Guitar tube effect box.

[RS]

On Fri, 31 Jul 2009 14:15:37 +0100, "ian field"
<gangprobing.alien@ntlworld.com> wrote:

"RS" <RS@nonspam.com> wrote in message
news:fg3475tf8ao17vuq3g6vhlousemd4n9onk@4ax.com...
On Thu, 30 Jul 2009 21:17:40 +0100, "ian field"
gangprobing.alien@ntlworld.com> wrote:

It may be a different thing entirely, but my compressor project is one
that's actually making decent progress at the moment.

Its another topic I'd welcome any insights - as yet I have no idea how to
design circuitry that distinguishes between attack and decay in a
compressor's gain envelope.

I presume that you're talking about detecting the slope of the
envelope: Up = attack, down = decay. That's easier to do in digital
domain, but in analog, you could try feeding the output of the
envelope detector into a differentiator (like a cap in series) which
will give you a positive pulse for attack, etc. Then feed the diff's
output to a schmitt trigger (opamp with some positive feedback would
do it). You'd get a 1 or 0 for attack vs decay. Problem would be in
tweaking time constants, especially for the differentiator, as the
envelope slope for attack will be sharper than the slope for decay.

At the moment I'm having problems with the variable gain amplifier bursting
into oscillation, I've tried to keep the feedback control as simple as
possible and I'm not sure if its the diode pump amplifier that's oscillating
or the whole loop

Tough to even guess without seeing a schematic. What are you using
for the gain control element?

Also, if it's high-frequency oscillation, that would imply that your
signal detector is not directly at fault, since it should have a long
time constant. Put a scope on the envelope detector, or take it out
of circuit and put variable DC on the gain controller to see if the
oscillation continues.

If you want to post a schematic, there are a couple binary groups for
electronics. Send it there, then post a notice here.

Another possibility, though I've never tried this: Feed the envelope
into something that will delay it (like a low-pass filter, if you can
live with the inconsistent phase delay). Then run both the delayed
and non-delayed signals into a comparator. This would probably get
tricky.

Delayed AGC would cause the strike of the note to be un-attenuated followed
by the quieter flat portion, I anticipate this being a big enough problem
without doing things that make it worse.

That would not contribute a delay. I was talking about comparing
current signal to the delayed signal to determine the slope. The
comparator would be triggered by the current signal.

Why are you trying to distinguish between attack and decay?

Some crazy notion I got doing an internet search for any info I could find
on guitar compressors.

"Distinguishing between attack and decay" implies some binary
switching mechanism. Compressors are generally designed to respond
more quickly to attack than to decay, but they don't work purely by
some binary switch between the two. It's usually something like a
forward diode that does a fast charge-up of a cap, with a resistor for
the discharge path. That automatically accounts for the faster
response to attack. The general topology is known as "peak hold."

If you rectify the audio signal, then feed that to a peak hold
circuit, you'll have a variable DC output that follows the envelope of
the signal. The peak hold 'release' mechanism has a time constant
long enough to smooth out ripple, else you'd just have a rectified
version of the audio.

Sounds like you would benefit from looking at existing compressor
circuits. "Hifi" or studio compressors work the same as guitar
compressors, so check those too.

Here's a start:
http://www.schematicheaven.com/effects.htm
http://www.schematicx.com/

 

[ian field]

"RS" <RS@nonspam.com> wrote in message
news:5id675hh9q1as4irf0m4oovq26qtegrt0b@4ax.com...

On Fri, 31 Jul 2009 14:15:37 +0100, "ian field"
gangprobing.alien@ntlworld.com> wrote:


"RS" <RS@nonspam.com> wrote in message
news:fg3475tf8ao17vuq3g6vhlousemd4n9onk@4ax.com...
On Thu, 30 Jul 2009 21:17:40 +0100, "ian field"
gangprobing.alien@ntlworld.com> wrote:

It may be a different thing entirely, but my compressor project is one
that's actually making decent progress at the moment.

Its another topic I'd welcome any insights - as yet I have no idea how
to
design circuitry that distinguishes between attack and decay in a
compressor's gain envelope.

