in 1 package and the 34063 is roughly the same.

That's because that's all you need... Of course the newer chips are much more advanced, but sometimes you really don't need that. MC34063 has been quite sufficient for many applications for a long time.

[c4757p]

At that output current

Whoa... er, I retract my previous suggestion. Somehow I missed "20-25A". I mean, I realized it was a high current, but that's a good bit more than what I was imagining. 

Yeah, I don't think I'd recommend either of those for this particular application.

[AL3xGR]

I guess I have to look for another IC..right?

[AL3xGR]

At that output current you definitely want synchronous rectification. At a low output voltage you would have high conduction time in the rectifier so losses at 25A output would start hitting 20W or so with even a Schottky.

The rectifier will be placed before the buck conveter , like figure 21 in TL494 schematic.
(I think you mean something else)

[c4757p]

Synchronous rectification is where you use a MOSFET instead of a diode as the rectifier, switching it externally to eliminate diode loss. And... I'm not sure how the rectifier can be before the buck converter? It is part of the buck converter!

Of course, there are different TL494 datasheets. The only one I found with a Figure 21 is the Motorola one, and Fig 21 is titled "Pulse Width Modulated Step–Down Converter". Is that the one?

[free_electron]

At that output current

Whoa... er, I retract my previous suggestion. Somehow I missed "20-25A". I mean, I realized it was a high current, but that's a good bit more than what I was imagining. 

Yeah, I don't think I'd recommend either of those for this particular application.

see what happens if you dis my suggestions? 

for that kind of stuff you want some chip that has some brains. sync rectification, current control , saturation detection etc etc etc... and you DONt want hysteretic for such applications . your post regulator is going to go crazy !

[gxti]

Actually, let's do talk about the input rectifier. OP says "a step-down transformer and then a buck converter", but 25A is quite a lot for a diode and a big electrolytic cap at 50/60Hz, no? High ripple isn't a deal-breaker if it's being immediately stepped down anyway, but is there an inexpensive solution for rectifying low voltage 50/60Hz more efficiently?

[Mar]

i think TL494 is better for high power applications than MC34063 ... i am not sure but as i remember 494 do the PWM driving with more advanced  options than MC ... and if u had an old ATX .. or even new cheap one .. modified it .. it has all u need TL949 > LM393 > p and N channel fet > shotkky >> caps >> protections >> etc ...

[AL3xGR]

Actually the first idea was a circuit like figure 21 of TL494 datasheet (Pulse width modulated step-down converter).
I would use a transformer, then a rectifier (which I could replace with a synchronous rectifier) and then supply the buck converter circuit with a DC voltage.I know that I need a very big transformer but I wanted to avoid high voltage switching and EMI (I read that isolated design is known to produce a lot and feed it back to the mains - This needs extra parts for filtering).And the transformer wasn't really a problem for me if I could find a core - I was planning to make the whole transformer myself.

But from what I see , an isolated step-down converter could be better,so I leave the first plan.
Do you have any schematics to suggest? ( Or ICs that can be found easily)

@ AcHmed99 Could you send me a schematic or the IC part # you are using? (If you don't mind of course).

[AL3xGR]

It's not the first time I'm dealing with high voltages. I have done some projects that work on AC mains.
I know how a PSU works too but when I saw the schematic in the datasheet I though this was possible too and that's why I asked for opinions.I have also seen someone's PSU with 40+ amps output based on TL494 (isolated though).

Of course I should have posted some schematics earlier...

[M. András]

the voltcraft laboratory psu-s vsp 2410series uses tl494 as control ics and an lm 324 opamp nothing else except a voltage ref ic 40v /10A output. it does have transistors to amplify the tl494 drive current so. you can search in the archives i posted photos of the thing a long time ago

[M. András]

the voltcraft laboratory psu-s vsp 2410series uses tl494 as control ics and an lm 324 opamp nothing else except a voltage ref ic 40v /10A output. it does have transistors to amplify the tl494 drive current so. you can search in the archives i posted photos of the thing a long time ago

Yes but he wants 20-25A not 10A there is a big difference. If the 494 is actually being used as a controller for a variable output 400W buck that is likely why they have a 10A current limit on it. If they didn’t then losses in the buck rectifier would get what I would consider unacceptable 20-30W. So you can add over twice the aluminum and or fans as your power supply has for cooling the rectifier, or you can add a half-bridge driver, or you can rig something together to drive the sync fet with delay to avoid shoot through or just get an IC that has all that.

