Power Supply, new design.

Status
Not open for further replies.
For the LM317k, the junction max temp is 125°C and the junction to case is 2°C/W. So according to audioguru in this post, my input voltage on the higher range will be 33.4V. So, the lowest voltage the higher range would have to cover is 13V or so. 30.6 Watts to be dissipated if current drawn is 1.5A.... so I would need a heat sink with a rating of... um... carry the 4... with a minimum of 1.1°C/W. If I did the LM338k,then the minimum heat sink rating would be 2.1°C/W.. or maybe try the LM138K, 2.9°C/W. But at 13V @ 1A I still would need a 2.6°C/W heat sink using the LM317k... Where can I find a TO-3 1.1°C/W heat sink at the best or a 2.6°C/W one?

So heat aside, what about the filter cap, someone said something about getting one with a larger capacitance? If I do, will my input voltage (the 33.4V @ 25V transformer output or the 15.7V @ 12V center tap) change? And what capacitance should it have?
 
Last edited:
But the LM317 reduces the current when it has 15V or more from input to output. You will have 20.4V.


The LM317k has a thermal resistance of 2 degrees C/W. For its chip to be at its max allowed temperature of 125 degrees and with an ambient of 30 degrees then the heatsink must deal with 95 degrees C.
With power at 30.6W then the heatsink and its thermal paste must have a thermal resistance of 95/30.6= 3.1 degrees C per W.

If the filter capacitance is too low then the DC voltage will have ripple which reduces its average voltage. For 1.5A a filter capacitor value of 4000uF produces ripple of 2V p-p if the mains is 60Hz. Use 4700uF.
 
audioguru said:
But the LM317 reduces the current when it has 15V or more from input to output. You will have 20.4V.
Click to expand...
Where will I have 20.4V? I don't quite under stand what you mean there.

The total thermal resistance would have to be no more than 3.1°C/W, yes. But the thermal resistance of the LM317k is 2°C/W, so my heat sink would have to have a thermal resistance of 1.1°C/W, right?

If the filter capacitance is too low then the DC voltage will have ripple which reduces its average voltage. For 1.5A a filter capacitor value of 4000uF produces ripple of 2V p-p if the mains is 60Hz. Use 4700uF.
Click to expand...
Alright, but a larger one won't effect the input voltage and thus mess up the heat dissipation calculations, right?
 
Hank102938 said:
Where will I have 20.4V? I don't quite under stand what you mean there.
Click to expand...
The input voltage is 33.6V and the output voltage is as low as 13V. Then the input to output voltage is 20.6V which is too much for an LM317 to have an output current as high as 1.5A.

The total thermal resistance would have to be no more than 3.1°C/W, yes. But the thermal resistance of the LM317k is 2°C/W, so my heat sink would have to have a thermal resistance of 1.1°C/W, right?
Click to expand...
Yes you are correct, I was wrong. That is a huge heatsink.

a larger one (filter capacitor) won't effect the input voltage and thus mess up the heat dissipation calculations, right?
Click to expand...
It won't affect the calculations since the calculations assumes an infinite filter capacitor.
 
Ok, but at 13V I think I can still pull about 1.25A with appropriate heat dissipation, right?

I have figured a 1.6°C/W heat sink will cover all the voltages at 1.5A, and 13V-18.6V at 1.25A (assuming they are limited to 1.25A or lower). So, you said a heat sink of that rating is huge... how bug is huge? If its no bigger than around 2in x 2in it'll fine. The box I have fits my transformer with some room to spare so a fairly large heat sink will fit. The question is where can I get one? Also, if I were to add a fan, would that help the heat sink dissipate more heat without getting as hot (like a 1.6°C/W sink would only increase 1°C per watt dissipated) or would it just keep the ambient temp at 30°C?
 
Here is a good deal on a heatsink that "looks like" maybe 2-3 degrees C per watt.
LARGE HEATSINK FOR TWO TO-3 DEVICES | AllElectronics.com
Mike has a good idea if you want the higher current. The maximum (infinite heat sink) power for that regulator is 50 watts as opposed to 20 for the 317 and it is only 1.4 degrees Junction to case. Seems the 2 might get you there.
 
Sorry, I was away and a few posts came in between my last post. Here is a heatsink with a graph that shows the effect with air. I have no idea how much air is represented by the graph, but i don't think a whole lot.

**broken link removed**
This one is available from mouser, if you decide to stick to the to220
 
Hank102938 said:
Ok, but at 13V I think I can still pull about 1.25A with appropriate heat dissipation, right?
Click to expand...
The datasheet says that you get 1.5A minimum when your output is no lower than about 18.6V.
If your LM317 is "typical' then the output at 13V will be about 1.4A.

I have figured a 1.6°C/W heat sink will cover all the voltages at 1.5A, and 13V-18.6V at 1.25A (assuming they are limited to 1.25A or lower). So, you said a heat sink of that rating is huge... how bug is huge? If its no bigger than around 2in x 2in it'll fine.[/quote]
2" x 2" is a tiny 5W heatsink. You need the heatsink to be at least 6 times larger and it must not be enclosed, it must be in free air.

The question is where can I get one?
Click to expand...
All electronic parts distributors have many heatsinks of many sizes.
 
Alright, this just popped in my head. What if I use two regulators? One that has a set voltage and the other that is variable. The set one brings it down for the variable one. I'm just trying to figure out how to manage the heat. My box is 8"x4"x3" (LxWxH) so a huge heat sink is out of the question... Any other ideas... If I have it admit that I'm not getting the max amperage out at all the voltages then I guess that's that... But I can defiantly put in a fan, I just don't know how effective it is.
 
Last edited:
Attached is the Vin-Vout versus maximum current graph from spec sheet of LM317.

With 30 volt Vin to Vout you will max out at 600-700 mA's.