I presume that you're talking about detecting the slope of the
envelope: Up = attack, down = decay. That's easier to do in digital
domain, but in analog, you could try feeding the output of the
envelope detector into a differentiator (like a cap in series) which
will give you a positive pulse for attack, etc. Then feed the diff's
output to a schmitt trigger (opamp with some positive feedback would
do it). You'd get a 1 or 0 for attack vs decay. Problem would be in
tweaking time constants, especially for the differentiator, as the
envelope slope for attack will be sharper than the slope for decay.

At the moment I'm having problems with the variable gain amplifier
bursting
into oscillation, I've tried to keep the feedback control as simple as
possible and I'm not sure if its the diode pump amplifier that's
oscillating
or the whole loop

Tough to even guess without seeing a schematic. What are you using
for the gain control element?

Its a BTL audio amp with DC volume control, it can be used single ended as
long as the peak current is kept below 100mA. It seemed a good choice at the
time as the loading of the diode pump is connected to a completely separate
output and can't distort the audio signal.

Unfortunately the output swing doesn't live up to the data sheet spec ( I
ran a test with the non DC controlled version of the same chip). It makes a
difference which output is used for which function - the first way I tried
suffered "motorboating" and the other way produced heterodyne whistles,
although this was eventually traced to an open circuit electrolytic in the
AGC damping.

Initially I had other problems which gave the impression that the chip
wasn't sensitive enough to get by with only a source follower impedance
converter input buffer, but since then I've had trouble with severe clipping
after re designing the input buffer to one with some gain.

Maybe I should've made my own vga out of a couple of dual gate MOSFETs!

 

[RS]

On Fri, 31 Jul 2009 21:30:13 +0100, "ian field"
<gangprobing.alien@ntlworld.com> wrote:

"RS" <RS@nonspam.com> wrote in message
What are you using for the gain control element?

Its a BTL audio amp with DC volume control, it can be used single ended as
long as the peak current is kept below 100mA. It seemed a good choice at the
time as the loading of the diode pump is connected to a completely separate
output and can't distort the audio signal.

Unfortunately the output swing doesn't live up to the data sheet spec ( I
ran a test with the non DC controlled version of the same chip). It makes a
difference which output is used for which function - the first way I tried
suffered "motorboating" and the other way produced heterodyne whistles,
although this was eventually traced to an open circuit electrolytic in the
AGC damping.

Doesn't exactly sound ideal for this app, especially with stability
problems.

Initially I had other problems which gave the impression that the chip
wasn't sensitive enough to get by with only a source follower impedance
converter input buffer, but since then I've had trouble with severe clipping
after re designing the input buffer to one with some gain.

Maybe I should've made my own vga out of a couple of dual gate MOSFETs!

You could take a look at ICs that were designed to do this. That's
usually a better approach for low quantity prototypes.

There's a company called "THAT corp" that usually makes an appearance
at audio industry trade shows. They specialize in this kind of thing.
They even have composite chipsets with RMS detectors.

http://www.thatcorp.com/Analog_Engine_Dynamics_Processor_ICs.html

VCAs that respond logarithmically:
http://www.thatcorp.com/Blackmer_IC_Voltage-Controlled_Amplifiers.html

Their app notes will also help:
http://www.thatcorp.com/appnotes.html

 

[ian field]

"RS" <RS@nonspam.com> wrote in message
news:2dm675dpcsv323b48fs4gqvf8q5505oq0k@4ax.com...

On Fri, 31 Jul 2009 21:30:13 +0100, "ian field"
gangprobing.alien@ntlworld.com> wrote:


"RS" <RS@nonspam.com> wrote in message
What are you using for the gain control element?

Its a BTL audio amp with DC volume control, it can be used single ended as
long as the peak current is kept below 100mA. It seemed a good choice at
the
time as the loading of the diode pump is connected to a completely
separate
output and can't distort the audio signal.

Unfortunately the output swing doesn't live up to the data sheet spec ( I
ran a test with the non DC controlled version of the same chip). It makes
a
difference which output is used for which function - the first way I tried
suffered "motorboating" and the other way produced heterodyne whistles,
although this was eventually traced to an open circuit electrolytic in the
AGC damping.

Doesn't exactly sound ideal for this app, especially with stability
problems.

Initially I had other problems which gave the impression that the chip
wasn't sensitive enough to get by with only a source follower impedance
converter input buffer, but since then I've had trouble with severe
clipping
after re designing the input buffer to one with some gain.

Maybe I should've made my own vga out of a couple of dual gate MOSFETs!

