How to smooth the audio switching pop or spikes?

[wuemura]

Hello guys, long time no see!

I've done a audio bypass switch using a pair of CD4053 and a CD4049, the two 4053 are biased with 1M Ohm resistors at the input and the 4049 are build to generate signal to the CD4053. All works fine, no noise (thanks god) but when i change the switches i can hear a clear "pop".

I was wonder if there is a way to smooth this "pop" sound when i change audio source.

Here are an sample of the pop:
Download

Here is the picture showing the first version at the top and the new version with the bypass circuit.

Thank you.

[audioguru]

Hi Wellington,
I have made many audio switching units with the CD405x multiplexer IC without pops. The trick is to DC bias the inputs, outputs and sources at the same DC level.

Please attach the schematic of your switching circuit to see.

[audioguru]

Hi Wellington,
Here are 3 causes of your pops:
1) Your CD4053 multiplexers have 3 different bias voltage dividers. With 5% 10k resistors their "6V" DC voltage could differ by up to 0.6VDC. Use only a single voltage divider. Test it by shorting together all 3.

2) You are switching when the program is playing. The existing program could have an instantaneous voltage of +9V or more and the other program could be +3V or less when it is switched. The CD4053 cannot blend the inputs nor be switched slowly. The only way to avoid a pop is to fade down the existing and new signals, switch when the programs are muted then fade the new program up. Too complicated.
You can buy a digital volume control IC that waits for a "zero crossing" in the signal before it steps the volume to eliminate a pop. Its circuit must be very complicated inside.

3) The outputs of the CD4053 multiplexers are not biased so the output capacitor that is not selected is not charged. It charges suddenly with a pop when it is first switched. Connect 1M bias resistors to the outputs and to the single "6V" bias point.

I notice that the coupling capacitors are 10uF. They will take 50 seconds to charge to 98% of their fully charged voltage by the 1M resistors when the power is first turned on. 1uF capacitors will charge 10 times faster and still be big enough for good bass since a 1uF coupling capacitor can drive a load of 37k ohms and have a response that is flat down to 30Hz.

[wuemura]

Hi Audioguru.

1) The first voltage divider is made to make the OPAMP work with non symetric supply, as you suggest me at the time to do it (R12/13 C5). The main problem that i've faced with this is the cross talking effect when, if, i dont use the other dividers (R17/18 & R32/33) maybe because the R5 (100K) at the pin 3 of the OPAMP, if i remove R5 the circuit doesn't work and if i join all at the same point i get this strange effect. So the idea was to isolate and avoid this effect.

2)Hahaha  ;D
No Audioguru, in fact, this is the same program from the same audio source, this is the main propose of the circuit, to use the OPAMP and make a calculation and reproduce only R - L as surround channel and R + L as center channel, this configuration is also know as Hafler Effect. It took me a long time to get to this configuration because the effect is not perfect, you get some "garbage" as a result if you play a bad MP3 audio for example, you can hear the compression artefacts mixed with what you want to hear.

If you heard the sample, you will noticed that the vocal is almost gone and the background is almost intact and very clear, using some acoustic technics when used inside the car you will only hear the back ground and the vocal is completly removed making a very impressive effect.

The configuration here with the CD4053 is made only to bypass the surround circuit since i have only one source of audio (my car head unit) and on power amp.

But, thank you for the comment, if you told me that i was changing from two different sources i guess i'm doing a good job ;)

3) We will have to solve how to make only one bias point, if i bias the output using the same configuration will do the trick??

The 10uF capacitors play a important role at this circuit and i can explain why:

a) My car audio system was build with focus on quality, so i use a pair of component speaker with a pair of passive cross over each component, a total of 4 passive crossover and a pair os 12' JL subwoofer.

b) The signal is digitaly crossovered to not allow 120Hz frequency and below (my head unit is Pioneer DEH-9880BT), tests made with capacitors from 1uF to 4.7uF do a strange "buzz" when i turn the circuit off.

c) Since i have subwoofer in the car is not interesting to me to make the surround channel reproduce low bass, and the second point is that part of the solution to remove part of the garbage from the surround channel is to not allow too many bass in the channel, including thouse 30Hz frequency.

d) As part of the design, when the surround circuit is turned on it will not kill all the bass but also will not remove all, maybe allowing to pass from midle to high frequency (200Hz and up, sorry can't precise that information), this is good for voice, guitar, symphony orchestra or what ever you hear at the car. You will notice in that sample that the backgroung is pushed forward, mixing that with the full configuration of the car you will hear the voice on the front and that beat song at the rear. Fading away that little voice echo and pushing forward the background, since is playing midle range frequency and up you can pin point exacly that the backgroung music is coming from the back and the voice is coming at the front.

There is another problem that i thought for a long, long time. The bias resistors are in a indirect way "mixing" the signals ( L  +  R ), when i say indirect is because is not my intention to mix the signal channels but also i can't make it without it. If a 1M bias resistor makes the 10uF charge 10 times slower, if i down the value from 1M to someting about 530K it will make the same capacitor charge 5 times faster, but also will also mix the channels evem more, correct?

Thank you.

[audioguru]

Hi Wellington,
The impedance of the capacitor to ground at the voltage divider's "+6V" point determines how much crosstalk there will be. The impedance of the 100uF capacitor at 100Hz is only 16 ohms so the level of crosstalk from the 91k plus 100k resistors biasing the opamp is only 0.000084 which is about -72dB and less at higher frequencies and probably cannot be heard.

