0
\$\begingroup\$

Does anyone have any experience with Analog Devices iCMOS high voltage analog switches or multiplexers like the ADG1434 or ADG1411?

The specs for iCMOS sound very impressive. But they've been available for many years and yet I can find virtually no first-hand accounts of anyone actually using these chips. There are no references to iCMOS-anything in AD's press releases for the last 2 years. The availability of these chips on Mouser and DigiKey is very limited.

I want to route audio using a microcontroller and in practice the signal would have to weave it's way through 8 of these switch elements. Assuming the load is >1000 x Ron and the signal level is within say 10Vpp, am I going to trip over something? I'm not an engineer so I'm thinking I should stick with relays but if these things really worked well, they would make things easier, more robust and slicker.

Improve this question
\$\endgroup\$
1
\$\begingroup\$

I've used the ADG1401/ADG1402 and the ADG1419 pretty extensively without issues.

The one thing that would give me pause in your application is that the resistance of the iCMOS switches is non-linear with voltage. While I wouldn't worry about using perhaps one or two switches, stringing a lot of them in series could produce some interesting voltage-dependent non-linearities.

enter image description here

With a non-linearity of ~0.5Ω per switch, eight switches would vary ~4Ω across the input voltage range. If the load you're driving is 100KΩ+, it shouldn't be a problem (though some crazy audiophiles will probably disagree).

Realistically, why do you have eight switching elements in the signal path? If I were you, I would try to use fewer larger multiplexers. Analog Devices make 8:1 and 16:1 multiplexers. They have higher on-resistance, but since you have fewer devices, it would balance out.

Improve this answer
\$\endgroup\$
2
  • \$\begingroup\$ There would be 8 switches because they are basically bypass switches for 4 circuits. So because of the "on resistance flatness" you describe, the 0.5Ω variation would add up. So the 4th circuit would see a 3.5Ω variation. But I'm not that concerned with the non-linearity because it's so symmetric it would really just be a sort of compression. It's the noise I'm worried about. Like some kind of capacitive feed-through or whatever. But if you've tried them with success, then I guess I have to try one. I'll report back for posterity. \$\endgroup\$ – squarewav Apr 8 '13 at 4:18
  • \$\begingroup\$ It occurs to me that in addition to the four switches (per channel) to bypass the four stages, you can have additional switches which bypass any two stages in a row, any three stages in a row, as well as all the stages with a single switch. This way you can minimize the bypass resistance. \$\endgroup\$ – Kaz Apr 8 '13 at 6:45
0
\$\begingroup\$

There is a THD+N (total harmonic distortion plus noise) line in the datasheet which claims that it is 0.025 into a 100 ohm load, 15V peak to peak. A diagram of the test circuit is shown.

Improve this answer
\$\endgroup\$
2
  • \$\begingroup\$ Yes, I can read a datasheet. So what are you saying? You wouldn't use this device for switching hi-fi audio? Are these chips designed for voice / telecom or what? \$\endgroup\$ – squarewav Apr 8 '13 at 2:36
  • \$\begingroup\$ Your question makes it look like you're concerned about a 10V peak to peak signal but it is specified as high as 15V, where the load is fairly low. So you probably don't have anything to worry about. The distortion plus figure is decent for the low load. For a much higher load like 10,000 ohms, the distortion should be a lot lower. \$\endgroup\$ – Kaz Apr 8 '13 at 2:44

Not the answer you're looking for? Browse other questions tagged or ask your own question.