Cambridge CXA81 MKII Amplifier Review

[Dougey_Jones]

Imagine buying this turd over a Schiit Kara with a pair of Topping B100's..

 

[Old_School_Brad]

Imagine buying this turd over a Schiit Kara with a pair of Topping B100's..

It appears you are a bit confused. Referring to this negatively, despite the clear and accurate measurements provided in the first post, doesn't seem very fair.

It’s easy to get caught up in groupthink, isn’t it?

 

[Yevhen]

Imagine buying this turd over a Schiit Kara with a pair of Topping B100's..

There's at least 5 years between the designs you compare

 

[NoMoFoNo]

Imagine buying this turd over a Schiit Kara with a pair of Topping B100's..

Well, the CXA81 can be bought from Cambridge on ebay, with warranty, for $800. The Kara alone is $700 and Schiit exacts a 15% restocking fee if one of their items is returned so that has to be considered as well. Oh, and let's remember Topping's VERY recent customer service and reliability SNAFUs. I would hesitate to buy any Topping amplifier any time soon.

Taking everything into consideration I'd choose the CXA81 over the Schiit/Topping combo you mentioned.

 

[Yevhen]

The Kara alone is $700

To be fair, he could take a cheaper preamp, but then there's probably no Phono, Remote etc. So overall, you are right, CXA81 isn't easy to beat, especially when you want a minimalistic look. It's also a better investment if you plan to sell it on eBay when it's time to update

 

[ryanmh1]

I'm going to return to this amplifier for just a bit, because it stands out on a chart I'm building, and its a relatively inexpensive name brand product. I said before that it was linear. This is also a good spot for this until I have the time to do a more detailed post about amplifier linearity. For those who are interested in this sort of thing, it hopefully it will be helpful and we can learn a thing or two from each other.

For the CXA81 Mk II, I don't think I fully appreciated just how linear it is. So far, I have only found an absolutely tiny handful that best it: Benchmark AHB2 and Topping B100 (and the other similar discreet Toppings), and Soncoz SGP1. Stereophile has reviewed and measured a few in the multiple kilobuck range, but out of the 20 or so I've catalogued from here, that's it. Hypex is a unique case I'll address at the end.

The chart that shows it best is this one, here:

Look how clean that is. The noise floor is somewhat higher because of a linear supply, but for most of us with normal speakers, that noise floor is not a relevant issue. The noise is not correlated to the signal, and will generally not result in audible problems. What matters is what happens once the distortion starts to come out it, and we can see how the amplifier circuitry itself is performing. If there is something audible on normal speakers (emphasis on the "if") this is one primary area you will find, particularly in a traditional AB amplifier. At 1kHz, distortion is somewhere below the noise. Likely no issues there. Distortion at 15kHz never rises above -77dB. Oddly, it even gets lower as power increases. I can only assume that's because as the signal continues to increase, the distortion is so low that the noise plays an ever lower role. It's relatively unusual. As frequency increases, distortion usually will increase as the feedback loses control of the amplifier's output, since there is ever less feedback at higher and higher frequencies. Generally, the same will happen as power increases if there is any nonlinearity affecting the output. It doesn't do that here. It flattens out, rises a bit, and then drops as power increases. The 5kHz curve is a typical curve for a well-controlled amplifier. But, let's just call this under .015% at any power level or frequency up to 15kHz. Not that impressive right? That's like a -77 SINAD! Average at best. WRONG! It's actually very good.

Let's compare this to a couple of old favorites, most of which are "better" in the rankings. First, Topping PA5 for a Class D chip amp. Remember this one for later when we look at a Hypex.

There is less noise, overall, but there is a significantly larger amount of distortion. By 5W, the amplifier has already lost the plot compared to the Cambridge. At full power, which we'll call 80W, we're at -63 or .07%. To be clear, that's still good.

And because I need another sort of decent AB amplifier (and not that many have been reviewed outside of AVRs and the like), the Schiit Aegir:

Yup, it is quieter than the Cambridge. Props where they are due, the Schiit boys got the linear supply in this thing worked out. It is exemplary. Not a lot of linear supply amps where the power supply is this clean, particularly for this cheap. But then watch what happens as frequency and output rise. By 5 watts, the amp is at .02% THD, or -77dB. And it keeps going up. At higher powers, it's maybe -73if we take an average. Still not bad, really, "by high end standards". But it's 15W.... This is interesting, though, because without that low noise, this would look better.

