voltage follower / adjustable voltage regulator

[L0D|Mr_B]

as the topic suggests, i'm looking for a device which operates on +12-0V rails, and when given and input voltage will provide the same voltage out (i know there will be a little clipping at 12V - not an issue) HOWEVER with a much larger current caperbility (potentially 3A). ive been searching for ages but cant find anything which fits the bill, anyone know of such a device? a 4 pin device is what ive been hunting for like a +V, 0V vin and Vout type layout.

reason - being driven from a 12V D-A with a +5V offset for fan control. will need many of these, and some will power potentially 3 fans in parallel.

B

 

[audioguru]

Hi,
If you make it linear then it's going to get hot and will need a metal tab and a heatsink. When you apply 6V to a 12V/3A fan, it might draw 1.5A. That same current will conduct through your linear control device which will have the remaining 6V across it. Therefore it must dissipate 9W. A small transistor is rated for only a few tenths of a Watt. You need a 3A power transistor that is mounted on a heatsink, and a small driver transistor. A 3A power darlington transistor will do the same thing.
A switched-mode power supply circuit can be arranged to follow the output of your D-A and with its high efficiency may not need a heatsink.

 

[Nigel Goodwin]

From the description given, this isn't a power supply at all (so it's rather misleading!). From what he says it appears he just wants to control the speed of a fan? - presumably a standard brushed DC motor? - as he's thinking of varying the supply to it!.

You don't need the complexities of a switchmode PSU, a simple PWM speed control circuit would be far simpler, and do EXACTLY what is required.

L0D|Mr_B - what are you feeding it from?, and do you have access to that device?, and is it a micro-controller you are programming? - if YES you should be able to provide a suitable PWM signal from the micro on a single pin. If it's a PIC it may have hardware PWM available as well, which would make things really simple! - some other micro's also have hardware PWM.

 

[L0D|Mr_B]

ok.

i'm building a control circuit to control the array of fans in my PC (i have 12 of the buggers - dont ask!). to control this lot i'm using a VB app ive already 80% coded (needs some tweeks which can only be done when hardware is available). device is to be powered from standard molex power pins (0v, 5v and 12v lines available). communication was to be using the parallel port, but using it serially (as i dont want to wire up so many wires), however i'm now thinking about USB, however i have no idea how to code for, or hardware required for USB. if anyone has any linkage, please fell free to post in this thead, even if it going sideways to this original post.

anyway.

i was thinking of putting some inteligence into the device, such as if no contact from the software after 30 seconds the device goes into "full protection" mode - all fans on 100%, but at the end of the day it doesnt need it - a bit overkilled, besides i have no abaility to program the device - writing the code no probem but i dont have time or money to make a programmer aswell

I have read up about PWM, but ive read in places this is not actually that good - can make fans have a "growling noise" and across 12 fans would be a nightmare esp. when playing doom3 LOL

my idea was first to have each fan connected to 2 relays, so i could put 0V, 5V 7V and 12V across each fan respectivley, but i want a little more control, so i want to use D/A'a to convert an 8 bit number to a voltage with a range from +5 to +12 (a 5V offset). my idea was to plug that voltage via some current buffering straight into the fans - hence my previous question.

if i havent answered what i was asked, sorry! atleast i may have cleared up what i'm doing given the above project profile, what you tink i should do. do you think PWM is the best way to go, or can a current buffering device be applicable here?

B

 

[ante]

Mr B.

With 12 fans running at full power I hope you have tied your computer down or it will run away like a hovercraft. :lol: This site has a lot about fan control: http://www.cpemma.co.uk/pwm.html

Ante :roll:

 

[Oznog]

PWM can make audible noise unless filtered, which should be done with an inductor. If you know what the max surge load you have is, you might be able to design a useful RC filter.

It's certainly possible though to just build a reg with an op amp and a power transistor, I think that's what you want. You may need to slow down the amp's feedback path to prevent oscillation. It will dissipate heat in a predictable way, it's 7 watts per amp when going from 12V to 5V. Now you CAN shift the heat away from the sensitive transistor by adding a resistor in series. If you know your circuit never needs more than 1 amp, and your 12V power is regulated and guaranteed not to drop, you can add 7 ohms of resistance if it's a MOSFET transistor or about 6.5 ohms if it's a bipolar transistor. The transistor will dissipate little or no heat at full power, its max heat is at the 0.5 amp point. You would need to select a resistor capable of handling the power dissipation.

