Author Topic: CMOS Bypassing doubts  (Read 579 times)


CMOS Bypassing doubts
« on: September 13, 2019, 04:34:35 AM »

HI, I have a few questions regarding CMOS switches. I've been reading alot about this topic from R.G.'s and the tone god's articles, as well as from various threads on this forum and I have a few doubts. First of all, what I want is to build a multieffects unit. My current plan has 16 effects with all of them being bypassable. I'd like to use 16 cmos bypasses but first I'd like to Check with greater minds the following:

-If I place a buffer before the 16 chained DPDT CMOS switches, and one at the end of the chain, this would help diminish the distortion caused by switch resistance right?, as basically no current would flow through my switches. Right?

-Do i need to place caps between each "DPDT"? (all articles I saw referenced a single bypass, and hence they placed caps at the input/output of the switch, but if I chain several "DPDTs" in parallel I think I would only need the coupling caps at the beggining (before 1st switch) and end (after the 16th switch), and at the input / output of each effect.

-the Resistances placed at 1/2 Vcc consume little to no current, as their role is only to pull the switches at half the supply voltage,  right?. I wanted to use a TLE2426 to power all the resistors, but i think it only supplies up to 20 mA.

-If I diminish the amount of caps between switches (see point 2) I should have lower treble losses right?

Thanks in advance for your help!


Re: CMOS Bypassing doubts
« Reply #1 on: September 13, 2019, 06:29:15 PM »
If you eliminate the coupling capacitors, you will increase the bass response, not the treble.  Eliminating the coupling capacitors at the output of each effect may allow switch pops due to slightly different DC voltages at the output of each effect but eliminating the coupling capacitors between the switches is a good idea.

Definitely use a buffer to drive the switches from a low impedance and another buffer at the end of the chain to keep the impedance up.  Don't mess with transistor buffers - an input op amp stage can give you a closed-loop output impedance on the order of an ohm to drive the switches.  The output buffer should be biased by at least one megohm to get a high impedance at the end of the switches.  The same low output impedance / high input impedance should be used for each effect, so a Fuzz Face or a Dunlop Wah pedal would not work here.  The switches are bidirectional so bias applied at one point in the series string biases everything in the chain that is connected in series where all of the switches are on i.e. bypassed.  You do not need a resistor to Vcc/2 at every switch junction.

The supply voltage should be as high as the circuitry will tolerate.  If you use 15 volts, the switch resistance will range from 75 to 90 ohms.  At 10 volts, the switch resistance will go between 100 and 120 ohms.  At 5 volts, the switch resistance will go from 180 to 320 ohms.  If you have a 1 megohm input impedance in the output buffer, the cumulative total from 16 switches will be on the order of 1000 to 2000 ohms, so this will be negligible with the appropriate buffers.  The curve for resistance vs. signal voltage is shown in the data sheets.