Discussion: does class-d needs speaker protection?

[shanghai2004]

Hi,

I hope to start a discussion if single chip class-D amps needs DC speaker protection with relay's when designing products for paying customers.

Almost all class-A and class-AB amps are being designed with DC speaker protection relays, but when I look at class-D finished commercial product, many have no relays.

This is probably only for cost reasons. Many are bridge designs, DC speaker relay protection needs more components and space for bridged amps.

Some IC vendors claim their internal protection also protects the speakers, but this seems to me unlikely to be full proof.

As recent single chip amps become high quality and powerful it's now more likely expensive speakers will be used. Would be bad if your amp fries your customer's $10k speakers...

On the other side, adding 2 relays & related electronics almost doubles the PCB area in many designs.

Any thoughts? Any experience with class-D single chip amps failing and putting DC out?

 

[us1070]

Whether it fail or not speaker protection is a must for your important speaker. And yes class d can also fail and can give high offset
DC can damage your speaker withing a seconds

I use my Apex AX14 without speaker protection that only because i donot care about my cheap hone driver (18$)

 

[gmarsh]

A blown output transistor gives the exact same effect, regardless of amplifier class.

 

[JMFahey]

As Enzo said, "the amplifier is that thingie between the power supply and the speaker"
Output devices control how much and in what way that supply voltage reaches the speaker, modulated by input audio and hopefully a faithful enlarged copy of what drives it.

Notice I didn't mention "class" at all, since that's not the main point.
Now if one output device shorts, it becomes an uncontrolled piece of wire, by definition you can't turn it OFF.
That full supply voltage usually makes interesting things inside your speaker, such as:

 

[nigelwright7557]

I have started adding a speaker protection to all my designs.

I use a PIC micro that monitors amp output.
If the PIC sees DC for more than 500mS it shuts off the relay.
I don't even need a2d on the pic as PIC inputs switch at 1.5 volts so this can be used as a comparator.

 

[00940]

Hypex seems to think that protection is a good thing; you can find protection circuitry on all their power supplies made to be used with their amplifiers.

 

[Globulator]

A blown output transistor gives the exact same effect, regardless of amplifier class.

I disagree, with part of your statement, and perhaps all. .

Because class D chips can have protection circuity they can monitor the outputs and detect fault conditions. Once a short (or DC input bias) event is detected the protection has a couple of options:

1) It can drive the other MOSFET, shorting out the whole bridge. Possibly a last ditch way of protecting stuff.

2) Or in the bridged design (Most are BTL!) it can simply switch off both MOSFETs on the other bridge, reducing current to zero until the fault is cleared. If the fault persists then it can resort to the crowbar (1).

I'm sure they can fail badly, but I'm also sure that effective protection can be built into a BTL chip class D amp too.

Certainly with a fault I'd prefer a TPA3116 to be in charge than a dumb classA/AB/B amp .
Additionally current sensors in the chips reduce the risk of the device blowing in the first place.

 

[ChocoHolic]

I am really scratching my head about this thread.
To me the answer is clear. No!
No matter which type of amp, the amp does not need a speaker protection.
But may be the speaker needs... ? Also, no!
Only humans who love the speaker or are economically responsible for the speaker need a speaker protection.

1) It can drive the other MOSFET, shorting out the whole bridge. Possibly a last ditch way of protecting stuff.

Valid point. But is this really possible only for classD, or just particularly simple in classD?

2) Or in the bridged design (Most are BTL!) it can simply switch off both MOSFETs on the other bridge, reducing current to zero until the fault is cleared.

Again. Valid point. But is this.... ?

 

[Globulator]

Valid point. But is this really possible only for classD, or just particularly simple in classD?

Again. Valid point. But is this.... ?

It's because class A-B is a simple, dumb analog circuit that doesn't 'know' if an output device is over current or has died, whereas class D is a complex chip with quite a bit of control and logic circuitry, some of which is specifically dedicated to protection.

To quote the TI datasheet for the 3116:

The TPA31xxD2 devices are fully protected against faults with short-circuit protection and thermal protection as well as over-voltage, under-voltage and DC protection. Faults are reported back to the processor to prevent devices from being damaged during overload conditions.

 

[ChocoHolic]

It's because class A-B is a simple, dumb analog circuit that doesn't 'know' if an output device is over current or has died, whereas class D is a complex chip with quite a bit of control and logic circuitry, some of which is specifically dedicated to protection.

