A version three of the
processor is being designed. Instead of being DIY, this version might be
made pre built - either entirely or as a module.
Design work is in the early
stages, but differences will include:
A 16-bit audio CODEC IC including
S/PDIF I/O.
Avoids the need for software oversampling. Higher
audio quality.
Avoids the need for software BMC encoding for S/PDIF.
Allows 44.1 or 48 kHz sample rate conversion and antialias
filtering to be added.
Lower CPU usage, allows better processing functions.
Allows optical S/PDIF audio input to be added,
which eliminates earth loop problems with external audio sources. Multiple
S/PDIF inputs allow input selection.
Allows high quality VOX operation to be added.
Crystal locked CPU clock.
Ensures high clock stability, and thus more accurate
sample rates for recording.
A faster SMD dsPIC processor
(~70MHz) or an STM32F429/STM32F439 DSP processor (180MHz).
Allows for more advanced software features such as
FFT based filtering and processing or more processor frequency bands.
Allows for higher sample rates.
More RAM allows for longer, more precise filters and
transforms.
Power consumption will be slightly higher.
A larger box, with a more complex
display and more options for user input.
This will allow easier user control and more features,
but will be less compact. A panel width of at least 120mm is desired.
Relieves crowding of the back panel connectors. Allows
standard microphone connectors to be directly fitted. Allows several TOSLINK
digital audio inputs.
Easier to work on, with room to spare.
Extra I/O ports will be available, allowing all kinds
of extra functions to be implemented.
Displays could be made more complex to make processor
adjustment easier. EQ graphs and spectrum displays become possible.
TFT touch screens are cheap. Using a touch screen
for functions such as EQ adjustment could make the processor easier to
control.
The ability to display full alphanumeric characters
and words would make adding new menu functions much easier.
For the prototype, it is planned to use a simple
electronics enclosure such as an aluminium or plastic box, with an LCD
touch screen mounted on the lid. A touch screen eliminates the need for
pushbuttons, and makes better use of the available space for user display.
The connectors can be mounted along the top edge of the PCB, and will be
accessible through the top edge of the box. This requires that the box
has fairly square sides.
The PCB assembly can be mounted together using
spacers, with the assembly fastened to the lid of the box. With careful
design, it may be possible to allow an alternative mounting arrangement
with the connectors at the back of the PCB. This would allow the assembly
to be shelf mounted if desired.
Multiple TOSLINK optical digital inputs could be
used. TOSLINK inputs are better suited for a processor, as it prevents
ground loop noise which would be heavily amplified by the compressors.
Using more than one input allows the modulation source to be conveniently
changed without changing connectors. The main microphone input will continue
to be analog.
Using a faster processor with more RAM, and a CODEC
chip with built-in S/PDIF produces a big speedup in the DSP firmware by
increasing raw CPU speed and removing the need for software oversampling.
This allows more complex processing, such as:
More bands in the multiband compressor or FFT based
processing.
More bands in the multiband clipper, reducing audible
distortion.
More stages of phase rotation.
A longer, more accurate Hilbert Transform filter,
producing lower distortion in the clippers.
Anti-alias filtering can be added to the digital output.
A higher sample rate, for increased bandwidth.
Alternative applications, such as guitar processing.
This higher level of performance should make the processor
more useful.
If the STM32F429 processor
is chosen, the DSP unit could be designed around a prebuilt development
board as a plug-in add on containing the 16-bit audio CODEC, OpAmps, and
PTT switching circuits only. This would greatly speed up hardware development.