comment on this simple psu circuit

[jarthel]

I do not have visio installed yet (or any graphics software) so I'm hoping you'll understand the ascii diagram


===== the circuit ===============

12V 3.12A transformer => bridged diodes => .22ohm 5W resistor => 10mF 16V => .22ohm 5W => 10mF 16V => LM1084 => 10mF 16V

=====================

LM1084 is similar to LM317 except that it has a lower voltage dropout and has a max current rating of 5A.

I have added the .22ohm resistors to drop some voltage to reduce the voltage that will bed dropped in the LM1084. It should also allow me to use a smaller heatsink. I will increase/decrease this .22ohm resistor as required.

The R2 (the one connected to the adj pin and to ground) in the LM1084 will be bypassed by a 47uF cap.

this circuit will drive an 8V 2.55A load.

thank you.

ps. please assume that the necessary resistors are connected to the LM1084 to produce the desired 8V.

[N-Channel]

Jarthel-

What is your load? At 12V (X1.41) you will have ~17V at the first filter cap, and 16.5V at the second one, and across your LM1084, you will have 7.84V @ 2.55A for a power dissipation of 20.01W.

Basically, you have an R-C-R-C filter before the regulator, which is good for low-pass filtering. BTW, are your caps 10mF or 10mF? Or 1mF (1000mF)? This will, of course, affect the cut-off frequency of your LP filter.

Have you considered a SimpleSwitcher IC, such as the LM2576 (3A) or MC33167 (5A)? Both are extremely easy to implement and would offer considerably better than the current 54% efficiency. This is just a thought, but with proper filtering and shielding, you can realize a high quality power supply without all the heat dissipation and thermal issues associated with a linear power supply. In their simplest form, they would require one diode, one coil, and two caps (input & output). Variable versions use the same voltage divider technique as the LM1084.

Of course, if noise is of paramount consideration, then a linear will most likely be the way to go. Anywho, if you're interested, let us know, and we can post some links to datasheets on many of the switching regulators out there.

Cheers,

Steve

[jarthel]

Originally posted by N-Channel
Jarthel-

What is your load? At 12V (X1.41) you will have ~17V at the first filter cap, and 16.5V at the second one, and across your LM1084, you will have 7.84V @ 2.55A for a power dissipation of 20.01W.

Basically, you have an R-C-R-C filter before the regulator, which is good for low-pass filtering. BTW, are your caps 10mF or 10mF? Or 1mF (1000mF)? This will, of course, affect the cut-off frequency of your LP filter.

Have you considered a SimpleSwitcher IC, such as the LM2576 (3A) or MC33167 (5A)? Both are extremely easy to implement and would offer considerably better than the current 54% efficiency. This is just a thought, but with proper filtering and shielding, you can realize a high quality power supply without all the heat dissipation and thermal issues associated with a linear power supply. In their simplest form, they would require one diode, one coil, and two caps (input & output). Variable versions use the same voltage divider technique as the LM1084.

Of course, if noise is of paramount consideration, then a linear will most likely be the way to go. Anywho, if you're interested, let us know, and we can post some links to datasheets on many of the switching regulators out there.

Cheers,

Steve

The load will be 8V 2.55A or 3.13ohms. The load will be tube heaters.

are you saying a 12V transformer is too much? maybe a 9V will do? I have asked about this in another forum and they have said 9V is "cutting it close". maybe they are wrong?

I have made a bridged rectifier powers supply before. It was a 24V 1A. And after the 1st cap, the voltage was around 30+V. but the current requirement is not as large as this new one.

I was looking at the hammond website and it says
============

average voltage after the 1st cap: .9 x Vsecondary
peak voltage after the 1st cap: 1.41 x Vsecondary.

================

At least according to them, 1.41 is only for peak and I would not like to build a PSU using peak values.

noise is not of importance but I already have the LM1084. and the switcher IC you mentioned aren't available locally.

[cerrem]

Go to Linear Technologies web site...
Look up the data sheets for LT1374 or LTC3415...
These simple to use bucks have internal switching FETS..
All you need to do is add the L and C ....
Game Over....

Chris

[jarthel]

Originally posted by cerrem
Go to Linear Technologies web site...
Look up the data sheets for LT1374 or LTC3415...
These simple to use bucks have internal switching FETS..
All you need to do is add the L and C ....
Game Over....

Chris

could we stick to LM1084? where I'm living, those parts are hard to come by even on the biggest online electronic in the country.

And I do not want to order something from mouser if this is the only thing I need.

[cerrem]

Basically your using the wrong tool for the job....
But if you want to continue to use this part....Then your best bet is to simply put a few RC in series prior to the reg... YOu can also put a bunch of power diodes in series since this will drop about .65v each prior to the reg.....

Chris

[N-Channel]

Jarthel,

I gotta concur with cerrem. Heat dissipation alone will be a problem. We don't want to sound like we don't want to help, on the contrary, it's feedback like this from forum members that can prove to be a BIG help, and perhaps enable DIYers to see something from a different perpsective.

If you don't want to order an LM2678, that's fine, especially when National Semiconductor will sample them for free. That's the best price anywhere!

[jarthel]

Originally posted by N-Channel
Jarthel,

I gotta concur with cerrem. Heat dissipation alone will be a problem. We don't want to sound like we don't want to help, on the contrary, it's feedback like this from forum members that can prove to be a BIG help, and perhaps enable DIYers to see something from a different perpsective.

If you don't want to order an LM2678, that's fine, especially when National Semiconductor will sample them for free. That's the best price anywhere!

I tried getting samples for free months ago and I never got it.

I'll try again.

[jarthel]

are you saying that with a step-down regulator, I do not need a heatsink??

[jarthel]

is this a suitable choke for the regulator?