MOSFET output stage

[RichD]

http://www.studiomaster.com/1984 - 1986.htm

"This was the amplifier pro sound companies were waiting for;
many buy up to 100 units. "

Did you use MOSFET on the output stage, and why?

 

[Phil Allison]

"RichD"
"Kevin Aylward"

Did you use MOSFET on the output stage, and why?

** The amp used Hitachi lateral mosfets - egs 2SK176 & 2SJ56.

For technical reasons, that have been stated here and are completely beyond
your infinitesimal comprehension.

You damn troll.



...... Phil

 

[Eeyore]

Did you use MOSFET on the output stage, and why?

I have used Mosfets in some of my designs. They had very superior HF
characteristics to the readily available bipolars of 1980. Notably they
also don't suffer SOA limitation problems at higher voltages.

They also typically exhibit much lower crossover distortion when
suitably biased than bipolars do.

Graham

 

[Mr.T]

I occaisionally hear artifacts in 16 bit 44.1 KHz, in music.
It is easy to make a test signal turn up severe artifacts with 44.1 KHz
sample - see what happens with a sinewave at a higher audio frequency that
is several Hz off a frequency that the sample frequency is a multiple of.

You should simply use a system that is not "broken" then.

Since I only occaisionally hear artifacts in music with 44.1 KHz 16 bit,
and when I do I usually find them minor, I would expect a sample rate
twice as high as that to be OK.

Of course it is, as is a *competently* designed 44.1 or 48kHz system.
Using either doesn't present much of a problem these days.

MrT.

 

[JosephKK]

I think it should be possible [I could] design powered speakers with a WiFi interface.

How would you synchronize the different channels?

Some time stamping and segment overlapping and a restricted RF
environment, plus NTP as needed for clock synchronization. Total
latency and latency variation are the primary issues. Maybe modified
bluetooth is a better idea (it really does support streams).

Each speaker would have its own IP address, or perhaps its own port on one IP,
and from the [new] mixer only digital Ethernet to a wireless access point.
No bandwidth problem I think.
56 Mbits / second, should be enough for a few channels.

The real 802.11G throughput is 2.8MB/s at the best. An uncompressed
audio channel takes roughly 100KB/s.

A bit more than that, the normal time 44.1 kHz 16 bits/sample CDA runs
somewhere around 1.5 Mbits/s. 48 kHz 24 bit audio (semi-pro and pro
level) takes a skosh more, about 1.15 Mbits/s per channel.

The big problem with WiFi for audio is the synchronization between the
different WiFi units while maintaining the reasonable delay. This is
hard (if possible at all) to attain with the WiFi equipment.

AFAIK the solutions for audio via Ethernet (CobraNet and such) used the
special protocol stacks and were not fully compatible with the standard
networking stuff. In the general, Ethernet is not good as the network
for the multimedia; it was not designed for that purpose.

This is very true. If you want well time constrained transport, use
appropriate (usually telephone like or telephone) technology. If you
also want to ship video use SMPTE standards. The bitrates can be
daunting (SD-SDI {standard definition - serial digital interface
[both electrical and optical]} runs at 270 Mb/s) but the standards
guarantee interoperability.

 

[JosephKK]

On a sunny day (Sun, 21 Sep 2008 13:41:43 -0500) it happened Vladimir
<[email protected]>:



Jan Panteltje wrote:


I think it should be possible [I could] design powered speakers with a WiFi interface.

How would you synchronize the different channels?

Yes, good point, timestamp would be one way, but that does not solve the delay.
the delay would be fatal in a live application.

Here is the idea: using the power frequency as the common timing
reference. In the local WiFi network, the ping time would be at the
order of 1ms, so all channels could be PLLed to the same half period of
the AC power without an ambiguity. With the sufficient amount of
buffering, that should allow streaming multiple synchronized channels.
Sooo simple... I bet somebody already got a patent on that.


Vladimir Vassilevsky
DSP and Mixed Signal Design Consultant
http://www.abvolt.com

Clever!
How about this: we give each speaker a GPS.
It will also send back its position, and the 'mixer' will
then calculate the optimum sound pattern for 5.1.
GPS also has a very precise clock.

Yes the satellites do have cesium clocks and rubidium clocks, the
system would not work without them. Moreover you do not get good time
solutions without serious long term reception. Your little handheld
does not have such a clock and requires about 45 minutes power on time
to synch up that well.

