From: info@turneraudio.com.au   
      
   On 03/30/16 07:27, Patrick Turner so wittily quipped:    
   - show quoted text -   
   sure, given big enough output components, 500VDC at 2A or even more is    
   not unreasonable.  "big fat MOSFET" devices switching high volts, or    
   IGFETs if they can switch fast enough.  Old-style computer power    
   supplies were upwards of 250W in their day, so just take that kind of    
   power and some large enough capacitors.    
      
   But I see no schematics I could build, anywhere.....   
      
   the microcontroller simply measures the voltage and generates the PWM    
   signal.  That signal would then be amplified and passed along to the    
   output stage.  So it doesn't really matter what the output stage looks    
   like as long as you handle 'crossover' properly and get it to    
   efficiently switch whatever supply volts you feed it with.    
      
   I guess the mosfets are either ON or OFF, and in either state have very low   
   Pdd, so hence the efficiency.    
      
   > I don't much understand SMPS, but I guess if the 240Vrms sine wave mains is   
   rectified to make +324Vdc, then it possibly can be switched to give a +/-   
   324Vpk square wave at say 100kHz, and then applied to HF tranny to raise   
   slightly to give between +   
   324Vdc to maybe +648Vdc and power a rectifier to make the Vdc rails that are   
   isolated from mains for safety.    
      
   normally an isolation transformer would be on the output, usually a fat    
   toroid with only a handful of turns on it running in the hundreds of    
   khz.  open up your typical computer power supply for examples.  So yeah.    
      
   The output of that could then be fed to "whatever".  What I would    
   probably do is use 250V out of that transformer with a very wide input    
   range, 100VAC to 250VAC for example, like a typical switcher would do.    
   You then take its output [typically running in a 'burst' mode], filter    
   that, and feed the main switcher with it.   
      
   Aha, "burst" mode, does that mean "pumped up until it bursts"? or "Bloody   
   Underpowered Revised Silly Technical?" or both?    
      
     A separate one for low volts    
   would be there as well, maybe +/- 15V for supplying the heaters and low    
   volts supplies, and +250v for high volts.     
      
   Yes but all the good tube amps with 6550 et all have B+ at least +430Vdc. Amps   
   with cathode bias and 845 need 1,250Vdc. So hence the need to be able to   
   easily get higher than your +250V, and then scale back to what's needed.    
      
   you just use a    
   diode+capacitor full-wave multiplier [because transformers are    
   expensive].    
      
   Well, it all has to work at maybe 500kHz. That's a bit more difficult than   
   +/-12Vdc because the higher the Vdc, the longer it takes for rise times. Such   
   a lot of RF energy lurking in the box....   
      
   Just need to watch ESR on any electrolytics, and filter    
   with bypassing ceramic capacitor.  multi-layer ceramic capacitors should    
   ALSO have a bypass, although I have not seen one with a published ESR.    
      
   The basic idea is to run a combination of 'burst mode' and PWM mode    
   driving the multiplier to regulate the output voltage.  This is not a    
   new idea (boost/buck converters do this all the time) but I have my own    
   twist on it.    
      
   We wait and see.   
      
      
   > SMPS are not well accepted by Hi-Fi ppl, but Halcro used them in their   
   hi-end 200W amps, and many professional PA amps used SMPS with quite high +/-   
   rails and very high Po for the weight of the amp.    
      
   exactly!  purists can have their iron weights if they want to, and tube    
   rectifiers to go with it.  I doubt they affect the audio quality.    
   [except for guitar amps, which have a distinctive sound when overloaded,    
   when a tube rectifier is used].    
      
   I agree that all the tubes need to sing for hi-fi is a battery like PSU, and   
   it don't matter what produces the Vdc. Guitar amps get some gravvely sounding   
   IMD when they clip the amp with deep overdrive and there's a lot of 120,100Hz   
   in series with 1/2    
   primaries in AB mode from low value reservoir caps and no LC filtering.   
   The B+ goes down on overdrive, and comes back up as notes fade, so sustaining   
   the sustain longer. Guitar amps used for recordings of Mozart make ol' Moetsy   
   sound real crook.   
      
   > The schematics are never in the public domain, all are trade secrets.    
      
   If I built this as a product, I'd probably open source the hardware and    
   sell the boards to make $, through someone like Sparkfun.  Having a    
   board made, even if you have the layout, generally costs more than    
   buying one from a company that builds them in bulk and sells them at a    
   profit.    
      
   [I figured output boards and a control board.  the FCC on the control    
   board would be expensive, but you could keep some frequencies low enough    
   (<9khz) to avoid FCC testing on the driver boards too.]    
      
   the software wouldn't have to be open source necessarily, but it could    
   be, and be licensed for free "for use with an approved board".    
   knockoffs will happen anyway, but at least this way you couldn't legally    
   ship product using plagiarized software.    
      
   [but I'd license it cheap to someone who DID want to ship it with their    
   own hardware, I mean why not]   
      
   Waddawee want? To build it in the shed. When? after we buy the bits easy, How?   
   afta readin an easy schematic.   
      
   Having to navigate around the restrictions and secrets seems like far too much   
   trubble, and that's why so few ppl in DIY land would ever build any SMPS. I've   
   known blokes to build heavy linear PSU for their PC, just a silent box, no   
   heat, no fans, and    
   unlikely to suffer if lightning brings a huge pulse down Cat-5 cables at phone   
   /data inputs. Lightning stuffed PSU 5 years ago in my PC, then the HD failed 3   
   months later, been OK since though. PIA wennit appens.   
      
   Patrick Turner.      
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)