CD4049 click noise

[Marvin777]

Hey guys!

Hope you can give me some tips about this problem.

I'm using a CD4049UBM on one of my designs to control a slow muting JFET and 4053 mux for switching signals.

So far it's been doing well but as one more buffer is connected (pin 11) a strong clicking noise is audible and I don't understand why.

For reference:

- 4049(2) goes to a JFET switching
- 4053 (9-10) goes to control pins 9 and 10 of a 4053 mux
- 4049 (12) floats or it is connected with a jumper to a 4053 control pin.
- I've changed original R27 and C14 values from original R.G. circuit to suit my switch bouncing.

I'm running out of ideas of why the inverter is clicking and I don't want to just discard the use of that last buffer.

Cheers!

[ElectricDruid]

I'm not sure I understand the problem. Why *wouldn't* it click? It's all digital square-wave signals. That's the point of it. If you feed sharp square edges to anything, pretty much, you get clicks.

So what does the rest of the circuit look like, and how have you prevented clicks on the other parts of it? Do the same for the last bit?

[Marvin777]

I'm not sure I understand the problem. Why *wouldn't* it click? It's all digital square-wave signals. That's the point of it. If you feed sharp square edges to anything, pretty much, you get clicks.

So what does the rest of the circuit look like, and how have you prevented clicks on the other parts of it? Do the same for the last bit?

Hey Tom, yes you are right, CD4049 switch as fast as hell and that may cause a spike current that translate to clicking noise bleeding through the audio path. That was one of my thinkings but was not sure about it, maybe using the least quantity of buffers is the right way.

I'm using the outputs of this hex buffers just to control a switching JFET like the one in the picture, a 4053 and that's it.

[dschwartz]

Probably the click comes from the 4053 switching

[ElectricDruid]

+1 agree with dschwartz.

You've got RC's to slow down the switch-on of the JFETs, but you can't do the same trick with the 4053, I don't think. Not totally sure. Do the 4053 control inputs have Schmitt triggers on them?

The switch switches instantly, and that translates into an abrupt change in the audio signal, which you hear as a click. The only way to remove that is to replace the fast switch with a slow crossfade, but you can't do that with a 4053.

[Marvin777]

+1 agree with dschwartz.

You've got RC's to slow down the switch-on of the JFETs, but you can't do the same trick with the 4053, I don't think. Not totally sure. Do the 4053 control inputs have Schmitt triggers on them?

The switch switches instantly, and that translates into an abrupt change in the audio signal, which you hear as a click. The only way to remove that is to replace the fast switch with a slow crossfade, but you can't do that with a 4053.

In order to isolate the problem, I cut the traces that goes to the 4053 and the jfet and the click stills there so 4049 is the suspicious one.
I've replaced the chip with a new one and nothing changes, so the buffer is not faulty.
This is my first experience with this kind of switching configuration (and maybe is the nature of the beast) but my early conclusions are more gates switching more noise.

[dschwartz]

Are you decoupling the 4049 supply?
I try to separate the 4049 supply from the audio supply by using a small resistance before the audio supply, and connect the 4049 straight to 9V with it's own capacitor.
That way the current spike caused by the 4049 switching doesn't creep into the audio supply..
Also try making the slow down cap of the jfet bigger..just in case. 2n2 or 4n7 ....until it stops clicking.
Also if your switch mechanical "click" can creep into the signal if the pcb is microphonic (ceramic capacitors)

[anotherjim]

Exactly what Daniel just said. Decoupling the power supply to the 4049 has a good chance of mitigating the current pulse effect.

Actually, since you have low current needs for the LED and all other loads are high impedance, the two inverters in the flip-flop and a third inverter providing the inverted control is probably all you needed from the 4049U. Also, invert the LED drive to run from +9v to the inverted control so it's on when driven low - 4049 can sink more current than it can source.

Probably the problem stems from the flip-flop using the fairly slow unbuffered version of the chip. The less snappy switch action of the unbuffered device means the RC timer in the flip-flop causes it to spend some time in the "both MOSFETs on" state when the capacitor voltage is between logic levels. This causes the following inverters to spend too much time in between also.

[R.G.]

If I were doing this circuit, I'd raise the values of R27 and R29 by a factor of ten, and lower the value of C14 by the same factor. Those values have worked well for me in a similar application.

You're getting good advice. The 4049 is a high current CMOS device that takes almost no current while static, but pulls pulses of current from it's positive supply and dumps those same pulses into its ground. The pulses are short, but intense. One way to help that is to put a local bucket of electrons right at the chip in the form of a decoupling cap as close to the chip's pins as you can get it. I always use 0.1uf ceramic. The pulses come from and go into the cap before getting spread back into the power supply for other things.

If it were me, I'd use fewer inverters. Connect the tails control signal right at pin 4, and whatever you're driving with your pin 12 right at pin 2 instead, and disconnect that pesky pin 11, or more properly, tie it to V+ or ground and leave pin 12 open.

If you try this, let us know what happens.

[amptramp]

Work I have done in the past shows that when a CD4049 goes from one rail to the other, the current drawn at the midpoint is about 8 mA.  That is a lot of current in a stompbox.  The advice about local decoupling is valid.

A CD4053 will snap suddenly from one input to another and there is no way to slow it down.  There is a way to synchronize switching so it happens when the two signal inputs are equal and therefore generates the least noise.  I have a thread about quiet switching:

https://www.diystompboxes.com/smfforum/index.php?topic=120006.msg1122270#msg1122270

Something in there may suit your purposes.

[dschwartz]

Btw, cd4016 can be switched slowly as they are unbuffered switches.

[Marvin777]

Thank you people for all the valuable advice on the subject, I have found a terrible mistake on my board that is causing the strong spike or most part of it. The analog audio input signal trace is running parallel (and close) to the trace that is carrying the pulse from 4049 pin 12 forming a big capacitor. Lifting this pin from the board stopped the click.

Besides that, I will try to make the changes that you guys suggested to increase the overall performance of the board:

1 - Isolate the 4049 supply from the analog supply by inserting a serial resistor to the 9V that goes to the analog part.
2 - Use the least amount of hex buffers
3 - Adjust values for R27, R29 and C14
4 - R.G. you mean putting a cap in parallel to the 4049 supply rails or in every output?

One way to help that is to put a local bucket of electrons right at the chip in the form of a decoupling cap as close to the chip's pins as you can get it. I always use 0.1uf ceramic. The pulses come from and go into the cap before getting spread back into the power supply for other things.

5 - 4053 is performing great and I'm confortable with the inaudible small clicks when switching signals that are not zero crossing.

I will give these items a try and let you know guys, thanks a lot.

[Slowpoke101]

R.G. certainly meant to say that the capacitor would go in parallel to the 4049 supply rails. A nice bucket of electrons (love that term ) immediately available to the 4049 should help satisfy its current hungry switching demands.