235 lines
11 KiB
Markdown
235 lines
11 KiB
Markdown
+++
|
|
date = "2014-11-24"
|
|
title = "Design: VoltMeister 100, a DIY Bench Power Supply"
|
|
tags = ["electronics", "voltmeister", "power supply"]
|
|
categories = [ "Electronics" ]
|
|
summary = "A bench power supply is one of the essential tools of any electronics hobbyist.Although you can buy a such a unit for less that € 50, it's way more fun to build one yourself."
|
|
aliases = [
|
|
"/2014/11/24/voltmeister-100-atx-bench-power-supply-part-1/",
|
|
"/2014/11/26/voltmeister-100-atx-bench-power-supply-part-2/"
|
|
]
|
|
[image]
|
|
+++
|
|
|
|
As an electronics hobbyist one of the most essential tools is a bench power
|
|
supply. I don't have one yet, so I'm currently stuck with using simple wall warts.
|
|
This is fine for powering an Arduino, but it gets more tacky when dealing with things
|
|
like op-amps.
|
|
|
|
Because I only have limited experience with Arduino and a rudimentary understanding
|
|
of electronics, I decided not to make my first project about using 230VAC directly
|
|
and creating an adjustable power supply. Instead, I opted for a more common approach:
|
|
use an existing ATX computer power supply to deliver the four common output voltages
|
|
of these supplies: 3.3V, 5V, and ±12V.
|
|
|
|
## The ATX Power Supply
|
|
|
|
Modern computers all comform to the ATX standard, set by Intel in 1995. Part of this
|
|
standard is the _power supply_. ATX power supplies are easily available and interchangable.
|
|
|
|
Older models ATX power supplies have a 20-pin connector, new ones have a 24-pin connector.
|
|
|
|
|
|
|
|
The 24-pin configuration adds four pins, one for each supply voltage and GND (Ground).
|
|
A 20-pin connector does fit a 24-pin socket. Yay for standards!
|
|
|
|
An ATX power supply offers a total of four voltages. However, they each have different power
|
|
ratings. There's probably a sticker on your supply telling you the exact details.
|
|
|
|
|
|
|
|
The supply I'll be using in this project comes from an old Asus S-Presso. As the
|
|
label indicates, it can supply different voltages with different power ratings:
|
|
|
|
* 3.3V at 17A
|
|
* 5V at 13A
|
|
* 12V at 16A
|
|
|
|
That's quite beefy (except the -12V), especialy if you're mostly toying around with
|
|
microcontrollers and microprocessors.
|
|
|
|
There are also two other notable voltages. -12V and 5VSB.
|
|
|
|
The -12V is ideal if you're doing things with op-amps that need a positive and
|
|
negative reference. But, as you can see, the -12V rail only supports
|
|
up to 0.3 amps. This means you can't combine the ±12V rails to create
|
|
a 24V output that draws lots of current.
|
|
|
|
The 5VSB provides 5V at a maximum of 2A _when the power supply is on stand by_.
|
|
When you have the power supply plugged into mains, but it's not running, the
|
|
power supply still delivers 5V over this _stand by_ line! Your computer normally
|
|
uses this to wake from sleep and such things.
|
|
|
|
## Design goals
|
|
|
|
I've set the following design goals for the VoltMeister 100:
|
|
|
|
* Use a small-size ATX power supply (Asus SL-22A)
|
|
* Output 3.3V, 5V, 12V and -12V
|
|
* Show stand-by and power-on conditions
|
|
* Limit output current to 2.5A
|
|
* Build a nice enclosure so it can sit on my bench :-)
|
|
|
|
The ATX power supply does a good job of limiting current and shutting down in
|
|
short-circuit situations. However, 16 amps on 12V is quite a lot of power
|
|
and will probably cause a lot of ICs to release their magic smoke. Since there
|
|
is nothing I can think of right now that'd need more than 1 amp from this supply,
|
|
I'm want to limit the output current the somewhat safer level of 2.5 amps (for each
|
|
voltage).
