PSU Regulator Question

[mwtheplumber]

Is it possible to use more than one regulator such as 7812 in parallel to increase the current available, always assuming the transformer is man enought to supply it?

 

[Hero999]

This question gets asked over and over.

I've never tried it. I've heard that it can cause stability problems.

The problem is the two regulators won't be exactly the same. One will have a slightly higher output voltage than the other and one will be able to supply more current thant the other. It's possible that each regulator might fight to keep the voltage at its level which could cause oscillation.

There again, it might be alright as when the weakest regulator starts to shut down the more powerful regulator will take the strain.

Why don't you do some experiments with a range of different regulators powering different loads and see what happens?

I'd be greatful if you posted the results here so we can see how it went.

 

[KMoffett]

Not a good idea. I cringe every time someone wants to do this. The purpose of a regulator is to hold the voltage between its output and ground terminals at a fixed level. As Hero999 said, no pair of regulators will have "exactly" the same output voltage. So when the lower voltage regulator shuts down in an attempt to get the voltage it sees at its output lower, you just lost half your current supply. ... and what was the purpose of adding the second regulator?

Putting two voltage regulators in parallel is like putting two constant current regulators in series!

Why not do what the manufacturers recommend, and add an external pass transistor?

https://www.electro-tech-online.com/custompdfs/2008/03/LM7805.pdf .....figures 14 & 15

Ken

 

[mwtheplumber]

Thank you all for your help

 

[audioguru]

Why don't people look at the datasheet for a part???

 

[KMoffett]

'cause there are people out here like you who are too nice and helpful the say "READ THE <$!#*&%> DATASHEET!". And I'm glad of that. I no longer watch the usenet newsgroups because of all the bile and venom.

But...please read the datasheet!

Ken

 

[Roff]

I haven't built and tested this, but the simulation works really well. The basic idea is to use one of the regulators as a master, and then use a diff amp to force the current through the slave regulator to be the same as that of the master. The op amp output forces the ADJ (or GND) pin on the slave to whatever voltage is required to make the currents equal.
If you use a slave with a nominally lower output voltage than the master, you can get away with a single supply. For example, the master could be a 7812, and the slave a 7805. They both wind up dissipating the same power, because the 7805 will have +7V on the GND pin.
If you use two identical regulators, e.g., 7812's, you will need a low-voltage, low-current negative supply for the op amp, because the slave might need to have the GND pin pulled negative by as much as a few hundreds of millivolts.
The op amp needs to have a common-mode range that includes the positive rail. I played around with other amplifiers, including a discrete PNP differential pair, but the op amp is the easier, and performs better.

 

[aussiepoof]

Why not use a switchmode IC which can output 3A or 5A??? One IC, one inductor, one schottky diode and one low-esr cap... so easy!

I've used National Semiconductor's SimpleSwitcher IC's to great effect.

P.

 

[Roff]

Why not use a switchmode IC which can output 3A or 5A??? One IC, one inductor, one schottky diode and one low-esr cap... so easy!

I've used National Semiconductor's SimpleSwitcher IC's to great effect.

P.

Some applications need the clean, no-switching-noise output of a linear regulator.
For whatever reason, this question comes up time and time again.

 

[KMoffett]

And lots of people have lots of 78XX regulators laying around just begging to be used.

Ken

 

[Hero999]

I haven't built and tested this, but the simulation works really well. The basic idea is to use one of the regulators as a master, and then use a diff amp to force the current through the slave regulator to be the same as that of the master. The op amp output forces the ADJ (or GND) pin on the slave to whatever voltage is required to make the currents equal.
If you use a slave with a nominally lower output voltage than the master, you can get away with a single supply. For example, the master could be a 7812, and the slave a 7805. They both wind up dissipating the same power, because the 7805 will have +7V on the GND pin.
If you use two identical regulators, e.g., 7812's, you will need a low-voltage, low-current negative supply for the op amp, because the slave might need to have the GND pin pulled negative by as much as a few hundreds of millivolts.
The op amp needs to have a common-mode range that includes the positive rail. I played around with other amplifiers, including a discrete PNP differential pair, but the op amp is the easier, and performs better.

Whilest I don't see why that wouldn't work, it's much easier to add a pass transistor as suggested on the datasheet. The only benifit I can see is that a pass transistor increases the dropout voltage significanty and your idea doesn't.

 

[Roff]

Whilest I don't see why that wouldn't work, it's much easier to add a pass transistor as suggested on the datasheet. The only benifit I can see is that a pass transistor increases the dropout voltage significanty and your idea doesn't.

If you add short circuit protection for the pass transistor, it raises the dropout voltage by another 0.7V on top of the ~1V added by the pass transistor, and the pass transistor is still not thermally protected, as an extra regulator will be.
With a multiple op amp package, I believe this scheme can be expanded to several current-sharing regulators with their outputs in parallel.
I'm not trying to promote the idea, just pointing out possibilities.

 

[mwtheplumber]

Well I am glad my ignorance has promoted some interesting debate

 

[aussiepoof]

Some applications need the clean, no-switching-noise output of a linear regulator. For whatever reason, this question comes up time and time again.

I understand that... my question was posed to mwtheplumber not as a generic "why oh why do people not do this?" you assumed it to be.

More helpful would have been to list those applications for which a switchmode might be inappropriate so that mwtheplumber might learn more... and make a decision.

P.

 

[Roff]

I understand that... my question was posed to mwtheplumber not as a generic "why oh why do people not do this?" you assumed it to be.

More helpful would have been to list those applications for which a switchmode might be inappropriate so that mwtheplumber might learn more... and make a decision.

P.

Be my guest.