DIY power supply

[Pirateguy]

what if i use them in high current config with the power transistor?

[MarkF]

I don't know what your goal is now since your AT Power Supply died.

You could try using two regulators in series to split the load as shown in the "Tracking Preregulator Circuit" example in the datasheet.
I would configure the first regulator to always be 6.5V above the output.  This would give you a 0-20V output with your 32V supply.
   Start with R1=240,  R2=1K,  R3=330,  R4=5K and see how it works.

   
   

[JS]

I don't know what your goal is now since your AT Power Supply died.

You could try using two regulators in series to split the load as shown in the "Tracking Preregulator Circuit" example in the datasheet.
I would configure the first regulator to always be 6.5V above the output.  This would give you a 0-20V output with your 32V supply.
   Start with R1=240,  R2=1K,  R3=330,  R4=5K and see how it works.

   
 

I think 330Ω for R3 is too high, it's supposed to meet the minimum load specs, but for both LM317 this time, if the regulation is supposed to be maintained without any load or very low loads. Also, this circuit doesn't share the thermal load evenly between the two ICs so I don't know if it's optimal for this.

JS

[Pirateguy]

well i'm pretty sure i can pick up an ATX from a thrift store around here if need be.
in fact i saw one the last time i was there.

i ordered the dps3012 for my variable supply, along with a case for it.
so now i just want to make the fixed output one.

for that i intend to use the large heatsink from my disassembled AT,
continue using the AT housing (it's all i got atm) and for the time i do
stick to 3.3v, 5v and 12v outputs. leaving the negative unconnected for now.

so some ideas i had:

-using the bypass transistor for current to take some strain off the regulators.
-using multiple regulators in parallel.
-cascading regulators; 32v PSU to 24v regulator, from the 24v to 18v regulator, from there to 12 and so on.
-i have a fairly heavy wall wart that switches between different voltage levels, could i hack that?
maybe i can draw a line from the transformer directly and add a rectifier circuit for each?

assuming one of those ideas will work, i am again left with the challenge of finding an enclosure.
any suggestions for that?
it would have to be fairly big so i have room to add more stuff, incl power source(s)...

[Wolfgang]

... why not just add a pass transistor and keep the LM317 cool ? Its not one of the xtra stable or noise free regulators on earth anyway ?
And if you want to be really nice you can add a current limit plus an overtemp protection plus a crowbar at the output to avoid surprises ...

[MarkF]

... why not just add a pass transistor and keep the LM317 cool ? Its not one of the xtra stable or noise free regulators on earth anyway ?
And if you want to be really nice you can add a current limit plus an overtemp protection plus a crowbar at the output to avoid surprises ...

You still have to dissipate all that power if you want a variable supply (say 32VDC input and 0-20V @ 1A output).  The power will have to go to the regulator or the bypass transistor.  I proposed the preregulator to split the load (granted not evenly to get a 20V output).  I thought it the easier approach.

LET'S SEE A CIRCUIT FOR ALL YOU PROPOSE.

[MarkF]

well i'm pretty sure i can pick up an ATX from a thrift store around here if need be.
in fact i saw one the last time i was there.

i ordered the dps3012 for my variable supply, along with a case for it.
so now i just want to make the fixed output one.

for that i intend to use the large heatsink from my disassembled AT,
continue using the AT housing (it's all i got atm) and for the time i do
stick to 3.3v, 5v and 12v outputs. leaving the negative unconnected for now.

That was your original approach and from what you've said and your knowledge at this point, that is probably you best course.

Don't discount the negative voltage.  Especially since it's available.  If you end up doing any analog projects you will probably need it.

[Wolfgang]

Well, I would say that a pass transistor is about as easy as another LM317. The advantage of a pass transistor is that it could be user with a much higher dissipation than even two LM317s. A proper cooler and a fan could easily do 100W, a TO220 LM317 can only do roundabout 25W. On top of that, a regulator that does not run hot is definitely more stable.

[Pirateguy]

quick note: i just got some lm338 regulators.

negative voltage is available?
do you mean to say i should go with an ATX?

[JS]

There's no negative version of LM338, if you need dual supplies using it you need them floating and then connect them in series at the output.

JS

[Cliff Matthews]

Hard to understand why TO220 338's exist.. what good is the 5amps when internal thermal protection limits the whole package to dissipating ~15watts on an ideal heat sink. Seems LM317 and 337 are fine with additional pass devices..??
For the OP, maybe someone can comment on the schematic half-way down on this link:
https://electronics.stackexchange.com/questions/186760/why-do-linear-voltage-regulators-have-minimum-output-voltage-0-v

[JS]

Hard to understand why TO220 338's exist.. what good is the 5amps when internal thermal protection limits the whole package to dissipating ~15watts on an ideal heat sink.

Why do you say so?
At 0.7ºC/W it's better than the TO3 package, the thing is the TO3 package has a bigger plate with two screws so the thermal resistance case to heatsink can be made lower, plus the bigger thermal mass can make it handle higher peaks better.

Still, if what you say it's true it could work in a 3V dropout application, which it can do over the whole temp range.

JS

[Pirateguy]

well wouldn't u know it, the thrift store no longer has the ATX i saw, nor much of anything else PSU wise.
i guess ill have to take inventory of the wallwarts n stuff i have...

[Pirateguy]

scored an ATX!