Use a single voltage divider with a 100uF filter capacitor and 1M resistors to each input and each output of every CD4053, plus the bias resistors for the opamp. Use a 470uF capacitor instead of 100uF as the filter for the voltage divider to decrease the amount of crosstalk by 14dB. Use a 1000uF capacitor to decrease the amount of crosstalk to -92dB at 100Hz.

With a single voltage divider then pops will only be caused when the switching is done during a high amount of signal.

[wuemura]

Hi Audioguru,

Thank you!

Back to design board.....

[wuemura]

Hi Audioguru,

Changes made, i removed R17/18/32/33 and capacitor C15/17, joined R21/22/26/27/28/29 to R6/C8 point,  added 4 more 1M bias resistors at each output and joined then also to R6/C8. Upgrade C8 value from 100uF to 470uF.

The "pop" is not gone but is in a level so low that you can't hear it depending on the audio volume, if you can hear it is more like a soft pop, much better that way.

But a few problems came with the solution:

1) There is a huge eco overlaping the sound with a few seconds delay.
2) When i turn on the device with effect on looks like the sound is charging someting and is noisy until it get fully charged, since i changed the C8 to a bigger value it must be it.
3) Sound "blow out" a little at high frequency.

By the way, i've removed C1/2/9/10, it did not alter nothing at the item 3 above.

Thank you.

[audioguru]

I don't know where the echo comes from. It takes a lot of hardware to make a delay then mix it to make an echo.

I don't know what takes a noticeable time to charge. R12 and R13 have a source resistance of 5k. They can charge a 470uF capacitor to 63% of full charge in 2.35 seconds. It that a problem?

[wuemura]

Hey, let's get a patent of this before some one do, we've build a echo machine without a echo machine  ;D

Joking  ;D

This echo effect is coming from the OPAMP, months before i design the board the way it is now, i've tryed to bias all together the way you said and had the same problem, i get the effect plus this echo overlaping it. My guess is R6, the circuit generate some small echo but depending on the configuration you get more or less of that.

That is why i've choose to make two voltage divider for each IC, maybe if we put a diode after R6 to avoid audio return?? (with the cost of 0.7 voltage drop :( )

You are right, it takes about 1.5 seconds to charge, the problem is this 1.5 seconds of noise.

I've recorded more two samples, one you will notice the echo and the other one (sorry for the lyrics) you will notice the noise, when he starts to sing is more or less ok, but when the voice decay you hear some discharge like noise.

DOWNLOAD 1
DOWNLOAD 2

Do you have more ideas?

Thank you.

[audioguru]

The echo is normal "ambience" in a surround sound system when the source is in stereo. It is caused by echoes in the recording studio or theater which have random phase, various levels and are delayed because they travel far. Poor quality electronic reverberation circuits make strange noises in a surround sound system when the main sound is cancelled by the left-minus-right circuit.

Prove that your circuit is good by feeding in a mono signal into both channels. You might need to carefully adjust the level of one channel for a perfect null (cancellation).

One day my mono FM radio went dead. It worked normally on all stations except my favourite one. I tried that station on my stereo and discovered that they were broadcasting the channels out-of-phase!
In-phase sounds were cancelled. Because stereo echoes have a random phase and various levels then they don't cancel.

[wuemura]

Ok, hold on a minute, dont go any were ;D

[wuemura]

Don't challenge me, i love a challenge ;D

So, circuit is fixed, return C8 to 100uF (bye noise), add a 4148 diode before R6 (OPAMP pin 3), and i play a bit with the levels by changing R1/2 values (still messing around).

So check now what i've got ;D

DOWNLOAD 3
DOWNLOAD 4

[audioguru]

It sounds good now. ;D

I don't know where you put the diode. It might rectify the sound if it gets louder than about 0.6V peak. With lower levels then it might mess up L-R.

[wuemura]

Cathode at OPAMP pin 3 Anode at R6.
I'm still messing around to fine tune this ;D

[audioguru]

The only thing your diode will do is become forward biased when the signal is loud and goes positive then it will charge the small capacitance at pin 3. Then pin 3's DC voltage rises and the opamp's output DC voltage also rises. It causes an odd action but I don't think it affects the sound.

If the input leakage current of the FET in pin 3 is positive-going then the diode will be reverse biased and cause the opamp's input to float up to +12V then the opamp won't work.

So the diode should be removed.

[wuemura]

You are right, here at my mini-lab it works ok but the pioneer has a 5v output and mess up everithing, diode removed :P

So, i've played with the levels and make tests in my car, so the R1/2 value now is 62K.

Few more samples, you will enjoy this one ;D

Download 5
Download 6
Download 7

[audioguru]

I played with surround sound about 30 years ago like you are doing now. I made a surround sound decoder system using Motorola's SQ Full-Logic ICs. Some vinyl records were made encoded especially for it. It also worked very well on ordinary stereo music and had knobs for the amount of effect and amount of automatic fade control.

[wuemura]

Hey Audioguru, is nice to talk with a guy like you because every mistake i do you say "don't do it, i've done it before"  ;D
30 year ago it must be a very interesting year to live in, i was born at 75 so i wonder how it wolud be to enjoy the 60's, Elvis Presley, ABBA shows and so on. Today kids just think in HALO, GTA, MSN... Not that i'm against it, but in my times we enjoy more the time that we had.

Some months ago i was thinking to add more features to this like a LCD, PIC, digital controlers and so on, but this surround is good the way it is, like my teacher said to me "keep it simple" ;D

Congratulations on your surround, it must be a very interesting pice of hardware!