|
|
|
|
## ATX Connector Pin-out
|
|
|
|
ATX Power supplies have a 20-pin or 24-pin connector. These are compatible
|
|
connectors and the 20-pin connector fits the 24-pin connector.
|
|
|
|
|
|
|
|
Let's walk through these quickly. All `COM` pins are common ground and can be used
|
|
with any supply voltage.
|
|
|
|
* Pins 1,2, 12 and 13 provide 3.3V
|
|
* Pins 4, 6, 21, 22 and 23 provide 5V
|
|
* Pins 10 and 11 provide 12V
|
|
* Pin 14 provides -12V
|
|
* Pin 9 provides 5VSB, always available when the supply is connected to mains
|
|
* Pin 16 is the power switch, connect it to GND to turn the supply on.
|
|
* Pin 8 (PWR_ON) supplies 5V when the power supply is on _and_ providing stable voltages.
|
|
* Pin 20 is marked as _not connected_. In the past the ATX spec placed an optional -5V here, but has since been removed all together. Don't rely on -5V to be available.
|
|
|
|
The most important thing to note here is the `PWR_ON` pin. It supplies 5V when
|
|
the supply is providing a stable output. `PWR_ON` does not come on instantly, as
|
|
it takes some time for the ATX supply to stabalize. Although this process is quite
|
|
fast, there is a noticable delay of about half a second between switching the supply
|
|
on and the `PWR_ON` going high.
|
|
|
|
## Schematic
|
|
|
|
All the hard work of converting mains 230VAC to a more suitable 3.3/5/±12V is done by
|
|
the ATX supply. The only custom things left to do are:
|
|
|
|
* Accept a 20 or 24 pin ATX connector
|
|
* Provide a connection for the power-on and standy-by LEDs
|
|
* Provide a connection for a power switch
|
|
* Provide connections for 3.3/5/±12V outputs and GND
|
|
* Limit current to 2.5A for each output voltage
|
|
|
|
The schematic for this is rather straight forward:
|
|
|
|
|
|
|
|
The power LED is directly hooked up to pin 8, `PWR_ON` with a series resistor to limit
|
|
the current through the LED and make it not annoyingly bright. The same goes for the stand-by LED.
|
|
|
|
The power switch is directly connected to pin 16 and GND, pulling pin 16 low when switched on.
|
|
|
|
The four output voltages, together with GND, are routed to output pins so they're easy to hook
|
|
up to the actual banana sockets in the case.
|
|
|
|
Each supply voltage is fitted with a 2.5A resettable fuse (or PTC). Anything up to 2.5A is fine,
|
|
above that the fuse will start to act as a circuit breaker. When the faulty situation is resolved,
|
|
the PTCs will reset and you use the supply again.
|
|
|
|
_Note: you may notice there is a 2.5 amp PTC on the -12V output. That's weird because the
|
|
power supply limits this voltage to 0.3amps anyway. Keep in mind that other power supplies
|
|
may be rated for higher current on the -12V rail and will need this PTC to keep me safe._
|
|
|
|
## PCB
|
|
|
|
Because the ATX connector (MOLEX 39-28-8240) uses a 4.2mm pitch spacing it does not fit
|
|
on a protoboard. So, let's design a PCB! Here's my the C revision of my PCB design:
|
|
|
|
|
|
|
|
I've created wide traces for the supply voltages, as up to 5A needs to be able to
|
|
flow through them. (5A is the trip-current of the PTCs). Each voltage, together with
|
|
GND is exposed through screw terminals. There are pinheads for the LEDs and the switch.
|
|
|
|
I've opted for SMD parts because I wanted like to give SMD soldering a try.
|
|
|
|
## Lesson learned
|
|
|
|
* ATX power supplies are easy to work with, but also quite powerful. Be very careful if you open one up.
