Simple smps for 100W audio amplifier needed.

Acca

New member
I Think Acca's smps that's what I need for beginning :) Not complicated, simple and cheap :)


MicrosiM - in audio i'm a little bit more experienced that in smps, my proposal is:

-unregulated ~ 300W power
- Output voltage to be changed by changing number of turns at the secondary side.
- PFC is not required, maybe we can focus on a good filter on the primary side ?
- +/-10% stability is for audio enough
- over current isn't a must, but good idea, more important is that te overvoltage protection works in 100%
- not heavy winding of the transformer(s)
- can't generate much heat - in amplifiers we have enough heat :)

Other technical details of the smps - I don't know, I'm not good at it :)

You ask - why not 1000W ? Because homemade amplifiers are most under 300W. Of course, if the 300W version will be stable, than more powerful one can be built.
My SMPS have all of this.
-300W unregulated
-probably stable voltage
-overcurrent can be easily added,I explained in my theme
-voltage change by number of turns.
-over voltage is not needed,because smps works on max duty,and output voltage is max.
-it don't generate much heat.
 

Acca

New member
As a start, I think the power supply specifications will look like

1- Unregulated, 800 ~ 1000W SMPS.

2- Output voltage to be changed by changing number of turns at the secondary side.

3- Must use simple and rugged AUX sypply circuit to feed the Oscillator and the Mosfet drivers.

4- Adding PFC will be a big +, but I dont think all DIY can get all parts for the PFC, unless we organize this as a kit, I dont know

5- Half bridge configuration is best option for this medium power SMPS.

6- Over current protection is a must.

7- Must have a Turn On Delay (Line in rush current limiting)

This is for now, hope to hear more

Regards
I have this in my repertoire.
But it's fullbridge-effidency is better,less heat
 

Acca

New member
At the beginning I want to say hello to everybody :)

I've spent some time by reading many threads but I didn't found anything for me.
Maybe I specify what I'm looking for:

- simple smps for a 100W audio amplifier
- has a single-sided board
- can be built and run by a smps beginner ;)
- 10% stability would be very good for me
- overvoltage protection if it's not complicated
- can run on a etd39 core, like this one:
View attachment 4233

Why that core ? Because I've bought one years ago and if it could be useful... :) But of course it's not critical.
It would be a great feature if I could change the output voltage only by changing the secondary winding (had not to correct the voltage output protection circuit - maybe a separate winding for this ? ).

I think it's all I need - sorry for my english and I hope you understand all this... I'm sure that You have many that kind of circuits.

Greetings :)

ETD39 is enough for ~400W
 

KX36

New member
MicrosiM's specs have diverged slightly from what I meant. Just to clarify; I would not classify an unregulated power supply as an SMPS. All these IR2153 (IIRC) supplies I'd consider to be linear supplies running off a high frequency square wave instead of a low frequency sinusoid and no more.

I know regulation isn't massively important to the audio community, but if you're going to do it, do it right. A proper SMPS sin't that difficult to design at <500W. The only real headache is the itteration involved in transformer design.

So I'll probably split my design into a different thread but to start me off, lets say you want a typical class AB transistor amp with 100W RMS audio output on I presume 2 channels (is that reasonable?), very simple by audiophile standards but in my inexperienced eyes, a start. What voltage rails would be most useful, and how much current would each rail need to supply?
 

Acca

New member
If you wish regulated,use tl494.I help you to calculate transformer.For <300W you can use one EE33,for more power use two indetical.
If response time is slow,it's better to use unregulated smps.If response time is good,a little better is regulated.But it's more complex.
For 100W amp on 4 ohms,we have +-35VDC.For regulated PSU,on one EE33 we must have 2*20 turns for primary and 2*18turns for secondary.For unregulated we have 2*9turns.
For 2 channels total power is 200W.For this 200W SMPS is enough.(classic transformer must use >300VA,because of power facor).
I have idea about this with TL494,it's like AT supplies.You can use flyback or forward topology if you wish,but:
Flyback transistor see higher voltage than rail voltage input (325VDC).It's probably cca. 600V or higher.
But it can be simple schematic.And core dimensions is higher than half bridge (HB)
With EE42 in flyback we have 150W max.With halfbridge power is bigger with this core - cca. 450W on same frequency.
HB is complex,but it's for higher powers.Forward converter with EE42 can put 200-300W on output.
HB can be realised in self osc. mode very simple (My Uni4 smps is like this),but unregulated.
For regulated psu we must use PWM IC.He "lock" frequency at desired.For HB we can use FET's or BJT.
Fet are voltage controlled device,and BJT is current controlled.Because of this,in BJT topology we have proportional base drive or full drivered BJT.
Prop. base drive is better choice,because base current is proportional to output current.Full BJT drive is worse choice,and switching times will be high.
For fets,we must use gapped core in drive,or it will go in saturation.Or we can use drive IC (like IR2110 or similar).
You should choose one of this.Then tell me for which topology you have chosen and we will elaborate it.
If you started new topic,send me a link.

