Bridging power amps

[emerth]

Noob question probably...

I understand how to bridge a pair of power amps: non-inverted input on one amp, inverted input on the other, and connect the speaker between the hot outputs of the two amps. Also that the power out scales by 4 not 2, thus I need to know the amps can handle that.

But I am not sure if I need to worry about the two amps' speaker ground outputs.

Should I connect the ground speaker outputs of the two amps or leave them unconnected? I *think* it does not matter but I am worried about whether ground would 'float' between the two amps, thus introducing a DC offset.

Thanks for any help!

 

[pacificblue]

Which speaker ground outputs?

 

[emerth]

Which speaker ground outputs?

A speaker is connected to an amp between the output stage (hot output) and ground, which I have called "ground output". So if am bridging, do I worry about the ground "output" or speaker connections?

 

[sregor]

Speaker grounds on the amplifier side should go back to power supply ground and should already be connected at that point. The speakers themselves are not connected to any grounds, but should be connected between the two hot signals of the amplifers being bridged. Best if nothing is connected to the grounds externally.

 

[emerth]

Speaker grounds on the amplifier side should go back to power supply ground and should already be connected at that point. The speakers themselves are not connected to any grounds, but should be connected between the two hot signals of the amplifers being bridged. Best if nothing is connected to the grounds externally.

OK, makes sense. Thanks for explaining it!

 

[SUZTEK]

Bridging

I have a bridging circuit built in all my amplifiers with a switching facility for normal or bridge.The advantage is that you double the output power.This increases the bass and overrall response for a well designed amp.The result is well worth it.The speaker connections are between the hot (+) speaker terminals.Leave the (-) unconnected as these are connected to the star point on the power supply between the main smoothing capacitors.The input ground from the pc board should be connected to the same star ground of the power supply.Any other grounding should also be routed to the star point at the power supply.
wILLIAM

 

[AndrewT]

I understand how to bridge a pair of power amps........................that the power out scales by 4 not 2,

No !!! Absolutely not.

The rule for a bridged pair of amplifiers is that the combination can deliver twice the power into twice the load impedance that the single amplifier can.

If you have a 100W+100W into 4r0+4r0 two channel amplifier then it can deliver 200W into 8r0.
It cannot deliver 400W into 4r0 and worse, it may not survive.

Do not believe the fools that tell you otherwise.
As a final comment, note that the total power delivered by that two channel amp is unchanged. as two channels it delivered 100W+100W = 200W to the loads.
As a bridged mono amplifier it is still able to deliver 200W to the attached load.

 

[AndrewT]

... switching facility for normal or bridge.The advantage is that you double the output power.

No !!!
the maximum total power that the amplifier is capable of delivering is exactly the same.
No magic, no extra power.

 

[emerth]

v^2 / R

Hmmm, bridging doubles the voltage applied to the speaker other things being equal. A first approx says power dissipated is (V^2) / R, neglecting reactance. So if (bridged voltage) = 2*unbridged, the V^2 term causes the power to increase from bridging by a factor of 4, not 2. What is the flaw in this?

 

[SY]

What is the flaw in this?

No flaw at all, as long as the load is the same AND the amplifier can supply the additional current. The real-world flaw is the last assumption.

 

[emerth]

No flaw at all, as long as the load is the same AND the amplifier can supply the additional current. The real-world flaw is the last assumption.

Yes, that's certainly true. But in my own defense, AndrewT dropped a clause in quoting my OP: "... Also that the power out scales by 4 not 2, thus I need to know the amps can handle that."

Thanks all for helping with my ground question.

 

[Nrik]

But I am not sure if I need to worry about the two amps' speaker ground outputs.

Should I connect the ground speaker outputs of the two amps or leave them unconnected? I *think* it does not matter but I am worried about whether ground would 'float' between the two amps, thus introducing a DC offset.

I am more worried that the input sockets will carry the ground current. If it is two separated cabinets (and therefore without common gound level) I would connect the two speaker ground terminals with a piece of speakercable.

