• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Testing and measuring bridged dual-mono amplifier

pma

Master Contributor
Joined
Feb 23, 2019
Messages
5,377
Likes
13,218
Location
Prague
Testing and measuring bridged dual-mono amplifier

Bridging a dual-mono amplifier that was not originally designed for this task as a bit challenging, especially if it only has unbalanced RCA inputs. This is the case of my A250W power amplifier, however I was curious how it would work in the bridged mode.

A250W is a dual-mono amplifier, two completely separated power amplifiers with separate power supplies, all in one 4U x 19” case. The inputs are only unbalanced RCA. To make a bridged amplifier, it is necessary to tight output “Lo” speaker terminals of both channels together and the load is there connected between “Hi” speaker terminals of the left and right channels. This will interconnect power signal grounds of both channels. Now, we must not connect the signal grounds of the left and the right RCA input connectors together, as it would create a ground loop including the high current speaker ground returns and thus destroy the S/N of the amplifier, resulting in hum and buzz. At the same time, we must invert signal for one of the inputs to get the doubled voltage swing between the “Hi” output terminals of the left and right channel. This is the trick of the bridged amplifier, we get twice of output voltage of the single amplifier channel, thus four time the power (in theory). But, now any of the two channels sees half of the load impedance and must deliver the two times more current. So, the output transistors are much more stressed.

My test setup looks as follows:

BTL_test_A250W.png


L and R channels of the dual mono amplifier are completely separated. They are then connected by the “COM connect” wire between the OUT Lo terminals of both channels. The load is connected between OUT Hi terminals.
RCA input of the left channel is driven directly, input of the right channel is driven via the 1:1 link transformer in opposite polarity, which assures inverted polarity at the OUT Hi of the right channel. Load is thus not tied to signal ground and the measurement must be differential, as shown in the schematics.


Measurements

I have measured THD vs. output level for 4ohm, 6.8ohm and 8ohm resistive load. Then I have also measured a burst power into 4ohm.

THD vs. level at 1kHz

A250W_bridged_thdlevel_1kHz_526W_4R_460W_8R.png


The amplifier keeps its low distortion even in the bridged mode. With 4 ohm load, it gave 526W at THD = 0.1% and 430W at THD = 0.008%. With 8 ohm load, it gave 460W at THD = 0.33% and 365W at THD = 0.002%.


Burst power

A250W_bridged_4R_100Hz-burst.png


Burst power was measured by 7 periods of 100Hz sine, total duration 70ms, rectangular window (no smoothing of signal rise and decay). Burst power measured seems to be 770W/4ohm.


Conclusion

This amplifier rated at 2 x 250W/4ohm and 2 x 150W/8ohm gives 1 x 526W/4ohm and 1 x 460W/8ohm with a stepped level method of sine input. Burst power, 100Hz/70ms, is 770W/4ohm. The limiting factor here are the power supplies of the two channels, each based on a 300VA transformer. But the power in the bridged mode is higher than for the similarly rated class D amplifier.

 
Last edited:
THD and THD+N vs. power into 6.8ohm load. 566W/6.8ohm/THD+N=1%. It starts to be not so simple to measure at these power levels ..... :).

A250W_bridged_thdpower1kHz_6R8_sm.png
 
Some more measurement : THD+N vs. power into 4ohm and 8ohm at 1kHz and 5kHz, @BW45kHz. We have "clean power" up to 400W and we get >700W/4ohm/THD1%. And some blown fuses inside the amp that had to be increased for bridged 4ohm measurements :).

A250W_bridged_thdpower_4R-8R_sm.png




Comparison of distortion and power in "normal" mode and bridged mode:

A250W_single-bridged_thdpower_4R.png



A250W_single-bridged_thdpower_8R.png
 
Last edited:
Burst power

View attachment 422029

Burst power was measured by 7 periods of 100Hz sine, total duration 70ms, rectangular window (no smoothing of signal rise and decay). Burst power measured seems to be 770W/4ohm.
Looks like there is a small ~2-3V DC/ULF offset shift in the amplifiers output. I would assume it recovers relatively quickly.
If it were being used to drive a tweeter directly in an active 2-way without any passive HP filter (say a cap) it has the potential of causing issues. Thou it is far more likely the 770 Watts itself would be far more damaging.
 
Output DC voltage (measured) is 0.0004V (0.4mV). What you see after the clipped burst end is a low frequency recovery time during the limited low frequency corner. Only the amplifiers that have frequency response flat from DC are free of the LF shift. It is a so called Overload Restoring Time. More on this topic is here.
 
Output DC voltage (measured) is 0.0004V (0.4mV). What you see after the clipped burst end is a low frequency recovery time during the limited low frequency corner. Only the amplifiers that have frequency response flat from DC are free of the LF shift. It is a so called Overload Restoring Time. More on this topic is here.
I should have put in "ULF" = Ultra low frequency :)

It has been quite some time since I performed 60268-3 and used the incorrect vernacular.
I did not intend to imply the offset was either permanent or related to the standard meaning of DC offset.
 
