Question regarding differential amplifiers

Legis

Member
2009-12-23 6:18 pm
Hi!

Fully discrete and differential amplifier Emotiva XPA-1 has 12pcs 10,000µF/63V capacitors, while I think it's maximum operation voltages are something like 100V. With 8 ohm load the maximum output is approx 600W/0,1-0,2% distortion, which means approx 1000-1100 watts of power draw from the wall, which equals approx. 90-92V voltage.

In order to cope with the voltage requirements, the capasitors could be wired in series-parallel combination, that would double their voltage rating and drop the capacitance rating to 30,000µF. But is this needed in differential amplifier?

So my questions are:

1) Does the differential amplifier (which has two amplifier boards, one wired to black speaker posts and one wired to red speaker posts,) mean that while the other amplifier goes into positive the other amp goes into negative, and by this they create a total of double the potential difference than a non-differentian amp would create? So if both amp's filter capacitors are run in 50 volts, the total potential difference is 100 volts? If my logic is correct, the 63V capacitors are adequate for the job of 100V operation voltages?

2) Does each of the amplifier boards see a halved impedance (like in bridged amp), for ex. when driving 8ohm speaker each of the amp boards see a 4 ohm load, or do they both see the full nominal impedance?

Regards,

Legis
 

Legis

Member
2009-12-23 6:18 pm
This is a bridged amp.

600W @ 8 ohms = 69.3vRMS = 98Vpk = ~200V pk-pk.

If the amplifier outputs are balanced then each must have +100V and -100V.

If the amplifier outputs are single ended then each must have 200V.

w

Each sees 4 ohms.

Thanks for clarifying. Does the damping factor remain the same as it would be if only one of those amps would drive a 8 ohm load, or is the damping factor halved in a balanced bridged amp?

As the amp is balanced, so each amp must have +100V and -100V. Is capacitors' voltage rating listed as a vRMS voltage or peak voltage, e.g. does 63V capacitor withstand 63VRMS or 63V peak? If the voltage rating is listed as a peak voltage rating, I think it leaves no choice but to wire the capacitors in parallel/series combination to cope with the voltage requirements in this case?

edit. what about the rectifier at the front of the capacitors, won't they convert the AC voltage to DC voltage for the capacitors? So the capacitors won't see the peak values, only the RMS voltage? And the 63V filter caps would still be adequate for the job? Little lost here :)
 
Last edited:

Legis

Member
2009-12-23 6:18 pm
A bridged amplifier running on ±57V will put out about 600W/8R and could use 63V caps (although a little more headroom might be nice).

Are you sure?

57 volts peak = 40,3 volts RMS
I = 57V/8R = 7,125 amps peak = 5,03 amps RMS
P = 40.3*5,03 = 202,7W.

I think, as wakibaki stated, that the RMS voltage would have to be ~70V, which would mean that the caps should be able to withstand ~100V peak voltages. I think wakibaki calculated the power into 4 ohms (because of the bridged amp) and used ~50% efficiency. With emotiva's ~65% efficiency the voltage requirements would be ~61Vrms/86,2V peak in order to get 600W output from the speaker terminal.




I decided to open the XPA-1.

Here's the backside of the PCB:
[IMGDEAD]https://i817.photobucket.com/albums/zz95/LegisActio/Emotiva%20XPA-1%20mono%20blocks/capacitors2.jpg[/IMGDEAD]



I think the capacitors are wired something like(, does it make any sense?):

[IMGDEAD]https://i817.photobucket.com/albums/zz95/LegisActio/Emotiva%20XPA-1%20mono%20blocks/wiring2.jpg[/IMGDEAD]


It seems so that the caps are not all wired in parallel configuration, but what kind of wiring is this?

XPA-1 uses this rectifier (RBV5006): RBV5006 datasheet Any comments on this, is it good?


Any comment are welcome :).
 
Last edited:

Legis

Member
2009-12-23 6:18 pm
A bridge amplifier doubles the voltage swing. Each half of the amplifier only need supply 35V RMS, so a ±57V supply can swing that if the supply is designed properly.

So 63v would have been adequate for parallel wiring, but it seems the caps have not been wired that way. Could you comment on the wiring of the capacitors? What combination are they wired in, and can you presume what total capacitance that wiring gives or is more information needed?
 
Hi,
if you want 600W into 8r0 when in bridged mode, this requires a pair of amplifiers that can provide 300W into 4r0.
Go back to the formula I posted earlier.

Vpk = sqrt(300*4*2) = 49Vpk.
Add on 8V to 11V for the PSU and you have +-57Vdc to +-60Vdc.

This supply must be designed for driving a pair of 4ohm amplifiers. The transient current demand will be enormous. The PSU must be designed to deliver that transient demand without the voltage rails collapsing. I would be aiming for a peak transient demand of 70Apk to 75Apk from the PSU and decoupling to power a reactive 8ohm speaker to 600W
 

wakibaki

Banned
2008-01-08 11:51 pm
This is a bridged amp.

600W @ 8 ohms = 69.3vRMS = 98Vpk = ~200V pk-pk.

If the amplifier outputs are balanced then each must have +100V and -100V.

If the amplifier outputs are single ended then each must have 200V.

w

Each sees 4 ohms.

AndrewT is usually pretty reliable.

Sorry, I seem to have slipped a cog when translating to bridged mode.

600W @ 8 ohms = 69.3vRMS = 98Vpk = ~200V pk-pk.

If the amplifier outputs are balanced then each must have +50V and -50V. This equates to 100V differential in one direction and 100V differential in the other direction for 200V pk-pk

If the amplifier outputs are single ended then each must have 100V. Which again equates to 100V differential in one direction and 100V differential in the other direction for 200V pk-pk

You would require to add 7~11V for PSU losses.

Sorry again for having created this confusion.

w
 

Legis

Member
2009-12-23 6:18 pm
I noticed that there is a dummy load in parallel with speaker load in Emotiva XPA-1:


[IMGDEAD]https://i817.photobucket.com/albums/zz95/LegisActio/CR.jpg[/IMGDEAD]


Why is it there? Are there any downsides regarding sound quality and/or is there some reason to bypass the dummy load?
 
Last edited:
if the red and black are the speaker terminals then what you have shown is the amplifier's output Zobel.

It provides a load at HF for the amplifier circuits.It helps maintain stability margins irrespective of whether a speaker is attached or not.

The output Zobel is usually located very close to the output stage.
However, more advanced version can have the extra stability components spread around for best effectiveness.
I use a Pi version of the Thiele Network and this requires a Zobel direct across the speaker output terminals.