 

[pacificblue]

There will be no significant ground currents. That is one advantage in a bridged amplifier. Ground is only a reference and the ground currents are all line level. The speaker current flows in through one amp and out through the other.

The grounds of both channels are already connected through the source. If you make another connection you may create a ground loop.

 

[SUZTEK]

Bridge mode:If a 4 ohms load is used, one has to make sure that your amp can safely drive a 2 ohms load as the amps in bridge mode only sees half the load impedence. Also make sure that the power supply can deliver the extra current required.
In a non bridge configuration assuming that the power supply is +/-20v and the load is 8 ohms, then we have P= 20X20/8= 50 Watts. In bridge mode P=40X40/8=200 Watts.
I have used my amps in bridge configuration very successfully and can vouch for the benefit of extra output power.
William

 

[AndrewT]

Bridge mode:If......
In a non bridge configuration assuming that the power supply is +/-20v and the load is 8 ohms, then we have P= 20X20/8= 50 Watts. In bridge mode P=40X40/8=200 Watts.

Balderdash !!!!
A bridged pair of amplifiers delivers double the power into double the load impedance. Never 4times into the same load impedance.

If you have an amplifier specified/measured as follows
100W into 8r0
175W into 4r0
290W into 2r0
and
Capable of driving a 4ohm or higher speaker load.

Then a bridged pair of these amps have the following spec.
200W into 16r0
350W into 8r0
580W into 4r0
and Capable of driving an 8ohm or higher speaker load.

Note:
that the 350W into 8r0 after bridging is more than 100W+100W into 8r0 but only 75% more, not 300% more.
and
that 175W+175W into a pair of 4r0 is more than the zero watts that a bridged amplifier that is incapable of driving a 4ohm speaker load.

As a final comment, note that the total power delivered by that two channel amp is unchanged.

my earlier comment still holds true.
The 350W into 8r0 exactly equals the total power output of the two amplifiers producing 175W+175W into a pair of 4r0 loads.

 

[asbjbo]

So, if we bridge a pair of LM3886 rated at 68 W into 4 ohms and 38 W into 8 ohms with +/- 28 V power supplies, we get 2x68=136 W into 8 ohms and 2x38=76 W into 16 ohms at the same distortion level, right?

That is a 136/38 = 3.6 x power increase into 8 ohms. Not quite 4x, but not too bad if the load happens to be a single 8 ohm driver. 28 V also happens to be the maximum recommended supply voltage for bridged LM3886s into an 8 ohm load, but you will probably run into heat management issues if you try to run it continuously at that power level. Don't try connecting the bridged amp to a 4 ohm driver, though.

Regarding the opening question: Don't worry about the negative loudspeaker terminals. Just leave them unconnected, since they both go to power supply ground anyway. You can read more in the AN-1192 application note.

 

[AndrewT]

So, if we bridge a pair of LM3886 rated at 68 W into 4 ohms and 38 W into 8 ohms with +/- 28 V power supplies,

yes, in theory you are right. But only for driving resistive loads.
You are also right to warn of the heat management issue. This combined with the reactive speaker loading means you cannot expect a bridged pair of 68W+68W into 4r+4r to effectively drive an 8ohm reactive speaker when hot.

68W into 4r0 requires outputs of 23.3Vpk and 5.83Apk, just to drive a resistor.
A speaker can demand many times this peak current on fast transients.
Allow ~ 3times, i.e. ~15Apk into a 4ohm speaker.
This is way beyond the capability of a 3886. A hot 3886 will limit at even lower currents. +-28Vdc for a bridged pair of 3886 driving an 8ohm speaker is not recommended for passing a good quality audio signal. You are condemning the pair of chipamps to failure at least in the sound quality stakes.

