CD4053 Pop Troubleshooting Help

[Fuzzy_BYU_ME]

Hey all! New to the forum, but have read several threads to help with projects so far.

I am trying to build an Arduino-based programmable loop switcher. I'm hoping to use the CD4053 chips to switch, in part because they are much cheaper than mechanical relays. I already have about 50 of them to use.

I followed the schematic found on the GEO FX page. Everything works well. However, I am getting some pop. I am running the FX loop through my OCD pedal, playing a Strat, into an Orange amp, and I am using an Arduino Uno to provide the 5V switch signal. The Arduino is also providing 5V power for the IC.

When the volume on the guitar is rolled down and the OCD pedal is turned off, there is no pop noise at all when the CMOS switches. When I turn my OCD on and the guitar volume is still rolled down, there is some pop. OCD back off but volume rolled up, very small pop. Roll the volume on with the OCD on, and the pop is quite noticeable.

My voltages are:

When control voltage is 0
Positive Power: 5.06 v
1/2 V: 2.44 v
Pin1: 2.35 v
Pin2: 2.38 v
Pin12: 2.36 v
Pin13: 2.32 v
Pin14: 2.33 v
Pin15: 2.32 v
Pin16: 5.06 v

When control voltage is high
Positive Power: 5.02 v
1/2 Voltage: 2.44
Pin1: 2.38 v
Pin2: 2.23 v
Pin12: 2.23 v
Pin13: 2.39 v
Pin14: 2.36 v
Pin15: 2.35 v
Pin16: 5.04 v

So I guess my question is... is there anything I can do about it? The switch isn't making any noise on its own, only when getting signal. Is this noise just something I'm going to have to live with? I tried adding capacitors and pull down resistors to and from the pedal signal (which GEO said can be omitted, but thought I'd try) and that seamed to make it worse. Any ideas?

[PRR]

What is the DC voltage at the OCD In and Out jacks, just a short cord, no other gear connected?

10mV of DC leakage will be a significant pop.

[Fuzzy_BYU_ME]

Switch goes out to the pedal on pin 13, comes back to the switch on pin 1.

So what is the best way to control the voltage and keep everything the same? The resistors I am using to bias to 1/2 voltage are all darn close to 1M ohm. Does this circuit really have that little room for error? Any ideas to make it more forgiving?

[R O Tiree]

How about providing a separate +5VDC supply for the switching system, rather than powering it from the Arduino?

[Fuzzy_BYU_ME]

Update:

I tried using a 2nd Arduino to add the 5V supply, and that made absolutely no difference.

I decided to try adding capacitors with pull up and down resistors to and from the pedal again (same setup as what comes from the guitar and to the amp) and that actually seamed to help this time. Maybe I didn't pay close enough attention last time...

With the extra caps and resistors, I tried several kinds of pedals this time. My OCD again, plus my RAT, Muff, true bypass Tremolo, a delay, a reverb, and a compressor. There is still a bit of a pop. Not nearly as bad as without the caps/pull resistors, but it's still there a bit. It's the worst with the compressor (guess that shouldn't be a shocker). I tried using a patch cable directly from pedal out to pedal in, and got pretty much no pop that way.

Any other ideas? Should I try maybe a bigger cap value, or maybe some smaller values for the pull up and down resistors? Is it possible to make this switch bring pedals in and out silently?

[Fuzzy_BYU_ME]

Oh, I forgot to mention this, in case it matters.

For the 3rd switch which is unused in the GEO article, I am running 5V into common, and when switched the 5V goes through an LED to tell me which position the switch is in.

[R O Tiree]

Using an Arduino to supply 5V is a bit of an atom-bomb to crack a nut... look up 7805 voltage regulators - cheap as chips and very stable output.

I would definitely look at PRR's answer above... if there is any residual DC bias to the input/output of the OCD, then it doesn't matter how well you have the voltages around your 4053 switches matched, there will always be a pop.  If there is no DC bias at either end of that pedal, it's time to really try to get all those pin voltages matched as accurately as possible.  For example, when logic is low, pin 15 should be the same as pin 2... it's out by 60mV.  When logic is high, pin 1 should read the same as pin 15... it's out by 30mV.  Both of those are enough for a pop.  Pins 12-14 are very concerning... there's a 130mV jump on pin 12 between logic high and low and that will definitely cause a pop.

The internal resistance of these IC switches is only about 150ohms or so, so putting V_Ref via a 510k resistor on each of those pins should result in an identical voltage at each of those pins.  The resistors are simply there to provide a bias voltage (so the switch doesn't pop) and they're big enough that cross-talk between channels is negligible. Visualise it like this:

V_Ref -> 510k ->150R -> 510k -> V_Ref

If V_Ref is the same at both ends of that chain, you've got a teeny little resistor in the middle of a whopping great 1M+ resistor.  You should not be able to measure any difference at all. 

