need pwm controller

The ucc35701 is a nice chip. I imagine he's surprised by its "voltage feedforward" capability and uncertain what to do with it. He could ignore it by setting it to a fixed value, but it's too valuable a capability to ignore. The Vff voltage sets Irt, and is used to instantaneously set the Ct capacitor charge rate, thereby proportionally setting the pulse width without using the feedback loop, which is useful for offline applications (in the longer term it has no effect, since Vrt/Irt cancel each other). A little power-line sag or dip comes along, no problem. Note, older-style non-PFC smps are especially susceptible to the line- dip issue, because they work off the very peak-ety-peak voltage, and respond to any lack of pure power-line sine-wave integrity.

While Yzordderrex could use the rf output as a voltage-feedback term, he may be wise to add a driver output voltage-limit path. If he wants more applications info, he can check out Unitrode's lower-frequency ucc3570 predecessor.

--
 Thanks,
    - Win

How do you plan to implement *voltage mode control* with no error amp ?

Graham

Thanks for replies.

I would use an outboard error amp. Hell, I would settle for a nice triangle generator in 8 pins.

As far as the feedforward pin goes I know what to do with it, and give it the respect it deserves. There is a group of... well, here, A thread I am involved in on another board. I am N9NEO. Guys there build these modulators that are totally open loop and require massive capacitor banks in order to get the 120Hz hum out of carrier. They DO NOT use the feedforward pin. I have already made the suggestion they use that pin.

Ok, so maybe I will take another look at the 35701 or 3570. I have already built a full bridge modulator that does full legal power and fits into a small cigar box. This forward switching converter is going to feed into an analog output section. Just keep a few volts headroom above the bipolar to run it. I plan to run the bipolar output section with enough bandwith to get rid of the 140kHz switching noise from the forward converter.

regards, Bob

John,

No Hum, but they pay big money to eliminate it. The hum would be noticable if there were 120Hz on the bus caps.

They run a big 60Hz transformer into an even bigger bank of caps. Some nice optos to drive the transistors fast in order to try and keep it all linear. Big butterworth filters on the output to get rid of the

85kHz switching garbage. Nice big signals with Hi-Fi audio. Can be found pretty much coast to coast near 3880-3990kHz

I suspect they just like to build these big transmitters reminescent of the old boatanchor days. They are a very nice bunch, but a little set in their ways. I am going to try and give them a modulator which isn't so far removed from the present technology.

regards, Bob

I read in sci.electronics.design that Yzordderrex wrote (in ) about 'need pwm controller', on Fri, 29 Apr 2005:

Good name for a non-humanoid autonomous robot.

How does this hum problem arise? Keeping hum out of audio amplifiers (which I suppose is what the modulator is, in a sense) isn't rocket science, at any power level.

--
Regards, John Woodgate, OOO - Own Opinions Only.
There are two sides to every question, except
\'What is a Moebius strip?\'
http://www.jmwa.demon.co.uk Also see http://www.isce.org.uk

I read in sci.electronics.design that Yzordderrex wrote (in ) about 'need pwm controller', on Fri, 29 Apr 2005:

Have they ever heard of a smoothing choke? Or maybe even better, a choke-input filter?

Low PSRR in the output stage? that can be fixed.

I'm in UK and I don't have an HF receiver any more.

--
Regards, John Woodgate, OOO - Own Opinions Only.
There are two sides to every question, except
'What is a Moebius strip?'
http://www.jmwa.demon.co.uk Also see http://www.isce.org.uk

the uc3843 is 8 pin curent mode controler but its easy to run it in voltage mode, you just feed the triangle waveform from the oscillator into the curent sense pin, its shown how in the datasheet, although if u want curent limit too you would have to combine the two signals.

Colin =^.^=

just wondered why you would be using forward conversion rather than flyback, is it posible the power output of the transmiter can be controled with current as efectivly as with voltage ?

current mode control would provide a hum free output as it is not influenced by line voltage.

obviously you would need to control the maximum output voltage, but this would not be dificult.

Colin =^.^=

Colin,

Going to use forward mode.

Flyback mode trying to get to 1500w pep would be unmanagable. Would need to store too much energy. Forward mode is probably a leap as well. I've already done it the right way (bridge) now I feel this perverse need to do it again. (It's actually the audience that I'm playing to.)

This is not just a power supply. This converter has to control the voltage at an audio rate. Need gain well beyond 5kHz - It's a modulator.

regards, Bob

col> > I'm looking for a very simple pwm controller. I would prefer just 8

cycle

thought

control

flyback,

with

influenced

this

yes i apreciated its a modulator, assuming the transmiter apears resistive modulating the current would efectivly modulate the voltage, so you could feed the input directly into the curent comparator. (with some limit on the peak voltage)

i agree 1500w probably is rather a lot for flyback, would mean a large transformer, although im not sure how feasable it would be to have a seperate inductor for energy storage in paralell with an ordinary transformer.

you can use a bridge with flyback too, just the use diodes in place of diagonaly oposite switches.

It was just an alternative idea, just for thought, as I read your link out of interest and noticed the need for very large capacitors to get rid of hum, but output power with curent mode control is unafected by line voltage so does away with the need for large caps.

Colin =^.^=

The LT1246 can be made to do a forward converter. The datasheet examples are done as a flyback but the basic PWM action is not dependant on the type of converter you are doing.

The "slope compensation" circuit can be copied to make a circuit with normal (non-current mode) operation if you want.

Using an extra part or two you can make it into a "constant off time" or "constant dead time" controller if you want.

It works ok up to about 750KHz and can drive the gate of a largish MOSFET.

Doing 1500W in any single ended topology converter is likely to be trouble. You may want to consider a multiphase design to reduce the size of the inductive elements.

--
--
kensmith@rahul.net   forging knowledge

In article , colin wrote: [...]

Yes, in designs like this, the supply input impedance of the output section is nearly resistive. Its value determined mostly by the load impedance connected to the output stage.

For this reason the OP will want to feedback both the voltage and current to the modulator section. This way the system can be made well behaved for modest mismatches and self defending for larger ones.

Breaking the inductor and transformer action apart makes for more troubles not less. Making 1500W out of, lets say 5 300W switchers would be what I would suggest if I have to do this with flyback designs

--
--
kensmith@rahul.net   forging knowledge

Thanks for the input Ken (and others)

Yes, the LT part could probably do a good job. I like the idea of constant off-time. I did some thumbnail simulations and could probably get the thing to soft switch over most of the modulation envelope. Looks like I would have problems at the bottom of the envelope at the low currents. Probably not a big deal down there anyway.

Multiphase might be trouble coordinating without current mode control, and probably harder to keep resonant. I might like the challenge of the 1500w peak with the one converter. No fun doing things the easy way. Heheh.

thanks Bob

Ken Smith wrote:

cycle

thought

control

examples

the

or

MOSFET.

size