Where are the simple diy SMPSe for Class-D-Amps?

[NebuK]

Heyyas,

after tinkering around with some gainclones, a dac and a small tube amp i began digging into class-d amps and active crossovers because i wanted to make diy plate amps for subwoofers ... Now after quite some while of reading, reading, asking a etech friend many hours of annoyingly stupid questions and then some more reading i think i've found everything i've wanted: I'm currently making a nice adjustable LR crossover pcb, with the simple yet nice class-d designs found here on the forum and/or the iraudamp designs about as much power as you'd want are at your hands with good enough sound ... just the power supply is giving me headaches.

I do know that most class-d amps (except the iraudamps) have a quite low psrr and therefor require a good power supply - yet i somehow refuse to build a efficient, "cold", modern, high-powered amplifier and then use a 1920s style transformer-buffer-regulator power supply which needs extensive cooling, is large and unefficient.

Therefor i've started looking into what SMPSs are out there ... many people seem to go for a meanwell or similar brand premade SMPS. Problem here's that you need two and they're quite costly. Also, if you'd want *very* high voltages for the simpler higher-power class-d designs you're stuck. Meanwells here in Germany go for ~80Eur (48v / 300w), so if you need two of them for a symmetrical supply its quite costly.

On the DIY front i also haven't had the luck to find designs that kind of match what i want. Most of them are very very complex making it hard for a beginner like me to design a PCB for them or they're quite low power or low-voltage.

So ... in the good hopes that i've overlooked something really neat somewhere - the question to you out there: Where are the SMPSs for "higher power" Class-Ds out there?



Thanks in advance, kind regards
- NebuK

 

[Th3 uN1Qu3]

Well, SMPS design isn't simple. But for starters, if you want to learn, you should pick up a cheapo PC power supply, which is controlled by TL494. You can learn by modifying that. If the controller is one of those obscure chinese types, you can patch a 494 in - but that's more advanced. I know all the cheap ATX PSU platforms inside out, so if you decide to go this route i can help you.

It goes without saying that you will need an oscilloscope - DON'T attempt to build a SMPS without a scope or you'll get nothing but smoke and headaches.

 

[Bibliophile]

It goes without saying that you will need an oscilloscope - DON'T attempt to build a SMPS without a scope or you'll get nothing but smoke and headaches.

And an isolation transformer.

 

[Th3 uN1Qu3]

Or an inverter and a battery.

 

[NebuK]

Heyyas,

thanks for your replies!

While i haven't got a scope myself a friend of mine has one, which is good enough for me right now.

@Th3 uN1Qu3: Thanks so much for your offer for help - but thats one thing i really don't understand about SMPS builders. Whenever i see a thread here its like "hey, i've got this $USECASE and i need a smps, so i'm doing it from scratch right now, could you look at the schematic?". Why - most of the time - are there no replies like "huu, why reinvent the wheel? $COOL_KNOWLEDGABLE_USER21 did just that over there!".

I'm quite a electronics noob, so designing a SMPS myself is way out of league. What i've found are tons of nice class-d designs which i could start off copying and learn by tinkering on them. I haven't found that for SMPSs. Noone seems to offer a robust yet (at least more or less) simple design that beginners could just build for a starter - and for learning. When you're not good enough yourself you copy others until you get to understand what the hell they (and you) are doing. I'm kind of missing this step for the supplies :/.

Thanks again, Regards
- Dario

PS: When tinkering around with ATX PSUs, what exactly did you mean? Is there some way the output voltage can easily be raised by altering a few parameters, maybe replacing a filter or two ... ? Sorry for asking so dumbly/boldly

 

[Th3 uN1Qu3]

I'm quite a electronics noob, so designing a SMPS myself is way out of league. What i've found are tons of nice class-d designs which i could start off copying and learn by tinkering on them. I haven't found that for SMPSs. Noone seems to offer a robust yet (at least more or less) simple design that beginners could just build for a starter - and for learning. When you're not good enough yourself you copy others until you get to understand what the hell they (and you) are doing. I'm kind of missing this step for the supplies :/.

Take this here schematic: http://www.smpspowersupply.com/ATX_power_supply_schematic.pdf It is representative for all cheap chinese PSUs floating around.

Also check out "Switching Power Supply Third Edition" by Abraham Pressman, Keith Billings and Taylor Morey. That book basically takes you from zero and explains everything for you, with as little math as possible. There are a number of inconsistencies such as a formula that you have to go back and look for (but if you're paying attention you'll find it), and some things that are better explained in other books, but to me it was the best starter book in SMPS design.

PS: When tinkering around with ATX PSUs, what exactly did you mean? Is there some way the output voltage can easily be raised by altering a few parameters, maybe replacing a filter or two ... ? Sorry for asking so dumbly/boldly

The output voltage can be raised to about 18v on the 12v rail without losing regulation by a trimpot in the feedback path of the TL494. If it's got another controller this doesn't apply, as the "all-in-one" ICs include over/under-voltage protection and will not allow you to do this. Best find a power supply with TL494 and LM339. To get 18v safely all you need is new secondary capacitors.

Then for more voltage you rewind the transformer and the inductors and replace the diodes. Or, for an unregulated supply (which i would do as first step, as the feedback loop is the most complicated part of a SMPS) you remove the inductors entirely. An unregulated SMPS will need larger capacitors, both primary and secondary to avoid mains hum getting in your amplifier. Primary i'd use 2x 1000u minimum, maybe larger (you may have to get creative about fitting them), and secondary 2x 10000u. Just like in a regular mains transformer based PSU. When you go regulated the capacitors shrink by a factor of two or more.

The transformers are usually E+I. For more power, you put two 'E's together for a larger winding area which enables you to use thicker wire. For audio amplifier application, a typical ATX PSU can put out 650W with a suitable transformer and primary caps if the primary transistors are TO-247, or 400W if they are TO-220. This might seem surprising considering that those pieces of junk usually blow at 250W out, but audio amplifiers require less current (but at higher voltage) than computers, so less current -> higher efficiency -> more power output from the same devices.

 

[NebuK]

Heyyas,

thanks for your reply. That ATX modding sounds quite interesting already ;P. I'm trying to acquire a few PSUs in the next days (friend of mine collects old hardware, so should be no prob ;P) and open them to see what they use.

I've also tried searching around to find voltage modification threads, information, anything that i could suck in and feed to my grey cells ... not much luck though. Do you (appart from that awesome schematic that will help tons, i believe!) happen to have some links to nice atx psu modding information and/or general information on them at hand?

Thanks once again! Regards
- Dario

 

[Th3 uN1Qu3]

Most ATX PSU modding that i know i've done myself, so can't really help you with links. I tend to document projects in detail and they end up left unfinished because of some minor glitches, but i've also done successful modifications that ended up undocumented.

Once i've been thru the circuit once, i don't need to document everything in such detail again, so i just skip that part. That's why circuit design books are sometimes hard to read, because it's difficult to explain something you've been doing for years. I've only been messing around with them for 4 years (i'm 20), and i already tend to overlook certain things when explaining to someone, i can only imagine how tough it is for an experienced engineer in his 50s or 60s to write a book about something he's been doing for 30 years. It becomes automatic.

I did have a few links but i can't find them now, and all of them had something i did not agree with... Best way is to learn. There's no better way of learning a circuit than blowing it up in 10 different ways and documenting all of them. I know all the typical failure points of ATX PSUs - how do you think i found out?