And finally, since we want to be mean to the Cambridge as well, the Benchmark AHB2:

Now, that's clean. But, all is not quite as it would appear. The measurements were taken in "low gain" mode. Why does this matter? It matters because it means that the amplifier has to do a lot less work. It effectively gets the benefit of having a $20,000 test rig do the hard work of voltage amplification for it. Let's see if we can fix that. Fortunately, this amp has been profiled a lot. According to measurements taken by the legendary Bascom King (https://www.soundstagenetwork.com/i...er&catid=97:amplifier-measurements&Itemid=154), the AHB2 clocks in at .002% THD+N into 4 ohms at 120W. That's minus -94dB. Same, but Bascom didn't specify the gain. Stereophile? (https://www.stereophile.com/content/benchmark-media-systems-ahb2-power-amplifier-measurements). Gain not specified. But, about .0025 at 60W. So -92. Amir's old review? Low gain specified there, too. So, you just compared 3 data sources, apples to apples, and lined them up. And we cannot confirm that any of the measurements were high gain mode, so we do not know whether this Benchmark is an apples to apples comparison. Now, somewhere I actually do have high gain results but cannot find them at the moment. The result is around -90, so it drops perhaps 3dB. Still very respectable.

Now that we know all that, let's compare:

Cambridge: -77 (or -80 at full power)
Topping: -63
Aegir: -73
Benchmark: -90

So, the Benchmark is more linear than the others. But it is not 25dB cleaner than the Cambridge as 1kHz SINAD would suggest. The Topping is actual less clean than the Cambridge by a significant margin, despite 1kHz SINAD results that would suggest the opposite. The Aegir is slightly worse, again contrary to 1kHz SINAD results suggesting the opposite. The difficulty with 1kHz SINAD is that it is can turn into something of a noise ranking. 1kHz also does not test amplifier linearity all that well. In other words, it doesn't really tell you which amplifier is likely to have more distortion overall, both harmonic and intermodulation. At 15kHz (or ideally 20kHz) the feedback is doing its job less and less well, and the amplifier is struggling harder. If the amplifier continues to do extremely well, it is highly likely it will have no problems. And so it is, surprisingly, for the Cambridge.

Last but not least, Hypex NCx500:

Look at this! It's less clean at full power than the Cambridge at 15kHz! Eh, sort of, but not really. As power increases, the feedback loop does lose a bit of control. But not nearly as much as it looks like. We're starting from a shockingly low baseline of -90dB up to 20W at 15kHz. Nothing else ever even gets that low except the Benchmark. It's a clean amplifier, but on the 15kHZ full power metric, it's technically worse than the Cambridge. So, one number does not tell the whole story. You still need to read the whole chart, since the Hypex is cleaner, but it has nothing to do with it being cleaner on 1kHz SINAD. The Hypex is cleaner on the multitone, and it's cleaner on the CCIF twin tone than the Benchmark. And yes, it also "cheats" with low gain a little bit, but so does the Benchmark. Odds are, it would do that on other tests too since it is so incredibly clean. No 1kHz component on the multitone, and almost no sidebands on the twin tone. The old ones weren't nearly this good. The new NCx and Purifi amps are really quite amazing. But, so is the Cambridge for a really simple design with a linear supply. At some point, this story could continue with the new Topping amps, which in turn are able to absolutely decimate the Hypex amps. Just in case you were thinking the Hypex results had anything to do with it being Class D. They don't. Hypex does this despite being a Class D, not because it is.

Cambridge's merit here is that it does this all in an integrated amplifier, with a brand name, a real warranty, and a nice package. And once you get past the supply noise, it measures up really well. And you can see that from simple chart plotting SINAD at 15kHz.

Moral: You need to learn to read charts to understand what they mean, and what they show the amplifier is doing if you want to have a chance of sorting this all out. If your goal is finding an amp that is clean and the least likely to "misbehave" you cannot rely on any single number, particularly any single number taken at 1kHz. Usually, you can rely on 20kHz much more confidently (or 15kHz, if that's what you've got). But even then, it still take sorting and verifying.

