Share |
Login Form
Newsletter



Receive HTML?

Latest Members


Need a handy low-cost benchtop supply? Use an old PC ATX PSU! Hot

 
User rating
 
5.0 (1)

It ain’t exactly DSP, but sure is useful!

Let’s start with the basics: for any electronics or embedded system tinkering, you have to have a good power supply. Now I don’t know about your personal lab equipment budget, but mine is rather, well… let’s just say that I’m married to the head of the finance department, and she’s not exactly an electronics engineer. If you’re anything like me, you’ve probably crafted a few linear regulated power supplies over the years, but it’s no trivial task to design and build a variable high powered lab supply. And also, if you’re anything like me, the majority of what you tinker around with is digital circuits and low-voltage analogue stuff, like audio preamps and such.

 

In most cases, I am messing around with a microcontroller and a few opamps, so typically it’s handy for me to have a 5V rail for the micro (and any glue logic I might have), and +/-12V rails for the opamps. Increasingly, the sample devices I’m working with (for example, the Freescale DSP56367 DSP, or the NXP LPC2101 ARM-7 microcontroller) require low voltage rails for internal core and IO, namely 1.8V and 3.3V. This is a pain in the butt if you’re having to build a supply with all these rails for each project, but most of them (except 1.8V) can be obtained from off-the-shelf PC ATX switching power supplies. This is great, because somehow (I don’t exactly know how) over the years I have accumulated several of them. They all have 3.3V, 5V, 5V Standby, +/-12V supply outputs.

 

Granted; on many, cross regulation of the 12V rails is not fantastic. For the vast majority of opamp circuits it is more than adequate. I have seen others converting ATX power supplies to benchtop lab supplies and putting minimum load resistors in them – in my experience this is not necessary since the cooling fan is generally enough of a minimum load, and most modern switching power supplies are designed to run down to zero load anyway. I thought I’d give it a go – I need to volts and these are just lying around.

 

One thing I didn’t want to do was simply hack up ATX cases to put binding posts on them, for a few good reasons:

  • You could really be messing with safety clearances inside the case. I don’t want to risk getting electrocuted because of the location of the 5V binding post output.
  • I want the ATX supplies to remain ATX supplies – there is a chance I’ll want to actually plug them into a motherboard again at some point, so I don’t want to hack off the ATX connector.
  • I wanted a more elegantly designed solution, that had not only the binding posts, but also some clip points I could clip alligator leads onto.

 

So I set about designing a PCB for adapting the ATX supply to benchtop use. As with anything, a good place to start are some specifications. The adapter shall:

  1. not alter the power supply itself,
  2. use an ATX connector,
  3. have binding posts for each ATX power supply voltage and standby,
  4. have a matching return (GND) binding post for each voltage output,
  5. be capable of handling heavy supply currents,
  6. have a switch circuit to make use of the ATX on/off controller,
  7. have LEDs to monitor standby and Power Good signal, and
  8. use through-hole PCB design so you, dear reader, can make one for yourself easily.

I had also considered a built-in panel meter so you could monitor the output voltage of each output, but quickly realized this would be unnecessary feature-creep for this, since the outputs are fixed and fairly well regulated, and most users (like me) can plug in their multi-meter and get a more accurate measurement anyway.

 

The switching circuit is a simple latch based on a two inverters from a 4049 HEX inverter, powered from the ATX standby 5V rail. The other gates in the inverter are used to drive the power on control signal to the PSU, and the LEDs. Pressing S1 turns on the power supply while S2 turns it off again. I used a yellow LED for D1 (indicates when Power Good goes low, about 300ms after power up), and green for D2 (indicates standby mode). I designed the PCB and had a batch made up for prototyping and testing. I must say I think they came out rather well. The front and back of the boards are shown below:

 ATX Adapter board photo showing front and back of board.

You can see clearly here the plated slots I designed in around the edge – these are for clipping alligator leads onto, and the silkscreen text indicates the output of each one. The plated-through holes designated K3 to K14 are for bolting the back conductive end of the banana socket / binding post onto. Notice the used of multiple vias in a circle around each of these and also the mounting holes – these offer increased strength for tightening up screws over them, and also for the binding posts a good low-resistance current path through the PCB.

 

Below is the back of the final assembled prototype. I got the ATX PCB mount header from Jameco Electronics – they’re only about $0.50 each. Alternatively you could de-solder one from and old dead Mobo (get out that blow-torch!) You can see the connectors I used did not have the optional extra two rows like what you find in modern motherboards. But I did design the PCB to use them if you want to.

 Rear view of final assembled prototype showing ATX cable connected.

…and of course, it wouldn’t be complete without a front view:

 Front view of adapter board with DMM leads connected

Here you can see all the components clearly. All the capacitors provide a bit of decoupling for the different supply outputs, except C2 which is for de-bouncing the on-off circuit. All the others are ceramic 100nF caps. R1 and R4 are LED dropping resistors and R2/R3 form the feedback paths for the on/off dual-inverter latch.

I have now been using this board extensively at work and at home, powering several development kits off a single ATX supply (at work I have a Nanoboard-II and two Nanoboard 3000 FPGA development boards running at once, as well as a few other things). It has really simplified things well for me!

Finally, I have a number of these handy boards made (obviously, it’s uneconomical to get just one made for me). If you want one, please send me an email and let me know: ben [at] jordandsp.com

P.S. Due to popular requests I've turned this into a complete kit, with everything you need to make it. Please email me if you would like to get one!

User reviews

Average user rating from: 1 user(s)

To write a review please register or login.
Overall:
 
5.0
 
 

Great way to reclaim old supplies

Overall:
 
5.0
Joshua Israelsohn Reviewed by Joshua Israelsohn
April 24, 2010
Comments (0)
Report this review
 

Ben-

This is a terrific post. With the countless tons of computer equipment that enter the landfill every year, it's good to see a simple and practical way to keep some fraction of it in service.

The only thing your design doesn't accommodate is the minimum-loading requirement for ATX supplies. A bit long-winded for this space, I've written up a discussion of this topic in a separate posting entitled "Further to Benjamin Jordon’s ATX PSU to bench-top supply conversion project"

Owner's reply

Thankyou Joshua for the addendum.

I do agree that in general, SMPS require a minimum load (in fact most small flyback supplies I've seen have the minimum load resistor in them - which leads to a whole other discussion about energy wastage of PSUs in "standby"... perhaps a good topic for the Green Power community).

And yes, the back of the binding posts are a handy place for this!.

Typically, I use mine with a load connected (three Nanoboards) all the time. And for breadboarding the particular supply I'm using with it (a small Morex MFM145) is well designed - it has low ripple under no-load conditions.

So yes, for random supplies it's a good idea to check ripple with a DSO if you can first, and decide whether or not the one you are using needs a minimum load for your purposes, or simple use linear post-regulation or zener clamps on your bread board ;-)

Ben.

 
 
Written by :
Benjamin Jordan
 
 






Latest Content
User rating
 
0.0 (0)