Recommendation for DSP/Evaluation Board

Hello,

I've been tasked with designing/building a signal processor box for
Lab/Student use at my university.

It needs to generate a summed waveform based on numerous front panel
control settings. There are nine harmonics, each of which can have its
gain and phase set by a pot adjustment, that are summed at the output
(there will be 18 pots total).

The base input is a sine wave from 20hz to 1khz. Can someone please
recommend a DSP/Evaluation module that would be well suited for this
application, taking into account reading in set values from 18 pots?
Note the pots will be read as user settings, not part of the analog
signal conditioning.

I could envision supplying a constant current to each pot and haveing
the voltages multiplexed, fed into an A/D, then stored digitally. The
pot readings would be used as part of the mathmatics to generate the
summed output waveform.

Thanks,

Strato

"strato" <strato79@comcast.net> wrote in news:1103136087.409110.297170
@z14g2000cwz.googlegroups.com:

> Hello,
> 
> I've been tasked with designing/building a signal processor box for
> Lab/Student use at my university.
> 
> It needs to generate a summed waveform based on numerous front panel
> control settings. There are nine harmonics, each of which can have its
> gain and phase set by a pot adjustment, that are summed at the output
> (there will be 18 pots total).
> 
> The base input is a sine wave from 20hz to 1khz. Can someone please
> recommend a DSP/Evaluation module that would be well suited for this
> application, taking into account reading in set values from 18 pots?
> Note the pots will be read as user settings, not part of the analog
> signal conditioning.
> 
> I could envision supplying a constant current to each pot and haveing
> the voltages multiplexed, fed into an A/D, then stored digitally. The
> pot readings would be used as part of the mathmatics to generate the
> summed output waveform.
> 
> Thanks,
> 
> Strato
> 
> 

I can't think of any board that has the right I/O for your application. 
Our dspstak boards separate I/O from the DSP portion. If you are up to 
creating the I/O section yourself, our dspstak DSP Engine cards might be 
a good choice.

-- 
Al Clark
Danville Signal Processing, Inc.
--------------------------------------------------------------------
Purveyors of Fine DSP Hardware and other Cool Stuff
Available at http://www.danvillesignal.com

Interesting little project!  Sorry I don't have a specific recommendation but do
have a few comments/questions.

"strato" <strato79@comcast.net> wrote in message
news:1103136087.409110.297170@z14g2000cwz.googlegroups.com...
> Hello,
>
> I've been tasked with designing/building a signal processor box for
> Lab/Student use at my university.
>
> It needs to generate a summed waveform based on numerous front panel
> control settings. There are nine harmonics, each of which can have its
> gain and phase set by a pot adjustment, that are summed at the output
> (there will be 18 pots total).

Too bad it's not 16 as that would fit in better with most common devices (muxes
or multi-channel ADCs).

> The base input is a sine wave from 20hz to 1khz. Can someone please
> recommend a DSP/Evaluation module that would be well suited for this
> application, taking into account reading in set values from 18 pots?
> Note the pots will be read as user settings, not part of the analog
> signal conditioning.

I'm not really sure what distinction you are trying to make between "user
settings" vs. "part of the analog signal conditioning".  Could you clarify?

> I could envision supplying a constant current to each pot and haveing
> the voltages multiplexed, fed into an A/D, then stored digitally. The
> pot readings would be used as part of the mathmatics to generate the
> summed output waveform.

My company makes a product with audio I/O plus 8 pot inputs
(http://www.symetrixaudio.com/SymNet_Web/products/8x8DSP.htm).  We also have
8-pot expander boxes
(http://www.symetrixaudio.com/SymNet_Web/products/controlio.htm).  Unfortunately
its not a development platform, i.e. we don't provide hooks to let you write
your own code for it.  :-(  Plus it would probably be overkill for this
application.  It does include sine wave modules that can be controlled from the
external pots with a wide variety of signal conditioning on the pot inputs.  But
I'm assuming you want to generate the harmonics from the base input, which we
have no easy way to do off-the-shelf.

You probably already know this, but to get a control voltage out of a pot, just
put a constant voltage (usually the full-scale voltage of your ADC) across the 2
end terminals and take the reading from the middle "wiper" terminal.  Since the
resistance between the two end terminals of the pot is a constant, you will get
your constant current just from a simple voltage source.

This would probably be a pretty easy software project if you used a PC with
on-screen sliders instead of hardware pots.  The PCs sound card provides easy
decent-quality audio I/O and you would have the ability to save and recall
settings, show the waveform on screen, etc..  Just an idea, though you may have
a reason why it needs to be a hardware project.

Do you have a budget for the project?

"Al Clark" <dsp@danvillesignal.com> wrote in message
news:Xns95C0833DC36EEaclarkdanvillesignal@66.133.130.30...
> "strato" <strato79@comcast.net> wrote in news:1103136087.409110.297170
> @z14g2000cwz.googlegroups.com:
>
> I can't think of any board that has the right I/O for your application.
> Our dspstak boards separate I/O from the DSP portion. If you are up to
> creating the I/O section yourself, our dspstak DSP Engine cards might be
> a good choice.

Maybe some microcontroller eval board for the pots?  I know a lot of
microcontrollers have multi-channel ADCs.  But probably not 18 channels worth.

"Jon Harris" <goldentully@hotmail.com> wrote in
news:32bgb5F3gnvktU1@individual.net: 

> "Al Clark" <dsp@danvillesignal.com> wrote in message
> news:Xns95C0833DC36EEaclarkdanvillesignal@66.133.130.30...
>> "strato" <strato79@comcast.net> wrote in
>> news:1103136087.409110.297170 @z14g2000cwz.googlegroups.com:
>>
>> I can't think of any board that has the right I/O for your
>> application. Our dspstak boards separate I/O from the DSP portion. If
>> you are up to creating the I/O section yourself, our dspstak DSP
>> Engine cards might be a good choice.
> 
> Maybe some microcontroller eval board for the pots?  I know a lot of
> microcontrollers have multi-channel ADCs.  But probably not 18
> channels worth. 
> 
> 
> 

You could do it with a few PICs or AVRs and connect via SPI on our boards. 
Maybe you could adapt a few Basic Stamp boards?

