Audio Power Amplifier's Delay in Start-Up Problem

[djnik1362]

hi everyone
I have a problem with my audio power amplifier circuit :

As you see in this schematic there is 470u capacitor and 100k resistor in output stage.
When i power this circuit for the first time the capacitor is fully discharged and the circuit works fine.
But if i reset the power it takes about 2 minutes to work again and i think it's all about output stage.
It is possible that input signal is not enough to reject charged DC in capacitor.

Load speakers are 3W and 4 ohm.

Can you help me to fix this problem ?

Thanks.

 

[KlausST]

Hi,

Are the speakerscontinously connected?
If yes, then the capacitors are charged very fast with the speaker resistance.

I'd check the bypass and shutdown connections.

A more detailed description and a complete schematic where we see the input and the shutdown connection can help.

What exactely happens during the 2 minutes? Noise? Current, voltage...

Klaus

 

[djnik1362]

Hi,

Are the speakerscontinously connected?
If yes, then the capacitors are charged very fast with the speaker resistance.

I'd check the bypass and shutdown connections.

A more detailed description and a complete schematic where we see the input and the shutdown connection can help.

What exactely happens during the 2 minutes? Noise? Current, voltage...

Klaus

Thanks for responding .

*Speakers are always connected .
*Inputs to this amplifier is coming from an analog switch which change audio path to this amplifier or external amplifier .
*During this transition speakers are completely silent and there is no noise at all.

I put output voltage on scope and see when i reset power this capacitors charged to about -10V
it eventually reach to zero and then after a loud click my speakers responding.

I change 100k resistors to 4.7k and my problem solved mostly but i don't know is this a good idea or not .
I also think about two MOSFTs in parallel with 100k resistors so in power up i discharge capacitor through these
MOSFET to ground .

I don't know what to do .

Thanks for your support.

 

[FvM]

Sounds confusing. If speakers are permanently connected, they would be parallel to 100k resistors. Also the capacitors would quickly discharge when the supply is turned off. Some information ís apparently missing.

 

[Audioguru]

I suspect that this amplifier is driving tweeters that have series highpass filter capacitors.

 

[djnik1362]

I suspect that this amplifier is driving tweeters that have series highpass filter capacitors.

Can you explain more ?

Speaker P/N is : AS07104PO-WR-R from PUI Audio Inc.

I put 100k resistors because i saw it in a reference design but actually i don't know
what is it's function . I guess maybe it functions as a high pass filter or capacitor dis-charger.

 

[Audioguru]

Your tiny speakers might sound like tweeters but they have a voice coil that is about 3 ohms so the 470uF output capacitors should charge or discharge in about 0.07 seconds.
Do the tiny speakers have other capacitors in series?

 

[djnik1362]

Your tiny speakers might sound like tweeters but they have a voice coil that is about 3 ohms so the 470uF output capacitors should charge or discharge in about 0.07 seconds.
Do the tiny speakers have other capacitors in series?

No capacitor , only cable .
Is it possible capacitor polarity make this happens ?

 

[KlausST]

Hi,

Definitely:
With the given circuit and the directely connected speakers... the 100k resistors will have no notable influence.

There must be an issue elsewhere.

*****
Capacitor polarity.
You need to connect the capacitors with the shown polarity. Otherwise you risk to kill the speakers and/or the capacitors.

Klaus

 

[chuckey]

Are you sure that the electronic volume control and mute are working correctly? The simple loudspeaker arrangement is unlikely to give you this sort of problem.
Frank

 

[djnik1362]

Hi,

Definitely:
With the given circuit and the directely connected speakers... the 100k resistors will have no notable influence.

There must be an issue elsewhere.

*****
Capacitor polarity.
You need to connect the capacitors with the shown polarity. Otherwise you risk to kill the speakers and/or the capacitors.

Klaus

I change the 100k to 4.7k and the problem almost gone . How is this ?

- - - Updated - - -

Are you sure that the electronic volume control and mute are working correctly? The simple loudspeaker arrangement is unlikely to give you this sort of problem.
Frank

Volume control come from a microcontroller DAC with a buffer .
Mute is come from that microcontroller too and is a CMOS level .

 

[Easy peasy]

Are you connecting the speakers from output to ground?

 

[KlausST]

Hi,

4 Ohm parallel to 100k gives 3.99984 Ohm. You see the nuance?

I expect abad solder joint, or something similar....

Klaus

 

[Easy peasy]

KlausST has a good point, a close visual inspection of the build so far has much to reccommend it..!

 

[djnik1362]

Are you connecting the speakers from output to ground?

As depicted in schematic there is connector that has two ground pins and a signal pin.
I've connected one of the grounds and the signal pin to the speaker with a 28 AWG twisted pair cable .

- - - Updated - - -

Hi,

4 Ohm parallel to 100k gives 3.99984 Ohm. You see the nuance?

I expect abad solder joint, or something similar....

Klaus

There are two speakers (Stereo Configuration) and both have the same problem .
Solder Joints are fine because after transient moment speakers works fine.

 

[Easy peasy]

perhaps you should connect a diode from pins 4 & 7 to the Vcc (two diodes, kathode to Vcc) this will return the charge on the caps to the Vcc at power down. I note the IC is supposed to take care of clicks and pops in the speakers at power up, due to the 470uF caps, but perhaps it is not so good for short power interruptions as you are suggesting, certainly putting 4k7 in parallel with the speakers should have no noticeable effect, perhaps you are putting them on the other side?

 

[KlausST]

Hi,

Datasheet says:

In order eliminate "clicks and pops", all capacitors must be discharged before turn-on. Rapidly switching VDD may
not allow the capacitors to fully discharge, which may cause "clicks and pops".

EasyPeasy's recommendation with the diodes is good.
I'd add a third diode from pin 9 to pin 6, cathode to 6.
1N4148 are O.K.

In normal operation the diodes do nothing. But when power is switched OFF then the diodes help to discharge the capacitors.
Ensure that VCC goes much lower than 1/4 of the typical voltage.

Klaus

 

[Easy peasy]

1N4007 diodes may last a bit longer 4148's are only 150mA ave, 500mA rep peak, (we use 4448's exclusively instead for a fast-ish but some what gruntier signal diode)

 

[Audioguru]

Millions of audio amplifiers use a coupling capacitor to feed the speaker because the power supply has only a single polarity. They do not discharge the capacitor when the power supply is turned off because the circuit uses the output transistors to keep the input to the capacitor always at half the supply voltage. Then when the supply voltage drops when the power supply is turned off the capacitor is discharged by the amplifier circuit.

If a diode is needed then surely the datasheet would say so.

EDIT:
Pins numbers were mentioned instead of pins functions. The BYPASS pin is pin 9, not the output pin that I was thinking about.
The 4.7uF capacitor on pin 9 might have a horrible tolerance and be actually be 20uF? Then the "wakeup time" would be about 2 seconds.
But the capacitor value is not changed by the power supply type as happens with the problem.

- - - Updated - - -

I put output voltage on scope and see when i reset power this capacitors charged to about -10V
it eventually reach to zero and then after a loud click my speakers responding.

I change 100k resistors to 4.7k and my problem solved mostly.

How can the output capacitor charge to -10V when there is no negative power supply?
Your schematic shows +12V and ground. Do you also have a -12V supply?

 

[Easy peasy]

just because the data sheet does not mention (or even know about) a remedy, does not mean that remedy should not be implemented.