You could take a look at ICs that were designed to do this. That's
usually a better approach for low quantity prototypes.

There's a company called "THAT corp" that usually makes an appearance
at audio industry trade shows. They specialize in this kind of thing.
They even have composite chipsets with RMS detectors.

http://www.thatcorp.com/Analog_Engine_Dynamics_Processor_ICs.html

VCAs that respond logarithmically:
http://www.thatcorp.com/Blackmer_IC_Voltage-Controlled_Amplifiers.html

Their app notes will also help:
http://www.thatcorp.com/appnotes.html

Do they have a UK distributor and what's their minimum order charge?

 

[RS]

On Fri, 31 Jul 2009 22:38:26 +0100, "ian field"
<gangprobing.alien@ntlworld.com> wrote:

"RS" <RS@nonspam.com> wrote in message

You could take a look at ICs that were designed to do this. That's
usually a better approach for low quantity prototypes.

There's a company called "THAT corp" that usually makes an appearance
at audio industry trade shows. They specialize in this kind of thing.
They even have composite chipsets with RMS detectors.

http://www.thatcorp.com/Analog_Engine_Dynamics_Processor_ICs.html

VCAs that respond logarithmically:
http://www.thatcorp.com/Blackmer_IC_Voltage-Controlled_Amplifiers.html

Their app notes will also help:
http://www.thatcorp.com/appnotes.html


Do they have a UK distributor and what's their minimum order charge?

They probably do have distribution in UK. I've picked up samples at
AES and other engineering shows, so you may be able to get them to
send a chip or two. If you're not prepared to deal with surface
mount, you may want to verify whether they have DIP versions or
adaptors.

 

[ian field]

"RS" <RS@nonspam.com> wrote in message
news:5id675hh9q1as4irf0m4oovq26qtegrt0b@4ax.com...

On Fri, 31 Jul 2009 14:15:37 +0100, "ian field"
gangprobing.alien@ntlworld.com> wrote:


"RS" <RS@nonspam.com> wrote in message
news:fg3475tf8ao17vuq3g6vhlousemd4n9onk@4ax.com...
On Thu, 30 Jul 2009 21:17:40 +0100, "ian field"
gangprobing.alien@ntlworld.com> wrote:

It may be a different thing entirely, but my compressor project is one
that's actually making decent progress at the moment.

Its another topic I'd welcome any insights - as yet I have no idea how
to
design circuitry that distinguishes between attack and decay in a
compressor's gain envelope.

I presume that you're talking about detecting the slope of the
envelope: Up = attack, down = decay. That's easier to do in digital
domain, but in analog, you could try feeding the output of the
envelope detector into a differentiator (like a cap in series) which
will give you a positive pulse for attack, etc. Then feed the diff's
output to a schmitt trigger (opamp with some positive feedback would
do it). You'd get a 1 or 0 for attack vs decay. Problem would be in
tweaking time constants, especially for the differentiator, as the
envelope slope for attack will be sharper than the slope for decay.

At the moment I'm having problems with the variable gain amplifier
bursting
into oscillation, I've tried to keep the feedback control as simple as
possible and I'm not sure if its the diode pump amplifier that's
oscillating
or the whole loop

Tough to even guess without seeing a schematic. What are you using
for the gain control element?

Also, if it's high-frequency oscillation, that would imply that your
signal detector is not directly at fault, since it should have a long
time constant. Put a scope on the envelope detector, or take it out
of circuit and put variable DC on the gain controller to see if the
oscillation continues.

If you want to post a schematic, there are a couple binary groups for
electronics. Send it there, then post a notice here.

Another possibility, though I've never tried this: Feed the envelope
into something that will delay it (like a low-pass filter, if you can
live with the inconsistent phase delay). Then run both the delayed
and non-delayed signals into a comparator. This would probably get
tricky.

Delayed AGC would cause the strike of the note to be un-attenuated
followed
by the quieter flat portion, I anticipate this being a big enough problem
without doing things that make it worse.

That would not contribute a delay. I was talking about comparing
current signal to the delayed signal to determine the slope. The
comparator would be triggered by the current signal.

Why are you trying to distinguish between attack and decay?

Some crazy notion I got doing an internet search for any info I could find
on guitar compressors.

