Author Topic: Lab Bench PSU  (Read 8689 times)

0 Members and 1 Guest are viewing this topic.

PotatoBox

• Regular Contributor
• Posts: 104
• Country:
Lab Bench PSU
« on: August 24, 2016, 04:25:22 am »
Hello everyone,

I have a 300w atx psu I salvaged from an old computer and I want to convert it into a variable bench power supply. Now, I wanted to use an LM317 to adjust the voltage on the 12v rail from the psu. In the texas instruments datasheet on page 10 shows the suggested schematic. Now, which resistors do I use for R1 and R2? it shows a 240ohm resistor for R1 but Ive seen people use different resistors for R1.

The following users thanked this post: montreuilg@sympatico.ca

gigabyte091

• Contributor
• Posts: 21
• Country:
Re: Lab Bench PSU
« Reply #1 on: August 24, 2016, 04:53:39 am »
I usually use 10k pot with LM317, if you have -12 or -5V on PSU you could use it and have output voltage adjustable from 0V instead of 1.25V.

The following users thanked this post: montreuilg@sympatico.ca

Mechatrommer

• Super Contributor
• Posts: 9512
• Country:
• reassessing directives...
Re: Lab Bench PSU
« Reply #2 on: August 24, 2016, 12:01:28 pm »
there!...

R2 = [(Vo / 1.25) - 1] * R1
ignore Iadj * R2 as starting point...
if something can select, how cant it be intelligent? if something is intelligent, how cant it exist?

The following users thanked this post: montreuilg@sympatico.ca

newbrain

• Frequent Contributor
• Posts: 811
• Country:
Re: Lab Bench PSU
« Reply #3 on: August 24, 2016, 02:59:22 pm »
Hello everyone,

I have a 300w atx psu I salvaged from an old computer and I want to convert it into a variable bench power supply. Now, I wanted to use an LM317 to adjust the voltage on the 12v rail from the psu. In the texas instruments datasheet on page 10 shows the suggested schematic. Now, which resistors do I use for R1 and R2? it shows a 240ohm resistor for R1 but Ive seen people use different resistors for R1.

A couple of thoughts:
* About the resistor: the LM317 datasheet specifies the minimum load current with a typical value of 3.5mA and a maximum of 10mA. This means that in the general case 240ohms are enough, but to be safe 120ohms should be used. LM117, OTOH, with its 5mA as maximum is guaranteed to work with 240ohms.
* Remember that the negative terminal of your supply (and the PSU chassis) is connected to earth ground, so your power supply will not be floating. This has both safety and practical implications. And no, cutting the ground connection is not a good idea.
Nandemo wa shiranai wa yo, shitteru koto dake.

The following users thanked this post: montreuilg@sympatico.ca

PotatoBox

• Regular Contributor
• Posts: 104
• Country:
Re: Lab Bench PSU
« Reply #4 on: August 24, 2016, 04:15:20 pm »
Sooo the resistor values are all dependant on the the maximun voltage I want in the circuit? But what if I wanted to have my psu output more than 10ma? Would I just use a higher resistor for R1?

The following users thanked this post: montreuilg@sympatico.ca

macboy

• Super Contributor
• Posts: 2007
• Country:
Re: Lab Bench PSU
« Reply #5 on: August 24, 2016, 04:56:05 pm »
Sooo the resistor values are all dependant on the the maximun voltage I want in the circuit? But what if I wanted to have my psu output more than 10ma? Would I just use a higher resistor for R1?
No, resistor R1 provides a minimum load when nothing is connected to the output. Notice that there is always 1.25 V across R1 (this is the reference voltage) so 1.25 / 240 ohm = 5.2 mA. That amount of current always flows through R1 (and R2) regardless of the output voltage setting or the connected load.

Any load you connect will simply add to this load. So if you connect some load that draws 100 mA, the LM317 will be required to supply 105 mA total.

The following users thanked this post: montreuilg@sympatico.ca

buffoon

• Contributor
• Posts: 8
• Country:
Re: Lab Bench PSU
« Reply #6 on: August 25, 2016, 12:36:34 am »
Sooo the resistor values are all dependant on the the maximun voltage I want in the circuit? But what if I wanted to have my psu output more than 10ma? Would I just use a higher resistor for R1?
A few things:
1) Keep R1 constant around 240ohm. The 5mA calculated by newbrain goes into the feedback loop. Use the equation mechatrommer posted above to choose R2 depending on your output needs (or use a pot to make a variable PSU).

