Grounding one half of a bridged amp?

Sangram

Moderator
Paid Member
2002-09-25 11:01 am
India
Apart from the fact that the output swing reduces by half, what are the potential pitfalls of such an approach?

I have a bridged amp I'm planning to use as a studio monitoring system with multiple sources, and incorporate a stepped attenuator. I don't want to take the hassle of doubling up all the switching and control circuitry, and I'm trying to avoid the use of a bridge chip and additional construction. I have some DRV134s though if it comes to that.

I was hoping the grounded side would act as a current sink, but was unsure if that actually happens. My Dynes work fine with the bridged amp in bridge configuration, FWIW, and I'm planning on keeping this partnership going if I can. I don't need the additional voltage swing for full power. But the Dynes don't work well with any of the single-ended chipamps I drive it with.

I'm not keen on using the output of one amp to drive the other as is also suggested sometimes (Rod Eliot has an example here: http://sound.westhost.com/project20.htm).

The sources will be a bunch of unbalanced soundcards. And a Delta66 soundcard, which is anyway psuedo-balanced, not a proper differential circuit.

The amp in question is a chipamp (BPA LM4780), but I posted this in SS because I figured it was a sufficiently generic question. Mods may move the thread if they feel it's better there.

Thanks for the help! :)
 
sangram said:
Apart from the fact that the output swing reduces by half, what are the potential pitfalls of such an approach?

A bridged amp is simply 2 amps working together, each sending out opposite signals.
So, if you ground one, you are short circuiting one of the two amps.
If you do, then the magical smoke will come out of the transistors and they won't work any more :(
 

Sangram

Moderator
Paid Member
2002-09-25 11:01 am
India
@poynton and myhrrhleine: Sorry, should be a little more explicit. I meant ground the input of the second amp, and connect the speaker across both amp outputs.

Kind of half-bridge configuration, if you know what I mean. The negative terminal is floating above ground but that half of the amp is only sinking the current, not providing any swing of its own.

@Andre. The amp is currently a balanced amp - each half of the amp is driven by one side of a balanced output from a balanced soundcard. In my mind it was (incorrectly) equivalent as the net result is similar - I was not looking at noise figures when I planned it, but output power only. Very wrong way to look at it, I know, but that's how it worked out for me.

(Un?)Fortunately the Delta 66 is not a pure balanced output. It uses a 'trick' of its own, but it seems to work fine in conjunction with this amp, even if it did produce a high DC offset. The 'return' is just a floated ground using a single capacitor and resistor, and this develops a DC offset depending on the length of the cable. My reasoning basically 'moves' the return circuitry from the preamp to the power amp.

I'm basically looking at educated guesses or actual experiences of having done such a thing. The basic principles are alluded to here: http://sound.westhost.com/project87.htm Look at the last part of the page titled 'Hey that's cheating!'. Extend this into a power amplifier situation and replace 'R3' with an amplifier which actually has no active drive.

Hopefully that'll help some of the confusion.
 
You won't lose any output power by turning a balanced input into an unbalanced input by connecting the return path (-) to ground. You'll just have to turn up the source some more, and you also lose the benefits of a balanced connection (common mode signal rejection).

It's perfectly fine to do though. I've done it many times to connect unbalanced sources to a balanced input. In fact, many cables made explicitly for this purpose do exactly that.:)
 

unclejed613

Member
2006-12-28 12:19 am
grounding the input of the second amp will simply turn it into a virtual ground. if you do this, put a scope on your output bases or gates. you will find that to maintain the virtual ground, the amp actually drives the output devices "in reverse". the second amp actually drives the output devices a little bit to maintain the output terminal at ground potential. you're relly not gaining anything by doing this, but you're wasting energy in the second amp. simply wiring the speaker between the first amp output and ground would give better results. the second amp also contributes noise and since the second amp also has a finite output impedance, using the second amp reduces the damping factor.
 

Sangram

Moderator
Paid Member
2002-09-25 11:01 am
India
Things are a lot clearer, at least to me.

Thanks for the replies.

@ Andre/TheMG: yes, that's what I was planning to do. The amp is actually two identical amps inside an LM4780 chip (but could literally be anything). The input of one amp is driven by the source. The other input would normally be driven by the return (-) half of a balanced source, but I was asking if it's OK to ground that - since my (supposedly balanced) source does exactly the same thing, using a resistor from ground to provide the return path.