You have checked that all the electro caps are -ve pin to incoming signal, +ve pin to the CD4053 pin?

The 3rd switch to power your LED should not make any difference to the voltages on the other pins.

[Fuzzy_BYU_ME]

R O Tiree,

Thanks for the great reply. I just double checked, and yes, the + end of all caps are in the direction of the CMOS.

I will give 510k ohm resistors a try in the next few days. Right now I'm using 1M. I'll let you know how that goes.

[Fuzzy_BYU_ME]

Update:

I've replaced all pull up and down resistors with 560k instead of 1M. I also replaced the voltage divider resistors with 10k instead of 1M. New voltages are as follows:

When control voltage is 0
Positive Power: 5.06 v
1/2 V: 2.531 v
Pin1: 2.47 v
Pin2: 2.49 v
Pin12: 2.49 v
Pin13: 2.47 v
Pin14: 2.49 v
Pin15: 2.49 v
Pin16: 5.06 v

When control voltage is high
Positive Power: 5.05 v
1/2 Voltage: 2.524
Pin1: 2.49 v
Pin2: 2.41 v
Pin12: 2.41 v
Pin13: 2.50 v
Pin14: 2.50 v
Pin15: 2.50 v
Pin16: 5.04 v

Clearly the voltages look a ton better than it was. But the pop is still really bad. It's looking like Pins 2 and 12 are the culprits (they are tied together to send the bypass the guitar signal with low signal voltage). What should I do about this? Should I try an even smaller resistor for that position? Everything else seems to look great. Any suggestions?

[idiot savant]

I took a look at the geofex article.

I would try removing the bias resistor on pin2, and pin12.

Think about it, in both switch cases you'll want 2 bias resistors in parallel. In the bypass state you'd have 3 using that extra one. That might cause a DC shift big enough to pop.

[Fuzzy_BYU_ME]

Update:

I've spent some time playing around with the circuit. I tried removing the pull up resistor on pins 2 and 12, but that made little difference, if anything made it slightly worse. The most helpful thing was to add some capacitors to and from the pedal. They reduced pop a TON. But still not eliminated. Pull down resistors at the pedals didn't seam to help, so I figure I'll save parts and not include any.

So, here's what I've got going on for my best results as of now: All pull up and down resistors are 560k. I am using 4.7u Electro caps, one at input, output, pedal send, and pedal return. The voltages are:

Logic Low:
Pin 1: 2.414 V
Pin 2: 2.494 V   *
Pin 12: 2.493 V *
Pin 13: 2.412 V
Pin 14: 2.494 V *
Pin 15: 2.495 V *

Logic High:
Pin 1: 2.464 V  *
Pin 2: 2.405 V
Pin 12: 2.403 V
Pin 13: 2.462 V *
Pin 14: 2.462 V *
Pin 15: 2.466 V *

Something interesting. For either logic level, the used pins (marked with *) are all darn close to each other. That's good. But there's some major voltage dropping between logic states. The pins not being used for each state are also way lower than they're supposed to be.

Just checked my Vref. When logic low, Vref = 2.536V. When logic high, Vref = 2.514V. I have a big 47uf Electro cap with the divider resistors, per the GEO schematic. Maybe this could be the root of my problem? Anything I can do about this? Maybe I can find some 2.5 V regulator to create my Vref? I'd really like to make these pops disappear, and I'm told it's possible, though I hoped it'd be easier than it's shaping up to be.

[ElectricDruid]

Since the 4053 switches instantly rather than slowly, if there's any sort of signal on the input when you switch (even just some background noise), you'll probably hear the switch. You can reduce it, but you *can't* switch audio instantly and expect it to be noiseless. It just doesn't work like that. Imagine if your guitar is halfway through a 1V waveform when it switches...that goes instantly to zero. That sharp edge as the voltage drops to zero is heard as a serious pop or crack or click or whatever you call it.
The only way around that is a "soft fade" of a few milliseconds or tens of milliseconds to eliminate the click/pop as the level shifts from one to the other. You can't do that on the 4053. The only solution there is to get the two levels as close as possible *before* the switch (which is what you've been trying) and then hope that when you switch they're the same.
This is why more modern audio switching chips have "zero crossing detection" circuits on the chip, so they can switch when the next zero crossing is detected and both in and out channels are at zero.
The alternative to something like that is something more like the Boss FET switching, which uses a slow (ok, slow*ish*) fade in/out rather than instantaneous switching.

HTH,
Tom