[Pirateguy]

the new plan:

from the wall:

ATX:

polyfused 3.3v
polyfused 5v
polyfused -5v (7905 from -12v rail)
polyfused 12v
polyfused -12v
USB (from 5v SB)
-variable (DPS3012, from 12v rail)

on the front panel:

Thermometer with sensor broken out to binding posts. link
Thermostat controlling the fans link or link
Func gen. link
Relay timer link
usb tester link

also i think i will add a double wall socket with a dimmer on one of them, and add a halogen ballast with 2 outputs, of which one has a dimmer.

and on the wall sockets i will have this:
AC volt/amp meter link

also since i cannot find a nice box to built it into, i found some help for making a custom box out of multiplex or something like that.

also i may want to put an adjustable load in there at some point, but i need to learn more about them.
am wondering if it might be better to keep that separate cuz of temperatures? seems like that kinda thing would get hot...?

[Pirateguy]

couple of quick questions:

when a polyfuse is triggered, what exactly does it take to reset it?
could i put in a NC button to interrupt the line to reset a polyfuse?

do polyfuses work any differently on negative voltage?

can i run a 7905 on the -12 rail normally as per datasheet, or does it need something anything else?

[exe]

I think polyfuse is just a PTC -- a resistor that quickly increases resistance as it heats up. It doesn't care which way current flows imho.

I don't think it can be reset with a button. The only way to reset it is to cool it down.

I suggest read a datasheet on it, it answers all your questions. I think this is a good start: http://www.littelfuse.com/products/resettable-ptcs/surface-mount.aspx . Be sure to see different plots and numbers in the datasheet. Like, maximum guaranteed current that will not trip it (and how it depends on ambient temperature!), minimum guaranteed current that will trip it, etc.

[drussell]

couple of quick questions:when a polyfuse is triggered, what exactly does it take to reset it?

Removal of the overcurrent and time.  (On the order of a few seconds.)

could i put in a NC button to interrupt the line to reset a polyfuse?

No, that is not how they work. 

https://en.wikipedia.org/wiki/Resettable_fuse

do polyfuses work any differently on negative voltage?

No, they are not polarity sensitive.

can i run a 7905 on the -12 rail normally as per datasheet, or does it need something anything else?

You most certainly can use a 7905 on a -12 volt rail.  You're going to be limited as to the current you can draw on your negative side due to the 0.8A maximum output of the power supply compared to the huge currents available on the positive rails, but within that limit you'll be fine.  (Obviously the -5 rail will have even less than 0.8A available since the linear 7905 is going to dissipate a good chunk of your available ~10 watts.)

I think polyfuse is just a PTC -- a resistor that quickly increases resistance as it heats up. It doesn't care which way current flows imho.

Correct.  They are actually known as PPTCs.  The extra P stands for Polymeric.  They are made from a special crystalline organic polymer that gives them the same kind of sharp current characteristic as a zener has on the knee voltage.  Instead of a slowly changing, smooth curve like a regular PTC has, a polyfuse stays basically the same resistance until it heats enough for the polymer to change from a crystalline state to an amorphous state, which raises the resistance very rapidly, limiting the current.  Once the current is reduced sufficiently, the polymer recrystallizes as it cools and the "fuse" is "closed" again (resistance returns to close to the original value, though it can take several hours to return fully to its original value.)

[Pirateguy]

right, so the only action i need to undertake is to disconnect the load and wait.

thanks

is 2,7 ohm enough for the load resistor?

[Pirateguy]

one day i'm looking for an ATX, the next a friend brings by 2 of them + another device that contains one
i got a small 150watt one, a large 500watt one and what looks like a 10watt one that has -24v on it for some reason
from a 'media center extender'.

anyhoo i was looking at this guide to converting atx psu and came across something odd.

Again the voltages that can be output by this unit are the same as before 24v (+12, -12), 17v (+5, -12), 12v (+12, 0), 10v (+5, -5), 7v (+12, +5), 5v (+5, 0). Note that some ATX12V power supplies with a 24-pin motherboard connector may not have the -5V (pin 20) white lead. In this case use the older ATX power supplies with a 20-pin connector above if you need the additional -5V supply.

this suggests a +5v and +12v rails can somehow be used to create 7v??

[MarkF]

this suggests a +5v and +12v rails can somehow be used to create 7v??

   +12V - +5V = +7V

You would be using the 5V rail as the reference (ground).  The ground output in this case would be -5V.  This also prevents you from using the remaining outputs in their normal way.  NOT something I would recommend on a routine bases.  I would stay away from the odd voltage combinations and just use it as designed.  i don't know why it was even brought up.  It just confuses the novice!

ONCE AGAIN.  DON'T GET CREATIVE.   

Just use the -12V, +12V, +5V and +3.3V outputs as designed.  In practice, you will seldom use or want anything else.

I would use the smaller 150 watt power supply.  There is NO way you would need the current capabilities of the larger one.

If the smallest 10W supply has a +24V output, I would use it for the adjustable supply instead of the +32V supply.  That is if you're still building the LM317/LM338 design we discussed earlier.  It would ease the power dissipation concerns.

[Pirateguy]

actually no i ordered one of those dps3012 modules.

and no worries, i already got creative


so i have a selector switch between .9A 6A and no poly fuse, and a bit with a
switch and momentary button and a 3.5mm jack on the ground in case i need
to cut power quick, or just want to give a quick jolt to see if something works or not.
the 3.5mm jack connects to another button or a foot pedal to do the same remotely.
also they are all connected to RGB LEDs to indicate power and which fuse setting they are on.

is there ever a reason to limit 3.3v or should i just skip the switch on that one and put a 6A fuse?

[ricktendo]

Don't forget to check out OpenDPS, you can remote control it over WiFi and upgrade over Serial!!!

https://github.com/kanflo/opendps

I wish I had the money to send one of each RD-DPS units to the developer, maybe somebody can contact him and/or setup a gofundme to collect money for the hardware for development!!!

[Pirateguy]

thanks

but i don't think the model i have has the wifi capability.

[ricktendo]

thanks

but i don't think the model i have has the wifi capability.

They are all WiFi capable, its a matter of adding a ESP8266 board.