MOSFET output stage

[Eeyore]

Indeed.

I don't actually like powered speakers though. They need *two* leads. Just
more hassle in setting up for the gig.

I know exactly what you mean. Far too much hassle and just more leads getting
in the way for live.

When The Horn had an upgrade last year, all the PA amps, crossover (controller
now actually) went into a back room in a dedicated amp rack with ventilation
and speaker lines (in 4mm2) were run in trunking to the appropriate points.
The monitors sit on a raised surface at the front of the stage and the runs to
them terminate in connectors mounted in purpose made pressed metal panels
mounted in the performer's side of the upright. This means you need only 1 or
2 m leads to wire them up. It's lovely.

Graham

 

[MooseFET]

If you really think that's important (and I don't), why not just put the
amplifiers near the speakers? That way, there won't be any nasty
stability problems to deal with.

You can do about as well if you characterize what a linear system does
and then correct for it. If a system is nonlinear it is harder to do.

If you want to go all the way to having feedback from the speaker, it
seems to me that moving the feedback detection as close to the sound
output as practical is the way to go. Sensing actual cone movement is
better than sensing the driven voice coil's voltage. Some years back I
saw a really nice design for a subwoofer that did this. Here is
basically what the design did:

There was no voice coil at all. The cone was moved by a fast servo
motor with a metal belt drive that drove a rod on the cone center..
This allowed motions of several inches but didn't have a very flat
frequency response.

A position sensor on the rod on the cone center and a pressure
transducer where combined to form the feedback signal. The position
sensor couldn't know about the cone flex and the pressure sensor
couldn't know the very low frequency components. The combined signal
covered the entire range of interest.

There was a really massive servo amplifier driving the motor giving
the system the ability to work up to several Hz.

It seems to me that this sort of thing could be done today and make it
up into the 10s of Hz.

 

[Vladimir Vassilevsky]

If you want to go all the way to having feedback from the speaker, it
seems to me that moving the feedback detection as close to the sound
output as practical is the way to go. Sensing actual cone movement is
better than sensing the driven voice coil's voltage.

The idea of the feedback directly from the cone is no new. The main
problem is economical: it is easier to make a reasonable conventional
speaker rather then trying to make a good speaker from the bad one by
the means of feedback and feedforward compensation.

Some years back I
saw a really nice design for a subwoofer that did this. Here is
basically what the design did:

There was no voice coil at all. The cone was moved by a fast servo
motor with a metal belt drive that drove a rod on the cone center..
This allowed motions of several inches but didn't have a very flat
frequency response.

A position sensor on the rod on the cone center and a pressure
transducer where combined to form the feedback signal. The position
sensor couldn't know about the cone flex and the pressure sensor
couldn't know the very low frequency components. The combined signal
covered the entire range of interest.

There was a really massive servo amplifier driving the motor giving
the system the ability to work up to several Hz.

It seems to me that this sort of thing could be done today and make it
up into the 10s of Hz.

In geophysics, they use the powerful hydraulic vibrators which operate
at the frequencies up to 100Hz. It is interesting that the flat response
and the linearity are important for geophysical application; so some
sort of compensation is applied. This technology can be used for the
woofers if we are after the power levels of 10kW or higher.


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

 

[Eeyore]

Wouldn't want them on a gig - very good for studio monitoring though.

Exactly. KRK have a very good name in that respect at the moment and I think
they do actives.

Graham

 

[Eeyore]

You can do about as well if you characterize what a linear system does
and then correct for it. If a system is nonlinear it is harder to do.

If you want to go all the way to having feedback from the speaker, it
seems to me that moving the feedback detection as close to the sound
output as practical is the way to go. Sensing actual cone movement is
better than sensing the driven voice coil's voltage.

Philips did that decades ago with a piezo transducer IIRC. No idea why it
didn't take off other than Philips aren't exactly reknowned for hi-fi.

