Amplifier Interference Issue

[Art]

Hi Guys,
I have a few projects where pics make sounds (as do a lot of us I imagine).
In a few projects I have used a small 1 Watt amplifier to drive a loudspeaker.

All is good, but I hear a little hum from the amp when it's supposed to be silent.
If the project has a flashing LED or the like, I then Hear a hum flashing at two different frequencies.
I assume this is interference from the pic itself.

A solution I'm thinking of is to leave the amplifier on all the time,
but to allow the pic to switch on the speaker only while a sound is being played.
The interference is inaudible while a sound is actually playing.

My question is.. What is a suitable circuit to achieve this?
The amplifier chip is a TDA2822M.
Cheers, Art.

[Byte_Butcher]

Is the noise coming form the output pin of the PIC and fed to the input of your amp, or is it being coupled in through the power supply?

If you disconnect and ground the inputs of your amps does it go away?

If you set the PIC pin that drives it to an input (during the time you're not using it) would that help?

steve

[mtripoli]

You've got a whole list of things to run down...

Where is your power coming from? A battery, wall-wart, PC? If it's anything other than a battery you need to see how quiet the supply is. Wall-warts are incredibly noisy and unless you take care you are going to get noise. What kind of decoupling do you have on the supply? You're going to need something fat and something skinny. Start with 100uF-470uF electrolytic on the rail. Pair that with a 0.01-0.1uF ceramic (maybe both). Skipping this and you might as well forget it. You can go further and use a series resistor in the supply but start with the caps.

How is the input connected? AC-couple (use a cap on the input to the amp). Something like a 0.47uF should do it. Then it doesn't matter if the PIC is an output or an input (the cap blocks DC. Regardless if it's an output high or low nothing is showing up at the amp. The only time the amp sees signal is when you are outputing "sound" as an AC waveform). Use a high value resistor to ground on the input of the amp (on the amp side of the cap). Something like a 1M ohm (the data sheet shows a 10K but this is to match the output from a preceeding pre-amp or line level. You don't need that coming from a PIC). This keeps the input at ground unless there is a signal.

Are your grounds tied together at one point (preferably at the supply)? If you have multiple ground paths then you're going to get into trouble. Tie the supply leads together at the supply. Use a single wire from the output of the PIC to the input cap.

Looking at the data sheet you should run this in BTL (bridge-tied load) configuration. This uses both amps in the package and gets rid of the big caps on the output (the 470's). This part uses a lot of caps, make sure you are using them. Keep the leads short and if you are using a "plug-strip" protoboard get rid of it and solder the stuff together. Flying leads between components are far better than proto-boards.

There are amps out there that have an enable pin. You can get one of those and turn the amp on and off as you need. If something is battery powered this saves on power. This only masks the problem however; the hum and noise are still present unless you get rid of it.

Mike Tripoli

[Art]

Ok, that gives me a lot to think about.
Unfortunately the protoboard has to stay (this is a one off).
I am using a plugpack, but the interference sound seems vary according to what the pic/s are doing, so I don't know what that means.
I don't think i've ever tested this particular project with a battery though,
so that might be interesting, thanks

I guess I'll revive this if I run into trouble.

[Archangel]

. . .but the interference sound seems vary according to what the pic/s are doing, so I don't know what that means.

That could be a variation in leakage across the diode's junction due to changes in load . . . Used to see that a lot in car alternators.

[Art]

Is that a rectifier diode in the plugpack power supply ?

Hey mtripoli,
How would you like to make a PCB for this:


Give me a bit and I'll whip up a schematic with the transistors the wrong way round
Nah just kidding

I'm not actually using the 7805s at the moment because I have a 2 amp 5 Volt regulated wall supply.
The plugpack is a charger for a Sony PSP, and I would have thought the caps would be in that.

So it's a talking clock with a display now, but I also want to add an FM radio that I'm making with a Philips FM radio IC tonight.

There should be some circuit to switch the input of the amplifier chip
between the pic and the radio chip.
Should I use signal transistors?

[mtripoli]

I don't know what the output of PSP supply looks like. I have never seen what a "plug-pack" (what we derisively call a wall-wart) looks like for the PSP. I would be very surprised if it has any kind of filtering at all. Usually these things are a transformer, a bridge (if you're lucky) and MAYBE a single cap (it could be a universal input switching supply as well. If so the thing most likely has tons of high frequency component on the DC. Some ceramic caps strategically placed may be all you need. LM78xx devices don't respond at these higher frequencies and will let a lot of this through). The manufacturer places the caps and related circuitry in the device being powered (for lots of reasons). So first, whenever you are using a "plugpack" put some big caps on the input of the supply to the board. 470uF-1000uF is not unusual. Use some ceramics as well. Watch the voltage; don't use a 6.3V cap on a "5V" supply (higher voltage never hurts). Typically voltage rating is under load, so a "5V" supply may actually be much higher than that.

