using two switch mode controlers to get a +- output

[carbon dude oxide]

hello, i ahve been looking around and have been unable to find any SMC that can do negative to negative 3A output and i was wondering would i be able to use 2 buck 3A switch mode controllers to give my +- voltage output?

i have drawn a crude diagram of this, it comes off with 2 isolated taps on a transformer and imagine there is bridge rectifier and all that jazz. but by having 2 switch mode controllers isolated from eachother and then connect them up so that the output of one is the ground of the other this giving a   + GND - output?



i think i may have explained this badly but yea if i ahve please let me know (using the same principle as putting two isolated power supply's in series)

[eman12]

I am wondering isn't a center tapped transformer helpful in this regard?

[carbon dude oxide]

yes it would work but im going to be drawing approximately 3A of current with a significant voltage drop when regulating at lower voltages (im building a variable PSU) and i would like to use switch mode supply as a pre regulator then use a linear regulator so im only dissipating a few watts as opposed to +30W

i would like the suppy to and a + and - supply so in order to pre regulate the - rail i would either need to find a negative to negative smc or find another solution

[TerminalJack505]

So far as I know, because you have two secondaries, you can just have the exact same circuit on both secondaries and then either leave them floating and isolated from each other or connect two of the outputs in any combination to get two positive rails with a common, a positive and a negative rail with a common, or two negative rails (this would be a weird case) with a common.  Just don't try to put them in parallel.

Just treat the two circuits as completely separate and isolated from each other.  Individual transformer secondaries are already isolated from each other.  It's basically like having two different transformers with a similar connection on the primary side.

[carbon dude oxide]

does anyone know a switch mode regulator that can handle 3A (probably 5A just to be on safe side that can have its output voltage controlled by an micro? or so that it tracks about 3V above the linear regulator?

i have found a few that work via a feedback loop which you change the input voltage via a pot but none that i can control from a micro, unless i feed a pos/neg voltage to the other end of the potentiometer to alter in input voltage?

[lgbeno]

Always thought that this was kinda cool.

http://www.ti.com/lit/an/slva369a/slva369a.pdf

[carbon dude oxide]

Ive got this design using a digital pot would this design work to allow me to control the voltage output of the smps?

[Paul Price]

No.

[carbon dude oxide]

why not?

[johansen]

4 phase and above motherboard buck controllers have a digital input for the output voltage, however the feedback network remains analog for most. good luck messing with those parts though, ball grid arrays on .5mm spacing if you're lucky they are that big., i have seen DAC's inside pwm controllers but i can't think of any that aren't surface mount.

If you want a two quadrant power supply the best thing to do is build a full bridge converter, powered from an isolated supply.

The "stiffness" of the power supply will be related to the resistances of the components you use, but 50 amps per volt isn't out of the question.

The way you could easily do this is feed a 0-95% duty cycle into one half bridge, and 95-0% duty cycle into the other. you then filter both half bridges like you would a synchronous buck, and you take the current from each one.
With 60 volts mosfets surface mounted (at under 10 amps they won't produce much heat) you should easily be able to get 25 amps per volt of "stiffness" from a 50 volt regulated power supply. (it is difficult to achieve regulation across 4 quadrants, so don't even bother)
You then change the voltage by changing the duty cycle going into each half bridge, zero output is 50%, 50%

regulation of current is much more difficult, because you have to know which direction the current is flowing to reduce the duty cycle.
i'm sure there's a very creative way to achieve current limiting in two quadrants, four is much more difficult.

[TerminalJack505]

Ive got this design using a digital pot would this design work to allow me to control the voltage output of the smps?

I've used a digital pot with a Simple Switcher before.  It worked.  I used a cheaper 5V digital pot and put a big resistor between the feedback and the pot to ensure the pot never saw a voltage larger than 5V.  I don't know if you can do this in your case or not but if you can then you can save some money on the pot.

You should also be able to do what Dave did with his power supply and use an op amp to drive the feedback pin.  Use a pair of standard resistors to feedback a scaled-down voltage to the inverting input and feed a filtered PWM voltage to the non-inverting input.  Take care of necessary compensation and whatnot and the op amp should servo the feedback pin to the appropriate voltage to get the desired output voltage.

