Looking for advice on selecting voltage regulators for a project

[Positron]

Hi,

I'm a bit of a newbie at electronics (I'm a software guy) so I'm looking for some advice on how to get the voltages I need for a circuit.

I'm building a circuit that needs a :

+5v supply  for digital logic (500ma to 600ma)
+/- 12v supply for audio opamps (50ma - 100ma)

I want the circuit to be powered by a 9v - 16v plug pack (whatever is lying about in the junk box)

I was think of using a LM2575 / LM2595 (or something similar) to get the 5v supply and then feeding that into a MAX743 to get the dual rail 12v supply.

I could also use a 7805 for the 5v supply, but they tend to get pretty hot even with a 9v supply so I'd like to avoid that.
I've used a MAX743 before and had no problems with it, but I've never used the LM2595, or two switchmode regulators together.

Could anyone tell me if this a sensible approach or if there's a better way to do it?

[Rerouter]

seems like a decent design, however it will kinda fall back to what level of noise you can tolerate, at 52Khz on the LM2575, and 150Khz on the 95 you would probably want to filter its output :/

[nctnico]

I'd go for two DC-DC converters. Nowadays they are pretty cheap.

[Kremmen]

+1 for filtering the output.
Does the +/-12V need to be carefully regulated? If not, why not save 1 regulator and take both voltages from a single one? Of couse you need a transformer based topology for that but it might still be cheaper that 2 separate supplies.

[Positron]

+1 for filtering the output.
Does the +/-12V need to be carefully regulated? If not, why not save 1 regulator and take both voltages from a single one? Of couse you need a transformer based topology for that but it might still be cheaper that 2 separate supplies.

I've never worked with mains voltage before, I think it's also illegal where I live.  I'd rather keep it on a DC plugpack.  It does need to be regulated (although I guess you could use a LM317/337 for that.)

I'd go for two DC-DC converters. Nowadays they are pretty cheap.

I had looked at DC-DC converters, you can even get dual rail ones.  Although the only place I could get them is from digikey/mouser that I've found.  I'll have to try to find a local supplier, as using one of those would reduce the parts count a lot.

I might be able to get access to a scope so I can make up some test circuits and play around with filtering.

Thanks for the responses!

[Kremmen]

While connecting to mains is not illegal here that was not what I meant. I meant using a transformer as part of the dc/dc topology so that the secondaries would produce the voltages needed. Of course only one of those can be strictly regulated hence my question.

Sent from my Lumia 800 using Board Express

[Positron]

While connecting to mains is not illegal here that was not what I meant. I meant using a transformer as part of the dc/dc topology so that the secondaries would produce the voltages needed. Of course only one of those can be strictly regulated hence my question.

Sent from my Lumia 800 using Board Express

Oh right, In that case I'm not entirely sure what you mean.

[Kremmen]

Ok, a recent example that i did myself: I applied a FAN7554 controller to drive a flyback "transformer" with multiple secondaries. The primary supply voltage was a nominal 24VDC and there were several secondaries at +8, +/-15, +30 and +5.3 volts. Only the +5.3 volt secondary was fed back to the controller to precisely regulate that voltage. The +8 V was fed into a vanilla 7805 to make a smooth 5 volts to a microprocessor, rest were used for various opamp and IGBT firing circuit supplies. So once you have a transformer in the circuit, you are basically free to have as many secondary voltages as needed from the one switcher. The drawback is that only one secondary voltage can be the feedback to control that one voltage precisely. But then the upside is that good quality opamps have very good supply rejection ratios and small variations are usually insignificant.

[Positron]

Cool, thanks for the info.  I hadn't heard about flyback transformers.  I'll have to do some more reading.

Is the transformer coil something you can buy or do you need to wind it yourself?

[Kremmen]

Cool, thanks for the info.  I hadn't heard about flyback transformers.  I'll have to do some more reading.

Is the transformer coil something you can buy or do you need to wind it yourself?