 

[SUZTEK]

Andrew, can you explain your deduction regarding the output power in a normal 100+100w amp driving a load of 8R and the bridging of the same amp which results in an output of 200 w into (16R). I am under the impression that in bridge mode the amplifier sees HALF the load impedence, certainly not doubled.
In a normal and bridge configuration,the voltage avaiable to the normal speaker is half the voltage avaiable at the bridged speaker.The normal configuration uses the signal output referenced to ground,whilst the bridged confi uses the two signals outputs with no ground reference.Therefore the instantaneous voltage to the bridged speaker is doubled.So ,the power at the speaker is 4 times..If the amp power supply is 40 v with ground as a reference (+20/-20v), therefore, in the normal mode the max voltage that can be applied is 20 v and in bridged mode the max voltage will be 40 v.
This is how i understand the difference.
Regards
William

 

[AndrewT]

Andrew, can you explain your deduction regarding the output power in a normal 100+100w amp driving a load of 8R and the bridging of the same amp which results in an output of 200 w into (16R). I am under the impression that in bridge mode the amplifier sees HALF the load impedance, certainly not doubled.
In a normal and bridge configuration,the voltage available to the normal speaker is half the voltage available at the bridged speaker.The normal configuration uses the signal output referenced to ground,whilst the bridged configuration uses the two signals outputs with no ground reference.Therefore the instantaneous voltage to the bridged speaker is doubled.So ,the power at the speaker is 4 times..If the amp power supply is 40 v with ground as a reference (+20/-20v), therefore, in the normal mode the max voltage that can be applied is 20 v and in bridged mode the max voltage will be 40 v.

so much in this statement is wrong.

A bridged pair of amplifiers delivers double the power into double the load impedance.

Forget all the other "claims" you have heard or seen. Some are correct but misunderstood, some are wrong and mislead the listener.
Just remember that rule I quoted above.

Now from following that rule you find that a 4ohm capable two channel amplifier that is 100W+100W into 8+8r0 and 175W+175W into 4+4r0 (and can safely drive a resistor of much lower resistance, possibly as low as 1r0 to 2r0) when bridged becomes a single channel amplifier that delivers exactly the same total power.
The 350W into 8r0 is exactly the same power output as 175W+175W (=350W)
The 200W into 16r0 is exactly the same power output as 100W+100W (=200W)
You gain NO EXTRA POWER by bridging a pair of amplifiers.

Now let's consider another case of the above example amplifier measured differently.
Measure the single channel output of that 175W+175W into 4+4r0 amplifier.
It might manage a bit more or even a lot more. I'll guess it measures 190W into 4r0 driven single channel.
Now when the two channels are bridged you do not get 380W into 8r0. You get 350W into 8r0. What happened to "bridging rule"? It predicted the wrong power output.

No it did not.
The amplifier when both channels are driven is capable of 175W+175W into 4+4r0. That is the total maximum power of this example amplifier.
When you bridge it into a double impedance load you still get the same total maximum power of 350W.

 

[asbjbo]

yes, in theory you are right. But only for driving resistive loads.

Your theory, not mine.

In the app note, National states "about 120 W of output power" for bridged LM3886's on +/- 28 V rails driving an 8 ohm load, even less than the 136 W predicted by the "twice the power into twice the impedance" rule and far less than the simplistic 4x38 = 152 W estimate. Also, their output power falls off a cliff at about 3 ohm load with +/- 28 V rails, so nominal 8 ohm speakers that drop to less than 6 ohms impedance at certain frequencies could certainly bring surprises, even in steady state.

Yes, the performance degrades rapidly with increasing temperature. According to the data sheet, peak current before the protection circuitry cuts in drops by a third as the heat sink warms up from 25 to 75 degrees, and peak pulse power dissipation for a 1 ms burst also falls from somewhere around 140 W to about 110 W for a single LM3886.

IMHO, getting useful performance from these ICs in bridged configurations is probably as much or more of a thermal engineering issue than an electronic engineering issue. I'm working on a pair of multi-channel amps, and I have no illusions about the inherent limitations in this design. Could be pretty good if operated within its limits, though. I'll put up a build thread when I have something to show & tell.