Why Do We Never See Voltage Regulators or OP Amps Used to Control Ripple in Flybacks?

Thread Starter

JackieTee

Joined May 28, 2019
11
I've been doing lots of research on designing a flyback converter for work (no, I cannot purchase an off-the-shelf one). In all of my research, it looks like nearly all design involves snubber/clamper circuits around the switch to reduce ringing followed by a carefully selected capacitor on the output to control Vout ripple. Why not put a voltage regulator on to reduce voltage ripple to 0? The regulator (in my mind) would be added on after the capacitor to help eliminate the ripple - not replace it all together. None of the texts or articles I read mention anything about including an OP Amp or Voltage Regulator and I am starting to figure there must be some important concept I am overlooking here.
 

Papabravo

Joined Feb 24, 2006
14,412
I'm guessing you are talking about using a LINEAR voltage regulator. If so, the reason for not using one is most likely power dissipation. The equation for that power dissipation is:

\((V_{IN}\;-\;V_{OUT})I_{OUT}\)

Now clearly the larger the voltage difference, the more each milliamp of current draw hurts you. The reason for using any SMPS is to efficiently handle large currents. Throwing a 50% efficient linear regulator on the output would be like loading a high capacity clip in your 9mm and shooting yourself in the foot.
 

mvas

Joined Jun 19, 2017
537
You can add a secondary Voltage Regulator, to reduce the ripple.
The voltage difference would be minimal = low power loss.
 

BobTPH

Joined Jun 5, 2013
2,541
I believe some of the better bench an lab supplies do just that. They use an SMPS to get the voltage near the target voltage followed by a linear regulator to smooth it further.

Edit: You can get a very good compromise between efficiency and clean output by using that combination. Consider a 0-30V 10A supply. If you feed it to a linear regulator, and you want 1V at 10A you are going to waste about 97% of the power. If you step down the, say, 32V input to 1.5V then feed it through a low dropout linear regulator you are wasting only 33%.

Bob
 

Papabravo

Joined Feb 24, 2006
14,412
I believe some of the better bench an lab supplies do just that. They use an SMPS to get the voltage near the target voltage followed by a linear regulator to smooth it further.

Edit: You can get a very good compromise between efficiency and clean output by using that combination. Consider a 0-30V 10A supply. If you feed it to a linear regulator, and you want 1V at 10A you are going to waste about 97% of the power. If you step down the, say, 32V input to 1.5V then feed it through a low dropout linear regulator you are wasting only 33%.

Bob
Everything is about compromise. A bench supply delivering 10A probably cannot tolerate much dropout voltage so I'd be surprised if that was the preferred solution. We have one poster saying he never sees this and another saying it could happen. They can't both be correct.
 

DickCappels

Joined Aug 21, 2008
6,590
Back in the old times when transistors were made of wood, I designed a 400 W steering magnet power supply for a particle accelerator. The specifications called for 0.01% random and periodic deviation over 24 hours - a tough specification in those days.

The solution was an SCR pre-regulator followed by am opamp driving a class A-B output stage. After going through a few hundred dollars worth of transistors it worked beautifully.
 

WBahn

Joined Mar 31, 2012
26,148
I've been doing lots of research on designing a flyback converter for work (no, I cannot purchase an off-the-shelf one). In all of my research, it looks like nearly all design involves snubber/clamper circuits around the switch to reduce ringing followed by a carefully selected capacitor on the output to control Vout ripple. Why not put a voltage regulator on to reduce voltage ripple to 0? The regulator (in my mind) would be added on after the capacitor to help eliminate the ripple - not replace it all together. None of the texts or articles I read mention anything about including an OP Amp or Voltage Regulator and I am starting to figure there must be some important concept I am overlooking here.
My guess is that commercial supplies don't (at least commonly) do that because people that are too sensitive to noise and ripple use linear supplies and most of the rest can live with the output of the switcher with some kind of custom post-conditioning if necessary (and the post conditioning can be a linear regulator). Another factor would be that adding the linear to the back end of the supply goes a long way to defeating some of the key benefits of the switcher in terms of efficiency, size, cost, and power dissipation. So the market for a hybrid supply isn't great enough for many manufacturers to produce it.

But using a switcher to get close the desired voltage and then following it with a linear (or even two) is a pretty common way to implement a reasonably low-noise power source.

Some of the chips I worked on were too sensitive even for this but eventually we had to deal with it. We did so by putting the switching controller onto the chip itself so that the switcher operated synchronously with the analog sections such that the random noise from a normal switcher was converted into fixed-pattern noise that we could calibrate for and almost completely cancel out.
 

BobTPH

Joined Jun 5, 2013
2,541
The other advantage of a hybrid supply is that, like an SMPS, it can provide near full power over the entire voltage range. A linear supply can only supply a max current no matter what the voltage is, so it reaches full power only at the max voltage. The 0-30V 10A supply I referred to is probably rated as a 300W power supply, but, if linear, it can only deliver 10W at 1V.

Bob
 

AnalogKid

Joined Aug 1, 2013
8,596
Why not put a voltage regulator on to reduce voltage ripple to 0?
Some do. Hewlett Packard and others used to (maybe still) make a high power rack mount lab supply that is a switching power stage followed by a tracking linear regulator output stage. Voltage drop across the linear part is held to a constant 2 or 3 volts. Basically it is two regulators in series, one adjustable to x volts and the other adjustable to (x-3) volts. At 60 V and 50 A that's still a lot of heat in the linear stage, but nothing like an all-linear approach.

ak
 
Top