Low power PWM controller, possibly the 555 (flame war coming)

[Klaus Kragelund]

Hi

I am working on a small 100mW forward converter and I need a very low power PWM control.

So the 555 comes up, along with some ultra low power opamps. But, isn't there a PWM chips out there with low operating current. Sofar I found the TPS5110 with sub mA current consumption:

http://www.ti.com/lit/ds/symlink/tps5110.pdf

Anyone used one with lower current?

Thanks

Klaus

 

[Klaus Kragelund]

I know it is an old song, but I am using PICs (PWM output, build in 2 comparators,
calibration in EEPOM) for all that.
Why go looking for chips when I have one in the box that can do almost anything, given some software capabilities.

Well, I have a Cortex M3 running at 4MHz, but the sw guys don't like
me using that in a SW loop.

I could generate a PWM to set the frequency, add F/F to set the
output, and a comparator to reset the output (current mode control).
Then add an voltage error amp to control the peak current level.

Going further, I could use the internal comparator for the reset of
the PWM for peak current control without CPU resources used, but the
comparator has 2-4us propagation delay, seriously lowering the
sensible frequency the converter can run at. I guess the delay is just
a level shift of the peak current level and that is regulated back by
the error amp anyway, so maybe it could work. Would however have a
minimum duty cycle of 20% (for 100kHz and 2us propagation delay)

For that solution I need to add a circuit to monitor the current and
reset the CPU if it hangs. That takes some more parts in the wrong
direction.

But, I would like to see if there is a PWM controller that has low
enough current anyway.

Regards

Klaus

 

[Klaus Kragelund]

That's a buck controller, technically not what you'd use in a forward

converter. Are you actually doing a non-isolated buck?



What are your input and output voltages?

Yes, its a buck, and it has to large max duty cycle. Just an example of a low current PWM.

My input voltage is from about 8V to 20V, output 3.3V. Transformer will probably be 1:1 so I can use off the shelf types approved for EN69650.

Regards

Klaus

 

[[email protected]]

Well, a forward, isolated Buck. Sorry if that was not clear.

 

[Tim Williams]

Do something like this, but configured for forward operation instead:
http://t3sl4.dnsdynamic.net/Images/Deadbug_Sch.png
obviously, replace the amp-gobbling TL431 with something more frugal.

You can save the bias current wasted in the switching transistor by using
a positive feedback winding, making the circuit a basic blocking
oscillator instead.

Tim

 

[josephkk]

Transformer? A buck uses an inductor!

What is the topology?

You clearly missed the isolation requirement implied by EN69650.

?-)

 

[[email protected]]

Well, I originally wrote it was a Forward, and that is commonly known as an isolated buck.....

Regards

Klaus

 

[Fred Bartoli]

[email protected] a écrit :

Well, I originally wrote it was a Forward, and that is commonly known as an isolated buck.....

The question being: why do you want a forward at this low power level?

 

[Klaus Kragelund]

Got a basic schematic of your topology? Is load regulation needed? Is

there isolated feedback?

Very basic schematics here:

www.electronicsdesign.dk/tmp/SED_Forward.pdf

Please note that it is simplified as I am not allowed to post details to the web.

Single switch forward with extra winding reset. Chosen since the input voltage is not steady 20V, but can drop very fast, which would pose a problem with half-bridge solution (capacitor midpoint slewing slowly) and since access to the current is easy (sense resistor)

I need high efficiency, so will probably convert this to ZVS topology lateron.
The FET is controlled by microcontroller PWM. This PWM is setup as high duty cycle (45%), but peak current is detected by the comparator which in turnresets the PWM (input to PWM module) before reaching 45%.

On the output I have classical TL431 regulation with optoisolated feedback to the primary side. (not shown). Larger caps needs to be placed on the output to ease the response time of the regulation loop (microcontroller measured error voltage)

I have a need for an extra voltage on the primary side for internal supply,but have not yet found off-the-shelf magnetics that facilitates this.