Cool stuff! How good is that transformer? It seems fine at 1 kHz, but how does it do lower down? I’m using a DRV134 to drive small B&O Class-D module in bridged mode for a sub. Works well and should not have any of the possible issues the transformer has.
 
Sure, below 500Hz distortion of the transformer starts to rise. THD vs. frequency, 360W/6.8ohm, with trafo inversion attached below.

trafo_bridged_thdfreq.jpg
 
Nice work Pavel!

In the best I used a center-tapped transformer or pair of op-amps (inverting and parallel non-inverting buffer) to better match path delay and reduce distortion from amplifier/path mismatching, plus usually had a fine gain tweak. But the resulting improvement was most likely inaudible.

Burst recovery is impressive, just nothing visible at all aside from the usual LF tail! My compliments to the designer. ;)
 
In a slightly different application it was not too uncommon to use a pair of closely matched transformers but only one inverted. So both signal paths have similar amplitude/phase transfer functions.
I have used a balun or pair of transformers if I did not have a well-balanced center-tap around. Some audio DI boxes had transformers built in that would produce decent in-phase and out-of-phase outputs. In my day job audio transformers were rarely a thing. We did have a bunch of these, along with the PSPL SE-Differential version that added an input splitter and semirigid coax non-inverting leg tightly phase matched to the inverting output. FR 200 kHz to 23 GHz so not much good for audio.

1739069989397.png
 
Nice work Pavel!

In the best I used a center-tapped transformer or pair of op-amps (inverting and parallel non-inverting buffer) to better match path delay and reduce distortion from amplifier/path mismatching, plus usually had a fine gain tweak. But the resulting improvement was most likely inaudible.

Burst recovery is impressive, just nothing visible at all aside from the usual LF tail! My compliments to the designer. ;)
Thank you Don. This measurement was just for fun, I wanted to know what I get from bridging that amp. Because that amp has not been originally designed to operate bridged, connecting the output common and input common of the two channels simultaneously results in severe degradation of S/N by hum components, loop current. Output commons must be interconnected to get the bridged signal power output and then the channels must be driven with opposite phase but the input grounds must not be interconnected to eliminate loops. They must be floating. So, it goes to transformer(s) or DRV134 (it has floating balanced outputs like pro devices). Two opamps in simple inversion circuit cannot be used as they have common ground. Center tapped trafo cannot be used as it has common center point. DRV134 cannot be used because of high output noise and 40 dB gain of the DUT. So what is left? @HaveMeterWillTravel suggestion is fine, two transformers, one of them phase inverted.
 
Last edited:
Now n days, I suppose one could use the available DSP functions to correct to some of the transformer non-linearities.
Not really, correcting this kind of distortion is hard to do with DSP. And if you have a DSP, you can probably get a balanced output anyway. Also a simple chip solution like DRV134 is much cheaper than a high-quality transformer.
 
  • Like
Reactions: pma
Also a simple chip solution like DRV134 is much cheaper than a high-quality transformer.
DRV134 is great, but in some cases the output noise is too high. If I put it in front of this bridged amp with 40 dB gain, the idle output noise of DRV134 sends THD+N to misery. Transformer distortion is inaudible, but the DRV134 noise is audible when 100x amplified.
 
Luckily there are alternatives that are lower noise.
 
Luckily there are alternatives that are lower noise.
Yes, but as explained several times, I need an alternative with floating outputs. Must not have a common ground for L and R channels. Can you suggest an alternative with floating outputs and very low noise? A link, please.
 
Thank you Don. This measurement was just for fun, I wanted to know what I get from bridging that amp. Because that amp has not been originally designed to operate bridged, connecting the output common and input common of the two channels simultaneously results in severe degradation of S/N by hum components, loop current. Output commons must be interconnected to get the bridged signal power output and then the channels must be driven with opposite phase but the input grounds must not be interconnected to eliminate loops. They must be floating. So, it goes to transformer(s) or DRV134 (it has floating balanced outputs like pro devices). Two opamps in simple inversion circuit cannot be used as they have common ground. Center tapped trafo cannot be used as it has common center point. DRV134 cannot be used because of high output noise and 40 dB gain of the DUT. So what is left? @HaveMeterWillTravel suggestion is fine, two transformers, one of them phase inverted.
Got it, thanks Pavel. I, and I am sure you, have built op-amp circuits for this that isolate the ground (through high-value resistors) or used differential (instrumentation) op-amps, but they have their own set of problems (noise, DC wander, CM rejection issues, etc.) As you and @HaveMeterWillTravel said transformers work fine and often better than op-amps. The results you got for a "fun" trial are great, and show what a good design can do even with a simple bridging circuit. Looks like you got over twice the power with roughly 2x THD/N, a win!

Curious: Output power limited by power supply or something else?

Love seeing real-world experiments!
 
  • Like
Reactions: pma
Power supplies. Transformers are 300VA in each channel and the voltage drops. But for reliable operation into complex load at very high power, number of output pairs should be increased.
 
Back
Top Bottom