Graham

 

[Eeyore]

The idea of the feedback directly from the cone is no new. The main
problem is economical: it is easier to make a reasonable conventional
speaker rather then trying to make a good speaker from the bad one by
the means of feedback and feedforward compensation.

The other problem is which part of the speaker cone do you sense ? They
'bend' and resonate etc etc in use.

Graham

 

[Jan Panteltje]

Exactly. KRK have a very good name in that respect at the moment and I think
they do actives.

Graham

I think it should be possible [I could] design powered speakers with a WiFi interface.
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.
You will have power cables to the speaker, but not a lot of audio wiring.

mmm maybe do the mikes too ;-)

 

[Vladimir Vassilevsky]

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

How would you synchronize the different channels?

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.

You will have power cables to the speaker, but not a lot of audio wiring.
mmm maybe do the mikes too ;-)

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.


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

 

[Jan Panteltje]

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.

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.

There is no reason not to use a compressed format, mp2 or AC3 would be cool.

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.

See above.

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.

OK, it looks like we have to scrap this idea, in its current form, but
the basic principle could work.

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

Good observation!

 

[krw]

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

How would you synchronize the different channels?

Don't. One WiFi channel.

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.

You're going to have to justify that number. 100MB/s? I can see
20KB, *maybe*. Five channels at perhaps three times this...

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.

Why is that so hard? All of the recievers get the same stream.
Each takes what it needs and throws the rest in the bit bucket.

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.

It may not be good, but easily could be good enough.

 

[Eeyore]

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

How would you synchronize the different channels?

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.

You will have power cables to the speaker, but not a lot of audio wiring.
mmm maybe do the mikes too ;-)

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.

i.e. latency. All the specialist solutions to this issue have addressed this problem
specifically.

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.

Cobranet causes trouble does it ? I know they have some kind of 'sync' signal.

Graham

 

[Eeyore]

There is no reason not to use a compressed format, mp2 or AC3 would be cool.

You vile vermin !

Graham

 

[Eeyore]

You're going to have to justify that number. 100MB/s?

He said kBps.

I can see 20KB, *maybe*.

44.1k samples / sec x 2 bytes each = 88.2kBps. Allow overhead for collisions etc. Actually if
you transmit the full AES/EBU SPDIF data it's THREE bytes.

100kBps sounds about right for one mono channel. Or 150 kBps with flags etc.

Graham

 

[Jan Panteltje]

You vile vermin !

Graham

mm actually, I had the intention to use separate amps in the speakers
for woofer, midrange, and tweeter, and do the filtering _in the mixer_,
so then you could compensate per speaker so to speak
Anyways, for people with supernatural hearing, like some animals, that
would perhaps not give full satisfaction.
Most people are satisfied with AC3 though.

 

[Jorden Verwer]

Actually, I do. By and large, they amount to the same thing. Its all low
frequency variations. For example, if one designs a chopper amp to get low
offset, it also kills/corrects for 1/f noise as well. If one has 1/f
problems in an system, one immediately thinks about using a chopper..well I
do any way...

Yes, that's exactly what I meant.

 

[Jorden Verwer]

If you look for an op-amp with an extremely low offset voltage, you
will find that it uses MOSFETs to obtain that extreme low offset.

Quite right. In hindsight, I could've been more clear about this not being
relevant here, I just didn't want any "You're wrong because MOSFETs have
lower offset"-type responses, so that's why I mentioned it...

 

[Jorden Verwer]

Radiation hardness

Boundary condition.

Safe operating area

Boundary condition.

Bandwidth

AFAIK, BJTs still hold the world record for cutoff frequency.

 

[Jorden Verwer]

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

Yuck, wireless...

 

[Jan Panteltje]

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

Yuck, wireless...

You mean you do not watch TV and listen to radio?
And have no cellphone too?

 

[Jorden Verwer]

Jan Panteltje write:

You mean you do not watch TV and listen to radio?
And have no cellphone too?

I mean I try to avoid wireless whenever possible. So when I'm at home, I
watch cable TV and listen cable radio and use my landline phone...