TDA car-audio chip amp: purpose of output caps

[13hm13]

Below is an typical schema from ones of the countless TDA car audio chip chip amps:



(The one I'm particularly concerned with at the moment is this TDA 7372 http://www.classiccmp.org/rtellason/...ta/tda7372.pdf )

A mild debate has arisen elsewhere as to what function(s) the OUTPUT CAPS serve (2200uF, C9-12 in schema above).

Are they to BLOCK DC and/or (as contended by another) "all push pull stages that operate only on a single rail power supply utilize a SERIES capacitor in series with the speaker, which acts a a voltage source during the negative half cycle." ?

Thx!

[Richard Crowley]

Yes, since the amplifiers are powered from a single-ended supply (+12V only), the output sits at 6V during "silence".  No speaker likes having a constant DC current which interferes with its ability to reproduce the waveform.  So, the most obvious function is that those are DC-blocking (aka AC-coupling) capacitors.

And well, if you hold your mouth just right and squint your eyes, you could say the the capacitor forms a voltage OFFSET (not exactly a "source") during both the positive and negative waveform excursions.  It shifts everything down by 6V.

Note: When I use the term "12V" I actually mean whatever voltage is being supplied to the chip.  As your diagram shows it could be something less than 12V all the way up to 16V or so.  And when I say "6V" I actually mean half whatever the supply voltage is.  "Nominally" 12V and 6V.

[13hm13]

I have seen several topologies in which a chip-amp using a SINGLE (+) supply drives, output transducer (headphones, speakers, etc) w/o ANY caps. In fact in audiophile wankdom, output caps are considered "bad". (e.g., better to use a servo, etc).

Below is another TDA chip amp (this time for portable/headphone apps) ... guess what ... no electro output caps ... just those two L/R 100uF caps for emergency DC protection.


It's used in a modern "iPod" device:

[ludzinc]

I have see several topologies in which a chip-amp using a SINGLE (+) supply drives output transducer (headphones, speakers, etc) w/o ANY caps. In fact in audiophile wankdom, output caps are considered "bad". (e.g., better to use a servo, etc).

Below is another TDA chip amp (this time for portable/headphone apps) ... guess what ... no electro output cap ... just that 100uF for emergency DC protection.

Um, no, there's two 100uF decoupling caps, one for left, one for right channel...

[Richard Head]

Below is another TDA chip amp (this time for portable/headphone apps) ... guess what ... no electro output cap ... just that 100uF for emergency DC protection.

Am I missing something? I see two 100uF caps in each output to the headphones, again for DC blocking.
A bridged output stage is "better" in that it eliminates the electrolytics (or tantalums) but is a bit more complex.
Another problem with capacitor coupled output audio amplifiers is the annoying click when the power is first applied. This is due to the capacitor charging to half the rail.
If a bridged output is used then you avoid that issue.

[Richard Crowley]

... guess what ... no electro output caps ... just those two L/R 100uF caps for emergency DC protection.

What does "emergency DC protection" mean? 
Those are conventional DC-blocking/AC-coupling output capacitors as surely as you are reading this.

[German_EE]

A number of years ago I had a stereo amplifier that had been constructed using potted modules from a company called ILP. Something in the power amplifier module failed, probably one of the output transistors, and it applied 12v DC to a very nice pair of headphones. The first sign that something was wrong was a loud click and smoke coming from one transducer.

In this case an 'emergency DC' capacitor' might have been a good idea.

[dannyf]

w/o ANY caps

There are BTL amps (bridged OTL amps) that in theory don't need the dc-blocking capacitor on the output. They are / were widely used in car amps.

[13hm13]

A number of years ago I had a stereo amplifier that had been constructed using potted modules from a company called ILP. Something in the power amplifier module failed, probably one of the output transistors, and it applied 12v DC to a very nice pair of headphones. The first sign that something was wrong was a loud click and smoke coming from one transducer.

In this case an 'emergency DC' capacitor' might have been a good idea.

Remember: the amp outputs AC-type (sinusoidal) signal -- music or audio -- and DC protection is built into the TDA IC (see datasheet). The in-series caps are there for added protection (there are also DC coupling caps often on the INPUT of amps). Certain "high-end" designs replace caps with active servo circuits.
The main point is -- and going back the OP -- that those in-series caps do not provide biasing.

[SeanB]

The only integrated stereo amplifiers that do not need a coupling capacitor are those which provide 3 output stages, using one as the way to connect the common of the headphones to the appropriate power rail. They float all three connections at half the DC supply rail, but have limited power capability.