 

[JosephKK]

With Quality of Service and no latency ?

Graham

With total throughput over 15 times what a single unidirectional
channel requires. 4 or 5 channels (per wifi channel, there are 11 in
802.11G) ought to be achievable on a good session, two channels most
of the time; and with adequate latency control.

Just the same i do not think packet orientated protocols deserve to be
used for decent streaming media.

 

[Jan Panteltje]

On a sunny day (Sun, 21 Sep 2008 13:41:43 -0500) it happened Vladimir
<[email protected]>:

Anyways, I repeated the test the other way around, now for a mp3 file:
-rw-r--r-- 1 user user 103760023 2008-09-09 19:03 instrumental.mp3

Reran the test, it seems a card in my network was configured for base
10 ethernet?, system reported 'network congestion'.
So with base 100, or at least the correct configuration, the same test:

On receiving side:
netcat -q 0 -l -p 1234 -u > q3.mp3

On transmit side:
-rw-r--r-- 1 user user 103760023 2008-09-09 19:03 instrumental.mp3
/home/user> date;cat instrumental.mp3 | netcat -u -q 0 10.0.0.150 1234;date

Wed Sep 24 13:03:32 CEST 2008
Wed Sep 24 13:03:57 CEST 2008

makes 25 sec, 4.150 MB/s = 33.2 Mbps.

-rw-r--r-- 1 root root 103760023 2008-09-24 13:00 instrumental.mp3
-rw-r--r-- 1 root root 103760023 2008-09-24 12:58 q3.mp3
grml: # diff instrumental.mp3 q3.mp3
grml: #
Zero errors at 33.2Mbps, signal strength was >90%, access point Linksys,
channel 7.


Of course UDP based protocols exist, like rtp, I can broadcast rtp
with the dvbstream program, from digital satellite or terrestrial,
It needs an entry in the routing table:
224.0.0.0 * 240.0.0.0 U 0 0 0 eth0

dvbstream -f 12188 -p h -s 27500 -v 166 -a 128
will broadcast RTL2 from sat, while
mplayer -ao oss:/dev/dsp1 -cache 2048 rtp://224.0.1.2:5004/
will play the rtp stream again.
Mixed results via wireless, while TCP over wireless from the same stream works
fine.. no idea why.
There also exists a utility 'dumprtp', it is part of dvbstream.
All this learns us that we can also design our own UDP based protocol,
while sending data you could perhaps run a CRC test over many packets,
and so once in a hundred packets ask for a retransmit... whatever.
I am sure with some careful config that 33.2 Mbps or there about can be used.

Just for fun, for just audio, 8 channels uncompressed sampled at 96 k, and
24 bit wide, makes:
8 x 96000 x 3 x 8 = 18 432 000 is only 18 Mbps

AC3 however, at a bitrate of say 500kbps has 5.1, we can carry 66
of those 5.1 channels.

33Mbps is about the size of a satellite transponder, and can carry
several TV programs, plus audio, plus teletext (ceefax), plus other services
like radio, and subtitles.

So, anyways, wireless is great for that sort of multimedia stuff.
Only limit I see is if you are in a crowded area where everybody uses
it, interference would slow the network way down, and UDP probably could
not be used at all.

I myself will stay with TCP as it, with 2.8MB/s is fast enough for audio and video,
and guarantees zero errors.

The average bandwidth of a sat TV channel here is 4500 to 6000 (wide screen) kbps,
you could get 2 in 22.4 Mbps.

 

[Eeyore]

I think it should be possible [I could] design powered speakers with a WiFi interface.

How would you synchronize the different channels?

Some time stamping and segment overlapping and a restricted RF
environment, plus NTP as needed for clock synchronization. Total
latency and latency variation are the primary issues. Maybe modified
bluetooth is a better idea (it really does support streams).

Barely enough reliable bandwidth IIRC.

Each speaker would have its own IP address, or perhaps its own port on one IP,
and from the [new] mixer only digital Ethernet to a wireless access point.
No bandwidth problem I think.
56 Mbits / second, should be enough for a few channels.

The real 802.11G throughput is 2.8MB/s at the best. An uncompressed
audio channel takes roughly 100KB/s.

A bit more than that, the normal time 44.1 kHz 16 bits/sample CDA runs
somewhere around 1.5 Mbits/s. 48 kHz 24 bit audio (semi-pro and pro
level) takes a skosh more, about 1.15 Mbits/s per channel.

