# High current Voltage Regulator

#### tim90

Joined Jan 15, 2016
20
Hi to everybody, I wanted to ask something regarding a project that I'm doing. I need a power supply for my circuit that must be equal to 3.3V and up to 25A. The main problem is that i am doing an anolog sensitive device in which i need a stable voltage.If I have a voltage difference more the 10mV, my circuit will change completely its behaviour. I'don't need a precise 3.3V but it must remain stable. If i have 3.2 it is good, but only if it remain stable and it is not easy to find a device able to ensure a stable voltage with this high current. Do you know some IC for my applications? my idea was to use a personal computer power supply to get 5V and high current but the i need a voltage regulator for the 3.3v and able to substain this high current, for example the lm 1084 but whit a maximum uotput current 5 times higher. If you have some ideas but instead of 25A i can reach only 8 or 10 or more amps, let me know, because i can start by considering for examples half of the maximum power of my circuit. Thanks a lot for who will answer.

#### #12

Joined Nov 30, 2010
18,190
I beat that quality with a 4 amp supply, but you are in the world where wire diameter and length matter. You need remote voltage sensing to an error amplifier. Look for (4 wire) remote sensing power supply theory.

#### Dodgydave

Joined Jun 22, 2012
8,797
Use a pc power supply, an atx will have a 3.3v rail at 25amp minimum.

#### GopherT

Joined Nov 23, 2012
8,012
All you need for $60. I am not sure how stable it is over surges and such but, other options are way more expensive. By the way, can you tell us what you plan to do with this? Since it is not a common need, you may be heading down a rabbit hole that just won't work out. I am guessing that a more easily implemented option exists. We can help you find one easier than we can test the regulation quality of the 3.3V source. http://www.newegg.com/Product/Product.aspx?Item=N82E16817148048&nm_mc=KNC-GoogleAdwords-PC&cm_mmc=KNC-GoogleAdwords-PC-_-pla-_-Power+Supplies-_-N82E16817148048&gclid=CjwKEAiAzuK0BRCW4tiLpJT-8TISJADV8cw9GITnfSFVrFS4HUIdbK0cg7EFGkcCHXawePyb14Hc1xoCzo7w_wcB&gclsrc=aw.ds #### #12 Joined Nov 30, 2010 18,190 Let me try again. There is no such thing as a 25 amp voltage reference. I could maybe design an analog supply for 25 amps, accurate to 0.3% from zero to full load, but I don't think you can buy them. #### ErnieM Joined Apr 24, 2011 8,018 A 10 mv change for a 25 amp load will happen if the wiring resistance is higher than 0.0004 ohms, total for both supply lines. So either use heavy bus bars or redesign your circuit so it is not so overly sensitive. #### GopherT Joined Nov 23, 2012 8,012 A 10 mv change for a 25 amp load will happen if the wiring resistance is higher than 0.0004 ohms, total for both supply lines. So either use heavy bus bars or redesign your circuit so it is not so overly sensitive. Or use a power supply with, as #12 suggested, a SENSE feature that can attach to the device. #### Lestraveled Joined May 19, 2014 1,946 The IR losses of a low voltage 25 amp power supple have to be considered. You can not get there with big wire alone. On your PCB you will have even higher IR losses. You will need to commit a entire plane to that voltage form and probably specify 2 Oz copper. Voltage sensing will give you a stable voltage at one point and one point only. If you do not map out your IR losses through out your distribution you will have problems. Consider using sub-regulators. #### ErnieM Joined Apr 24, 2011 8,018 Or use a power supply with, as #12 suggested, a SENSE feature that can attach to the device. A sense wire will regulate at a single point (actually a pair of points). I don't know of many 82 watt elements that exist as a pair of points. #### GopherT Joined Nov 23, 2012 8,012 So either use heavy bus bars or redesign your circuit so it is not so overly sensitive. I think this is the solution. #### ScottWang Joined Aug 23, 2012 6,873 All you need for$60. I am not sure how stable it is over surges and such but, other options are way more expensive. By the way, can you tell us what you plan to do with this? Since it is not a common need, you may be heading down a rabbit hole that just won't work out. I am guessing that a more easily implemented option exists. We can help you find one easier than we can test the regulation quality of the 3.3V source.

