As shown in the diagram below the receiver consists of an rf "front end" chip and a digital processing chip connected to a PC I/O interface. The front end connects directly to the 10 MHz clock and sends a 40 MHz signal to the correlator chip. The correlator chip appears to the PC as as a set of about 100 read and/or write registers. The rf signal is first down converted to about 175 MHz is sent to a simple bandpass filter and sent back to the front end. The second downconverter stage is at about 35 MHz. A SAW filter is used to filter and return the signal to the front end. In the final stage the signal is down converted to 4.3 MHz and a 2 bit A/D converter (sign and magnitude) is used to transmit the signal information to the GP2021 correlator chip. The sample timing is provided by the correlator chip which is simply the 40 MHz clock divided by 7. The PC interface consists of an ISA I/O board from JDR electronics set up to address two 16 bit ports. One port (0x304) uses the lower 8 bits to send the register address. The other port (0x308) used used to read and write 16 bits of data. The correlator chip interface can be set up in a number of ways. I set it up as an Intel 186 interface. This allows an indeterminate amount of time between latching the address and transferring the data.
Hardware Block diagram
There are a number of ways to get access to this kind of receiver hardware. My original board was double sided and just barely worked. I tried to build a new board (again double sided) and was not able to get it to work.
I eventually "hacked" a Canadian Marconi SuperStarTM by taking all of the digital processing chips off the board and connecting the ISA signals directly into the GP2021. Although this is a great way to destroy a perfectly functional receiver, it seems to work pretty well. Luckily I found that the GP2021 interface is set up as a '186 so it only required getting access to the data, address and I/O lines. This is how I did it - How to hack a SuperStarTM
Another way is to make up your own board. Although the documentation from Mitel appears to be very complete it appears that in order to make it work one needs at least a 4 layer board to handle the rf and digital requirements and especially to avoid the digital signals interfering with the down converter. If anyone has the desire and expertise to go this route please let me know.
Last but not least, Mitel on their website offers the detailed design of the "Orion" board. Without the digital processing chips this would be an ideal board to start with. Unfortunately I have not been able to determine if the file formats are compatible to actually have someone actually produce a board. In addition the board uses the GP2015 which I am not able to buy in small quantities.