Cambridge CXA81 MKII Amplifier Review

[Yevhen]

Maybe because possibly the amplifier , as many do- reaches full rated power at 0.3V or so?
I do not understand why integrated amps often have a gain that cause clipping at so low level… 0.3V, maybe it is a volume control thing ?

How many integrated amps are really designed for 2V input ??

I think they increase the gain on purpose to make sure you get loud enough from the headphone port of a smartphone or MP3 player. I don't see the point in buying an all-in-one integrated amp with DAC etc and then trying to connect an external preamp with the 2V out. The smartphone-friendly port with a higher gain is much more useful.

 

[restorer-john]

Maybe because possibly the amplifier , as many do- reaches full rated power at 0.3V or so?
I do not understand why integrated amps often have a gain that cause clipping at so low level… 0.3V, maybe it is a volume control thing ?

How many integrated amps are really designed for 2V input ??

It was due to amplifiers having much higher sensitivities for full rated power at the time. 300mV certainly wasn't unheard of and it made sense- 3mV x100 (40dB) in an RIAA stage, reel to reel decks had 300-500mV outputs and often had output level controls, cassette decks and tuners were all 150mV to maybe maximums of 500mV.

In the 1980s, when manufacturers were going for the best S/N ratios they could get, they stumbled on the great idea of bypassing the preamplifier altogether, using a high gain power stge with a FET front end and a 50K-100K volume pot. Basically a straight line through (often DC). Those amplifiers effectively had power stages that would produce full rated power from a 150-200mV line source. I have models from the early 80s with S/N of 110dB or more due to the complete abolition of the active preamp stage for line inputs.

Many of those designs be they mid range or TOTL did not offer preamplifier outputs or power amplifier inputs due to the levels internally being quite different to what we were used to and what was "standard".

Some even fitted a so-called preamplifier output set of RCAs that were derived from the speaker outputs of the power amplifier, padded down with resistors.

 

[mike435]

I was looking more closely at the SINAD graphs and how it peaks at 1.6V. If you look at its neighbor on the SINAD chart - the Denon AVR-X3700H - the chart there looks very similar and peaks at 1.1V. But the key difference is the Denon has the ability to disconnect the backend amplifiers from the DAC, which then flattens out the curve and eliminates the drop-off at higher outputs. I think I can make an educated guess here that the same thing would happen to the CXA81 if it had the ability to disconnect the amplifiers. It seems like it must be feedback from the amplifier circuits causing the drop-off in SINAD at the pre-outs. If that's the case then I would give more credit to the DAC frontend of this integrated amp. I agree no one who buys this device would ever use the pre-out with a dedicated amp - maybe the sub-out, but not the pre-outs.

All that being said... I also get Amir's point that this unit is a dedicated stereo unit at a relatively higher price point, and it should be held to a higher standard than a home theatre AVR. Seems like Cambridge did make a very competent unit here, when comparing to other integrated devices overall, but it indeed cannot compete with dedicated separates at today's benchmark levels.

 

[KU713]

Maybe I'm naive, but companies should only care about the characteristic sound they want to get from their product. If the human ear cannot hear the noise in silent mode, as is the case with active monitors, and the sound of the amplifier is pleasant to the ear, so why couldn't it be a good product?

What if it is possible, from companies that want to prove the best results from sinad, thd, etc., just to have the best results for this forum, so that the device is so filtered by all possible circuits, but then in fact all life is lost from the sound ?

Why is it that when you review active speakers, you never wonder what kind of amplifier is in it and how it measures? Suddenly it all doesn't matter and everyone is silent from the speaker, which beeps and hisses.

 

[restorer-john]

It seems like it must be feedback from the amplifier circuits causing the drop-off in SINAD at the pre-outs.

It's an overall strange phenomenon.

You can drive an amplifier well into "clipping" with no load connected and that should have zero affect on the separate regulated rails running the preamplifier and digital stages. Without a load, there's essentially no current being drawn from the HV rails, no modulation of those supply rails and there should not be any effect whatsoever on the D/A or preamplifier front end.

The preamplifier outs should never be tested with a connected amplifier being driven into gross clipping into a connected load. I sincerely hope that is not the case.

 

[amirm]

I bet that in a DBT test with properly matched level nobody would tell a difference from the Purifi, if played below clipping.

Then they should charge a lot less for it because there are clearly cheaper gear that would meet the same requirement you state.

 

[amirm]

Maybe I'm naive, but companies should only care about the characteristic sound they want to get from their product. If the human ear cannot hear the noise in silent mode, as is the case with active monitors, and the sound of the amplifier is pleasant to the ear, so why couldn't it be a good product?

They don't want just "good." They want to convince you it is the best there is. This is the second bullet for the product:

• ESS ES9018K2M SABRE32 DAC – reference level digital/analogue conversion

That statement is not remotely backed by my measurements. Yet can convince a buyer that it is true and purchase the product.

Then there is this bullet:

• Toroidal Transformer – for incredible sound-staging

We know that is not true either.

Even the first bullet is wrong:

• 80 watts per channel – power to drive the most demanding speakers

Most demanding speakers require a lot more power than 80 watts.

Finally your statement that human ear can or cannot do that needs evidence. Hook up a 100 dB sensitive speaker to this and you will hear noise.

 

[amirm]

What if it is possible, from companies that want to prove the best results from sinad, thd, etc., just to have the best results for this forum, so that the device is so filtered by all possible circuits, but then in fact all life is lost from the sound ?

They would need to prove this in a controlled test. If they can, they will have a huge winner on their hand. Without that, we can say with confidence that what you say is not true. Whether I test cheap or expensive audio product, the "life" is in the source content, not in reproduction. And that lower distortion and noise have never, ever sounded like something is wrong with them, let alone lacking life.