The other method uses one stereo amplifier chip per speaker, driving in a bridge configuration. This requires a separate connection to each side of the speaker, and you cannot have a common connection for any speakers.

All of them have a problem where if the IC fails ( and yes, even with SOA, overtemp, overcurrent and overwhatever protection they can and do fail, often quite explosively) they often first fail and deliver the one supply rail to the output direct, and as typically only one side is monitored for current limiting ( often only the high side) and typically the high side will fail shorted first, then a very high DC current flows through the speaker until something either cooks itself into a short and blows the fuses or the power supply, or the speaker burns out.

[mikerj]

I have seen several topologies in which a chip-amp using a SINGLE (+) supply drives, output transducer (headphones, speakers, etc) w/o ANY caps. In fact in audiophile wankdom, output caps are considered "bad". (e.g., better to use a servo, etc).

If you have balanced supply rails, i.e. a +ve, -ve and 0v rail then you can dispense with the output capacitor.  Obviously this isn't easily available in an automotive environment. 

The other way to do this is with a BTL (Bridge Tied Load) with two push-pull outputs which are nominally biased at half the supply rail.  This gives you up to twice the voltage swing and potentially up to 4 times the output power of a single ended output from the same supply rails.  Most car head units and portable audio electronics (MP3 players, phones, tablets) now tend to use this kind of output stage since it makes best use of limited supply rail voltages.

Audiophile wankdom is just that however;  the same people explode with joy when they listen to a tube amp that uses an output transformer.

[NiHaoMike]

It's possible to use a charge pump built into the chip to derive the negative rail.
http://www.maximintegrated.com/en/products/analog/audio/headphone-amplifier-ics/directdrive-headphone-amplifier-ics.html

The main strike against capacitive coupling is when OEMs skimp on the caps (or intentionally undersize them for saving space), causing the frequency response to fall off too soon. The solution I like is to add series resistance. It flattens the frequency response at the expense of volume, but as most headphone outputs are able to go way louder than necessary, that's usually not a problem. It can be done without modifying the device and also boosts SNR.
http://home.comcast.net/~machrone/playertest/playertest.htm

[N2IXK]

A number of years ago I had a stereo amplifier that had been constructed using potted modules from a company called ILP. Something in the power amplifier module failed, probably one of the output transistors, and it applied 12v DC to a very nice pair of headphones. The first sign that something was wrong was a loud click and smoke coming from one transducer.

In this case an 'emergency DC' capacitor' might have been a good idea.

Most decent direct coupled audio power amplifiers have a "speaker protection" circuit built in. It monitors the amplifier outputs for DC offset, and drives a relay that only pulls in when the DC output is zero. The speaker terminals are connected to the amplifier outputs through contacts on the relay. If an output transistor shorts, one supply rail fails, or any other fault occurs, the DC output level at the output shifts, and the speakers are disconnected from the potentially damaging DC signal.

Amplifiers like this usually incorporate a time delay feature, so that the speakers are initially disconnected from the amp at power up, and the relay pulls in a couple seconds after the power is applied. This eliminates the loud "thump" caused by transient voltages when the power is first applied.

[NiHaoMike]

A more modern way of doing that (in speaker amplifiers) is to have the main and aux rails be operated independently. Then the logic and drivers power on first, after which the main rails are ramped up. Many modern designs even vary the main rails in response to the volume setting as that boosts efficiency and resolution at low volume settings. Fault protection is then just a matter of switching off the drive signals and main voltage rails.

[13hm13]

If you have balanced supply rails, i.e. a +ve, -ve and 0v rail then you can dispense with the output capacitor.  Obviously this isn't easily available in an automotive environment. 

But THOSE caps have NOTHING to do with power supply rails. In an emergency, THOSE caps will block a sustained DC current from damaging the speaker/headphone/transducer.
Further, THOSE in-series output caps -- because they are IN the direct path of the audio signal -- are often NONpolar electrolytics, esp. in high-quality audio amps. BlackGate brand caps, no longer made, are  popular DC blocking non-polar electros...

As far as " [pos/neg dual voltage] Obviously this isn't easily available in an automotive environment.  "
No! Cheap n' easy for any battery powered device. You can use a rail-splitter like Texas Instruments’ TLE2426 and virtual ground. More info:
http://tangentsoft.net/elec/vgrounds.html

[Richard Crowley]

In an emergency, THOSE caps will block a sustained DC current from damaging the speaker/headphone/transducer.

What does "emergency" mean to you?  That circuit has SUSTAINED VCC/2 voltage whenever it is powered up. Even with no signal.
That is called "nominal operating conditions as designed".  Dunno what "emergency" means here?

