Voltage regulator in parallel

[tim90]

Hi to everybody. I wanted to ask if the circuit below can works or not. I need a precise 3.3 V and a current of 20A. I wanted to use the pc power supply but it is not so precise and stable. So I thought to use to voltage regulator (of 10A each) in parallel like in the circuit below. Can it works?

The pc power supply has dirìfferent 5V output pin, in the schematic above i used 2 of them.

 

[audioguru]

The output voltages of the power supply ICs will be a little different so one will power the load or be overloaded and the one with a slightly lower voltage will do nothing.
Also your circuit is missing the extremely important input and output capacitors and voltage-setting resistors shown on the datasheet of the LT1038. The datasheet shows how to parallel two LT1038 using two feet of wire as resistors.

 

[zahwi]

Hi to everybody. I wanted to ask if the circuit below can works or not. I need a precise 3.3 V and a current of 20A. I wanted to use the pc power supply but it is not so precise and stable. So I thought to use to voltage regulator (of 10A each) in parallel like in the circuit below. Can it works?View attachment 96882


The pc power supply has dirìfferent 5V output pin, in the schematic above i used 2 of them.

Your setup will work because when one supply will reach its limit the output voltage will drop a little and the other parallel supply will give the rest of the current you need.

 

[MrAl]

Hi,

You can not depend on luck for doing this kind of thing. It's hard to figure out the exact output resistance of each regulator, so one may supply more than the other even though it may not be everything. The only way to be sure is to measure, and that could mean over the entire range of temperatures that might be encountered.

Current sharing resistors are a good idea, but again a couple measurements tells all. If you can turn the load current up gradually that would be a good idea too for testing the setup.

 

[jpanhalt]

You might find this ap note from Linear interesting, as it addresses the issues raised by Audioguru: http://cds.linear.com/docs/en/application-note/an2f.pdf

John

 

[zahwi]

It is not important at all that the supplies share equally. There is nothing wrong with one supply giving 10A and the other one 0A, or one gives 10A and the other 5A.
Power supplies don't care about socialism.

tim90, I've seen this question before, are you the engineer that needs super stable 20A for analogue circuit?
This is only an opinion; 20A stable supply don't exist in real world because the voltage drop on the lead will ruin the stability. You have to isolate the sensitive circuit and supply it with separate power supply, or redesign your circuit.

 

[MrAl]

It is not important at all that the supplies share equally. There is nothing wrong with one supply giving 10A and the other one 0A, or one gives 10A and the other 5A.
Power supplies don't care about socialism.

tim90, I've seen this question before, are you the engineer that needs super stable 20A for analogue circuit?
This is only an opinion; 20A stable supply don't exist in real world because the voltage drop on the lead will ruin the stability. You have to isolate the sensitive circuit and supply it with separate power supply, or redesign your circuit.

Hi there,

Isnt that just a little bit of an over simplification?

You're right about what you said though, that there is nothing wrong with one reg giving 10 amps and the other giving 0 amps, but except for a redundant application we dont put two 10 amp regulators together so that they can supply, well, 10 amps. It would not be right to say that we should try to get 20 amps either, but say at least 18 amps should be reasonable dont you think? Then ideally, we want each one to supply 9 amps, not one 10 amps and the other 8 amps either. That requires some care in the design.

But hey, we are only talking statics here so far and have not even begun to talk about the circuit dynamics. What if one puts out 10 amps and the other 8 amps, and then 100us later they both put out 9 amps, then another 100us later the opposite one puts out 10 amps and the one that was putting out 10 amps now puts out 8 amps? One could go into current limit, then pop out again, then pop back into current limit, etc. That means they oscillate, at least a little. It really depends on their exact internal circuitry and how they start up and react to changing loads and changing line conditions. Can you really say with certainty how all this is going to pan out?

To add to the mystery, every application that involves regulators always has some sort of matching resistances placed somewhere in the circuit to help mitigate the balance in current sharing.
Take a look at the link John posted in post #5 and see what you think.

 

[zahwi]

Hi there,

Isnt that just a little bit of an over simplification?

Hi,
The OP already has a 3.3V 20A PS but he wants linear regulator because he needs better stability for his experiment. He is not developing a commercial power supply.
The LT1038 that he uses can regulate over 14A when the input is 5V.
In theory the chip that gives the higher voltage may give all the 20A, you cannot match 2 chips by any means. In real life and by the specs of the device the output drops at higher temperature, it means that the other chip will come in as soon as the first one gets hot. It is likely that the chips will give similar current regulated by their temperature.
When one chip goes into current limit the output drops and the other chip shares the current.
With the recommended caps by the datasheet there will be no oscillations of any kind.
All that I said is based on my experience using linear regulators. I don't read articles in this subject because I think that my understanding of this subject is ultimate. I know that what that I'm saying makes you laugh, that's ok. I'm about your age and not much time left to repeat learning things, I enjoy more learning new things and that is good enough for me.

 

[jpanhalt]

Jim Williams was a recognized expert in power supplies and voltage regulators. Linear Technology is a leading and well-regarded manufacturer in that field. The fact that the OP posted his question here implies that he is not of equal experience and expertise to Jim Williams and Linear. Unless the OP wants to get into a long development cycle, he would be best advised to follow the lead of experts and not anonymous individuals on a social media site who have no recognized track record in that area. That is not to say the advice of others to use untested designs with unbalanced regulation is wrong, it is just not as sure a bet.

John

 

[audioguru]

I don't read articles in this subject because I think that my understanding of this subject is ultimate.

HEE, HEE, HAW, HAW, HO, HO!

 

[Pommie]

Too young to know that you can never know it all.