2) What will be your output voltage and current? Keep in mind that the voltage doesn't just disappear. The power dissipated as heat will be (Vin - Vout) * current

The following users thanked this post: montreuilg@sympatico.ca

PotatoBox

• Regular Contributor
• Posts: 104
• Country:
Re: Lab Bench PSU
« Reply #7 on: August 25, 2016, 01:51:16 am »
Sooo the resistor values are all dependant on the the maximun voltage I want in the circuit? But what if I wanted to have my psu output more than 10ma? Would I just use a higher resistor for R1?
A few things:
1) Keep R1 constant around 240ohm. The 5mA calculated by newbrain goes into the feedback loop. Use the equation mechatrommer posted above to choose R2 depending on your output needs (or use a pot to make a variable PSU).

2) What will be your output voltage and current? Keep in mind that the voltage doesn't just disappear. The power dissipated as heat will be (Vin - Vout) * current

Well I want to have a variable output voltage between 1.5-12 volts.

The following users thanked this post: montreuilg@sympatico.ca

newbrain

• Frequent Contributor
• Posts: 811
• Country:
Re: Lab Bench PSU
« Reply #8 on: August 25, 2016, 09:41:32 pm »
Well I want to have a variable output voltage between 1.5-12 volts.
Take into consideration that the LM317 will not work correctly with Vout-Vin smaller than about 2.5÷3V.
So the maximum regulated voltage you can expect to obtain is ~9V; of course you can get the 12V directly from the ATX PSU
Nandemo wa shiranai wa yo, shitteru koto dake.

The following users thanked this post: montreuilg@sympatico.ca

PotatoBox

• Regular Contributor
• Posts: 104
• Country:
Re: Lab Bench PSU
« Reply #9 on: August 27, 2016, 01:14:42 am »
Should I try to use an LM2576 instead?

The following users thanked this post: montreuilg@sympatico.ca

newbrain

• Frequent Contributor
• Posts: 811
• Country:
Re: Lab Bench PSU
« Reply #10 on: August 27, 2016, 08:50:44 am »
Should I try to use an LM2576 instead?
Using a LM2576 will reduce the drop out voltage, according to the datasheet, Figure 3, it will be about 1V for a 1A load at 50°C. $\Omega$
But:
• This is a switching regulator: PCB layout is very important and the design process is more complex, see the chapters 8.2 and 9 in the DS.
• It still won't get you 12V. You would need a buck-boost type of switching regulator.
I think the LM317 idea is good, for a beginner's project, my initial warnings were for a couple a things that might go unnoticed (everybody uses 240$\Omega$ with the LM317, but that's not guaranteed to work with very small loads).
When you need 12V just get them straight from the ATX PSU...
Another alternative: there are buck-boost modules for sale on eBay o AliExpress.
They usually work almost as advertised, but the learning value of just connecting a couple of wires is next to zero...

The next step for LM317: thermal design!
How much current do you need? Do you need that maximum current at the minimum voltage?
Nandemo wa shiranai wa yo, shitteru koto dake.

The following users thanked this post: montreuilg@sympatico.ca

PotatoBox

• Regular Contributor
• Posts: 104
• Country:
Re: Lab Bench PSU
« Reply #11 on: August 28, 2016, 04:58:17 am »
Should I try to use an LM2576 instead?
Using a LM2576 will reduce the drop out voltage, according to the datasheet, Figure 3, it will be about 1V for a 1A load at 50°C. $\Omega$
But:
• This is a switching regulator: PCB layout is very important and the design process is more complex, see the chapters 8.2 and 9 in the DS.
• It still won't get you 12V. You would need a buck-boost type of switching regulator.
I think the LM317 idea is good, for a beginner's project, my initial warnings were for a couple a things that might go unnoticed (everybody uses 240$\Omega$ with the LM317, but that's not guaranteed to work with very small loads).
When you need 12V just get them straight from the ATX PSU...
Another alternative: there are buck-boost modules for sale on eBay o AliExpress.
They usually work almost as advertised, but the learning value of just connecting a couple of wires is next to zero...

The next step for LM317: thermal design!
How much current do you need? Do you need that maximum current at the minimum voltage?

So, are you saying I should go with the lm317 over the lm2576? Arent the lm2576 used in those buck converters?

