Dual polarity supply -- no center tap

How about a full-wave voltage doubler? Ground one end of the secondary, connect A of one diode and K of the other to the other end. Makes two half-wave rectifiers of opposite polarity, total voltage ~2.5 to 2.8 x VRMS.

Cheers

Phil Hobbs

t -

Phil are you sure you get the doubling effect AND two polarities?

you mean a circuit like this,

But most likely I just misunderstood.

George H.

You mean just like a simple half wave with another rectifier in parallel but reversed?

I never tried that -- didn't even think of it ! -- (or have even seen it in many schematics for that matter). (Of course, I wasn't looking for it, so maybe that's the reason).

I'll be darn -- except for the ripple compared to a bridge, it will do the job just as well?

Is it a good idea to match the rectifiers as far as forward and reverse resistance?

Using an over rated diode in a half wave is always a good idea too, isn't it?

Thanks.

On a sunny day (Wed, 30 Sep 2009 09:50:44 -0700 (PDT)) it happened mkr5000 wrote in :

My bass amp:

I'll be darn -- don't know why I haven't considered that before and for a power amp you don't want a regulated supply.

Is the added ripple a problem?

Or do the caps near the chip take care of it?

Well, technically, you might _want_ a regulated supply, depending on your style. But you don't _need_ them if your amp has good enough power-supply immunity.

Ripple can be a problem if the amp doesn't have good supply immunity; the filter caps are supposed to filter the ripple. There's some formula for calculating their values for x amount of ripple, but I just put in the biggest (i.e., most capacitance) caps that will fit. (and that I can afford. ;-) )

Have Fun! Rich

After looking at that circuit, do you still misunderstand? The _total_ voltage is 2.8 * Vpk, but that's split between the two supplies. And it's only two half-wave rectifiers, back-to-back, so your ripple per current will be way more than a full-wave. (I'd have to look up the numbers, but I'll leave that as an exercise for the student. ;-) )

Cheers! Rich

The OP was talking about a bipolar supply based on a centre tapped transformer and bridge rectifier, I thought, and wondering if it were possible to get +- voltage without it. A full wave doubler (or equivalently two half wave rectifiers) gets you a bipolar supply with one winding.

You still get the doubled voltage difference, the only difference is what you call ground.

Cheers

Phil Hobbs

On a sunny day (Wed, 30 Sep 2009 10:26:16 -0700 (PDT)) it happened mkr5000 wrote in :

No problem, that chip has good supply rejection, typical 70 dB.

The RMS diode current in a capacitor-input power supply (i.e. no inductor before the filter cap) is much higher than the average current because a whole cycle's worth of charge has to be delivered in a narrow pulse near the peak of the AC cycle. So yes, you want to use a diode rated for considerably more than the steady DC current. How much more depends on the size and ESR of the filter cap.

Cheers

Phil Hobbs

BTW, 70dB is not much at all for audio purposes.

There is a problem. The consumption of the audio amp is not symmetrical. There is going to be significant DC current in the transformer, and the rail voltage unbalance as well. I would unrecommend using that kind of voltage doubler for audio power amp. Use the traditional diode bridge with center tap, or the single supply bridged amp topology.

Vladimir Vassilevsky DSP and Mixed Signal Design Consultant

To a first approximation, just double up on the capacitance values to keep the ripple voltage the same. The nice thing about this application is that the currents from the + and - supplies will be about the same, so you get full use out of the xfmr core, unlike typical half-wave applications.

No need.

If it fries, it wasn't rated high enough. Turn-on surge can be pretty brutal.

er

ary,

wo

Yup I understand now. I'd just never seen this configuration for a power supply before.

George H.

On a sunny day (Wed, 30 Sep 2009 13:31:23 -0500) it happened Vladimir Vassilevsky wrote in :

Yea, well, its been working OK since year 2000, and it is actually 75 dB supply rejection (looked it up). I have even used it as driver for piezos at 40 kHz. And run it full power sine at many times. It is my universal amp, use it whenever I need some amp, else it is on the bass speaker. That 'significant DC' is only a little little bit hey:-)

But I will not argue with any audiophiles about stuff like that. Somebody send me an email 'your amp has big hum'. I dunno if they ever build it, but it has no hum. All I could say is: Well theoretically maybe, better check your PCB though. Mine has no hum (that I can hear).

But you can do the math right, 4700 uF, recharged every 40 mS, you know the load, the current, so the ripple, 75 dB attenuated of that, and that is the output ripple. And that ripple is only that big when driving full just below clipping. No way you can hear that in the presence of such a strong signal, as everything else in the room is then already vibrating... as I mounted the bass speaker against the floor, at 20 Hz even the chairs jump up and down at full output. Thank you for your contribution :-)

Font check: iiiii This should mmmmm line up

Two bridges like this will give you both polarities with 120Hz ripple instead of the 60Hz you normally get from a double.

----------+-----!!------ / \\ / \\ / \\ GND / \\ (-V)--+ >! +--+----+ >! +--- (+V) \\ / \\ / \\ / \\ /

----------+----!!-------+

The down side is the extra capacitors it takes.

Watch current levels - this configuration is only good for up to 38% of the rated AC current .

You're sure it isn't 37% or 39%? (It does sort of depend on the copper losses, leakage inductance, core behaviour, and filter cap value, after all.) Where do you get that number?

You always suffer a bit from average vs RMS current in the transformer, but remember you're using both half cycles, exactly like a bridge rectifier. If you're quoting (|I_plus| + |I_minus|)/I_rms, I think it's probably better than that, but if you're quoting just one of the two, I think that's probably a bit optimistic.

Cheers

Phil Hobbs

If you double the capacitance with the two half-wave circuits, you get the same ripple with only 1/4 as many diodes, and the + and - supply voltages will match better. Transformer utilization is the same as well, if the two supplies are loaded about the same, but if the + supply is lightly loaded your approach is probably superior. (Switching the bridges around would suit a lightly loaded - supply.)

Cheers

Phil Hobbs

The ripple in that case is still 60Hz and the two ripples have their peaks out of step. It is the same amplitude however.

If the diodes are in the form of complete bridge packages, The package count comes out good for my version.

I used the other version when I needed to make a lot of +12 and not much -12 in a supply. It worked nicely.