The output voltage would be determined by the PWM signal and the two feedback resistors.  For example, if your feedback resistors scale the output voltage down to 20% the output would be 5 * the PWM voltage.  So, in this case, your 5V MCU can control a 0V to 25V output.

I haven't done this with a Simple Switcher so caveat emptor.

[carbon dude oxide]

The switcher that i currently have requires that 1.21V should be going into the pin when the output voltage is the correct regulated one.

How would i use an op amp to output 1.21V when the output voltage is correct compaired to my refferance voltage from my avr?

[TerminalJack505]

I don't have a schematic for this particular case drawn-up but it would look something like the attached circuit.

That's for an LM317 so it won't be quite the same.  The 10k resistor may not be necessary between the output pin and the adjust pin, for example.  (It depends on how the reference in the regulator works.)

But notice how the MCU is out of the control loop.  The MCU gives the op amp a reference voltage and the op amp does the control.

[carbon dude oxide]

Ok i see, so you filtered pwm or dac output is put into the op amp as a reffereance voltage which changes the voltage on the end of the feedback potential dividor loop thus ajusting the reffereance voltage. I shall have a look in the morning as it is quite late now (well early)

[carbon dude oxide]

ok, im gna give this curcuit a go on LT spice (or attempt to )

[carbon dude oxide]

ok first circuit on lt spice and it worked, it gave the 2.Vref output like it says how would i go about changing the lm317 with my switching circuit?

[Paul Price]

I built this one a few years ago.. Dual track/independent output PWM control by MCU. With 3-amp Radio-Shack 24V transformers, and suitable heatsinks it delivers 0 to 24V at 1.5 amp both positive and negative outputs with short circuit protection.

The whole output transistor/current limit circuitry for the negative supply output of this supply could be replaced with a negative power supply equiv. for the LM317, but I had a lot of spare parts on hand but I didn't want to buy a LM337.

[TerminalJack505]

ok first circuit on lt spice and it worked, it gave the 2.Vref output like it says how would i go about changing the lm317 with my switching circuit?

I think the circuit would be similar for the Simple Switcher but you'd likely get rid of the resistor between the output and the adjust pin (the feedback pin of the simple switcher.)  The LM317 doesn't have a ground pin so its internal reference is relative to the output voltage.  The simple switcher does have a ground pin so I'm guessing it's internal is relative to ground.  Hence, no need for that resistor.

If you have a Simple Switcher then I'd first get it working with just a pot.  Switchers with an adjustable output can be tricky.  Especially with a wide output range.  Be sure to put a decent size load on the output.  Keep capacitance on the output the minimum required by the switcher's datasheet.

Once you have that working I'd replace the pot with the op amp and feedback network.  The op amp doesn't have to be anything special.  Any jellybean part should work.  I used a TL081. 

I'd start with larger values for the compensation cap and the resistor going to the feedback pin.  Those (and the two 10k resistors) control the bandwidth of the feedback structure.  You'll get less bandwidth with large values but you'll also be less likely to have oscillation.

The two 10k resistors control the gain.  If you need a different output range then adjust these accordingly.  Keep the total value of these to around 10k to 100k.

I came to the values for the LM317 circuit using a network analyzer.  If you have one of those it will save you a lot of time.  Otherwise you'll have to tweak the values and observe the output and watch for any instability.  If you have a signal generator then you can feed a sine wave into the V_ref node to sweep through the entire range of output voltages.  If you see any fuzziness or distortion (use an oscilloscope, obviously) then your compensation network needs some tweaking.  Remember that you will need a negative power supply if you expect to get to voltages lower than the internal reference (1V21.)

[carbon dude oxide]

yea i need to get myself an oscilloscope shoudent be too long now a rigol ds1052 should be fine for this right?

[TerminalJack505]

Yes, a DS1052 would be fine. 

By the way, the network analyzer that I used was the one on the Analog Discovery.  It's a great little tool for the money.

[carbon dude oxide]

im oing to ahve to wait a month or two before i can get one unfortunately :/ i may ahve to put this project on hold untill then

[TerminalJack505]

Yep, I'd put it off until you get a scope.  This definitely isn't a ready-bake sort of project.