|
|
* I should mark the +/- for the LED and terminals more clearly
|
|
* Consider the connection points on the PCB, it turned out that it would have been easier if the LED/Button connections were on the same side as the voltage output terminals.
|
|
|
|
## The enclosure
|
|
|
|
I think that in a professional design process there is some tension between the PCBs you design, the
|
|
components you use and the enclosure you need to all fit it in. I tried to be smart and opted to buy a
|
|
'universal enclosure' that would fit my ATX power supply and leave some room for a small PCB.
|
|
|
|
Because of the dimensions of the ATX power supply (roughly 140x50x80mm), I needed something that would fit that.
|
|
Note that most dimenstions specified when shopping for enclosures are _outside_ dimensions.
|
|
|
|
In the end I opted for a nice Hammond enclosure, large enough to fit my power supply snugly and not
|
|
too expensive. It even has nice aluminium front and back panels. For those wondering, it's
|
|
the [1598ESGY](http://octopart.com/1598esgy-hammond-46506) from Hammond.
|
|
|
|
|
|
|
|
Another thing I learned is to take a detailed look at the specification drawings.
|
|
|
|
|
|
|
|
Notice the two supports circled in red? Yeah, those are used to screw the top and bottom parts of
|
|
the enclosure together. And guess what, my ATX power supply does _not_ fit between those
|
|
two supports and is now protruding about 10mm from the back. Booh!
|
|
|
|
This is really a problem. I don't want to have an ATX power supply hanging out of the back.
|
|
|
|
At first I thought of just buying a larger case, but I would not be that easily defeated.
|
|
|
|
Because I'm curious by nature, I opened up my ATX power supply to see what's inside and
|
|
remove some of the cables I would not need.
|
|
|
|
The PCB for the ATX supply is pretty tightly packed, but I noticed that transformer on the side was
|
|
not mounted on the PCB. There was a nice cutout on the PCB to allow room for the transformer.
|
|
|
|
Putting one and one together I decided to remove the ATX enclosure and see if I could fit
|
|
the PCB in my small Hammond case. The transformer could be moved just a bit so it would
|
|
not interfer with the support inside the case.
|
|
|
|
This worked brilliantly. All I had to do was make cutouts for the powercord and 120/230V
|
|
selector.
|
|
|
|
The only reason I did this is because there is a clear separation between the high
|
|
voltage parts of the supply and the control (for the fan, I presume). Because the
|
|
orientation was just right, it would also shield my own PCB nicely.
|
|
|
|
## The build
|
|
|
|
Here you can see the inside of my bench power supply.
|
|
|
|
|
|
|
|
The bottom half is filled with the ATX PCB. You can see the ground wire floating around,
|
|
this has been connected properly tot he front and back panels.
|
|
|
|
The transformer is leaving a bit of a gap and has been placed just on the other
|
|
side of the support.
|
|
|
|
On the top you can see my PCB with the ATX connector attached to it and wires going
|
|
out to the LEDs, the power switch and the different output voltages.
|
|
|
|
## The result
|
|
|
|
And, this is the end result. Suffice it to say that my precision drilling skills need
|
|
some more love. But other than that, I'm quite happy with how it turned out. It's my
|
|
first project, after all.
|
|
|
|
|
|
|
|
## Conclusion
|
|
|
|
Building an ATX-based bench power supply like this is nice. If you need
|
|
a stable and reliable power supply, you're probably better off buying a cheap
|
|
linear power supply online for like € 50.
|
|
|
|
I did this project to learn something. And I did. I learned how to SMD solder components,
|
|
I learned what to look out for when selecting an enclosure for your project. I also learned
|
|
about ATX power supplies and how you don't want to mess with them. These units handle
|
|
pretty beefy currents. If you don't know what you're doing you could end up hurting yourself.
|
|
|
|
What's next? The VoltMeister 200, of course!
|