-- 
Al Clark
Danville Signal Processing, Inc.
--------------------------------------------------------------------
Purveyors of Fine DSP Hardware and other Cool Stuff
Available at http://www.danvillesignal.com

Jon,

Thanks for your reply. To answer a few questions:
1) Agreed 16 inputs would be easier to accomodate.

2) The pots are used to, program if you will, the amplitude and phase
shift settings. Their values will be read and stored digitally. They
are not part of an any analog circuity in the signal path. Instead of
pots, I could perhaps use thumbwheel switches, but these are not the
best for student interface and constant cycling.

3) The professor sponsoring this effort, wants to have this run in a
standalone box that can output to an o'scope, this means no interface
to a PC etc.

4)Constant voltage would be okay if I use a well regulated P/S. I was
just thinking a programable current may be easier to achieve for A/D
input scaling. I would probably build this little piece of the circuity
myself.

5) I do have a budget. We are looking to build 10 units for under 20K.
I don't think this could be done on the outside for this low of a
budget. Most companies would prbably want more than 20K in just
engineering, and rightly so.

strato wrote:
> Hello,
> 
> I've been tasked with designing/building a signal processor box for
> Lab/Student use at my university.
> 
> It needs to generate a summed waveform based on numerous front panel
> control settings. There are nine harmonics, each of which can have its
> gain and phase set by a pot adjustment, that are summed at the output
> (there will be 18 pots total).
> 
...
It's a very old design, but  Motorola published a 10-band stereo Graphic 
Equalizer circuit many years ago, featuring the DSP56001. They used a single 
8-bit ADC (ADC0804) for the front panel sliders (20 in total so you get two 
spare!),  which were read via software multiplexing from the DSP. I believe they 
used this in their first demos of the new chip.

You can find the document online from several sources, e.g:

http://www.soundart-hot.com/english/developers2.htm

Possibly one of the current Motorola 56~ development kits could be used as the 
basis for an implementation; running nine sinewave oscillators should be a 
doddle even for an original 56K! The Chameleon (see URL above) is itself a very 
interesting product, and if you have the budget for it it will do a great deal 
more than 9-partial additive synthesis. Use it with a MIDI controller box for 
the pot controls, etc.

Richard Dobson

"strato" <strato79@comcast.net> wrote in message
news:1103139929.619764.235150@z14g2000cwz.googlegroups.com...
> Jon,
>
> Thanks for your reply. To answer a few questions:
> 1) Agreed 16 inputs would be easier to accomodate.
>
> 2) The pots are used to, program if you will, the amplitude and phase
> shift settings. Their values will be read and stored digitally. They
> are not part of an any analog circuity in the signal path. Instead of
> pots, I could perhaps use thumbwheel switches, but these are not the
> best for student interface and constant cycling.

You could also consider using encoders.  This way the I/O would be digital.  You
would need 2 digital input pins per encoder, so you may still need some external
multiplexing as most DSPs don't have that many GPIO.  But digital mulitplexing
is generally easier and cheaper than analog.

> 3) The professor sponsoring this effort, wants to have this run in a
> standalone box that can output to an o'scope, this means no interface
> to a PC etc.

How about a stand-alone laptop?  :-)

> 4)Constant voltage would be okay if I use a well regulated P/S. I was
> just thinking a programable current may be easier to achieve for A/D
> input scaling. I would probably build this little piece of the circuity
> myself.

I was assuming a good power supply was available.  I'm no analog expert, but
I've always found it easier to create a constant voltage source (e.g. 3-pin
linear regulator) than a constant current source!  :-)

> 5) I do have a budget. We are looking to build 10 units for under 20K.
> I don't think this could be done on the outside for this low of a
> budget. Most companies would prbably want more than 20K in just
> engineering, and rightly so.

Are you doing the DSP/hardware development yourself or looking for someone else
to do it?

Richard,

Thanks very much for the information, it is very much appreciated.

I wasn't planning on haveing nine oscillators. There is only one
sinewave input (or perhaps this could be stored as data in a lookup
table). The signal is altered mathmatically based on the user input pot
settings, and then output as a single waveform. The output waveform
represents the addition of nine derived harmonic signals, each of
which's phase and amplitued is programmed via the pots. I imagine the
output will just run as a continuous loop, once things are setup.
Strato

Just curious as to how you were planning on generating the 9 derived harmonics?
All the ways I can think of involve PLL-like techniques, i.e. detecting the
frequency of the incoming signal, and then generating the harmonics based on
that.  So my method would take 9 oscillators, or at least a fancy 9-output
oscillator.  FYI, I think Richard was saying that on a modern DSP, 9 oscillators
would be not be problem computationally.  I would agree with that.

"strato" <strato79@comcast.net> wrote in message
news:1103142707.351786.37890@c13g2000cwb.googlegroups.com...
> Richard,
>
> Thanks very much for the information, it is very much appreciated.
>
> I wasn't planning on haveing nine oscillators. There is only one
> sinewave input (or perhaps this could be stored as data in a lookup
> table). The signal is altered mathmatically based on the user input pot
> settings, and then output as a single waveform. The output waveform
> represents the addition of nine derived harmonic signals, each of
> which's phase and amplitued is programmed via the pots. I imagine the
> output will just run as a continuous loop, once things are setup.
> Strato
>