THE ORIGINAL PROBLEM
as posted on uk.comp.sys.mac
I have been asked by the owner of a white iBook to to help set it up for portable stereo interview recording. (This was the one I previously mentioned suffered copious amounts of hum when used with the power supply plugged in.) The input device is an 'iMic' adaptor which converts audio to USB
We became a little suspicious of the design quality of the iMic when we found that you had to switch the input attenuator to 'Mic' for line inputs and to 'Headphone' for mic inputs - but we persevered.
With a semi-pro quality electret microphone which delivered a sensible signal level, I had to keep the on-scren gain controls of 'Coaster' right at maximum to get a reasonable input reading. Under those conditions, the signal to noise (mainly hiss) ratio was 35 dB - that's similar to a 78 and considerably worse than a pre-war AM radio.
Unplugging the mic caused the hiss level to rise slightly and the iMic became microphonic, producing 'boings' and 'clunks' if you tapped it. When you handled it, the hum level rose, even though everything was running on batteries.
I tried pre-amplifying the mic to about -13 dBm and, with the 'Coaster' controls turned down, achieved a S/N ratio of better than 60dB - perfectly respectable for the job in hand. The noise was obviously in the iMic, not in the microphone.
In desperation, I dismantled the iMic and discovered that, from the audio point of view, it's design and construction is utterly appalling. Neither the mic socket not any of the input wiring is even screened, the silvery appearance of the casing is just non-conductive paint. I have seen bettter design in a 'throwaway' cassette player.
For a few pence on an item costing 60 UKP (although it is only 35 USD) they could have put it right and produced a good quality job. Low noise amplifiers are so cheap, I am surprised anyone even bother to specify a different one. It also needs a bit of bent tin screening and the switch wiring sorting out - and this is 'Revision 3' I see from the board markings!
A box containing the circuit board
from the iMic with an additional pre-amp and volume control
This was designed to allow the Griffin
iMic to be used with a microphone input and to achieve a
reasonable level of quality. The chip used in the this
device for converting audio to USB signals is the Philips
UDA1325 which has the capability of 24-bit conversion, far
higher quality than the 30dB signal to noise ratio measured
from the iMic on its high gain setting.
The iMic also had no blocking capacitors
to prevent the input bias of the microphone pre-amp from
being short-circuited by a D.C. path through the microphone.
As a consequence, it would not operate with many types of
good quality microphone.
A box containing the circuit board from the iMic with an additional pre-amp and volume control
This was designed to allow the Griffin iMic to be used with a microphone input and to achieve a reasonable level of quality. The chip used in the this device for converting audio to USB signals is the Philips UDA1325 which has the capability of 24-bit conversion, far higher quality than the 30dB signal to noise ratio measured from the iMic on its high gain setting.
The iMic also had no blocking capacitors to prevent the input bias of the microphone pre-amp from being short-circuited by a D.C. path through the microphone. As a consequence, it would not operate with many types of good quality microphone.
In addition, the necessary D.C. blocking capacitors could be added and the whole equipment mounted in a screened enclosure to reduce pickup of electrostatic noise.
A large volume control knob was also incorporated to allow for easy adjustment of input levels during 'live' recordings
The original intention was to use the volume control to vary the gain of the pre-amp, thus providing sufficient gain to cope with low input signals but avoiding overloading with larger signals. To achieve this with a sensible relationship between knob position and gain would have required either a specially manufactured potentiometer or extra circuitry, for which there was insufficient space. Eventually a normal potentiometer circuit was used with fixed amplifier gain of about 30dB . This was found to give a sensible compromise between gain and overload over the range signal levels from a domestic-quality electret microphone in the sort of 'interview' situations this unit was intended for recording.
The diagram shows one channel, the second channel is identical. R12 and C6 are common to both channels.
Two 'Line' sockets were provided, connected in parallel. This is to allow a pair of headphones or powered loudspeakers to be used to monitor the sound from the tape recorder or other input device in cases where the Mac does not allow monitoring during the recording process.
The value of R8 sets the gain of the stage, in the event of other values of gain being found necessary, R8 can be altered as follows:
The microphone and headphone sockets were removed from the iMic board to reduce its overall height. Connecting wires being soldered to the vacant solder pads.
Two small pieces of board were cut and soldered to the main board to support it in the vertical guide slots of the box. One end of the iMic board was then supported by a short piece of stiff wire soldered from its earth plane to earthed tracks on one of the vertical boards. The other vertical board was cut down to support the opposite end of the iMic board which overhung it. When the lid of the box is in place, this arrangement allows very little unwanted movement of the boards.
A sheet of Mylar film was glued to the inside of the box lid and another was interposed between the pre-amp board and the iMic board to prevent short circuits. The space available was very restricted and considerable difficulty was encountered in getting all the components into the box without damage.
Veroboard was used for the construction of the pre-amp. It had to be checked at every stage of the construction process as it is particularly prone to short-circuits between tracks - if multiple faults had been allowed to accumulate, they might have proven almost impossible to locate and eliminate when construction was complete.