SMPS both rails stabilization

Status
Not open for further replies.
you could probably get it most easily in DSP, but then you have to spend time programming.

Linear Tech's application note #70 "A monolithic switching regulator with 100uV noise" explains the attempts that have been made in the past -- when you look at the time, trouble and real-estate consumed, an SO-8 device which does all the work is a small investment for a DIYr.
 
Notice on how to have 97.5% effiency...

Hello Again...Sorry for my english! 97.5% effiency was obtened with direct drive mosfet (ie with electronic drive directly mosfet, very fast rise time, negative discharge...) and 3F3 toroidal transformer in resonant mode, exactly 49.7% duty cycle (cue here...) at 62.5Khz, crystal controled SG3525...Try it you will know why after 😱) !For the amplifier, a full bridge standard class d with negative side tighted to ground for bridging purpose with SGS thompson STW50NB20 mosfet, direct drive too with differential output filter....That give 170A curent capacity in 50% duty cycle full bridge topology, so make calculus on how current you need to get 4Kw in 4 ohms at 180V power supply and with RDS of mosfet you will see that very small heat sink will be needed....Last trick now....When I talk about direct drive, forgot the IR2110 serie...That just for kid's....Use TC4429 driver and 6N137 optocoupler with 4.7 ohms resistor and a 1N4936 reverse diode ( dont forget dead time in your electronics!) and use good differential common mode filter at input and output over 1Mhz....That's all for now! My news 4Kw per channel amplifier in 1 Rack unit will be avaible soon! www.d-amp.com

Thank's

Fred
 
Fredos, you put here a very valuable information, at least for me. I have also cores made of 3F3 material but even if my frequency is 250kHz, I need two of them to be stacked to suffice the power and that forms a 7cm high and 8cm in diameter transformer, so it could not fit into 1U... I wonder how can it fit into 1U case with 4kW at 62.5kHz... What is your core diameter and height?
Have you ever heard of or tried a planar transformer?
 
Sorry Fred but there again some questions arise. Datasheets for SG3525 claim minimum of 0,4-0,5us dischadge time for each half-period, which yields maximum of 47% duty cycle each output signal for 62,5kHz (16us period). I tried it a lot and affirm 47% is real. Then, your "0.497" gives a deadtime of only 50ns! Which is less then intrinsic rise/fall time of driver, not mentioning MOSFETs real application switching - how about shoot-thru :hot:
So do you confuse this value or mystify?

I also know well about STM FETs. Actually in my similar class-D amplifier I have used their W34NB20 type. But the best efficiency what I've got from it, was about 89% together with power supply. And, they are not that easy to drive; anyway these are one of the best of 200V FET series at present but still worse than 150V or 100V series by their switching parameters :xeye:
Well, I never came across the switching amplifier efficiency of some 97% (excluding your claim 🙂), even alone, without converter. So... that sounds just like a dream to me.

For "97% efficiency" Google returned me not many relevant results and here is one of them
http://www.powersupply.com/
So they say they are the best and make power supplies "with up to 97% efficiency".
And here is the best what Google found for switching amplifier
http://www.soundlabsgroup.com.au/lcaudio/lc_audio_zappulse.htm
Well-known product though, so, "approximately 96% @ 400W RMS into 4 Ohms"

I don't want to sound dogmatic, but I have to base on and compare to my own experience. Saying again, entire system 97% efficiency is rather a dream for power designers... So, supposing that by some means your claims are brought to life - you're going to be a No.1 guru. As the first step, open the new threads in D-class and power supply forum and say aloud, I DID IT ! :clown:

By the way, I asked before about 4 cooling fans. Personal computer consumes average power of about 60-100W and is satisfied with 2 fans. So why 4 fans needed for 50W dissipation, especially assuming easy possibility of heatsinking? Yeah I'm kind of sceptic today 😎
 
Some information here for your ''pumping'' PSU

I just made a cut and paste of other post I've made...

Hello guy

You can see news picture of my 8000Kw amplifier on my web site (www.d-amp.com). Thank's to all guy who have reply to job offer, I got my guy!

For all other who ask me question about design, most of them was about pumping effect on single ended amplifier...I can give you my trick on my first series, I'm now in 4 quadrant output stage, so no more pumping effect.

The simple way is too add a small switching power supply who return (clamp) the amplifier bus voltage to the bulk capacitor of the main switching power supply. The trick is that if you have a voltage doubler main switching power supply (means 340V), and for example +/- 90V amplifier power supply, you just design a 180V (sum of the +/-90V) to 330V power supply. If amplifier rail go over 186V, the impédance of the main line will ''clamp'' the voltage trought switching. A simple IR2153 with a pair of IRF640 and a TN29/19/7.5 toroidal will do the job.

In that way, you will clamp amplifier rail with very high effiency! And best, return this energy to power supply...Think about this!

Another trick I have use in my model 2000, is to simply add 2 more turn on the output filter (speaker side) with 2 MUR840 diode to clamp to the power supply. This will saturate the coil when speaker feed back their energy to the amp and force amplifier to compensate (if you use analogue feedback instead of ''digital'' feedback!).

So thank's to Dan Blouin to join the D-Amp team!

Have a nice day!

I share some trick, so dont think I'm here to take you idea and the money!

Now, just to know if someone have built amplifier with analogue feedback and have some waveform probleme ( on the scope) over 10Khz....But amplifier pass distortion mesurement...Weird...I work on this now....

Bye

Fred
 
fredos,

This is probably off topic, but since we're drifting...

You mentioned something a few posts back about using a crystal to control the oscillator of the SG3525. I haven't the time to try now, but can you give some insight as to why you aren't just using the internal oscillator of the chip? Is there a stabilization/accuracy problem with the internal oscillator? I've seen this done before, but could not see the purpose.
 
carvinguy said:
fredos,

This is probably off topic, but since we're drifting...

You mentioned something a few posts back about using a crystal to control the oscillator of the SG3525. I haven't the time to try now, but can you give some insight as to why you aren't just using the internal oscillator of the chip? Is there a stabilization/accuracy problem with the internal oscillator? I've seen this done before, but could not see the purpose.
i'm also interested in crystal controlled , but i dont know how they work 🙁 can anybody explain that?
 
Hi Everyone,

I think I could throw some light on the use of crystals ..

The Crystal controlled oscillator is usually designed with CD4060BC [ripple carry binary counters] and the output is fed to "SYNC" pin of SG3525 IC ... The time period of internal oscillator of SG3525 must be adjusted 2 times that of crystal oscillator in order to ensure good working...
the crystal oscillator ensures:

1 Precise Oscillation i.e. High Precision Clock signal helps in predicting various necessary parameters involved in desgning ..

2 Eliminates the Beat Frequency Oscillation, when 2 or more supplies are working in close proximity and sharing a common ground and keeps the beat frequency to infrasonic range below 20 Hz so that it is unaudible to the ears .

3. Helps to synchronize multiple SMPS with One Master Clock..


regards,
K a n w a r
 
Status
Not open for further replies.