Hi guys,
my 50 cents:
1. Can a zero cross detection work at all? At 20 kHz every 25us has a zero crossing. With 3ms relay delay this is impossible to hit the zero crossing. At 20Hz it would be 25ms. I have a chance there. With small frequencies the ZCD will work a little. Not with high frequencies. y=a*sin(2*pi*relay_delay/period) >>0
2. Yes, you can use semiconductors for switching. The 4066 isn't that bad. But you only get good audio and not high end audio. You then have maybe 0.01% distortion. With a relay it would be close to 0%. It all comes down to your quality expectations.
3. Solid state switches have a problem with high levels. You don't get 10Vrms very quietly. Crosstalk is audible here.
my 50 cents:
1. Can a zero cross detection work at all? At 20 kHz every 25us has a zero crossing. With 3ms relay delay this is impossible to hit the zero crossing. At 20Hz it would be 25ms. I have a chance there. With small frequencies the ZCD will work a little. Not with high frequencies. y=a*sin(2*pi*relay_delay/period) >>0
2. Yes, you can use semiconductors for switching. The 4066 isn't that bad. But you only get good audio and not high end audio. You then have maybe 0.01% distortion. With a relay it would be close to 0%. It all comes down to your quality expectations.
3. Solid state switches have a problem with high levels. You don't get 10Vrms very quietly. Crosstalk is audible here.
Last edited:
Zero cross, not zero level. Even at max Chili the waveform is crossing zero many-many times a second. Changing gain at these points may be less obtrusive than changing on peaks. (Especially a mute which becomes a max-size CLICK.)
Here is Tarzan Boy, a remarkably stupid and frantic ditty from the last century. The full 3+ minutes is non-stop. I clipped out 1/6th of a second for study, so we can see the waveform. We see zero-crossings at least every 5 milliSeconds.
Here I cut gain to zero, first on peaks, then on zero crossings. There is less disturbance if we can wait for a zero-cross.
Probably not. I thought maybe you would use a FET for the fast-switch and back it up with mechanical contacts for that all-metal sound. You can't hear distortion in a 10 milliSecond interval. Also there is always "distortion" for ANY change of gain-- we accept and even expect a short unpure waveform for the sake of long-term ear-comfort.
OK, if the zero crossing thing isn't a critical requirement, how about the idea of bypassing a volume IC as shown in attached diagram? Step 0 to 4 shows the sequence when volume is changed. "100%" means full volume without attenuation, "NEW" means the new volume after user has stopped changing volume. Would that be a stupid idea?
Attachments
OK, I ordered PCBs with the attached schematic, let's see if that works well. I used 6 stages of relay attenuation, for volume outside the range I could always use the chip attenuator, but I think 6 stages should suffice. I am not sure if I need the bypass relays in parallel, but since they do not hurt, so why not I thought. Sorry for being OT, maybe I will start a new thread if the project goes well...
Attachments
I ran into the same problem of pops in the audio with a relay based attentuator and found the problem was acute when switching 7 relays off and 1 (MSB) on, going from 01111111 to 10000000. I was doing the switching with discrete digital logic circuits so the relays were changing in microseconds. I ended up using an Arduino processor to control the relays so I could introduce a 2 to 3 millisecond delay switching each relay sequentially, LSB to MSB turning the volume up and MSB to LSB turning the volume down. I least that is how I remember it, been a while. You can find my thread for the volume control in this forum. https://www.diyaudio.com/community/...input-mute-remote.326029/page-28#post-7272132