That assumes you have enough heat sinking not to cause it to reduce due to overheating.
 

Attachments

  • LM317 for Vin_Vout delta.pdf
    169.1 KB · Views: 236
Last edited:
Supply

Your kind of boxed in by your existing parts and your specs. You could add what is called a pre-regulator (say another LM317) but it needs voltage to operate. Remember how you need at least 2.5 volts more into the regulator than you get out? You could add a "pass transistor" but that has a similar problem when used with the 317. Probably the best you can do given all the constaints is to look around for the biggest heatsink that will fit your box, use the TO3 package and add a small fan if you have room. If you want to start over with just your transformer and bridge here is a supply I've been thinking about building. 0 to 20 Volts at 2 Amps with adjustable current limit.
 

Attachments

  • Lab.PNG
    52.8 KB · Views: 344

I agree about being boxed in. I'm willing to start over with just my trans. and bridge since they were really the only expensive parts. I really want a supply that can give me 24V and under at 1A... that would be perfect. And adjustable current limit would be awesome (where the level of awesome is dependent on the cost). Any thoughts or schematics?
 
Last edited:
If the heatsink and a fan are inside a closed box then there is barely any cooling.
 
This site doesn't show when I reply and double posts are not warned.
I deleted my double post.
 
Last edited:
Ok, I just finished coming up with a new design (see attachment).
With this, the first regulator has two set output voltages, one around 22V and the other around 6.5V. This takes some strain off of the second regulator. I've calculated and believe the maximum power that would ever be dissipated by one regulator at 1.5A would be about 14.55W. So, that way I can use a 5°C/W heat sink on each LM317k and be able to get max amperage out at all voltages from 1.25V-30V. So the question is, will it work like I think it will?
 

Attachments

  • Dual Reg.png
    37.6 KB · Views: 312
Last edited:
An LM317 is supposed to have a 120 ohms resistor (not 220 ohms) from its output to its ADJ pin to prevent the output voltage from rising without a load. A more expensive LM117 can use a 240 ohm resistor.

The second regulator needs a capacitor to ground at its input to prevent it from oscillating.

EDIT: Your voltage calculations are wrong.
The first LM317 has an output that is 6.93V and 22.84V. Then the second LM317 will have a max output of 4.93V and 20.84V.
The 5k pot will have most of its rotation unuseable.
 
Last edited:
audioguru said:
An LM317 is supposed to have a 120 ohms resistor (not 220 ohms) from its output to its ADJ pin to prevent the output voltage from rising without a load. A more expensive LM117 can use a 240 ohm resistor.
Click to expand...
So it has to be exactly 120 or could it be lower? What if I wanted to use one that is 100ohm so I could use a more common 2.5k pot?

The second regulator needs a capacitor to ground at its input to prevent it from oscillating.
Click to expand...
How large...?

EDIT: Your voltage calculations are wrong.
The first LM317 has an output that is 6.93V and 22.84V. Then the second LM317 will have a max output of 4.93V and 20.84V.
The 5k pot will have most of its rotation unuseable.
Click to expand...
I know the voltages were just a guess from an earlier calculation I didn't write down. I also forgot to add in a switch to bypass the first regulator to get the remaining range up to 30V. Yes the pot will have a lot of unused rotation when the input of the second regulator is dropped down by the first one.... but there wasn't really any other way to do it...

But this design isn't exactly ideal nor will it be user friendly with three different switches... someone mentioned using a transistor...
ronv said:
You could add a "pass transistor"
Click to expand...
Would that be a better solution or not?

Whats some other designs I could try if I use the same transformer and rectifier?
 
Hank102938 said:
So it has to be exactly 120 or could it be lower? What if I wanted to use one that is 100ohm so I could use a more common 2.5k pot?
Click to expand...
It can have a value lower than 120 ohms but then the current is higher and the pot might burn up.

How large...?
Click to expand...
It is shown in the datasheet. Don't you have the datasheet?

I know the voltages were just a guess from an earlier calculation I didn't write down.
Click to expand...
The calculation uses grade 4 simple arithmatic.
 
audioguru said:
It can have a value lower than 120 ohms but then the current is higher and the pot might burn up.
Click to expand...
True....

It is shown in the datasheet. Don't you have the datasheet?
Click to expand...
Yeah, 0.1uF I guess...


The calculation uses grade 4 simple arithmatic.
Click to expand...
Yeah, I know because I did the calculations, I just guessed at the numbers since I had calculated it before I posted and didn't write the exact voltage down. Just don't worry about it...

But anyway, whats my options here??? The circuit I just came up with, like I said, isn't exactly ideal. Surely someone has an alternative design or something? I don't know what influenced me to make this thing, but it's starting looks like just using batters is the best way...
 
Last edited:
Pre-regulator. But if you look at my post above it won't work for you. The lowest you can adjust the regulator is 1.2 volts. It needs 2.5 volts to regulate (each) for a total of 5 volts. So on 12 volts your high voltage is 13.8 (output from your filter cap) minus 5 (two regulators) is 8.8 volts. On the low side it is 2 X 1.2 (2 Minimum adjustment voltage) or 2.4 volts. It would help you a little on the 24 volt setting.
 
Status
Not open for further replies.

Similar threads

T
  • Question
Power problems
Replies
11
Views
2K
texas dave
T
  • Question
Bench Power Supply
2
Replies
20
Views
5K
Tony Stewart
Handy Bench Power Supply
Replies
8
Views
3K
schmitt trigger
S
A Simple Power Supply Project
Replies
1
Views
2K
Tony Stewart
  • Question
Powering RPI5 8gb from 5v supply
Replies
3
Views
2K
aliarifat794
A
Menu
Log in
Register
Cookies are required to use this site. You must accept them to continue using the site. Learn more…