The computer psu can be allowed the voltage tolerance ±5%, then V=3.3V*5%=165 mV, how can the 165mV used for a stable power needs a 10 mV shifting?

Maybe you have some special skills to make it more stabilize, I only know it needs a special original design and big big capacitors.

#### tim90

Joined Jan 15, 2016
20
The wire resistor is not a problem. I need to compute a current by software and I consider 3.3v minus the voltage drop acros a given resistance, dived by the resistance. So evene if my wire have a given resistance, i will consider it when i compute the current . I don' t know if it is clear.

#### ErnieM

Joined Apr 24, 2011
8,018
No that is not clear, you've omitted what we like to call "the details."

The drop across a resistor divided by the resistance IS the current, so why put the 3.3V into the equation.

Please provide a sketch of just what you have in mind.

#### ScottWang

Joined Aug 23, 2012
6,873
If you want o get help then attach a block diagram and labeling the details.

#### tim90

Joined Jan 15, 2016
20
In the picture you can see only one part of my circuit.I have 16 of thi branches, so 16 r_sense, 16 r_x and 16 current regulator. For each breanch i need 1.5 A at maximum power, so 24A. The resistors r_1,r_2 etc have a value between 0.35 to 0.5 ohm and this value depends on the temperature. For example 0.35 ohm means a temperature of 35°C and 0.5 ohm means 95°C. I need to know the temperature, so i need to measure the value of this resistances. To measure the resistance i use an ADC to measure the voltage dropp across the resistor and then I divide the adc value over the current. My problem start here. I need to know exaclty the current.The current regulator is controlled by the freescale board and regulate the current as a function of another parameter that i can not know. So the only way that I have to know the current is to measure the voltage drop across the resistor r_sense, so (3.3V - Vr_sense) and then i divided this value for r_sense and i will have the current. I am using this technique for only one branches and it works. Now i need a precise 3.3V and high current power supply to compute the current in each breanches. If instead of heaving 3.3V i have 3.25 volt i get an error on the current and so an error on the temperature higher then 10°C. Basically If the 3.25v remain the same, it is ok, because i can measure the current correctly considering ((3.25v-Vr_sense)/r_sense). I hope i was clear now. Sorry for my bad english. I repeat, if you know the possibility of a precise and stable power supply that instead of 24A provide less current (but more then 8 A), let me know, I can try to controllo only some of this branches for now. Thank you very much and sorry for my confusion

#### Attachments

• 31.8 KB Views: 17

#### ErnieM

Joined Apr 24, 2011
8,018
You are making things harder than need be with that arrangement. Put the sense resistor to ground, then the variable resistor above that.

This way you are basing your calculations on a current set of measurements, only assuming the sense resistor holds tolerance.

#### tim90

Joined Jan 15, 2016
20
Ok, i'm really sorry, i had to put from the beggining the entire schematich. In the picture you can see one branche and you can understand why i used that configuration. R_heater is the r_1 of the previous picture. what do you think about this circuit?How could i solve this problem?

#### tim90

Joined Jan 15, 2016
20
The freescale sets the output of the DAC, as the output of the DAC decrease, the current increase, in this way it is possible the regulate the current with a voltage signal. I need R_sense in the top part because it creates the voltage drop to compare the input of the operational amplifier.

#### ErnieM

Joined Apr 24, 2011
8,018
The first thing that comes to mind is to use a "high side current monitor" chip across Rsense. These chips are made for this purpose, they take a high line connected resistor and sense the differential voltage across it, the output that reading (typically amplified so you can use a very small sense resistor) and produce an output referenced to ground.

That eliminates the need to read the supply voltage, as would any differential amp put there. You could use an op amp of an instrument amp but these current sense amps are purpose made just for this.

Either way would then need to reverse the op amp inputs to use that.