"Distinguishing between attack and decay" implies some binary
switching mechanism. Compressors are generally designed to respond
more quickly to attack than to decay, but they don't work purely by
some binary switch between the two. It's usually something like a
forward diode that does a fast charge-up of a cap, with a resistor for
the discharge path. That automatically accounts for the faster
response to attack. The general topology is known as "peak hold."

If you rectify the audio signal, then feed that to a peak hold
circuit, you'll have a variable DC output that follows the envelope of
the signal. The peak hold 'release' mechanism has a time constant
long enough to smooth out ripple, else you'd just have a rectified
version of the audio.

Sounds like you would benefit from looking at existing compressor
circuits. "Hifi" or studio compressors work the same as guitar
compressors, so check those too.

Here's a start:
http://www.schematicheaven.com/effects.htm
http://www.schematicx.com/

Referring to the schematics I have saved from those and other sources, I
find that most commercial boxes use the now obsolete 3080 chip, searching
for alternatives all I've found so far is the LT1228 - which costs nearly Ł9
each from Farnell.

 

[RS]

On Fri, 7 Aug 2009 15:59:16 +0100, "ian field"
<gangprobing.alien@ntlworld.com> wrote:

"RS" <RS@nonspam.com> wrote in message

Sounds like you would benefit from looking at existing compressor
circuits. "Hifi" or studio compressors work the same as guitar
compressors, so check those too.

Here's a start:
http://www.schematicheaven.com/effects.htm
http://www.schematicx.com/

Referring to the schematics I have saved from those and other sources, I
find that most commercial boxes use the now obsolete 3080 chip, searching
for alternatives all I've found so far is the LT1228 - which costs nearly Ł9
each from Farnell.

The 3080 is an 'OTA' (operational transconductance amp):

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

There are more modern versions: the LM13600, and the more recent
LM13700.

http://en.wikipedia.org/wiki/LM13700

Those can be had for about $2 apiece. Circuit parameters, including
the voltage control feed, will need to be trimmed if you're starting
with a 3080 circuit. But that's the fun part, right?

I was originally suggesting looking at the circuits just to see how
the envelope detection and gain control was done. You could also look
at the "THAT Corp" chips as a more integrated solution, and they have
schematics and app notes.

 

[ian field]

"RS" <RS@nonspam.com> wrote in message
news:5uvo75p3tb6h189lasc31n4jctrmvog8fr@4ax.com...

On Fri, 7 Aug 2009 15:59:16 +0100, "ian field"
gangprobing.alien@ntlworld.com> wrote:


"RS" <RS@nonspam.com> wrote in message

Sounds like you would benefit from looking at existing compressor
circuits. "Hifi" or studio compressors work the same as guitar
compressors, so check those too.

Here's a start:
http://www.schematicheaven.com/effects.htm
http://www.schematicx.com/

Referring to the schematics I have saved from those and other sources, I
find that most commercial boxes use the now obsolete 3080 chip, searching
for alternatives all I've found so far is the LT1228 - which costs nearly
Ł9
each from Farnell.

The 3080 is an 'OTA' (operational transconductance amp):

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

There are more modern versions: the LM13600, and the more recent
LM13700.

http://en.wikipedia.org/wiki/LM13700

Those can be had for about $2 apiece.

The NS web page shows last time buy as 1998.

 

[ian field]

"RS" <RS@nonspam.com> wrote in message
news:5uvo75p3tb6h189lasc31n4jctrmvog8fr@4ax.com...

On Fri, 7 Aug 2009 15:59:16 +0100, "ian field"
gangprobing.alien@ntlworld.com> wrote:


"RS" <RS@nonspam.com> wrote in message

Sounds like you would benefit from looking at existing compressor
circuits. "Hifi" or studio compressors work the same as guitar
compressors, so check those too.

Here's a start:
http://www.schematicheaven.com/effects.htm
http://www.schematicx.com/

Referring to the schematics I have saved from those and other sources, I
find that most commercial boxes use the now obsolete 3080 chip, searching
for alternatives all I've found so far is the LT1228 - which costs nearly
Ł9
each from Farnell.

The 3080 is an 'OTA' (operational transconductance amp):

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

There are more modern versions: the LM13600, and the more recent
LM13700.

http://en.wikipedia.org/wiki/LM13700

Those can be had for about $2 apiece. Circuit parameters, including
the voltage control feed, will need to be trimmed if you're starting
with a 3080 circuit. But that's the fun part, right?

