Searching \ for 'class D audio amplifier' in subject line. ()
Make payments with PayPal - it's fast, free and secure! Help us get a faster server
FAQ page: piclist.com/techref/io/audio.htm?key=audio
Search entire site for: 'class D audio amplifier'.

Truncated match.
PICList Thread
'class D audio amplifier'
1999\09\15@005326 by tec

flavicon
face
I've been looking for information on high power class D audio amplifiers
and now I'm looking at the feasiblity of using a PIC as the controller.

Preliminary requirements are:
- class D 200-250Khz switching freq.
- single channel
- 200W RMS into a 4ohm load
- 12V supply voltage
- freq. response 100-10KHz
- THD 5%

I would like to do this without using a switching power supply to get
the positive rail, but instead using a full-bridge to drive a step-up
transformer that is connected to the speaker.  The output would be low
pass filtered and compared to the incoming audio signal to generate an
error signal to adjust the duty cycle to allow the output to track the
input.

Driving the full bridge. Can the PIC port pins drive MOSFETs or IGBTs
without expensive drive circuits or significant power loss at 200KHz and
high currents? What about implementing a ZVS (zero voltage switching)
resonant design to reduce the drive requirements for switching the FETs
by turning them off and on when the current is minimum?

EMI.  By varying the switching frequency in code from say 200Khz to
230Khz, would I see much reduction is EMI?  Or is this not a drastic
enough spread to see a noticable reduction in emissions?

Suggestions and comments are greatly appreciated.

Thanks,

Todd

1999\09\15@164245 by l.allen

picon face
{Quote hidden}

The ability of the PIC to drive the MOSFETs can be worked out.
Look up the specs on the mosfet, what is its gate input capacitance?
Xc=1/ (2pi x freq x c).... if this is less than 200 or so ohms.. no
dice, I am picking the PIC hasn't a prayer of directly driving  power
MOSFETs at 200kHz.
Changing the frequency from 200kHz to 230kHz will not help EMI.
200kHz is a bloody fast switching speed for a micro, thats only 5uS
per cycle to do the computing.
Gate capacitance is the same no matter what the voltage across or
current through the channel.
There are some class D amplifying chips out there like Texas Inst
TPA005D02 etc
_____________________________

Lance Allen
Technical Officer
Uni of Auckland
Psych Dept
New Zealand
_____________________________

1999\09\15@170156 by Richard Prosser

flavicon
face
And following on from Lance's Comments:-

You also might want to look at the drive voltages.
Most high current MOSFETs require at least 8V to turn them on - a PIC will
run to about 6 which is not really close enough. Logic level MOSFETS are OK
at 5V drive but are generally lower power.
And then if you're using an H-bridge you'll need to do a level shift anyway.

FET driver chips are easy to use however & the appnotes give a good starting
point for H bridge drives etc.

I'm not sure about IGBTs but I think they need about 4-5V and 200kHz or so
is, I think, getting a bit quick - I'm not too sure on this one.

I guess you'd be using the pwm output so it carries on in parallel with any
required processing.

from the EMI viewpoint - the risetimes are more of a problem than the actual
switching frequencies
Also, parasitics will add in ringing etc. which can be a major hassle -
particularly as you're driving a transformer.

Best of luck

Richard

{Quote hidden}

1999\09\15@171820 by Marcelo Yamamoto

flavicon
face
I had some problems using logic level mosfets as they turned on by itselves.
the environment were very noisy. The International Rectifier has some
mosfets drivers that are not so expensive less than a US$1.00, and they work
well in a half bridge config (they have a high side driver built in).

Marcelo


>And following on from Lance's Comments:-
>
>You also might want to look at the drive voltages.
>Most high current MOSFETs require at least 8V to turn them on - a PIC will
>run to about 6 which is not really close enough. Logic level MOSFETS are OK
>at 5V drive but are generally lower power.
>And then if you're using an H-bridge you'll need to do a level shift
anyway.
>
>FET driver chips are easy to use however & the appnotes give a good
starting
>point for H bridge drives etc.
>
>I'm not sure about IGBTs but I think they need about 4-5V and 200kHz or so
>is, I think, getting a bit quick - I'm not too sure on this one.
>
>I guess you'd be using the pwm output so it carries on in parallel with any
>required processing.
>
>from the EMI viewpoint - the risetimes are more of a problem than the
actual
>switching frequencies
>Also, parasitics will add in ringing etc. which can be a major hassle -
>particularly as you're driving a transformer.
>
>Best of luck
>
>Richard

1999\09\15@184635 by D. Schouten

picon face
Hi,



> I've been looking for information on high power class D audio
amplifiers
> and now I'm looking at the feasiblity of using a PIC as the
controller.
>
> Preliminary requirements are:
> - class D 200-250Khz switching freq.
> - single channel
> - 200W RMS into a 4ohm load
> - 12V supply voltage
> - freq. response 100-10KHz
> - THD 5%
>
> I would like to do this without using a switching power supply to
get
> the positive rail, but instead using a full-bridge to drive a
step-up
> transformer that is connected to the speaker.

You can eventually use this approach, however don't forget to reset
the transformer in the dead times of the pulses. I'm not sure if you
could use
a ferrite transformer if your duty cycle passes the 50% (saturation).
You could also use a common 50Hz/60Hz iron core transformer, but then
you need to prefilter the carrier frequency on the primary side.