 

[amirm]

Why is it that when you review active speakers, you never wonder what kind of amplifier is in it and how it measures? Suddenly it all doesn't matter and everyone is silent from the speaker, which beeps and hisses.

I comment on active speaker hiss and in some cases like JBL, it definitely has been objectionable to the point that I won't listen to the speaker. I also routinely note them running out of power and compressing/distorting and deduct points for that as well. Keep in mind that bi-amped active speakers can do things that passives cannot. If their bass driver amp runs out of power, the high frequency harmonics do not bleed into the tweeter as they do in passives. In that sense, some of their sins is less audible.

Regardless, whataboutism doesn't work when the competition in the same situation does better. Tell me what I am getting extra for higher noise and distortion. If it is nothing or something negative, then that is a reason to shop elsewhere.

 

[KU713]

Thanks for the lesson. And is it really so difficult for these professionals on the market to give circuits so that everything with the measurement is in order? Why would you even dare to design a poorly measured product in today's advanced age? As I look at your product measurements, even a very expensive product does not necessarily measure excellent. So in what is it?

 

[amirm]

Certainly design talent and experience is needed to get state of the art performance. That said, it think most designers can get close to matching that if it were made a target/priority for them by company management.

 

[Rjbell78]

Being from the UK I never trusted this brand. I saw it as an in house brand of our big hifi store richer sounds which their staff pushed on unsuspected customers incessantly along with other brand names they bought over the years like jpw, gale to name a few. But unlike the these brand, richer sounds with the help of what hifi magazine have managed to dupe internationally and not just the UK with the brand Cambridge audio.

 

[Yevhen]

Certainly design talent and experience is needed to get state of the art performance. That said, it think most designers can get close to matching that if it were made a target/priority for them by company management.

I wonder if we measure what we think is state of the art properly. I believe all the measurements are done with continuous wave signal. So, with the very deep (and slow) feedback in the power amplifier you can correct all nonlinearities (predistortion).

Is there any way to measure transient response and make sure it is linear? I think that the state-of-the-art Topping class B PA might have worse transient response compared to PAs with no feedback. Unless the feedback loop and PA section in Topping is very wideband ( low time constraint).

I think the same thing is happening with the speaker designs. The three or four way speaker will have much better IMD but the one-way speaker will have much better transient response.

 

[LTig]

I wonder if we measure what we think is state of the art properly. I believe all the measurements are done with continuous wave signal.

If you measure the frequency response (amplitude and phase) you can deduct the transient response (and vice versa).

So, with the very deep (and slow) feedback in the power amplifier you can correct all nonlinearities (predistortion).

There is no "slow" feedback. In fact the more negative feedback you apply the faster the amplifiers transient response and the wider its frequency response, at the cost of lower gain. See e.g opamps with 120 dB open loop gain and more, which can amplify signals in the MHz range with appropriate feedback applied.

I think the same thing is happening with the speaker designs. The three or four way speaker will have much better IMD but the one-way speaker will have much better transient response.

Nope, see coax drivers.

 

[Yevhen]

If you measure the frequency response (amplitude and phase) you can deduct the transient response (and vice versa).

I think it's only valid for an ideal transistor (variable resistor Vs input voltage or current).

The gain and phase of the real world semiconductor transistor partly depends on the power/current delivered before and after (future) the measurement.

In the radio frequency world, this phenomenon is called memory effect and significantly affects the possibility to predistort GaN transistors.

 

[Yevhen]

There is no "slow" feedback. In fact the more negative feedback you apply the faster the amplifiers transient response and the wider its frequency response, at the cost of lower gain. See e.g opamps with 120 dB open loop gain and more, which can amplify signals in the MHz range with appropriate feedback applied.

If there is a memory effect in audio transistor, the high gain multistage PA with deep negative feedback will have more memory distortion compared to a single stage PA.

Just googled for it. It seems like there already are some articles: http://peufeu.free.fr/audio/memory/memory-1-theory.html

 

[Yevhen]

Nope, see coax drivers.

I believe the transient response of the passive 2 way coax speaker will still be affected by delay in the crossover

 

[LTig]

I think it's only valid for an ideal transistor (variable resistor Vs input voltage or current).

This is valid for the measured amplifier as it is, or any other device (system theory), as long as you drive it within its linear range (not clipped).

The gain and phase of the real world semiconductor transistor partly depends on the power/current delivered before and after (future) the measurement.

Almost all audio amplifiers use more than 1 transistor and negative feedback which counteracts such behaviour.

In the radio frequency world, this phenomenon is called memory effect and significantly affects the possibility to predistort GaN transistors.

Luckily the frequency range of audio signals is far below radio frequencies.

 

[Yevhen]

This is valid for the measured amplifier as it is, or any other device (system theory), as long as you drive it within its linear range (not clipped).

But it n class B, AB and D you do not drive it in linear range. You switch it on and of randomly. And depending on the quality of semiconductor and number of traps you'll have different gate lag.

I wonder it this is the reason why majority of musicians prefer tube amps. They should have a very linear transient response.

 

[LTig]

If there is a memory effect in audio transistor, the high gain multistage PA with deep negative feedback will have more memory distortion compared to a single stage PA.

Nope, see below.

Just googled for it. It seems like there already are some articles: http://peufeu.free.fr/audio/memory/memory-1-theory.html

If the simulations in this article is correct, then the thermal drift signal at the output is just a small wandering DC offset (corrected by large negative feedback, as most power amps have a DC gain of 1) while the audio signal itself is still corrected by negative feedback.

If in doubt you can run a null test with music.