No. Such chips are just a specific implementation. Everybody is free to do a sophisticated implementation of protections in class A-B as well.
In the end it's all about money, size & weight.
In class A-B the additional size of the protection was just a minor concern.
Usually no large efforts were spend and no risks were taken in order to get rid of it.
For obvious reasons ClassD immediately motivates people
Size zero, be sexy, be rich. Get respect from the beautifulpeople.
to get rid of the old fashioned large protections.

 

[Globulator]

Such chips are just a specific implementation. Everybody is free to do a sophisticated implementation of protections in class A-B as well.

Have you ever seen a Class A-B implementation without a relay? Far from being sophisticated they are usually a simple DC sense with turn on delay. Or a fuse.

Class D you can actually turn the MOSFETs off using digital logic gates, a far better solution. Did you read the TI section I quoted above? It seems TI spent a great deal of thought and put it into protection. I doubt you could blow a speaker with a TPA3116 however hard you tried.

Class D does therefore plainly offer scope for far better protection than Class A-B.

 

[ChocoHolic]

Are you avoiding on purpose to get my point?
There is no doubt that many implementations (not only TI) of classD do offer
smaller and advanced protections compared to most AB-implementations.
My point is that you can implement ClassD and Class AB as well completely without or with very sophisticated protections. ...and that there is much more motivation in classD for doing it smart...

 

[ChocoHolic]

Have you ever seen a Class A-B implementation without a relay?

Yes. Many. One example:
http://www2.st.com/content/ccc/reso...df/jcr:content/translations/en.DM00102150.pdf

 

[Globulator]

Ah a chip-amp, yes, good call, I was thinking of only discrete amps.

I'm not 100% sure what your point was, but if it's that Class A-B chip-amps can have just as good protection as Class D chip amps then I agree.

I was however referencing my answers back to the original post of the thread, re class D and speaker protection, which is why it may have seems as if I didn't get your point I guess.

Out of interest, has anyone known a class-D fry a speaker?

 

[plasnu]

Almost all class-A and class-AB amps are being designed with DC speaker protection relays,

Really? I can't find any relay inside the Pass Labs or First Watt amps. I think Goldmund doesn't have relay either....

 

[gmarsh]

Class D you can actually turn the MOSFETs off using digital logic gates, a far better solution. Did you read the TI section I quoted above? It seems TI spent a great deal of thought and put it into protection. I doubt you could blow a speaker with a TPA3116 however hard you tried.

Class D does therefore plainly offer scope for far better protection than Class A-B.

You can't turn off a blown MOSFET.

Seriously though, there's very little difference between a class AB and a class D amplifier. Both amplifiers have output devices in a half bridge pulling one end of the speaker to either one rail or the other. In a class AB amp, they operate in linear mode and command the output voltage directly. In class D, the devices switch the output between the two rails using PWM or some other modulation scheme, and a LC reconstruction filter restores the audio.

In either topology, if one of the output devices fails short, it's going to pull the output to that rail. It's possible that the other device will pull the rail in the other direction and short out the supply.

Also, not every class D amplifier is a TPA3116, there's a hundred different ways to build them with different protection capabilities. Just like you can build a class AB amplifier a hundred different ways.

 

[gmarsh]

Also, I've blown a speaker with a failed TPA3118 amp.

 

[JMFahey]

You can't turn off a blown MOSFET.

That's the point

Seriously though, there's very little difference between a class AB and a class D amplifier. Both amplifiers have output devices in a half bridge pulling one end of the speaker to either one rail or the other. In a class AB amp, they operate in linear mode and command the output voltage directly. In class D, the devices switch the output between the two rails using PWM or some other modulation scheme, and a LC reconstruction filter restores the audio.

In either topology, if one of the output devices fails short, it's going to pull the output to that rail. It's possible that the other device will pull the rail in the other direction and short out the supply.

Again, that's the point

there's a hundred different ways to build them with different protection capabilities. Just like you can build a class AB amplifier a hundred different ways.

AGAIN that's the point.

 

[shanghai2004]

Really? I can't find any relay inside the Pass Labs or First Watt amps. I think Goldmund doesn't have relay either....

That's interesting... Do they use another way to protect the speakers in case of failure?

Or they assume if you can afford their amp, you also can afford a new set of speakers after a failure

 

[eduard]

Hello,
The risk of destroying my beloved and expensive Altecs does keep me from using solid state amp which dont have dc protection. Maybe the single ended Pass design with an output transformer. A very decent output transformer is still cheaper than my Altecs.
Greetings, Eduard