The source of the confusion was the way I was using the terms 'bridged' and 'balanced'. Basically I was using the signal ground as a common ground for both amps, and using the source and return paths to drive the two halves of the amp. Basically doubling the output swing.

@unclejed: That was the answer I was really looking for, thanks very much. I was hoping that the virtual ground would actually help instead of hurt the noise characteristics of the whole setup, but you confirm it is otherwise. I will probably have to press my DRV134s into duty - I was hesitating due to the possibility of sonic degradation because of them. I was hoping that the first approach would have some benefits as a few headphone amps seemed to use it - the Beta22 for example.

Anyway, thanks a lot, I think I have an answer! But if you have any other insights, would be much appreciated!
 

unclejed613

Member
2006-12-28 12:19 am
the bridged configuration will have the cumulative noise of both amps whether you use the second amp driven 180 deg out of phase or use it as a virtual ground. another drawback of using the second amp as a virtual ground is that the virtual ground may suffer from output impedance variations in the crossover region, adding crossover distortion. the output impedance goes up in the crossover region, resulting in less current sinking ability. the increase in impedance goes up with frequency because of feedback loss above the input stage corner frequency. the crossover distortion increase would be more noticeable with a virtual ground configuration than with bridging because the second amp will spend more time in the crossover region than with bridging.
 

Sangram

Moderator
Paid Member
2002-09-25 11:01 am
India
Sorry for the confusion Andre, I'll try and clarify.

I have a 4-chip stereo BPA of the LM4780. This is configured with each chip having outputs paralleled, then two chips are used to drive the speaker in a bridged configuration. A stereo amp thus has 4 chips. Hope it's clear so far.

To simplify circuitry and connections, I use a balanced soundcard to provide the input through a TRS connector. The ground (S) is used as a common signal ground for both amps, and the hot (T) and cold (R) of the soundcard are used as signal input for each amplifier. The load is connected between the outputs of each amplifier.

I have two soundcards with this capability. One is an E-mu 1212m. This has an active 4-pole filter on the output, which results in a true differential output. So far no problem. The second card is a Delta66, which actually uses a resistor and capacitor to 'float' the return signal off the signal ground. It is actually a single-ended output which uses a trick (I linked to it earlier) to create a 'balanced' output.

When I drive the amplifier using the Delta, I usually get a pretty high output offset, as the impedance of the cold line is high. I wanted to instead drive the amplifier using the hot (T) input, and instead of connecting the cold (R) to the other amplifier, ground it instead. The load would remain connected as is, i.e., between the outputs of the amplifier.

As unclejed pointed out (I vaguely knew about it but did not know the implications of such a thing), this results in a virtual ground amplifier. The speaker's return load current is not absorbed by the power supply ground directly, but through the second amplifier. That amplifier also works a little bit (I do not have an oscilloscope to test this but I see it clearly, as the feedback network will attempt to correct the output terminal).

I basically posted to figure out the tradeoffs of such an approach, and though it is a chipamp I am applying this to, I guess it's a broader question, so I posted it in SS.
 
If I understand correctly, you have a stereo balanced amplifier with balanced inputs.

You can wire the input + to signal, and input - to gnd (short gnd and - input) and use the outputs as normal or you can connect the "zero" output directly to power supply 0 Volt (directly to PSU capacitors). Either way, you will loose at least half of the amplifier's rated power.

Hope this help.

André
 

KSTR

Member
Paid Member
2007-07-17 2:35 am
Central Berlin, Germany
^^^ My thinking, too. Why not, for example, build a super-balanced line receiver (one NE5532 dual will do that, and please don't ever think that this or a DRV134 would degrade the sound of a LM4780 in standard config ;-) and then you have always a balanced output and can hook up on the input whatever you have, balanced or not.

- Klaus
 

Sangram

Moderator
Paid Member
2002-09-25 11:01 am
India
Originally posted by KSTR <snip>please don't ever think that this or a DRV134 would degrade the sound of a LM4780 in standard config ;-) <snip>[/B]

I guess you're right - I just wanted to avoid building another circuit, another power supply, however simple. Which was really the whole point of the OP.

I guess I'll first try without, and if it is not too great (my monitors should be able to tell me some of the story) I'll ditch the whole thing and go discrete or single-ended, or press my DRV134s into duty.