Multiple identical circuits on single PCB
There has been some discussion on this recently. I recently had to make
a 16 channel differential amplifier and adopted the following procedure
which I found worked satisfactorily. I wanted the artwork for each
amplifier section is identical and did not want to repeat the layout
procedure 16 times. All 16 pairs of inputs were from a single IDC
connector and the 16 outputs went to a DB25 connector.
Procedure was as follows:
1) Create schematic for single channel (channel 1) – normal
procedure
2) Include power connector on schematic (will be common to all
channels)
3) Add test pins for channel 1 inputs and outputs
4) Add IO connectors to schematic (IDC and DB25 in this case)
5) Connect test pins to required connector pins for channel 1
(rats nest shows this)
6) Add test pins for all other channel inputs and outputs (no
schematic - just the test pins)
7) Connect them to the required connector pins on the schematic
8) Schematic now has complete channel 1 and test pins for all
other channels connected to the right connector pins.
9) Create netlist and associate footprints as usual
10) Import netlist into PCBNEW
11) Do the layout for channel 1 and place IO connectors as required
12) Route channel 1 inputs and outputs to the connectors
13) Test pins for channels 2 to 16 not yet positioned or routed
14) Block copy channel 1 layout 15 times to create artwork for all 16
channels
15) Delete IO test pins on all copied channels
16) Replace them with the test pins (see 6 above) as created for
channels 2 to 16.
17) Route them to the connector pins (as specified by rats nest)
18) Manually connect/route power busses (I put them along edged of
board) to each channel
19) Component numbering for all channels will be identical, e.g.
feedback resistor on first op-amp will be labelled R1 on all channels .
I like that – makes debugging much easier !
20) Board was 280x120mm, with 48 surface mount chips – is up and
running !
Procedure may sound rather complex but it is really quite straight
forward once you get the idea. Anybody got any better suggestions ?
Bill Randall. University of Cape Town