Often the transformer is a custom part specially made. In case of multiple unregulated secondaries that is usually the case because you need to carefully match the turns to the desired voltages. If you only need one secondary voltage, or if you are in luck, a standard part could work. But read the following before running to buy one.
In case of flyback topology you should understand that the "transformer" isn't a true transformer at all but rather a coupled inductor with capability to store the primary energy in the magnetic field of the inductor. So when the primary side switch conducts and current flows through the primary coil, the "transformer" core magnetizes and stores the energy in the core air gap (either a true air gap or distributed gap in a powdered iron core - more common). When the switch stops conducting the secondary side diode becomes forward biased and the stored energy is released into the secondary side circuit. This is quite unlike a regular transformer that does not store the energy it passes through - the core just provides the flux linkage between primary and secondary windings.
The point of the above longish explanation is that you cannot use a normal transformer as a flyback transformer. It must have an air gap to store the magnetic energy and regular transformers don't have that.

[Positron]

Ah righto.  Interesting.

That sounds a bit complex for what I'm doing (And over my head), so I might give that a miss.  Thanks for the idea though!

[xquercus]

I'm building a circuit that needs a :

+5v supply  for digital logic (500ma to 600ma)
+/- 12v supply for audio opamps (50ma - 100ma)

I want the circuit to be powered by a 9v - 16v plug pack (whatever is lying about in the junk box)

One of the challenges here is that you want to use 9V-16V input and want +/-12V (thats 24V) for your opamps.  This moves the project out of the realm of simply using linear regulators.  A couple thoughts...  First, do you need to supply the opamps with that dual supply configuration?  Can you use a single 12V rail and a virtual ground configuration using a voltage divider?  If so, that might be the way to go.  Just feed your 12V linear regulator with at least 14 V (maybe even less) and you should be all set.  Alternatively, if you really need the dual rail opamp configuration, are you willing to build a power supply?  1A-2A 18V transformers are relatively common inside consumer products and often have a center tap (even if it's unused).  Even if you buy one, it shouldn't be all that expensive (maybe $10).  After rectification, you can feed the approximately +14V you'll have into a 7812 and the -14V into a 7912 (negative voltage regulator in the same family).  Just a thought.

I could also use a 7805 for the 5v supply, but they tend to get pretty hot even with a 9v supply so I'd like to avoid that.

Hot is fine (and unavoidable).  Just not too hot.  Mount a heatsink to the regulator package.  A piece of aluminum will be fine.

-Jeff

[vxp036000]

It seems to me that it would be easiest to use a 24 V wall-wart DC supply and use linear regulators to get 5 V for the digital circuitry.  And why do you need +/- 12 V for a couple op-amps?  Are these power op amps?  Also, you don't really need a regulator for the op-amp, just use large value filtering caps.  If I were designing something like this, I would use a run of the mill power transistor instead of op amps.  Way simpler. 

[Positron]

It seems to me that it would be easiest to use a 24 V wall-wart DC supply and use linear regulators to get 5 V for the digital circuitry.  And why do you need +/- 12 V for a couple op-amps?  Are these power op amps?  Also, you don't really need a regulator for the op-amp, just use large value filtering caps.  If I were designing something like this, I would use a run of the mill power transistor instead of op amps.  Way simpler.

Regulating 24v down to 5v with a linear regulator will generate an insane amount of heat.  Half the point of this project is to play about with switch mode regulators, It's a bit of a learning exercise as I've never used them before.

The power from this circuit is feeding another PCB that requires the +/- 12v voltages for audio DAC and opamps, so I don't want to go outside spec (even tho it may work out)

[T4P]

+/-12 is sufficient for sufficient for something like 22Vpp but that's just a guess because RRIO opamps can go all the way and +/-9 is quite limiting
And no, power opamps go even further.

It seems to me that it would be easiest to use a 24 V wall-wart DC supply and use linear regulators to get 5 V for the digital circuitry.  And why do you need +/- 12 V for a couple op-amps?  Are these power op amps?  Also, you don't really need a regulator for the op-amp, just use large value filtering caps.  If I were designing something like this, I would use a run of the mill power transistor instead of op amps.  Way simpler. 

Tell me how every commercial design has a regulator for the opamp.

[vxp036000]

What's the advantage of a power op amp over a simple single transistor audio amplifier?  Now you don't need to worry about +/- supplies.  It's not like you need such precise control over the gain. 