The following users thanked this post: montreuilg@sympatico.ca

spacedementia87

• Contributor
• Posts: 30
• Country:
Re: Lab Bench PSU
« Reply #12 on: August 28, 2016, 10:10:15 am »
I have seen posts on this forum before telling people not to use ATX supplies as a bench supply.

Safety, accuracy all come out as reasons, but not here.

Why not?

Sent from my ONE A2003 using Tapatalk

The following users thanked this post: montreuilg@sympatico.ca

newbrain

• Frequent Contributor
• Posts: 811
• Country:
Re: Lab Bench PSU
« Reply #13 on: August 28, 2016, 01:04:42 pm »
So, are you saying I should go with the lm317 over the lm2576? Arent the lm2576 used in those buck converters?
It's up to you and the confidence you have in your skills. Since you started simple, and this is the beginner section I imagined you were after a simple starter project.
Yes, they are used in buck (==down) converters, that means the output voltage will be lower than the input one.

I have seen posts on this forum before telling people not to use ATX supplies as a bench supply.

Safety, accuracy all come out as reasons, but not here.

Why not?
If using an ATX PSU involves modification of the internals (holes and wiring changes) and removal of the ground connection (as seen in some horror video on YT), I'm all against it.

If the supply is simply used as the input source for a regulator stage as seems to be the case here, I do not see major problems, when done carefully.

One of the things that might be overlooked is that all the output voltage are referenced to earth ground, that's why I pointed it out in my first answer.

Another point is that an ATX supply can provide a lot of current, especially on the 12V rail, though good ones have protections (don't know about the OP one).
The post regulator (LM317) section should at least be fused appropriately.
Nandemo wa shiranai wa yo, shitteru koto dake.

The following users thanked this post: montreuilg@sympatico.ca

PotatoBox

• Regular Contributor
• Posts: 104
• Country:
Re: Lab Bench PSU
« Reply #14 on: August 28, 2016, 05:52:34 pm »
Well, what I might actually do is scrap the atx psu idea and just make a power supply from scratch with a transformer and such. Get a better learning experience out of that.

The following users thanked this post: montreuilg@sympatico.ca

newbrain

• Frequent Contributor
• Posts: 811
• Country:
Re: Lab Bench PSU
« Reply #15 on: August 28, 2016, 07:50:35 pm »
Well, what I might actually do is scrap the atx psu idea and just make a power supply from scratch with a transformer and such. Get a better learning experience out of that.
I definitely agree.
A small linear PSU is a typical initial project, a good learning value and a useful tool (I still have mine, it might be 30 yo, and still goes strong...).
Nandemo wa shiranai wa yo, shitteru koto dake.

The following users thanked this post: montreuilg@sympatico.ca

PotatoBox

• Regular Contributor
• Posts: 104
• Country:
Re: Lab Bench PSU
« Reply #16 on: August 29, 2016, 02:06:12 am »
Well, what I might actually do is scrap the atx psu idea and just make a power supply from scratch with a transformer and such. Get a better learning experience out of that.
I definitely agree.
A small linear PSU is a typical initial project, a good learning value and a useful tool (I still have mine, it might be 30 yo, and still goes strong...).

The good thing is, I could just salavge most of the parts from the atx psu i already have!

Mechatrommer

• Super Contributor
• Posts: 9512
• Country:
• reassessing directives...
Re: Lab Bench PSU
« Reply #17 on: August 29, 2016, 03:17:53 am »
The good thing is, I could just salavge most of the parts from the atx psu i already have!
for linear? not really, not even the transformer. been there... well there are some you can salvage, like some misc resistors, wires, and the metal case (if that suits your taste)..
if something can select, how cant it be intelligent? if something is intelligent, how cant it exist?

PotatoBox

• Regular Contributor
• Posts: 104
• Country:
Re: Lab Bench PSU
« Reply #18 on: August 29, 2016, 05:20:45 am »
The good thing is, I could just salavge most of the parts from the atx psu i already have!
for linear? not really, not even the transformer. been there... well there are some you can salvage, like some misc resistors, wires, and the metal case (if that suits your taste)..

Linear... What does that mean?

Jay_Diddy_B

• Super Contributor
• Posts: 1851
• Country:
Re: Lab Bench PSU
« Reply #19 on: August 29, 2016, 08:30:34 am »
Linear... What does that mean?

In this context it means:

Using a LDO style regulator, like a LM317 or similar part. The alternative is to use a switching power supply design. The construction of a switching supply is more difficult, especially for a beginner.