How about the LM4250? Its called a "programmable OP-AMP" instead of an OTA,
the input structure is near identical to the 3080 but the control transistor
also controls the bias current for the output transistors.

 

[RS]

On Sat, 8 Aug 2009 16:55:37 +0100, "ian field"
<gangprobing.alien@ntlworld.com> wrote:

"RS" <RS@nonspam.com> wrote in message
news:5uvo75p3tb6h189lasc31n4jctrmvog8fr@4ax.com...
On Fri, 7 Aug 2009 15:59:16 +0100, "ian field"
gangprobing.alien@ntlworld.com> wrote:

Referring to the schematics I have saved from those and other sources, I
find that most commercial boxes use the now obsolete 3080 chip, searching
for alternatives all I've found so far is the LT1228 - which costs nearly
Ł9
each from Farnell.

The 3080 is an 'OTA' (operational transconductance amp):

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

There are more modern versions: the LM13600, and the more recent
LM13700.

http://en.wikipedia.org/wiki/LM13700

Those can be had for about $2 apiece. Circuit parameters, including
the voltage control feed, will need to be trimmed if you're starting
with a 3080 circuit. But that's the fun part, right?

How about the LM4250? Its called a "programmable OP-AMP" instead of an OTA,
the input structure is near identical to the 3080 but the control transistor
also controls the bias current for the output transistors.

The 4250 was designed to be set at a given gain and left there. You
want something that can be dynamically controlled.

Ebay (the universal price guide) lists LM13700's, and some normal
channels have them. I was assuming that you just wanted to pick some
up for prototyping, but were reluctant to spring for 3080's (which can
also be had, though they tougher to find). If you're thinking
production, then you may want to take your time to come up with a
discrete design for the VCA.

Anyway, I can't help much with prices or sourcing. I consider the time
of design and prototyping the major cost, so I just pay what I need to
pay for parts.

Have you checked the THAT Corp chips?

 

[ian field]

"RS" <RS@nonspam.com> wrote in message
news:56gr751kmm1a4pvcr6kaljahftsfpqbgka@4ax.com...

On Sat, 8 Aug 2009 16:55:37 +0100, "ian field"
gangprobing.alien@ntlworld.com> wrote:


"RS" <RS@nonspam.com> wrote in message
news:5uvo75p3tb6h189lasc31n4jctrmvog8fr@4ax.com...
On Fri, 7 Aug 2009 15:59:16 +0100, "ian field"
gangprobing.alien@ntlworld.com> wrote:

Referring to the schematics I have saved from those and other sources, I
find that most commercial boxes use the now obsolete 3080 chip,
searching
for alternatives all I've found so far is the LT1228 - which costs
nearly
Ł9
each from Farnell.

The 3080 is an 'OTA' (operational transconductance amp):

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

There are more modern versions: the LM13600, and the more recent
LM13700.

http://en.wikipedia.org/wiki/LM13700

Those can be had for about $2 apiece. Circuit parameters, including
the voltage control feed, will need to be trimmed if you're starting
with a 3080 circuit. But that's the fun part, right?

How about the LM4250? Its called a "programmable OP-AMP" instead of an
OTA,
the input structure is near identical to the 3080 but the control
transistor
also controls the bias current for the output transistors.

The 4250 was designed to be set at a given gain and left there. You
want something that can be dynamically controlled.

Ebay (the universal price guide) lists LM13700's, and some normal
channels have them. I was assuming that you just wanted to pick some
up for prototyping, but were reluctant to spring for 3080's (which can
also be had, though they tougher to find). If you're thinking
production, then you may want to take your time to come up with a
discrete design for the VCA.

Anyway, I can't help much with prices or sourcing. I consider the time
of design and prototyping the major cost, so I just pay what I need to
pay for parts.

Have you checked the THAT Corp chips?

I keep hearing pricey and single source.

 

[ian field]

"RS" <RS@nonspam.com> wrote in message
news:56gr751kmm1a4pvcr6kaljahftsfpqbgka@4ax.com...