I’m not sure if the 494 is used as a controller for a variable output buck or it’s being used to control a variable output forward converter in your supply.

That seems to be a pretty versatile supply for the price. Have you had any reliability issues with it like the poster in your thread did.

about the noise dunno i have no scope unfortunatly but apart from that its fine but for a bit less i would rather get now a tti 400 watt psu

[AL3xGR]

Ok. I will search for another controller soon. (At least now I know what I need )

(AC voltage in my country is 220V)

[Paul Price]

Consider this:  A typical available 450W PC( no cost, free, pulled from a discarded computer) switching power supply is a high-current  efficient power supply that could easily supply the 8 to 20 amps you want at >16V regulated output.

It is entirely feasible to control the whole step down circuitry of the switcher through the single optoisolator that is fed by a single TI431 (can be adjustable) shunt regulator that is built into these power supplies. You can discard all the other 3.3V and 5V  low-voltage circuitry and wiring except for the 12V output's built-in rectifiers and filter caps and the PWM drive circuitry that controls the HV half bridge transistors on the AC mains side of the optoisolator.
 
Output voltage current and voltage control can be accomplished by feeding your control signal to the single optoisolator or use another  optoisolator to control the TL431 shunt regulator voltage feeding the LED input of the optoisolator.  You can easily hijack full control of the power supply at this point of the feedback loop. The result: you have a simple, highly efficient, high current, adjustable and well regulated power supply.

In addition, many of the older supplies use TL431 or MC34063 controllers that you pull and get for free. You might not even need them to get the power supply you've described, but but might want to use them to create  other voltages or constant current source outputs that you might want to add.

Series connect or parallel the DC output of two (to maybe 4 free switchers) and you have 3 to 24V at >20amps and the need to experience the pain to wind your own inefficient stepdown transformer is eliminated. You can control and match the output of each supply with optoisolators.
.
The voltage output control could be as simple as a single potentiometer.  I use a PIC microcontroller to PWM the raw output to convert down to whatever regulated current and voltage regulated output I desire and the MCU's A2D conversion time is fast enough to monitor and control the current and voltages of the PC supply, as well as monitor the whole system and give an accurate 7-segment readout of voltage and current and respond to push-button switches to otherwise control functions. In my supply I use the raw output of the PC supply to feed sequentially pulsed 4 PWM paralled buck converters so as to competently control the whole system's current and voltage output and even the fan. The MCU's does this with precision and achieves reliable operation at high current levels.

I just use two potentiometers, one each for voltage and current control. The pot outputs  feed into A2D inputs on the MCU to set the desired levels to control my added PWM buck converters and the PC supplies. I used the salvaged toroid cores from discarded PC power supplies to transform the 12V switcher's output to the rectifiers to create a 24V-30V rectified output. I used other cores salvaged to create four 6-amp PWM post PC buck convertors. I did had to buy some new 35V and 50V capacitors to be able to handle these higher voltages. Also I had to buy some power N-chan MOSFETs. The heatsinks and fan I got for free from the PC swithers. I found a free defunct UPS supply and used its cabinet to house the whole circus.

The MCUs controls the high current PWM buck convertors by using the MCU PWM outputs that feed into integration capacitors at the comparator inputs to set the comparator threshold voltages needed achieve very fast feedback control. The raw PC output voltage is also set by this method at the PC supply's optoisolator. The LM393 and LM339 comparators were also free from discarded power supplies, although they are cheap enough to buy new.

If you are feeling ambitious after you've learned you can forget about trying to wind your own giant paper weight transformer., you can easily add a .005 ohm current sense resistor to the power supply's negative return, add a fast opamp to raise the current sense voltage to a voltage sufficient for feedback to the optoisolator control loop and add diodes to switch between constant current/current limited/constant voltage mode...bingo, you have a 20Amp capable constant voltage/constant current/current limited adjustable 3-24V power supply.

[AL3xGR]

This could be an alternative,but I can't find discarded parts here...