Mike.

 

[zahwi]

Jim Williams was a recognized expert in power supplies

John,
Do you think we can get Jim Williams to come on the forum and give the perfect advice to the OP?!

 

[cowboybob]

In this case, it's worth a shot...

 

[jpanhalt]

John,
Do you think we can get Jim Williams to come on the forum and give the perfect advice to the OP?!

Do you know where he is located?

John

Spoiler

Eternal Care

 

[zahwi]

Do you know where he is located?

No, but since his article was written 30 yeas ago he probably needs to be resurrected.

 

[jpanhalt]

Your more recent publication on the subject would be more interesting.

John

 

[MrAl]

Hi,
The OP already has a 3.3V 20A PS but he wants linear regulator because he needs better stability for his experiment. He is not developing a commercial power supply.
The LT1038 that he uses can regulate over 14A when the input is 5V.
In theory the chip that gives the higher voltage may give all the 20A, you cannot match 2 chips by any means. In real life and by the specs of the device the output drops at higher temperature, it means that the other chip will come in as soon as the first one gets hot. It is likely that the chips will give similar current regulated by their temperature.
When one chip goes into current limit the output drops and the other chip shares the current.
With the recommended caps by the datasheet there will be no oscillations of any kind.
All that I said is based on my experience using linear regulators. I don't read articles in this subject because I think that my understanding of this subject is ultimate. I know that what that I'm saying makes you laugh, that's ok. I'm about your age and not much time left to repeat learning things, I enjoy more learning new things and that is good enough for me.

Hi again,

I am happy to hear that you have experience with these regulators, i have not tried one myself yet although i have tried many others over the years. I understand what you say about the linear nature of the output and how it can partly adjust itself sometimes.

My experience was with power supplies that range in power from about 5 milliwatts up to about 30 kilowatts 3 phase synthesized sine wave, as well as line-tied solar array converters up to maybe 5 kilowatts, 90 percent efficiency 1 percent sine distortion. We did a 40v 100 amp DC power supply at least one time that had to parallel two sections to reach to the 100 amp spec. This was many years ago however although some were more recent, like the 5 milliwatt one was for an RS232 power supply that had to run off of the port power itself.

In spite of my 40 years of experience over power supplies, computer programming, etc., i still take what the manufacturer says on their data sheet as my best source of information unless i know of some specific error or oversight, and rarely play around with ideas that might run contrary to the spec's or data sheet recommendations. I know a lot of hobbyists like to experiment though and i dont want to take that way from anyone, but i like to stick to the most trusted sources for the most part so i know i can get something that is going to run for a long time without a problem, even though it is mostly a hobby now for me too. I will sometimes but still very rarely go with the consensus even if it does not match up with known real data or spec's, but that is very very rare for me because i know it adds extra risk just from not knowing what might go wrong. So you could say that i place higher priority on caution than on simplicity.

 

[zahwi]

Hi again,

Hi Al,
You have far more experience than I do in power supplies. I'm quite sure that this will be taken in consideration by the OP.
My experience is with linear regulators not including the LT1038. As I've already said my experience is enough for me.

For me datasheet isn't only the best information, it is the only information. When I work for a customer my design is to the requirements of the customer. When I'm asked my opinion about the OP idea or design I give my opinion.
In this thread the OP asked about an idea and gave a block diagram. To my opinion it was enough to tell him that his idea will work. If he has more questions he can ask.

When you asked me questions I answered them knowing that you are very experienced, I read many of your posts and I know a little about you. My guess is that what you know about me is only my name. I also know that it is not likely that we will ever agree on how to advice the OP or how to design the right circuit. I chose to make you laugh by saying something outrages.

 

[MrAl]

Hi Al,
You have far more experience than I do in power supplies. I'm quite sure that this will be taken in consideration by the OP.
My experience is with linear regulators not including the LT1038. As I've already said my experience is enough for me.

For me datasheet isn't only the best information, it is the only information. When I work for a customer my design is to the requirements of the customer. When I'm asked my opinion about the OP idea or design I give my opinion.
In this thread the OP asked about an idea and gave a block diagram. To my opinion it was enough to tell him that his idea will work. If he has more questions he can ask.

When you asked me questions I answered them knowing that you are very experienced, I read many of your posts and I know a little about you. My guess is that what you know about me is only my name. I also know that it is not likely that we will ever agree on how to advice the OP or how to design the right circuit. I chose to make you laugh by saying something outrages.

Hello again,

Oh i thought you were kidding about that, but i love comedy so anything funny i really like. I watch comedy a lot more than anything else, including some cartoons just for the fun of it
Maybe we could get together and write an episode of South Park about power supplies and and how they take on a life of their own and take over the whole town <chuckle>. They say that in shows like that you can write almost anything because it's fiction and you can do anything in fiction.

 

[spec]

Hi to everybody. I wanted to ask if the circuit below can works or not. I need a precise 3.3 V and a current of 20A. I wanted to use the pc power supply but it is not so precise and stable. So I thought to use to voltage regulator (of 10A each) in parallel like in the circuit below. Can it works?View attachment 96882

The pc power supply has different 5V output pin, in the schematic above i used 2 of them.

Your setup will work because when one supply will reach its limit the output voltage will drop a little and the other parallel supply will give the rest of the current you need.

zahwi's statement is correct and the parallel configuration will work fine.
It is not necessary for both chips to exactly share current, as he quite rightly says.
In view of the currents involved it will be best to have a good layout using thick conductors and a star configuration.
As each chip will be dissipating 17W a substantial heat sink will be required.
Good decoupling will also be required.

http://cds.linear.com/docs/en/datasheet/1038fa.pdf