 

[Nigel Goodwin]

It will dissipate heat in a predictable way, it's 7 watts per amp when going from 12V to 5V. Now you CAN shift the heat away from the sensitive transistor by adding a resistor in series.

It seems rather silly pumping all this heat into the case, when it's intended to run cooling fans :lol:

However, many computer fans are brushless, and don't accept PWM control well - they may not even be very happy about low power feeds?.

 

[Endolith]

a 4 pin device is what ive been hunting for like a +V, 0V vin and Vout type layout.

I've been wanting something similar, but without the need for high current output. Just a tiny 4-pin FET-input voltage follower for buffering high output impedance bias voltages, to be used as references for other circuitry. I can't find anything like it.

 

[Sceadwian]

12 fans in your PC? There was a post like this about a month ago, there is simply put no rational explanation for having that many fans in a PC. Like Nigel says, if you're linearly controlling the fans you're actually adding heat to the inside of the case not cooling anything, learn about air flow and and plan the inside of the case accordingly. I'd really like to know your reasons for 12 fans, I have 3 in mine. CPU, GPU, and power supply fan, that's it, and all three fans are PWM controlled not linear, they're all brushless too.

 

[crutschow]

I've been wanting something similar, but without the need for high current output. Just a tiny 4-pin FET-input voltage follower for buffering high output impedance bias voltages, to be used as references for other circuitry. I can't find anything like it.

Use a FET input op amp connected as a follower.

 

[Endolith]

Use a FET input op amp connected as a follower.

Those have at least 5 pins.

 

[crutschow]

Those have at least 5 pins.

Darn. I never was very good at math.

 

[Endolith]

I just think it would be useful to have an extremely tiny voltage follower with 4 pins (power, ground, in, out), that's tailored only for a gain of 1, with low noise, low offset, high speed, good supply rejection, high input impedance, etc. The nearest you can get is a 5-pin SOT-23 op-amp, which is compromised by applications other than buffering.

There's probably a good reason why they don't make them, though.

 

[Sceadwian]

You can use a regular opamp as a unity gain voltage follower, and also as any number of thousand of other general purpose uses. The only thing you can do with a unit gain voltage follower would be use it as a unity gain voltage follower. It's one more part you have to stock, it takes up a negliable amount less space and quiet honestly has no real benefit over tieing an opamp to be a voltage follower in the first place. Just because that's the only function you want doesn't mean it's particularly useful to have a chip that does that one function.

 

[Endolith]

It's one more part you have to stock,

After scouring websites to find the right part for an application, I am certain that this is not on the list of priorities for a chip manufacturer. They stock huge numbers of parts, with very small variations between them, even dozens of variations of the same part number, as long as people are buying them in quantity.

it takes up a negliable amount less space and

4 pins could be made to take up about 2/3 as much space as 5 pins, which is not negligible:

**broken link removed**

An array of several of them would take up much less space than the equivalent op-amps, like an analog equivalent of the digital buffer ICs:

**broken link removed**

quiet honestly has no real benefit over tieing an opamp to be a voltage follower in the first place.

Couldn't it be designed without the high-gain stage or feedback? It could be internally simpler than an op-amp and cost a lot less. Something with a single application can be made smaller and cheaper and more easily than something with a multitude of applications.

 

[Sceadwian]

Endolith, such buffer chips should exist. I doubt you're gonna find any 4 pin versions though. Again considering the pint count and the size of the IC there is absolutly no advantage in size. If you need something that specific and pin count and space is such a high requirement you'll need to use an ASIC anyways.

 

[Endolith]

Again considering the pint count and the size of the IC there is absolutly no advantage in size.

I was thinking the main advantage would be cost, actually, giving better performance than an FET buffer with less cost than a full op-amp.

 

[Sceadwian]

One shot/prototype and hobbyist applications don't need to cut costs that much, opamps are cheap enough as it is.

 

[crutschow]

Way back when, National used to make the LM110/210/310 voltage follower, but I believe it's no longer available. Probably didn't sell enough to keep in production.