On Sat, 8 Aug 2009 16:55:37 +0100, "ian field"
gangprobing.alien@ntlworld.com> wrote:


"RS" <RS@nonspam.com> wrote in message
news:5uvo75p3tb6h189lasc31n4jctrmvog8fr@4ax.com...
On Fri, 7 Aug 2009 15:59:16 +0100, "ian field"
gangprobing.alien@ntlworld.com> wrote:

Referring to the schematics I have saved from those and other sources, I
find that most commercial boxes use the now obsolete 3080 chip,
searching
for alternatives all I've found so far is the LT1228 - which costs
nearly
Ł9
each from Farnell.

The 3080 is an 'OTA' (operational transconductance amp):

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

There are more modern versions: the LM13600, and the more recent
LM13700.

http://en.wikipedia.org/wiki/LM13700

Those can be had for about $2 apiece. Circuit parameters, including
the voltage control feed, will need to be trimmed if you're starting
with a 3080 circuit. But that's the fun part, right?

How about the LM4250? Its called a "programmable OP-AMP" instead of an
OTA,
the input structure is near identical to the 3080 but the control
transistor
also controls the bias current for the output transistors.

The 4250 was designed to be set at a given gain and left there. You
want something that can be dynamically controlled.

In an audio compressor the gain changes relatively slowly, what makes
varying the gain problematic?

 

[RS]

On Sat, 08 Aug 2009 14:30:37 -0400, RS <RS@nonspam.com> wrote:

On Sat, 8 Aug 2009 16:55:37 +0100, "ian field"
gangprobing.alien@ntlworld.com> wrote:


"RS" <RS@nonspam.com> wrote in message
news:5uvo75p3tb6h189lasc31n4jctrmvog8fr@4ax.com...
On Fri, 7 Aug 2009 15:59:16 +0100, "ian field"
gangprobing.alien@ntlworld.com> wrote:

Referring to the schematics I have saved from those and other sources, I
find that most commercial boxes use the now obsolete 3080 chip, searching
for alternatives all I've found so far is the LT1228 - which costs nearly
Ł9
each from Farnell.

The 3080 is an 'OTA' (operational transconductance amp):

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

There are more modern versions: the LM13600, and the more recent
LM13700.

http://en.wikipedia.org/wiki/LM13700

Those can be had for about $2 apiece. Circuit parameters, including
the voltage control feed, will need to be trimmed if you're starting
with a 3080 circuit. But that's the fun part, right?

How about the LM4250? Its called a "programmable OP-AMP" instead of an OTA,
the input structure is near identical to the 3080 but the control transistor
also controls the bias current for the output transistors.

The 4250 was designed to be set at a given gain and left there. You
want something that can be dynamically controlled.

PS: The LM4250 is not a VCA. Though I slipped and said 'gain' above,
the external control pin sets slew rate and power consumption. The pin
is meant to be set once with a resistor. Usually for micropower apps,
not variable control of gain.

 

[RS]

On Sat, 8 Aug 2009 20:23:08 +0100, "ian field"
<gangprobing.alien@ntlworld.com> wrote:

"RS" <RS@nonspam.com> wrote in message
news:56gr751kmm1a4pvcr6kaljahftsfpqbgka@4ax.com...

Anyway, I can't help much with prices or sourcing. I consider the time
of design and prototyping the major cost, so I just pay what I need to
pay for parts.

Have you checked the THAT Corp chips?

I keep hearing pricey and single source.

You're going into production? If so, that's probably beyond the scope
of a usenet discussion. You should consider hiring an engineer with a
lot of experience with this stuff, to minimize costs across the board.

 

[ian field]

"RS" <RS@nonspam.com> wrote in message
news:d6or75pcdj56fclhi27d2lijv9p9ruahu5@4ax.com...

On Sat, 8 Aug 2009 20:23:08 +0100, "ian field"
gangprobing.alien@ntlworld.com> wrote:


"RS" <RS@nonspam.com> wrote in message
news:56gr751kmm1a4pvcr6kaljahftsfpqbgka@4ax.com...

Anyway, I can't help much with prices or sourcing. I consider the time
of design and prototyping the major cost, so I just pay what I need to
pay for parts.

Have you checked the THAT Corp chips?

I keep hearing pricey and single source.

You're going into production? If so, that's probably beyond the scope
of a usenet discussion. You should consider hiring an engineer with a
lot of experience with this stuff, to minimize costs across the board.

Its not for production but I would prefer to avoid discontinued parts.