16-05-2020, 04:15 PM
A good way to improve crosstalk on switches is a T section. That's 3 switches, 2 in series and a third that earths the junction between them when the switch is off. This works with mechanical switches, electronic switches such as the humble 4053 and other topologies such as a pair of back biased transistors. I think some high quality switching chips do it internally.
I've used T switches for video but it should be equally applicable to audio. Obviously video needs lots of bandwidth while audio needs dynamic range.
One attachment is a slightly speculative tracing of a video switch in a 16x4 audio/video switch at the Vintage Wireless Museum. I don't think it quite does the T topolgy but it's not far off. When an input is selected using the 4051 8 way analogue switch the emitter follower that drives it conducts normally. Otherwise it's effectively switched off. I'm guessing there may have been resistors from the emitters to +ve to force the emitter followers to switch off. It works well enough for our purposes. The audio switcher is a bit hairy as it runs 405x analogue switches at +/-18V which is right at their absolute maximum rating. Done to get dynamic range but seems like poor design.
The other attachment is the video crosspoint used from the 1960s to the early 1980s by Michael Cox Electronics. When S3 is turned off, S1 and S2 work as cascaded emitter followers. When S3 is turned on both S1 and S2 are turned off. Additionally the signal path between them is effectively shorted out by S3. What isn't shown is that there are multiple crosspoints on one o/p bus so the S2 of one of them is always conducting.
I've used T switches for video but it should be equally applicable to audio. Obviously video needs lots of bandwidth while audio needs dynamic range.
One attachment is a slightly speculative tracing of a video switch in a 16x4 audio/video switch at the Vintage Wireless Museum. I don't think it quite does the T topolgy but it's not far off. When an input is selected using the 4051 8 way analogue switch the emitter follower that drives it conducts normally. Otherwise it's effectively switched off. I'm guessing there may have been resistors from the emitters to +ve to force the emitter followers to switch off. It works well enough for our purposes. The audio switcher is a bit hairy as it runs 405x analogue switches at +/-18V which is right at their absolute maximum rating. Done to get dynamic range but seems like poor design.
The other attachment is the video crosspoint used from the 1960s to the early 1980s by Michael Cox Electronics. When S3 is turned off, S1 and S2 work as cascaded emitter followers. When S3 is turned on both S1 and S2 are turned off. Additionally the signal path between them is effectively shorted out by S3. What isn't shown is that there are multiple crosspoints on one o/p bus so the S2 of one of them is always conducting.
www.borinsky.co.uk Jeffrey Borinsky www.becg.tv
