The 4066 4PST analog switch is widely used in vintage audio gear, but is notoriously BAD, both in THD and on-resistance. Analog Devices has a new switch out that is vastly superior---the ADG6412. It DOES require an adapter board to drop it into the 14-pin DIP footprint of the 4066, but a clever fellow has made some available in this thread:
https://www.tapeheads.net/threads/adg6412-possible-new-4066-replacement.103635/
https://www.tapeheads.net/threads/adg6412-possible-new-4066-replacement.103635/
Few things to note while using 4016/4066.
- Try to use 74HC/HCT series instead of the older CD series ICs.
- Use a power supply that is high enough to ensure adequately low distortion for analogue applications.
- Use a dual symmetric supply or alternatively center the signal about Vcc/2.
- Minimise the current through the switch to alleviate the non-linear effects.
jean-paul says
Try micro relays. Nothing wins from good mechanical contacts. 4066 based
devices sound mediocre at best. A simple test can quickly reveal this.
While I mostly agree with jean-paul, in this case I think the test is warranted
but predict it won't be as cut and dried as he believes. ADIs chips tend to be
excellent performers and this solid state switch may tie or beat the relays.
Having spent 50 years replacing faulty parts I don't have that much respect for
relays. A good solid state switch fails so rarely that I will always use them in any
design. Plus there is no issue with snubber diodes on the relay coils and the switching
time of the solid state has no comparison at all.
The ADG6412 looks to be an excellent device and gets as close to a relay as any yet.
Just my 2 cents.
G²
When everything has to be app (or at least PIC or other micro) controlled, switches just don’t work. You have to get out of your chair. 😀 And I don’t like wires carrying signal flying all over to get to front panel switches - it just invites trouble. I vote for relays in preamps - easy to keep the digital hash out of the audio path and they play nice with tubes if you’re into that. For audio source selection, you don’t need nanosecond switching time.
If the switch topology works in shunt-only, bipolar transistors are pretty hard to beat. On resistance can be pretty damn low. Granted, I don’t use them in super-fi preamps, but in PA work they often have to deal with full output of a mixing desk levels without a ton of distortion. And two stages cascaded will drop well over 100 dB.
If the switch topology works in shunt-only, bipolar transistors are pretty hard to beat. On resistance can be pretty damn low. Granted, I don’t use them in super-fi preamps, but in PA work they often have to deal with full output of a mixing desk levels without a ton of distortion. And two stages cascaded will drop well over 100 dB.
Regarding that last point, there are at least four ways to work around the distortion of CMOS electronic switches (transmission gates):
1. Minimize the current through the switch, for example by putting a buffer right behind it.
2. Minimize the signal swing with respect to the supplies when the switch is on. This usually means that one side needs to be connected to a ground or a virtual ground.
3. Do both if you can. For example, when an analogue multiplexer is used to switch between different feedback networks of an op-amp in the inverting configuration, you could connect the common terminal straight to the negative input of the op-amp.
4. A bit more exotic: drive the switches from a current source.
1, 3 and 4 are meant to make the circuit insensitive to variations of the on resistance of the switch, 2 and 3 are meant to keep the on resistance variations small.
The only relays I would use for signal are reed relays which are sealed
so if built correctly, they almost never fail or cause any signal disturbance.
And of course reed relays have no issue with the control vs signal grounds.
Are you forgetting about cleaning "good old switches"? Like I said I serviced
electronics, mostly broadcast but some some consumer also so I've seen
more failures than most folks.
G²
No I did not forget but I have never cleaned or replaced an Elma switch for instance and the same goes for Nais and Tamakisawa sealed relays. I have cleaned cheap switches and I have replaced many Chinese relays (and many wrongly applied relays).
Maybe we should compare what is available in good "classic" stuff to good modern stuff. One does not need much to outperform a 4066 though
Maybe we should compare what is available in good "classic" stuff to good modern stuff. One does not need much to outperform a 4066 though
Last edited:
"Regarding that last point, there are at least four ways to work around the distortion of CMOS electronic switches (transmission gates'
NONE of your solutions offer a drop-in replacement for the 4066. The ADG6412 does, WITH an adapter. The MAX 4066 also will work, and is better than the stock 4066s, but NOWHERE near the improvement that the ADG6412 gives.
NONE of your solutions offer a drop-in replacement for the 4066. The ADG6412 does, WITH an adapter. The MAX 4066 also will work, and is better than the stock 4066s, but NOWHERE near the improvement that the ADG6412 gives.
Dear dotneck335,
The point that MarcelvdG is trying to make is that the 4066 need not be replaced as long as it is used in one of the configurations mentioned in post #9. A classic example that has minimal switch current is a programmable gain amplifier in which the switches are used at the opamp input vs. its output.
The point that MarcelvdG is trying to make is that the 4066 need not be replaced as long as it is used in one of the configurations mentioned in post #9. A classic example that has minimal switch current is a programmable gain amplifier in which the switches are used at the opamp input vs. its output.
I have to confess when I worked on consumer electronics is wasNo I did not forget but I have never cleaned or replaced an Elma switch for instance and the same goes for Nais and Tamakisawa sealed relays. I have cleaned cheap switches and I have replaced many Chinese relays (and many wrongly applied relays).
Maybe we should compare what is available in good "classic" stuff to good modern stuff. One does not need much to outperform a 4066 though
not the high end gear. I worked in a small store in a small college
town and we sold a lot of Japanese Marantz, Sony, Japanese H-K.
IOW I did not run into gear with high end switches. At this point
in my life I'm satisfied with a modified Technics SA-160 receiver
for my headphones. The mods include OPA2134 opamps and larger
value coupling caps to get lower frequencies and modded the
gyrators in the 5 band EQ to take 80Hz down to 25Hz and the
320 Hz filter down to 160Hz. It's good enough and I like it. No
arguments at all on the 4066 switches.
G²
Faders have a problem with dirt ingress due to the basic design - a better design would have the track facing down, not up, which would need a more involved mechanism - not sure if that's been tried.
BTW there are hundreds of good analog switches available, but performance can achieved by choosing the right topology as mentioned, just as with RF signals and PIN diode switches - use several imperfect analog switches and maybe an extra buffer to perform like a much better than one expensive one.
BTW there are hundreds of good analog switches available, but performance can achieved by choosing the right topology as mentioned, just as with RF signals and PIN diode switches - use several imperfect analog switches and maybe an extra buffer to perform like a much better than one expensive one.
