=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
    Date: Mon, 07 Feb 2000  16:41:49
    From: Nikolai Golovchenko 
      To: PICLIST@MITVMA.MIT.EDU
 Subject: Re: Optimal Frequency for PWM?
--------------------------------------------------------------------------------

Hello David,

I've done this amplifier too. AFAIR it's called D-class amplifier. It
usually has two stages: analog to digital modulator and half-bridge
switch. The speaker is connected directly to the half-bridge.

1)Modulator.
===========
The modulation frequency should be in the range of several MHz to
achieve high quality. I wouldn't recommend PWM, because the needed PWM
resolution (16 bit) will require the minimal pulse width step be lower
then 1us/65536=15ps. What kind of power switch can manage that ???
Instead, you can use the pulse position modulator (or whatever it's
called):

      +-------------------------------------O--CCC--+
      |                                     |       |
      |                                     R      ===
      |  +------+                           R
      |  |      |                           R
      +--| -    |               +------+    |               digital
analog   |      |---------------|D | Q |----o------------------
---------| +    |               |  |   |
         |      |               |  |   |
         +------+             +-|C |   |
                              | +------+
         Low delay comparator |   D-latch
                              |
                              |
                           1-10 MHz clock

(LP filter (R, C) is tuned to have the edge frequency of 20 kHz.)

This is the simplest modulator I know, it has digital output
syncronous to clock and it works well.

2)Half bridge switch
====================
I'm not an expert on MHz switches (my amplifier worked at about 500
kHz), but there is a couple of problems you should be aware of. First,
examine the speaker for capacitance. Having significant capacitive
load means that switch will experience big peak currents. They will
heat the switch easily. Besides, these currents will generate a lot of
RF interferences, which will inject noise to the analog part of the
amplifier and elsewhere. However, this capacitance can be compensated
by a choke in series. And this choke must keep its inductance under
the maximum load currents. The best choke in my case (10W speaker) was
air core coil. (it won't saturate, but wire should be thick,
prefferably silver coated to conduct high frequencies well).

Finally, I'd like to add that digital amplifier is not a trivial thing
to build, especially high power. This has a lot of problems dealing
with losses at high frequencies switching, RF interference, stabilty
of clock frequency and supply voltage, and more. If you cope with most
of them, the amp will pay off with good sound quality and low power
consumption.

Good luck
 Nikolai


                           
On Sunday, February 06, 2000 David Harris wrote:
> Hi -
> Would you have a circuit diagram?
> David

> Dave VanHorn wrote:

>> > Months ago I saw a drawing from some audio chip manufacturer who uses
>> > PWM/PCM audio output modulation going directly to speakers without any
>> > additional filtering (Philips?).
>>
>> I did that in '90, with a 74HC74, 74CH14 "drivers" and a 3 MHz clock.
>> Worked great, ate essentially no power, and had suprising fidelity.
>> All digital amplifier.