Further, THOSE in-series output caps -- because they are IN the direct path of the audio signal -- are often NONpolar electrolytics, esp. in high-quality audio amps.

Typically not. In unipolar situations (as in a vehicle amp), there is no negative voltage, so there is no requirement for (more expensive) non-polar caps. The output from that amplifier is ALWAYS POSITIVE. So a simple polarized cap can be used.  There is no advantage to using a non-polar cap where an ordinary polarized cap would be sufficient.

[Zero999]

Are they to BLOCK DC and/or (as contended by another) "all push pull stages that operate only on a single rail power supply utilize a SERIES capacitor in series with the speaker, which acts a a voltage source during the negative half cycle." ?

I'd say both are correct: the capacitor both blocks DC and provides a negative voltage source during the negative half cycle.

[13hm13]

For the in-series caps in question ...
"...provides a negative voltage source during the negative half cycle." is incorrect!
Just take apart your avg. portable (= batt. pwr'd) MP3 player, like "iPod". Mine uses an AD8397 opamp with SINGLE supply voltage (around +5V), and the output pins of the opamp are DIRECT-COUPLED to output (e.g., voice coil of headphone).
Remember (for the umpteenth time)... (a)  the AC (music/signal) voltage swing can happen (swing) entirely above the X-axis (domains with + 'y'); and (b) the elastic (compliance) nature of dynamic driver pulls it back ("negatively"), so no reverse bias needed.

 Electrostatic speakers/headphones, as noted, are another story -- they have their own NEGATIVE/POSITIVE power supply.

Please blow the dust off your high-school physics textbooks , brush up on fundamentals ... and then bring intelligent/informed discussion back to this thread ... otherwise, you only embarrass yourselves and the EEV community! SIGH!
LOL

[Marco]

The voltage swing doesn't have to be negative (relative to the other connection of the speaker) to pull the cone back, but so there is no DC bias. A DC bias voltage reduces the dynamic range and burns power in the coil for no sound output.

As for how you can do that without a DC blocking capacitor, multiple ways have already been mentioned. If you measured the supply voltages on the AD8397 correctly then presumably your "ipod" has a virtual ground at 1/2 Vcc for the headphone jack's shared ground.

[dannyf]

otherwise, you only embarrass yourselves...

Sounds like you could use some humility.

[13hm13]

As for how you can do that without a DC blocking capacitor, multiple ways have already been mentioned. If you measured the supply voltages on the AD8397 correctly then presumably your "ipod" has a virtual ground at 1/2 Vcc for the headphone jack's shared ground.

There is no virtual ground. Jesus H. Christ!
Folks ... this is the last time I'm gonna repeat this  ...and shame on you degree'd EE  college-boyz that a ghetto sh*t plumber (my real-life entrepreneurship biz is plungin' toilets and snakin' drains ... not shit*tin' ya ') - - and one who only dabbles in electronics as a "hobby"  -- has to refresh y'all EExperts on basic electronics.
The amp is outputtin' an "AC" signal to the transducer's coil lead (this can be "-" or "+" dependin' on which way you want the transducer phased). The other lead (let's say "-") is ground or sink.
Electronics for Dummies: from the get go (= by design) there should be no DC present on the amp output unless somethin' is eff'd up.
For those who still don't get it ... get educated for Chrissake...

[David_AVD]

So 13hm13, you're saying that an audio amplifier (using a single rail with the output @ half rail) that is capacitively coupled to the load doesn't have the load swinging above and below the 0V line?

[Richard Crowley]

Semantics.   Subjective perception.   Violent agreement.   

[BradC]

...and shame on you degree'd EE  college-boyz that a ghetto sh*t plumber (my real-life entrepreneurship biz is plungin' toilets and snakin' drains ... not shit*tin' ya ') - - and one who only dabbles in electronics as a "hobby"  -- has to refresh y'all EExperts on basic electronics.

I'd stick to plumbing if I were you. (And I'd have a closer look at the output stage in your iPod). Despite what you might want to call it, a capacitor in series between the amp and the transducer is there to block DC. It's there to block DC _all_ the time, not just in what you call "an emergency". A single ended amp in a vehicle does put out AC, but it alternates between 0 & 12 volts and therefore the average is 6V, so unless you have the other end of the transducer sitting at 6V you _NEED_ a capacitor in there to stop the transducer smoking.

Bah, I don't know why I bother, you'll only tell me I'm wrong too..

[Richard Head]

When the lower output transistor switches to ground the DC potential across the series output capacitor provides the source voltage to drive the speaker in the -ve direction.
Now who's the dummy?