The big problem with WiFi for audio is the synchronization between the
different WiFi units while maintaining the reasonable delay. This is
hard (if possible at all) to attain with the WiFi equipment.

AFAIK the solutions for audio via Ethernet (CobraNet and such) used the
special protocol stacks and were not fully compatible with the standard
networking stuff. In the general, Ethernet is not good as the network
for the multimedia; it was not designed for that purpose.

This is very true. If you want well time constrained transport, use
appropriate (usually telephone like or telephone) technology. If you
also want to ship video use SMPTE standards. The bitrates can be
daunting (SD-SDI {standard definition - serial digital interface
[both electrical and optical]} runs at 270 Mb/s) but the standards
guarantee interoperability.

I spoke to CobraNet some time in the past and they have a time synchronisation element they
call 'the conductor'. I gather it does not integrate well with networkds already carrying
modest traffic and a seperate cable run is advised in such cases IIRC.

Graham.

 

[Eeyore]

"Don Klipstein" wrote


Given the false claim that you've posted below, I somehow find that easy to
believe.

It's a common problem among vinylphiles and other digiphobes. They believe
some totally false, but possibly intuitively satisfying (to them) urban
myths about digital, and since they believe them, they hear them. One more
reason why only carefully bias-controlled listening tests can be trusted.

So why do top-end studio use 24 bit 192 kHz like this from my old friends and
colleagues at Prism Sound ? One of the best companies I ever worked for btw.
http://prismsound.com/music_recording/products_subs/ada8xr/ada8xr_home.php


Graham

 

[[email protected]]

I occaisionally hear artifacts in 16 bit 44.1 KHz, in music.

It is easy to make a test signal turn up severe artifacts
with 44.1 KHz sample - see what happens with a
sinewave at a higher audio frequency that is several
Hz off a frequency that the sample frequency is a
multiple of.

First, would you care to restructure that into a comprehensible
sentence.

Second, what exactly do you mean by "higher audio
frequency?" I have several consumer-grade, semi-pro
and professional A/D - D/A system here and not a single
one of them show any artifacts that you allude to at any
frequency below 1/2 the sample rate, and that includes
44.1 kHz sample rate systems. I have a couple that do
show problems with signals ABOVE 1/2 the sample rate,
but they are either badly implemented or broken.

Since I only occaisionally hear artifacts in music
with 44.1 KHz 16 bit, and when I do I usually find
them minor, I would expect a sample rate twice as
high as that to be OK.

This is why proper listening and engineering tests are
done with methods that attempt to minimize the effects
of expectation.

 

[Phil Allison]

"Eeysore"

So why do top-end studio use 24 bit 192 kHz like this from my old friends
and
colleagues at Prism Sound ?

** Ignoratio elenchi .....

Yaawnnnnnnnn.....

http://en.wikipedia.org/wiki/Ignoratio_elenchi



...... Phil

 

[Eeyore]

"Eeyore" wrote

But one with a pretty fair track record for hearing urban myths like the
crossover distortion that you claim exists in some power amplifiers.

BECAUSE IT'S SO AUDIBLE IT STICKS OUT LIKE A SORE THUMB.

Your precious ABX testing guarantees only a 'lowest common denominator' result.

Anyone that can't hear the distortion of QSC USA or MX series must have severely
damaged hearing.

You were in the forces weren't you ? Explains it all. Hearing damage.

Graham

 

[Eeyore]

"Eeysore"

** Ignoratio elenchi .....

Yaawnnnnnnnn.....

http://en.wikipedia.org/wiki/Ignoratio_elenchi

How do you KNOW that clock accuracy isn't a factor. I saw some DREADFUL clock
jitter on early digital kit and that's just as bad as amplitude inaccuracy.

Graham

 

[Phil Allison]

"Eeysore"

BECAUSE IT'S SO AUDIBLE IT STICKS OUT LIKE A SORE THUMB.

** Bollocks.

Your precious ABX testing guarantees only a 'lowest common denominator'
result.

** Gobbledegook plus a massive non-sequitur.

Anyone that can't hear the distortion of QSC USA or MX series must have
severely
damaged hearing.

** Shame how the rest of us do not have any of the defective examples YOU
claim YOU came across that lacked forward bias current in the output
devices.