Mike Tripoli

[Byte_Butcher]

hmmm.... using a wall wart for power.....

Many "modern" wall warts now contain switching supplies instead of the traditional xformer/rectifier setup in order to meet minimum efficiency standards
that now apply in many areas.
That may well be the source of your noise. Like Mike said... an electrolytic cap of a few hundred uF, with 1 or more small (.01-.1uF) ceramic caps in parallel may help a lot.


steve

[Art]

Thanks Guys,
a 1000uF cap across the supply almost eliminates the noise
That's a bit big, but I'm sure I'll find a combination that works well.

[mtripoli]

That's a bit big, but I'm sure I'll find a combination that works well.

No it's not... 3x220uF is better, 2x470uF is ok... but 1000uF is not big...

[Art]

I just mean a bit big looking on the circuit,
and it might prevent PCB stacking with tight spacing (if I go that way).
I'd rather use a couple of smaller ones.

This one is to end up a talking clock radio.
At last, a project I actually use in day to day life

[Ioannis]

It is a big circuit and I can understand how difficult might be to change the connections.

But in most cases that involve audio amps or pre-amps, the problem arises from the ground loops. And I see a lot on your photo Art.

So, I would suggest to take cables from all the minus (ground points) of the circuit parts and connect them to a single point where your (-) of the power supply is.

This wll also help you reduce the spikes on your digital power rail.

I suppose that you are having different power line for the analog and digital parts, right?

Ioannis

[sayzer]

..
So, I would suggest to take cables from all the minus (ground points) of the circuit parts and connect them to a single point where your (-) of the power supply is....

Ioannis

And, from right there, have a bobbin serially connected to GND, thus, all of your circuit will go to GND via that bobbin.

The value of the bobbin is best found by experimenting.


-------------------------------------

[Art]

there are 5 PCBs in the project picture, they are just close together.
Each has a gnd and + 5 Volt rail, except the amp which takes power from
whatever power is input before the 7805s, or from + 5 Volts in the case
that I can supply the circuit with regulated 5 Volts like on a desk at home.

All of those ground rails are connected together.
It would have been exactly the same rail if the tracks ran vertical up one big board.
Is there any difference?

[Ioannis]

When you say "together" do you mean in a spagetti form one after another or in a star form?

The enter of the star is where the power supply (-) should be connected.

All other connections and configurations are creating ground loop currents.

Also if possible use this kind of connection fr the (+) of the other pcb's too. The star topology is very important to minimize interferance. Thena good decoupling locally is also very helpful.

Ioannis

[sayzer]

[Acetronics2]

Hi, Art

Nice piece of wrapping ... impressing !!!

Seriously, I do not see the unforgettable ceramic capacitor ( 0.1 - 0.22µ multilayer ) which ever should be placed CLOSEST as possible to the Pic supply pins ... ( lower and upper right )

Also, you should try to place the Xtal and its caps on the other side of the pic, close to the pic pins ... ( nice 8 Mhz emitter !!! ... lower right )

there's lots of noise to catch here ...

a big reservoir cap ( L.S.R. ) is also a must to place close to the multiplexed LED displays:
they generate lots of noise ...

Alain

[Art]

Thanks

The 0.1 uF caps are there right next to every pic. They are blue.

Hi, Art

Nice piece of wrapping ... impressing !!!

Seriously, I do not see the unforgettable ceramic capacitor ( 0.1 - 0.22µ multilayer ) which ever should be placed CLOSEST as possible to the Pic supply pins ... ( lower and upper right )

Also, you should try to place the Xtal and its caps on the other side of the pic, close to the pic pins ... ( nice 8 Mhz emitter !!! ... lower right )

there's lots of noise to catch here ...

a big reservoir cap ( L.S.R. ) is also a must to place close to the multiplexed LED displays:
they generate lots of noise ...

Alain

[Art]

Bah! bit of a nightmare project.

I think the noise stopped when I was using a capacitor is because I was holding
it with my fingers.

I have tried a transistor OR gate to ground the input pin of the amplifier.
The amplifier is connected to an FM radio circuit, and a pic pin that produces sound.
The OR gate can receive a busy signal from either the pic or radio circuits
to make it stop shorting the amplifier input to ground.

[Acetronics2]

Thanks

The 0.1 uF caps are there right next to every pic. They are blue.

Hi,

Blue or not ...

They still are TOO FAR from the pic supply pins ...

Alain

[Art]

Hi,

Blue or not ...

They still are TOO FAR from the pic supply pins ...

Alain

You think I'd hear less noise from the amp if I moved the 0.1uF caps closer to each chip?

I had another idea. What if next time I used a copper clad board of the same
size underneath the proto PCB, so I could drill one layer down to a ground plane at any location?