Oakley midiDAC User Manual - page 6
The PCB has four mounting holes, one in each corner. However, using the midiDAC with the
recommended pots and brackets, will give you sufficient support without the need for
additional mounting hardware. The pots are Spectrol 248 series or BI TT equivalents,
supplied by Farnell, CPC and Rapid Electronics here in the UK. The pot mounting brackets
are specially made for the Oakley modular system and are available only from us.
Circuit Description
The midi data is electrically isolated by U7, a high speed logic output opto-coupler. The
output of U3 is pulled up via R44 and drives two circuits. One is the PIC, the processing
engine of the midi interface. The other is the midi THRU circuit. The latter is a circuit that
simply copies the data seen on the MIDI input port and presents it to the midi thru output
socket if one is fitted. U6 is a simple logic inverter gate, and two of these inverting gates in
series produce a buffered version of the opto’s output signal. Although U6 hardly affects the
signal at all, it does give it a current boost allowing it to drive the midi lines via the standard
220R resistors.
Notice the midi out connector requires the middle pin to be grounded for shielding purposes.
This is not allowed by the midi specification for the midi input socket, so grounding should
only be provided on the thru socket.
The heart of the midiDAC is a preprogrammed PIC16F628. This is where Trevor’s firmware
is located. X1, a 4 MHz crystal provides the necessary timing for the PIC’s internal oscillator.
For more details on the operating system of the PIC see the ‘Firmware Data’ section.
Three of the PIC’s output lines directly drive a 12-bit DAC, U9. The DAC is driven serially,
and this is different to the old issue tbDAC where we used a parallel loading DAC. Although,
the DAC is a 12 bit device we actually only use 7 bits. The other 5 bits of data are held low at
the appropriate time in serial data stream.
Why use only the top seven? Firstly, midi data is arranged, in the main, in blocks of seven bits.
For example there are only 127 notes that a normal midi keyboard can send out. Secondly, the
PIC does not perform any CV scaling or tuning. This is sometimes used on other midi-CV
convertors to generate ADSR and pitch bend information that is then merged in the digital
domain to the pitch data. 14 or 16 bit DACs are required for this. We do all of our CV
processing in analogue hardware. Thus slide time and pitch bend can be simply controlled by a
pot or a trimmer.
So why not use an eight bit DAC? 8-bits, although it gives us 256 steps to play with, the
accuracy of the steps is only plus and minus 1/512 of the highest output voltage of the DAC.
That is an error of 0.2%. This may not sound much, but it does matter. In musical terms, this
means that a semitone between one pair of adjacent notes, will be different to a semitone
between another pair. Tim Orr, of EMS fame, reckoned that at least 10-bit accuracy was
required for users not to hear any difference in the steps. I have chosen to use 12-bits,
because 12 bit DACs are cheapish and easily available. Errors in a good 12-bit DAC will be
negligible compared to VCO tracking errors.
6