Portable Headphone Amplifier: Volume Pot Problem...

[Relayer]

Hey guys,
I have a problem with a Portable Stereo Headphone Amplifier circuit which was obtained via a magazine issue I have from Silicon Chip Magazine here in Australia.
Unfortunately I am no audio guru, though I have built pre-amps and amplifiers in kit form, as well as guitar effects units, but I've never done or tried designing something myself. I just don't have the knowledge.
I used to be able to repair audio gear in my early technician days, but that's as far as I got.

OK, here's my problem...
Please refer to the link below:

Headphone Amp

The circuit works extremely well. Sound is quite good for a dual-opamp IC. No hum, very low noise and its virtually distortion free using the SGS-Thomson TDA2822M IC.
My only qualm with it is the range of the volume control pot. It is woeful.
You only need to turn it by one quarter and its already at maximum volume.
I assembled my test circuit on a Breadboard and kept the components close to each other as possible.
I've tried higher value pots, right up to 1M ohm, but they absolutely make no difference at all. I then turned to using Linear pots instead of Log types, and a 10k Linear pot worked much better. I can now wind it up half way before I hit full volume.

What I really need is a solution so that I get reasonable travel with the volume control from no sound what-so-ever, to maximum volume at the end of the pots travel.
Any help with the above would be most appreciated.
Thanx in advance.
Regards,
Relayer

 

[Audioguru]

A log pot should work perfectly but you probably have its input and its slider swapped by mistake.
When a linear pot is set to halfway then its output should be half its input that sounds almost the same as max.
When a log pot is set to halfway then its output is about one-tenth the input that sounds half as loud as max.
When a linear pot is set to one quarter then its output is fairly loud.
When a log pot is set to one quarter then the output sounds much less than at max.

A log pot is marked "A" and a linear pot is marked "B".

Usually a volume control is turned down maybe to three-quarters so that it can be turned up when the input signal is too low.

 

[Relayer]

Hello Audioguru,
Thank you for the quick reply.

A log pot is marked "A" and a linear pot is marked "B".

I am aware of the designations and I have a fairly reasonable stock of pots, linear as well as log types,
dual and single gang.

I've checked and double checked my breadboard circuit and can find nothing wrong with the input wiring
and components. Though I have to admit that I'm only using one channel for my testing. i.e. Left Channel.
I am in the process of making a small PCB to accommodate a dual-gang pot that can be inserted onto a
breadboard with out the pins contacting with each other.

One strange thing though: even though I'm only using the left channel only, I still get sound out of the right
channel, even though its not physically connected at the input or output.
I can only surmise that its the TDA2822 that's causing the mixture somehow. I have to admit I am pretty puzzled
about this occurrence.
Any further help would be appreciated.
Thank you once again.
Regards,
Relayer

 

[Audioguru]

A solderless breadboard has intermittent contacts and stray capacitance between all the rows of contacts and wires that are all over the place.
An output from an amplifier that has no input could be that stray capacitance feeds it or the grounds between the channels has some resistance in it because nothing is soldered.
The channel separation in the datasheet shows 50dB typical which is 1/320th.

 

[KlausST]

Hi.

I agree with audioguru.

You say a linear pot is better than a log pot.
This really sounds like wrongly connected pot.
10k is a good value. I recommend to use a log type.
Do a measurement:
Turn it to center position.
It has three conections, mark them randomly with A, B and C.
Then measure The resistances:
* AB
* BC
* CA
You should three values:
* about 10k (the highest value) --> mark the "not used pin" as "W"iper. ( when 10k is across BC, then A is the wiper) it is the pot output and needs to be connected to the TDA2822 input.
* about 9k --> one connection is the "W"iper, the other is the audio "IN"put
* about 1k(the lowest value) --> one connection is the "W"iper, the other is GND. It needs to be connected to the audio signal ground.

ALL GNDs need to be connected:
* audio input GND (best located at the pot GND)
* both channel pot GNDs
* amplifier GND
* power supply GND
* headphone GND

If you are still in doubt with your circuit, then upload a photo of your connections.

Good luck.

Klaus

 

[Audioguru]

When a linear pot is turned down from max to halfway then the loudness is almost the same as at max. When it is turned down more to one quarter then the loudness is still too loud. When it is turned down more nothing much happens until the sound suddenly turns off like a switch.
A log pot adjusts volume perfectly because our hearing's sensitivity to loudness is logarithmic.

 

[FvM]

I notice that the implemented way to reduce TDA2822 gain from original 40 dB to about 20 dB isn't suggested by the datasheet. The required input voltage for full output might overload the input stage.

Apart form this point I don't see a possible reason for cross talk. I'm however sure that it can be found by well considered tests and measurements.

 

[Audioguru]

I also wondered about the added negative feedback instead of having an input attenuator. The amplifier might oscillate with more feedback. The datasheet for the LM386 little amplifier says it oscillates when its gain is reduced to 9 or less with added negative feedback. Luckily the TDA2822M is obsolete and ain't made anymore.

 

[Relayer]

Hey guys,
Thank you very much for all your input.

Audioguru said:

A solderless breadboard has intermittent contacts and stray capacitance between all the rows of contacts and wires that are all over the place.
An output from an amplifier that has no input could be that stray capacitance feeds it or the grounds between the channels has some resistance in it because nothing is soldered.

KlausST said:

I agree with audioguru.

I decided to dig out a larger unused breadboard I had stashed away some time ago.
I reassembled the circuit, plus I made a proper jig for the dual-gang pot I was using,
so it could be plugged into the board snugly.
It does seem that both your suggestions were spot on. The circuit is working flawlessly
on both channels, though I have to reduce the input source a fair bit to reduce full
volume distortion.
Unfortunately I have a frequency generator in the works, but I've been a bit slack in
assembling it.

Anyone have an idea on the set voltage output of an Ipad or mobile phone?
I've heard, but not confirmed, that its around 700mV at 1kHz. Is this correct,
or there's no true standard being used at all?

Thanks again guys for all your help, its very much appreciated.
Regards,
Relayer

 

[Pjdd]

It seems to me that the amp is too sensitive even with the added negative feedback. But adding more feedback might cause instability. I suggest reducing the signal strength at the input side by placing a resistor before the volume control.

You can find the correct resistor value by trial and error. If you're using a 10k pot, I suggest you start with a 22k resistor before each pot. That is, one 22k resistor between AUDIO IN and C7 and another between AUDIO IN and C8. This will reduce the signal level to about one-third when it reaches the pot. If it's still too sensitive, try 33k and so on.

This is how it works:
The amp with a 9V supply will have a maximum undistorted output of around 2.5Vrms. The circuit looks like it has a gain of 11. Therefore, an input of 2.5V/11 or 0.23V will be enough to drive it to full volume. Now, suppose your phone puts out the nominal line level of 0.775V, that's more than 3 times the voltage needed to drive the amp to full power.

Of course, the output from the phone is not constant and obviously changes with the music level. And the output voltage of your phone at average music level is unknown. The 0.775V line level is just a starting point. This is why I suggest reducing it to 1/3 to begin with.

 

[Audioguru]

If you use a log volume control pot then when it is set to halfway, its output is about 1/10th its input which should work fine if the pot is connected correctly.