The reason I question the use of a regulator is two-fold.  First, most any ripple on the supply voltage can be taken out with a simple filtering cap.  Second, a regulator is going to try to maintain a constant voltage output.  That's not necessarily a good thing.  When your op amp gets warm, it draws more current.  With a simple resistive supply, the supply voltage drops, protecting the device.  Instead, with the regulator, the voltage remains the same.  Now the op amp dissipates more power, heats more, draws more current, and we have thermal runaway. 

Next problem.  Suppose your op amp fails with a short.  Your regulator is going to dump the maximum amount of current into the short, causing catastrophic heating and all sorts of other wonderful things.

As far as dropping 24 V to 5 V with a linear supply, it doesn't need to dissipate a lot of power.  What does the digital circuitry draw, a few mA?

[Kremmen]

Sorry but that is a very confused post. This is why:

What's the advantage of a power op amp over a simple single transistor audio amplifier?  Now you don't need to worry about +/- supplies.  It's not like you need such precise control over the gain. 

The answer to this question would be a long one if i went through every reason in detail. Basically the reasons are the same as with any op amp against a single transistor. The very reason op amps exist is to overcome the limitations and shortcomings of a earlier designs.

The reason I question the use of a regulator is two-fold.  First, most any ripple on the supply voltage can be taken out with a simple filtering cap.  Second, a regulator is going to try to maintain a constant voltage output.  That's not necessarily a good thing.  When your op amp gets warm, it draws more current.  With a simple resistive supply, the supply voltage drops, protecting the device.  Instead, with the regulator, the voltage remains the same.  Now the op amp dissipates more power, heats more, draws more current, and we have thermal runaway.

True to some extent. The filter caps will never completely eliminate the ripple, but wit op amps it would not be critical due to their generally high PSRR. Another advantage compared to single transistors by the way.
Maintaining a constant voltage is always a good thing in a DC voltage supply. In fact that is one measure of their quality. The difference in current draw of a cold vs warm op amp is not significant and definitely not a factor in PSU design. That is not a way to provide overcurrent protection. High internal resistance just causes the PSU voltage to sag under load causing extra distortion. Definitely useless. Thermal runaway is a totally different concept, has nothing to do with the PSU as such.

Next problem.  Suppose your op amp fails with a short.  Your regulator is going to dump the maximum amount of current into the short, causing catastrophic heating and all sorts of other wonderful things.

As far as dropping 24 V to 5 V with a linear supply, it doesn't need to dissipate a lot of power.  What does the digital circuitry draw, a few mA?

Well who cares if the amp already shorted. OK kidding but hey, fuses were invented a long time ago. Thats the way you do it, or in case you wish to be sophisticated, use a current foldback PSU.

There are reasons not to regulate a PSU but yours are not really any of them.

[vxp036000]

I still question the justification for op amps/ regulation in an audio amplifier because I have designed and built high performance audio amps without regulators and without op amps.  The only justification for op amps is in circuitry requiring very precise voltage and current control, neither of which is necessary in an audio amplifier.  I would also point out that a well designed common collector amplifier can have a very high PSRR.  This is why I never understood the point of a power op amp in this sort of application.  I might also add that power op amps are generally a bit pricier than a discrete transistor.

My example of thermal runaway is not imaginary, I've seen it happen with op-amps.  Any transistor that gets hotter will have a higher base / gate voltage, further increasing the current draw.

As for the voltage sagging under load, find a decent DC supply and focus on designing the amplifier.  Bias circuitry doesn't need to have built-in regulation if you're using a decent supply.

[Kremmen]

I still question the justification for op amps/ regulation in an audio amplifier because I have designed and built high performance audio amps without regulators and without op amps.  The only justification for op amps is in circuitry requiring very precise voltage and current control, neither of which is necessary in an audio amplifier.  I would also point out that a well designed common collector amplifier can have a very high PSRR.  This is why I never understood the point of a power op amp in this sort of application.  I might also add that power op amps are generally a bit pricier than a discrete transistor.

My example of thermal runaway is not imaginary, I've seen it happen with op-amps.  Any transistor that gets hotter will have a higher base / gate voltage, further increasing the current draw.

As for the voltage sagging under load, find a decent DC supply and focus on designing the amplifier.  Bias circuitry doesn't need to have built-in regulation if you're using a decent supply.

