Clicking Noise in Analog Multiplexers 4

[2sb18]

Hi guys/gals,

I'm making a signal router for musicians. I'm using analog multiplexers (MAX307/MAX308) instead of relays. I'm getting clicking noises when I switch between different routing schemes. Does anyone have any suggestions as to how I could reduce these.

Thanks,
Steve

 

[itsmoked]

They are caused by the fact that discontinuos changes or steps are caused when you switch. These act like music with a zillion different notes all at once. (orchestra falling down elevator shaft). These are usually hard to get rid of because getting rid of them gets rid of the music too.

If this router has amplification, one way is to turn down the output amplification just before switching and then turning it back up after. This can happen very quickly!

 

[sreid]

Two other thoughts;

1) Slow down the control signal speed (RC filter) to minimize capacitive feed through spikes.

2)Most audio relays/switches are form D (make before break). See if a slight overlap (1- 10 ms?) for signal on/off helps.

 

[BrianG]

Are you using the analog switches with d.c. isolation capacitors between pre-amplifier stages? if the impedances are high at the point you have inserted the muxes then you could get quite big transfers of charge when you switch, which makes the thump/click/plop noise. If you buffer the input to the muxes so that they are driven from a low source impedance this should help.

 

[2sb18]

Thanks BrianG,

Yes, now I am going to be putting op amp buffers before and after each of the switches. That should clean things up a bit.

S.

 

[elfgriper]

2sb18,
I don't think that slowing up the switch signal will work on this device. It looks to be driven by logic gates which will switch when the threshold voltage is reached. You might try using a single fet for each switch and ramping the control signal as stated above. If using a low res on fet you can put the fet across the buffer amp such that turning it on reduces the gain of that circuit to very close to zero. gain = Rfet/Rin
-elf

 

[itsmoked]

elfgriper great idea.

2sb18; I can not see how op amp buffers will help in the least...

 

[2sb18]

Thanks guys for the excellent advice. As for itsmoked's comment that buffers won't help, I would like a second opinion on that. I found a maxim application note (AN730) that states that the THD of a multiplexer is roughly:

THD = Rflat(ON) / Rload

Since an op amp buffer would increase the Rload, wouldn't this reduce THD?

I am going to take itsmoked other advice about decreasing gain to zero while making the transition and then bringing it back up. Right now the signal chain is:

Input buffer -> Multiplexer -> Buffer -> Attenuator -> Output buffer

Are there too many buffers?

Thanks for all your help guys,
Steve.

 

[BrianG]

The original question in this thread seems to have got somewhat confused! The problem was to avoid clicks and thumps produced when switching muxes. This problem is normally caused by charge transfer from one signal source to another and can be avoided, or at least minimised, by ensuring that there are no d.c. offsets or differences in potential at each source - including charged up isolation capacitors.

If you are now saying that you want you actually want is to switch muxes whilst these are carrying signals, then with basic muxes you will still risk getting clicks and thumps due to sudden connection/disconnection part way through the signal waveforms.

To avoid this situation you require programmable attenuators (either digital or analog)to avoid sudden "chopping" of the signal waveform. A relatively fast envelope shape over a few tens of milliseconds should produce a clean signal changeover (off to on, or on to off) without audible "artefacts".

NOTE: You may be able to dispense with the separate muxes by summing the individual attenuator outputs together into a single amplifier input.

 

[sreid]

brianG has an excellent point that might be generalized to the fact that in most posts we have no idea what the circuit is we are trying to "analyse."

Could the click problem simply be that in audio applications Form D switch contacts (make before break) are used to avoid the "click" or "pop" and for solid state switching the control circuits must electronically provide the overlap.

 

[2sb18]

i did confuse the issue when i brought up THD since my original post was about clicking noise. sorry about that. even though it's true what sreid is saying about not being able to see the circuit you're trying to analyse, the advice i receive on eng-tips still continues to point me in the right direction. thanks for all your help guys.