Hacking SMPS PSUs?

Status
Not open for further replies.
Jump to Latest
S

Spasticteapot

Member
Joined 2006
#1
Computer power supplies rated for 400+ watts (okay, maybe 250 watts sustained - yes, the cheaper manufacturers pad the specs) can be had for $15. They include a "sleep" power output, and are easy to turn on and off with a microcontroller. Plus, most of them have anti-short protection, and many other useful features - even on the cheaper models. (Hurray for huge demand and huge production!)

So, why not modify one for use with an amplifier?

I'm a bit clueless when it comes to anything like this, but I can say that I'll ship a decent computer PSU to the first person who can seriously say that they'll figure out how to hack it into something usable for, say, a Gainclone. (30-0-30 rails).
 
#2
Hi

In principle, it isn't too complicated: basically, you've rewind the main secondary of the power transformer, adapt the main choke in the same way, replace the ouput diodes and caps with suitable types, and scale the opto-coupled feedback loop to the new voltage.
There are some snags however: the transformer is almost invariably glued/impregnated, which makes the disassembly horribly difficult (in my experience anyway); another potential problem is the frequency response of the servo loop: for amplifier types other than class A, the current drawn from the supply will vary enormously with the musical contents, and this may well upset completely the regulation. It will probably be necessary to add big caps at the outputs to offset this problem.
LV
 
#4
Elvee said:
Hi

In principle, it isn't too complicated: basically, you've rewind the main secondary of the power transformer, adapt the main choke in the same way, replace the ouput diodes and caps with suitable types, and scale the opto-coupled feedback loop to the new voltage.
There are some snags however: the transformer is almost invariably glued/impregnated, which makes the disassembly horribly difficult (in my experience anyway); another potential problem is the frequency response of the servo loop: for amplifier types other than class A, the current drawn from the supply will vary enormously with the musical contents, and this may well upset completely the regulation. It will probably be necessary to add big caps at the outputs to offset this problem.
LV
Click to expand...


Desolder the transformer, then heat it in the oven -- it will easily come apart.

To satisfy the varying current demand you have to store energy in the inductors -- so the design gets a bit more complex -- I am pretty sure that Texas Instruments has a "how-to" on their website if you search under the TL494 -- you can bracket the current demand around the energy used while a class-B amplifier is coasting, and that of the peak current demand. As mentioned, when you change these parameters you also have to change component values in the servo loop to maintain phase-gain relationship.

Note that the switching transistors have to be re-rated for the much higher voltages -- you will also have to be ultra-careful with filtering the input to the power supply -- and the outputs.

Sound complex -- well it is and it isn't -- there are a lot of web articles on modding AT and ATX supplies.
 
#5
jackinnj said:



Desolder the transformer, then heat it in the oven -- it will easily come apart.

To satisfy the varying current demand you have to store energy in the inductors -- so the design gets a bit more complex -- I am pretty sure that Texas Instruments has a "how-to" on their website if you search under the TL494 -- you can bracket the current demand around the energy used while a class-B amplifier is coasting, and that of the peak current demand. As mentioned, when you change these parameters you also have to change component values in the servo loop to maintain phase-gain relationship.

Note that the switching transistors have to be re-rated for the much higher voltages -- you will also have to be ultra-careful with filtering the input to the power supply -- and the outputs.

Sound complex -- well it is and it isn't -- there are a lot of web articles on modding AT and ATX supplies.
Click to expand...

Seeing as how I'm still not totally sure of how ATX power supplies work...mind explaining it in "stupid people lanuage"?

I was thinking along the lines of using the inductor/transformer section, but replacing the low-voltage switching regulators with more powerful switching regulators.

Also, 48v supplies for industrial computing are not uncommon surplus finds. Anyone know how to add a split-voltage "ground" to produce 24-0-24?
 
Status
Not open for further replies.
Top Bottom