> Driving the full bridge. Can the PIC port pins drive MOSFETs or
IGBTs
> without expensive drive circuits or significant power loss at 200KHz
and
> high currents?

As been told earlier by Lance and Richard, I don't recommend it too.
At first the PIC I/O has to deliver quite some peak currents (high
di/dt) to the FET gate at 200kHz (It's not for nothing that even small
FET drivers can deliver peak currents of up to 0.5 Amps), using logic
level FETs or not. And secondly you need to create a high end (level
shifted) drive for the two top FETs anyway. So using two half-bridge
drivers or one full bridge driver is something to go for (check out
Harris [now Intersil] or IR).

> What about implementing a ZVS (zero voltage switching)
> resonant design to reduce the drive requirements for switching the
FETs
> by turning them off and on when the current is minimum?

With ZVS you turn on the FET when no drain-source voltage is present.
With ZCS (zero current switching) you turn on the FET when no
immediate drain to source current would flow right after turn on. IIRC
ZVS works in serie resonant mode and ZCS in parallel resonant mode.
Using ZVS wouldn't reduce the drive requirements much. You only loose
the gate 'plateau' at Vthreshold in your rising edge, which isn't much
of an advantage though. The switching technique for ZVS is more
extensive than the common 'hard' switching approach. You will need
some kind a phase shifting scheme used in some PWM controllers for
SMPS's (check out Unitrode UC3875,79...).
You also need some knowledge on how to model your full bridge in order
to calculate the right resonant tank components. One big advantage of
ZVS is reduction of EMI.

> EMI.  By varying the switching frequency in code from say 200Khz to
> 230Khz, would I see much reduction is EMI?  Or is this not a drastic
> enough spread to see a noticable reduction in emissions?

The spreading is too small for noticable reduction. The advantage of
varing switching frequency is too spread out it's fundamental
frequency energy which is, together with the higher harmonics of
course, the biggest problem in SMPS EMI filtering.
In your application the problem is that you can't spread the switching
frequency too much on the low end side, since you would loose accuracy
in the higher audio frequencies.

But, just like Lance said, I find it a very optimistic project to do
with PIC. If it's just to build a class D amp for hobby use and not as
a design challenge, I would build an analog type. Have you checked out
the Intersil webpage already. I'm not sure if it's still there, but
there are some app/design notes available from which you can build a
200W, 20Hz..15kHz class D amp which is relatively simple. I've build
it too for a subwoofer.

Occasionally I work for a company which have build a 'software based'
high power class D amp. But they used a big TI DSP for that.

Good luck!

Daniel...

1999\09\16@015450 by tec

flavicon
face
First of all thank you all for the great feedback!

I have already looked at:
1. TI TPA00502.  It is only 2 watts per channel and I need 200 watts.  I
really only need one channel. I have not looked at driving external FETs
off the outputs to get 200W and I haven't received feedback from TI on
this either.

2. Linfinity LX1720. It uses external FETs and is therefore a canidate,
but it is also two channels.

3. Harris (Intersil) HCA125ACREF. It's 200 watts!!!  I find it very funny
that you mention this one.  They want a $10K license fee, plus royalty.
:(   If you've got data on this I'd like to see it.  I don't beleive that
they are making info available other than a brief description, but that
wouldn't matter if they don't want to sell the controller.

4.  Tripath Class-T.  Very cool!!!  A class D with a DSP that implements
adaptive switching to 'learn' the charicteristics of the FETs to prevent
switching problems that cause distortion.  It also implements spread
spectrum switching up to 1.5Mhz with an average of 600-700kHz to reduce
EMI.  Very nice but I don't need hi-fidelity for a siren.

5.  Unitrode UC3879.  I had already looked at this.  Do you think it would
work well as a class D audio amp?

Of course, I would still use a PIC to generate the siren sounds and do
other slow tasks.

Thanks again,

Todd


D. Schouten wrote:

{Quote hidden}

1999\09\16@125332 by D. Schouten

picon face
Hi,


> 3. Harris (Intersil) HCA125ACREF. It's 200 watts!!!  I find it very
funny
> that you mention this one.  They want a $10K license fee, plus
royalty.
> :(   If you've got data on this I'd like to see it.  I don't beleive
that
> they are making info available other than a brief description, but
that
> wouldn't matter if they don't want to sell the controller.

When Harris was experimenting with Class D amplification some years
ago, they published a couple of design notes plus a 200W reference
design using standard Harris parts. I guess that in these days they
weren't really aware of the great commercial strength the concept had
in it. Today however they do. IIRC they now call it COOL AUDIO and you
may only use it on license bases.
That was the reason I said 'I'm not sure if these papers are still
online'.
I probably have some PDF's somewhere on this topic. I could mail you
them if you're interested.

> 5.  Unitrode UC3879.  I had already looked at this.  Do you think it
would
> work well as a class D audio amp?

Actually I didn't advice that part for building a class D amp, I only
pointed that way for more info about phase shifted PWM technology used
for ZVS.
However there may be some possibilities to make a class D amp with
this part. The drawback of this part is the lack of flexibility to
access the onchip connections which are handy in a class D amp. It's
optimised for building power supplies. A Harris full bridge driver
(HIP4080) with build in error amp could do the job very well.

> Of course, I would still use a PIC to generate the siren sounds and
do
> other slow tasks.

That's indeed less demanding :)

Bye,

Daniel...

More... (looser matching)
- Last day of these posts
- In 1999 , 2000 only
- Today
- New search...