Thanks

Klaus

 

[Klaus Kragelund]

That seems awfully complex for a 100 mW converter.



I've done similar things with ISDN line transformers, but much simpler

circuits.

Sounds interesting, how can you make it much simpler? (notice that either the diodes or the FETs are mounted on the output, the FETs are for syncronous rectification)

A flyback converter would be simpler, too.

Yes, that would be simpler, but a flyback must be gapped, so no off-the-shelf magnetics in small sizes are available.

Regards

Klaus

 

[[email protected]]

Sounds interesting, how can you make it much simpler? (notice that eitherthe diodes or the FETs are mounted on the output, the FETs are for syncronous rectification)




Yes, that would be simpler, but a flyback must be gapped, so no off-the-shelf magnetics in small sizes are available.

Regards

Klaus

considered push-pull something like a MAX845 ?

you must have a 3.3V available for the MCU, can you use that?
probably a lot more to choose from at that voltage, 20V is a
bit to high for most of the low power parts aimed at battery supply

-Lasse

 

[[email protected]]

it looks interesting. I have not been able to find any information on this exact topology, but it looks like a LLC converter without the high side switch. Do you have any more information on this?

Regards

Klaus

 

[[email protected]]

Max845 takes up to 5 mA just for the chip and that's all my power budget and more to it.. Nice chip, though, I might find another one with lower current consumption. Anyway, how do they avoid flux walking..... The dead time they have inserted?

I could run at a 3.3V which I o generate, but then I have two converters in series, and that's bad for efficiency. Also, at 3 volt the output diodes means I loose a lot of efficiency.

Regards

Klaus

 

[[email protected]]

Yes, the compound you drew. It's a very nice circuit and avoids the need to drive a high Side FET and has zero voltage switching possibilities. Maybe a circuit added to monitor the ringing and turn on the FET at the right time to lower losses

Regards

Klaus

 

[Tim Williams]

You couldn't find any references to this exact topology because I just
made it up! But I'm sure it's been invented and used somewhere,
somehow, before now.

The tubophiles have been calling it "parafeed" for years. I forget what
RDH4 calls it. It's probably older still.

Tim

 

[Tim Williams]

Except that this is a flyback switcher, not an audio amp.

Except if you overdrive an SE amplifier. Now what?

Tim

 

[Tim Williams]

Now come up with an idea of your own.

Well that was easy. You really are a terrible troll, you know that?

Tim

 

[[email protected]]

Max845 takes up to 5 mA just for the chip and that's all my power budget and more to it.. Nice chip, though, I might find another one with lower current consumption. Anyway, how do they avoid flux walking..... The dead time they have inserted?

I could run at a 3.3V which I o generate, but then I have two converters in series, and that's bad for efficiency. Also, at 3 volt the output diodes means I loose a lot of efficiency.

Regards

Klaus

have you looked at linear? they have lot of sub mA micropower buck/
boost/ext.

or something like http://www.linear.com/product/LT8300 flyback but
they suggest an number of trafos in the data sheet

-Lasse

 

[Klaus Kragelund]

have you looked at linear? they have lot of sub mA micropower buck/

boost/ext.



or something like http://www.linear.com/product/LT8300 flyback but

they suggest an number of trafos in the data sheet

That looks like a very interesting part indeed, would almost match my needs. Only downside is the price, over 2USD in large quantities, but it saves many parts from my schematics

Regards

Klaus

 

[Klaus Kragelund]

Our little ISDN transformers, from Talema, are good for EN60590,

whatever that means. They have 4 windings, 1:1:2:2. That has all sorts

of possibilities. Like make a blocking oscillator flyback converter

(near zero control power needed!) with some simple, low current opto

feedback. A suitable SOT23 mosfet would cost a few cents.

Do you have a part number for that part?

We use royer converters a lot, in designs that needs to be self sustained. Would probably be a good choice here too, thanks for the suggestion.

Regards

Klaus