I have no argument with unregulated supplies because careful regulation is not usually needed. I too have built numerous amps over the years and always with unregulated supplies. My reply concerned the specific points you made in your previous which i felt were somewhat inaccurate.
I have built amps from discretes and op amps and haven't really seen big differences in principle. Of course there are crap op amps but then crap discrete circuits dive up as well every now and then. Properly implemented both will work nicely.
Regarding themal runaway - sure, but if we wish to be painfully precise it is the carrier density i.e. conductivity of the silicon that goes up, not necessarily base/gate voltage (why would that happen?). My point was that runaway is an overload or load sharing problem, not so much related to the PSU. I don't know if you can effectively mitigate runaway by using an unregulated supply (from your first post) at least i never seen that or tried it.

[vxp036000]

Hmm, you bring up an interesting point with thermal runaway.  I always assumed that it was caused by resistance in the emitter (or source) increasing, resulting in the base (source) voltage going up, but I'm probably mistaken. 

For a bipolar device, Ic = Is e^(Vbe / Vt) or Ic = u k T ni^2 e^(Vbe / Vt).  Since u is roughly u0 T^(-3/2), we have
Ic = u0 T^(-3/2) k T e^(Vbe / Vt), which simplifies to Ic = u0 T^(-1/2) k e^(Vbe / Vt).  Now, ni^2 is roughly T^3 e^(-Eg / kT), so Ic = u0 T^(5/2) k e^(Vbe / Vt - Eg / kT).  So now it becomes obvious that the collector current goes up with with temperature because the carrier concentration increases.  Yep, I was clearly mistaken in my original understanding.  Thanks for pointing that out :-)

I have no argument with unregulated supplies because careful regulation is not usually needed. I too have built numerous amps over the years and always with unregulated supplies. My reply concerned the specific points you made in your previous which i felt were somewhat inaccurate.
I have built amps from discretes and op amps and haven't really seen big differences in principle. Of course there are crap op amps but then crap discrete circuits dive up as well every now and then. Properly implemented both will work nicely.
Regarding themal runaway - sure, but if we wish to be painfully precise it is the carrier density i.e. conductivity of the silicon that goes up, not necessarily base/gate voltage (why would that happen?). My point was that runaway is an overload or load sharing problem, not so much related to the PSU. I don't know if you can effectively mitigate runaway by using an unregulated supply (from your first post) at least i never seen that or tried it.

[Kremmen]

Well, i didn't proof read your typing (what this board needs is a LaTex renderer), so i assume they are right. But you see, the temperature related fractional power terms are the consequence of carrier density changes. How else would there be a change in current?
The product of carrier concentration, or the law of mass action can be written

n_en_h = ½(K_BT/pi h²)³ (m_em_h)^3/2 e^?e_g/K_BT

?e_g is the band gap.
You can see that the carrier density is proportional to the 3th power of temperature. So the conductivity goes up with temperature. If you have hot spots or unevenly heated devices the hot silicon hogs the current due to higher carrier concentration, other things being equal.

Edit: Oops, this forum doesn't like greek letters. the ? above should be capital delta.

[jahonen]

Well, a quiescent current plot of TI OPA2134:



I don't think that thermal runaway is a problem there. I have never seen an opamp spontaneously doing that, although I have used a lot of them. I think that the chip designers know what they are doing.

±12 volts is perfectly ok voltage for an audio opamp, which are not typically a rail-to-rail ones. That said, I have used MAX743 based switch mode converter to power a headphone amplifier TPA6120A2, works just fine. No problems with noise, although I have another LC-filter in additional of the mandatory ones. There still remains some MHz-range noise, but I don't think that is a severe problem.

Regards,
Janne

[SeanB]

Opamps generally have some thermal protection built in, though it is not meant to be used regularly, it generally is there for only extreme cases. It is not tested during production, and is only tested on sampling.

That said I have run 741's as replacements for OP22's, with a supply of +-22V, and they worked well.

Power IC's though need to be kept well within the absolute max voltage and current limits, and heatsinked well enough that the die does not exceed 70C. Running near the max, or at high temp will eventually let the magic smoke out, and may damage both the equipment and the load.