Computer SMPS and Gainclone

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
no you can not do that because both voltages are referenced to the same ground. if you connected them together then you would probably blow the 12 or 5 volt lines (or both) or at least short circuit protection would trigger.

also you need complimentary positive and negative power rails for LM 38xx type amps.

If you want to use a computer PSU for an amplifier you could use the TDA7374 which is designed for car radios. so it will work on 12 volts. Dont expect any astounding sound quality out of it either (esp with comp PSU)
 
If you dont decide to use the power supply itself, you could do as I plan to do and use the metal box from the psu to house your gainclone (minus psu bits inside, of course). It should have a nice dual pole switch and some vent holes - and as its metal you can mount your chip(s) to it as long as you insulate them :)
Then just spraypaint it your desired colour....

I'll post some pics of my one once I've finished. :smash:
 
You can disconnect the Ground reference point (chassis connections) on a computer power supply and supply it with +12v from another computer power supply. As you can have these amplifiers in single ended mode you could chain 3 or 4 computer power supplies in series.

for 3 computer powersupplies its potential from ground on PSU#1 to +12 on PSU#3 would be 36 volts, and depending on the PSUs's used you would have roughly 36 volts @ 10 amps.. 360 watts. You can also adjust the output voltage on the power supplies, up to around 14 volts. giving you 42 volts. ofcourse 42 volts into a single ended amplifier @ 8 ohms is only going to give you around ~20 watts audio power :)
 
li_gangyi said:
the negative 12V is rated at an amp due to it being provided by a 7912 regulator...I dun think u'll be able to power ur amp that way...


Replace the 1-amp rectifiers for the -12V side with some 10-15A rectifiers and you should be able to realize some real current for the -12V output, thus giving a near-symmetric (+/-) output for your LM3886.
 
Ever since that was posted, PC power supplies have gone further.
Some now have the -12V from a seperate winding on the main transformer. I guess the best way would be to strap a few of these together, and have the chassis connections isolated, don't know how safe that will be though. I guess with all the stray impedances, you could get a rude shock if you were to touch the chassis (of the SMPSU).

Your note of using a 10A reg seems intriguing, care to share how that can be implemented? Don't know if the -12V line will like it that much. They seem to be wound with finer wire than compared to the main +12 and +5 windings.
 
Forgot about that. This applies for all AT & ATX Supplies deriving the -12V output from the same windings as the +12V line.

If yours is one of the newer ones with a separate -12V winding, then some more delicate work will be needed, perhaps a small daughterboard mounted very near the +12V rectifier.
 
Cheapo computer switching power supplies are in general very shoddily made, and I would not trust one to run anything of mine. The are patterned after a cookie-cutter half-bridge design using a TL494 (or equivalent) controller and bipolar switches. They have minimum load regulation issues and have next to no overload protection (the TL494 has no provision for overload protection, and no one attempts to do a work-around on the cheap supplies as it would cost a few cents extra on a design that is already pared down past the bone). I would strip one for parts and heat sinks (woefully undersized), but that's about it.
 
Wrench,

Before you dismiss the whole class of CPU switchers, take a look at some of the modded AT & ATX PSUs in this forum. While it is true that there are many cheap designs, either poorly designed or having empty spaces on the pc board where parts were deleted (like AC line filters, or output chokes), or undersized components (like a T-80-26 toroid where there should be a T-106-26 unit, or maybe 100 mF / 200V caps for a 250W design) there are some really well-designed and well-built ones.

I am working on an old AT (if I ever get it finished) that has the full 9 yards: AC line filter (both common- and differential-mode), foldback current-limiting, oversized DC bulk caps (470mF /200V), L-C-L-C output filtering for both the +5V and +12V lines, not to mention output voltage adjust and current limit adjust.

I agree that there are far better controllers, like the SG3525, UC1846, UC1856, MC33025, etc., but the TL494 has been around for decades, and has a well-proven design. OK, I'm trying to make a point here, but I got lost in my own ramblings :xeye: :eek: :confused: :whazzat: so I'll stop here. I guess it's just a good idea to consider every approach when choosing a power supply for your gainclones.

Cheers,

Steve
 
There are some reasonably good power supplies for PCs made by first/second tier suppliers (mostly in name-brand computers), but the no-names are mostly garbage and not to be trusted. The basic TL494/bipolar half-bridge design was widely promulgated by the Taiwan government in the 90s (I saw the course book in my FAE's car during a trip there a few years back), so every wanna-be supplier makes them. Even with the better suppliers, I would worry about minimum load issues, as the newer PC supplies address this issue at low/zero load by shutting down the main supply and relying on the regulation of the standby supply. Generally, the control loops are closed around the wrong output (+5V) for gainclone applications, and it would take a lot of intelligent hacking and jumpering to move the loop where it needs to be. I'm not even mentioning layout issues. What I'm trying to say is, unless you're extremely knowledgable/lucky, the effort may well be akin to polishing a piece of excrement - it looks the same afterwards, and you get a lot of nasty stuff under your fingernails... I used a toroidal mains transformer for my OPA541 subwoofer amp so that I could forget about the power supply and focus my attention on the business end of things. I design and build switchers for a living, but I don't necessarily want to shoehorn them into every application.
 
It's good he got the thing to work for him. However, I wouldn't call it a universal endorsement. It's a project that should be approached with a healthy dose of caution/skepticism. One caveat - the supply should be brought up/debugged with a dummy load equivalent to the quiescent current draw of the amplifier chips in question before it's turned loose on the actual amplifier. This allows one to sort out regulation issues without blowing the daylights out of your pride and joy. BTW, your friend did pretty much the same thing I would do if I were trying to adapt one of those things to analog duty. For those who want to try, try to pick a supply that actually has an EMI filter and safety marks (though those can be counterfeited by the unscrupulous). These kinds of things make it more likely that the manufacturer was at least trying to exercise due diligence.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.