A little bit of metal work is required, linear supplies need a heatsink for the main power device.

Using a 60 Hz transformer, instead of an off-line switching power supply. This is much safer. There is less exposed high voltage.

I am sure that many members of the forum built these supplies as one of their first projects. When I started electronics, (shortly after electricity was invented  ), the low cost bench power supplies, we find today, were not available and there were relatively few surplus power supplies (hp, lambda etc.) and these were expensive.

Regards,

Jay_Diddy_B

setq

• Frequent Contributor
• Posts: 444
• Country:
Re: Lab Bench PSU
« Reply #20 on: August 29, 2016, 05:01:30 pm »
If you're going to do this, please wear safety glasses. ATX supplies can quite happily dump 200W into your load suddenly on one rail. I really wouldn't use one myself. A little TO220 regulator is going to spread its guts across the room pretty sharpish as are your probes, solder joints and anything conductive that temporarily acts as the load by accident. All it takes is one little mistake.

janoc

• Super Contributor
• Posts: 3057
• Country:
Re: Lab Bench PSU
« Reply #21 on: August 29, 2016, 07:35:23 pm »
If you're going to do this, please wear safety glasses. ATX supplies can quite happily dump 200W into your load suddenly on one rail. I really wouldn't use one myself. A little TO220 regulator is going to spread its guts across the room pretty sharpish as are your probes, solder joints and anything conductive that temporarily acts as the load by accident. All it takes is one little mistake.

Yeah, ATX supply without some additional current limiting is a disaster waiting to happen, even though they are protected against short circuit. However, 20A on a 5V rail, no problem! And the same or more on the 12V rail - that's getting into the arc welder territory already.

They also produce a rather dirty output, with a lot of switching noise and not great regulation. For PCs it is good enough, because it is assumed that pretty much everything has local regulation and filtering, but good luck testing some noise sensitive circuits with it ...

I have one, but I am using it only to power high current things like motors or my fume extractor fan, not as a regular bench supply.
« Last Edit: August 29, 2016, 07:40:22 pm by janoc »

Mechatrommer

• Super Contributor
• Posts: 9512
• Country:
• reassessing directives...
Re: Lab Bench PSU
« Reply #22 on: August 30, 2016, 03:16:47 am »
The good thing is, I could just salavge most of the parts from the atx psu i already have!
for linear? not really, not even the transformer. been there... well there are some you can salvage, like some misc resistors, wires, and the metal case (if that suits your taste)..
Linear... What does that mean?
i mean... if you want to salvage ATX PSU parts to make a linear regulating bench power supply.
if something can select, how cant it be intelligent? if something is intelligent, how cant it exist?

Brumby

• Supporter
• Posts: 9809
• Country:
Re: Lab Bench PSU
« Reply #23 on: August 30, 2016, 06:14:06 am »
The good thing is, I could just salavge most of the parts from the atx psu i already have!
for linear? not really, not even the transformer. been there... well there are some you can salvage, like some misc resistors, wires, and the metal case (if that suits your taste)..

Linear... What does that mean?

There are two basic common designs for power supplies: Switch mode and linear.

In general:

With switch mode, the incoming power is turned into DC which powers an oscillator.  Sounds like going in a circle, doesn't it?  But the oscillator runs at a much higher frequency like 50 kHz to 1 MHz (pick a number).  This higher frequency allows for the use of a much smaller transformer.  The design also can handle a wide variety of input voltages ... and frequency is hardly relevant.  Small and light, this type of supply is usually cheap and easy to make and tends to not waste a lot of power.  It does have a down side, though ... and that is, because of the high frequency and switching mechanism, you can get a lot of interference.  For some things, this doesn't matter, for others, it does.

Linear design has the absolute benefit of the lowest levels of interference possible - but it requires a (sometimes big and hefty) transformer designed to operate at the nominated supply voltage and frequency and capable of delivering the current required.  Voltage and current control is done by 'linear' circuitry - that is, continuous and smooth changes, much like adjusting the flow of water from a tap.  These control mechanisms often have a significant issue with power dissipation, so heatsinks become another necessity - and expense ... but there's no beating a linear power supply for quality of the power it can deliver.

The following users thanked this post: montreuilg@sympatico.ca

PotatoBox

• Regular Contributor
• Posts: 104
• Country:
Re: Lab Bench PSU
« Reply #24 on: August 31, 2016, 02:29:31 am »
Could I just rip a transformer from an alarm clock?

Smf