Regards
 

KX36

New member
Depending on power level, I'm likely to either do a UC3844 based forward converter or UCC28083 based push-pull. Unlikely to do a voltage mode control unless i end up using a half bridge as CMC is probably more suitable for the above topologies. In CMC, current limting is intrinsic, it's simpler to compensate than VMC, it potentially has a better transient response and it more easily accepts a wide range of input voltage. There are disadvantages to CMC too; peak CMC requires slope compensation above 50% duty cycle and it's suceptable to noise on the current sense circuit.

If I were to do VMC, I'd use a UC2525A (higher spec replacement part for UC3525 or SG3525, not much more money) before using a TL494. TL594 is usable too with some external circuitry for cycle-by-cycle current limiting etc, but it's really quite obsolete in SMPS. Compared to TL594, TL494 has no UVLO and a less precise regulator, so I don't see why you'd use it when it's only £0.10 cheaper. If I were to do VMC, I'd probably also either include some form of voltage feed-forward circuit or include a boost PFC pre-regulator or both. Then I'd probably use the UCC3580-4, which IIRC is a well suited chip, it's a bit more expensive but auidiophiles like expensive parts, they've been proven in listening tests to sound better ;).

I know relatively well how to design a transformer thanks. I'll get out the core and material datasheets, do the calculations and see if the winding fits the window correctly ;). It's an itterative process that there aren't many easy ways around and there's no one-size-fits-all transformer. Transformer saturation isn't really an issue in SMPS (barring topology specific factors like flux walking which should be dealt with appropriately, e.g. using CMC for push-pull or full bridge coverters). At the high frequencies used, the limiting factor in transformer design generally is the acceptable losses rather than saturation.

Switches will be MOSFETs, not BJTs or IGBTs. There's no good reason to use anything other than FETs here. Secondary rectifiers are likely to be Schottky diodes. There's also no reason to use quasi-resonant control at this power level. We favour flexibility in design and load conditions over getting the highest efficiency so typical hard switching with appropriate snubbers and clamps should be fine. Switching frequency is likely to be around the 100kHz-200kHz mark.

[P.S. Have you noticed that I like TI/Unitrode parts yet? Other Mfgs like linear, ON semi etc make good parts too, but I always find myself drawn back to the TI parts. The above parts are for the most part quite generic, you might use more external components but you can tweak them more to your needs IMHO]

Code:
[U]Current Prices on RS Electronics (UK):[/U]

TL494		VMC	£0.34 (was £0.56)
TL594		VMC	£0.65
UC2525A		VMC	£1.63
UCC3580-4	VMC	£4.83
		
UC3844		CMC	£0.39
UCC28083	CMC	£1.97
 

Neftali

New member
Troubleshooter

Hello "mayster"

It all depends on your amplifier´s type and its output impedance. For a Class AB amp, at 2-Ohms output impedance, you will need a dual 20VDC (+ voltaje accross ouput devices) @10A power supply; for a Class AB amp, at 4-Ohms output impedance, you will need a dual 28VDC (+ voltaje accross output devices) @7 A power supply; for a Class AB, at 8-Ohms output impedance, you will need a dual 40VDC (+ voltaje accross output devices) @5 A power supply; for a Class-AB amp, at 16-Ohms output impedance, you will need a dual 57VDC (+ voltaje accross output devices) @3.54A power supply. Which is your case?

I will go with the most-common 100WRMS @8 -Ohms, Quasy-Complementary Output Stage (BJT) Audio Amplifier. OK? The Quasy Complementary Output stage demands for an extra base-emitter junction, so we will need an extra 700 mV added to the 40VDC and if your amplifier is equipped with emmiter resistors (0.22 -Ohms), then we will need another 770mV to compensate for the voltaje drop through these emitter resistors. The total voltaje for your power supply will be 41.47 VDC and the current will be 3.54A.

Dual 42VDC_3500mA SMPS.jpg

Coupled Inductor Dual 42VDC_3500mA SMPS.png

Good luck with your project!!!
Troubleshooter
 

Neftali

New member
I know you were aiming at a more sofisticated SMPS; nevertheless, I believe in simplicity. Power Integrations has developed these small devices and they are perfect for an aplication the size you need (they are not small at all).

The differences between an ETD39 and an E42 ferrite cores are minimal, so maybe you can try to build this transformer (coupled inductor) on an ETD39 core. If you go through space limitations as you switch cores, you can give up the primary shielding windings (9 turns of 4 trands of #25AWG wires connected to pin 1 and going nowhere). Wind all parts on the same direction.
Best of lucks
 

mayster

New member
Hi !

Thanks for many anwsers :)
My personal life had changed for past two years, I have a break in electronics. But when I finish some things, there'll be a comming back ;)
I read the thread and almost cry that can't build anything :)

Hope to back soon :)
 

Neftali

New member
On my previous answer I included 2 attachments. One for the schamatic of a SMPS I designed according to the data you provided, and the other one, was a diagram on how to build the transformer. This is a seconda attempt.Audio-Compressor-AGC-Schematic.jpgCoupled Inductor Dual 42VDC_3500mA SMPS.png
 

Attachments

  • Dual 42VDC_3500mA SMPS.jpg
    Dual 42VDC_3500mA SMPS.jpg
    19.5 KB · Views: 85
Top