Shame you were too lazy and dumb to give the bias trim pot a tweak.

Wanker.

You were in the forces weren't you ? Explains it all. Hearing damage.

** So what explains YOUR obvious brain damage then ?

Was it too much LSD or is it simply congenital ASD ???




..... Phil

 

[Phil Allison]

"Eeysore"

How do you KNOW that clock accuracy isn't a factor.

** More and worse ignoratio elenchi

GIANT yaawnnnnnnnn.....

http://en.wikipedia.org/wiki/Ignoratio_elenchi


Peeeeeuuuuukkeeeeee......



...... Phil

 

[Rich Grise]

Depends on the phone. If I make a phone up out of a good vocal mic and some
studio monitors, it will be pretty good. These days really good electret
mics cost pennies, while good earphone elements are relatively small and
cheap compared to speakers. Most of the inherent losses in modern phones are
in the communications channel, which is wildly bandwidth-reduced. As
bandwidth becomes cheaper, there is a possibility that good-sounding
telephones will become commonplace.

If NASA can send broadcast quality video down from the shuttle or ISS,
howcome their audio still sounds like a fast food clown?

Thanks,
Rich

 

[Eeyore]

"Eeysore"


** Bollocks.

Deaf ****.

** Gobbledegook plus a massive non-sequitur.

Read how it works.

** Shame how the rest of us do not have any of the defective examples YOU
claim YOU came across that lacked forward bias current in the output
devices.

The classic QSC arrangement with grounded collectors has no qiescent current in
the output devices. They are biased 'virtually on the threshold' at about 0.5V
Vbe typically. This is actually quite clever and quite intentional and I've
used the same method myself. It avoid the huge gm jump at crossover by putting
the load current in that area through the DRIVERS alone which are highly
degenerated with about a 10 ohm emitter resistor to make gm look fairly
constant.

Fortunately there are experts here who will understand exactly what I mean by
the above and ignore your huffing and puffing. You're a tech. I'm a designer.

Don ? Kevin ?

Shame you were too lazy and dumb to give the bias trim pot a tweak.

You mean the WRONG tweak.

Wanker.

Careful with your language when you're talking to your betters.

** So what explains YOUR obvious brain damage then ?

Was it too much LSD or is it simply congenital ASD ???

Clearly you're stupid as well as deaf.

Graham

 

[Vladimir Vassilevsky]

Rich Grise wrote:

If NASA can send broadcast quality video down from the shuttle or ISS,
howcome their audio still sounds like a fast food clown?

This is for the presence effect; otherwise you will think of ISS or
Shuttle as if it is something routine.

Vladimir Vassilevsky
DSP and Mixed Signal Design Consultant
http://www.abvolt.com

 

[Eeyore]

"Eeyore" wrote


Everyplace but on test equipment, and in careful listening tests.

It stands out like a sore thumb on test equipment too. It's HIGHLY visible and
occurs typically at critical listening levels (around the 100mW - 1W area).

Let the record show that I have a number of QSC amps on hand to test, have
done so, and reported the results to Graham on Usenet. He has no such
resources at hand.

Uh ? I don't have access to it at the moment but my measurements have been
made on such amplifiers using AP test Equipment - the industry standard and
capable of incredible resolution. I may also shortly get my hands on a Prism
Sound dScope 3.
http://ap.com/products/index.html
I personally love the Portable One to bits since it's so intuitive and fast to
use and totally self contained. With System Ones you often had to turn off the
PC monitor to get accurate results.
http://www.prismsound.com/test_measure/products_subs/dscope/dscope_home.php

Horsefeathers.

As I read how ABX testing works, if say, you had a group 30 listeners and 3
were consistently about to determine A from B and the other 27 couldn't, then
you would say the products were indistinguishable, i.e. discard the results of
the 3 that could determine a difference.

That's what I mean by 'lowest common denominator'. All it proves is that most
people have crap hearing.

What the test should do is affirm there IS a difference on the basis of the 3
who can hear a difference.

If I have misunderstood these principles, my apologies.

However I WILL NOT engage in futile discussion about things *I* know I *can*
hear. Some of my colleagues recently used me as the 'professional ears' to
track down a curious hum that a studio mix engineer had reported but they
couldn't hear themselves. I eventually found it. It was a slightly humming
wall wart in the back of an equipment rack. If it was much louder than 10
phons I'd be surprised.

Graham