 

[taotone]

I'm going to return to this amplifier for just a bit, because it stands out on a chart I'm building, and its a relatively inexpensive name brand product. I said before that it was linear. This is also a good spot for this until I have the time to do a more detailed post about amplifier linearity. For those who are interested in this sort of thing, it hopefully it will be helpful and we can learn a thing or two from each other.

For the CXA81 Mk II, I don't think I fully appreciated just how linear it is. So far, I have only found an absolutely tiny handful that best it: Benchmark AHB2 and Topping B100 (and the other similar discreet Toppings), and Soncoz SGP1. Stereophile has reviewed and measured a few in the multiple kilobuck range, but out of the 20 or so I've catalogued from here, that's it. Hypex is a unique case I'll address at the end.

The chart that shows it best is this one, here:

Look how clean that is. The noise floor is somewhat higher because of a linear supply, but for most of us with normal speakers, that noise floor is not a relevant issue. The noise is not correlated to the signal, and will generally not result in audible problems. What matters is what happens once the distortion starts to come out it, and we can see how the amplifier circuitry itself is performing. If there is something audible on normal speakers (emphasis on the "if") this is one primary area you will find, particularly in a traditional AB amplifier. At 1kHz, distortion is somewhere below the noise. Likely no issues there. Distortion at 15kHz never rises above -77dB. Oddly, it even gets lower as power increases. I can only assume that's because as the signal continues to increase, the distortion is so low that the noise plays an ever lower role. It's relatively unusual. As frequency increases, distortion usually will increase as the feedback loses control of the amplifier's output, since there is ever less feedback at higher and higher frequencies. Generally, the same will happen as power increases if there is any nonlinearity affecting the output. It doesn't do that here. It flattens out, rises a bit, and then drops as power increases. The 5kHz curve is a typical curve for a well-controlled amplifier. But, let's just call this under .015% at any power level or frequency up to 15kHz. Not that impressive right? That's like a -77 SINAD! Average at best. WRONG! It's actually very good.

Let's compare this to a couple of old favorites, most of which are "better" in the rankings. First, Topping PA5 for a Class D chip amp. Remember this one for later when we look at a Hypex.

There is less noise, overall, but there is a significantly larger amount of distortion. By 5W, the amplifier has already lost the plot compared to the Cambridge. At full power, which we'll call 80W, we're at -63 or .07%. To be clear, that's still good.

And because I need another sort of decent AB amplifier (and not that many have been reviewed outside of AVRs and the like), the Schiit Aegir:

Yup, it is quieter than the Cambridge. Props where they are due, the Schiit boys got the linear supply in this thing worked out. It is exemplary. Not a lot of linear supply amps where the power supply is this clean, particularly for this cheap. But then watch what happens as frequency and output rise. By 5 watts, the amp is at .02% THD, or -77dB. And it keeps going up. At higher powers, it's maybe -73if we take an average. Still not bad, really, "by high end standards". But it's 15W.... This is interesting, though, because without that low noise, this would look better.

And finally, since we want to be mean to the Cambridge as well, the Benchmark AHB2:

Now, that's clean. But, all is not quite as it would appear. The measurements were taken in "low gain" mode. Why does this matter? It matters because it means that the amplifier has to do a lot less work. It effectively gets the benefit of having a $20,000 test rig do the hard work of voltage amplification for it. Let's see if we can fix that. Fortunately, this amp has been profiled a lot. According to measurements taken by the legendary Bascom King (https://www.soundstagenetwork.com/i...er&catid=97:amplifier-measurements&Itemid=154), the AHB2 clocks in at .002% THD+N into 4 ohms at 120W. That's minus -94dB. Same, but Bascom didn't specify the gain. Stereophile? (https://www.stereophile.com/content/benchmark-media-systems-ahb2-power-amplifier-measurements). Gain not specified. But, about .0025 at 60W. So -92. Amir's old review? Low gain specified there, too. So, you just compared 3 data sources, apples to apples, and lined them up. And we cannot confirm that any of the measurements were high gain mode, so we do not know whether this Benchmark is an apples to apples comparison. Now, somewhere I actually do have high gain results but cannot find them at the moment. The result is around -90, so it drops perhaps 3dB. Still very respectable.

Now that we know all that, let's compare:

Cambridge: -77 (or -80 at full power)
Topping: -63
Aegir: -73
Benchmark: -90

So, the Benchmark is more linear than the others. But it is not 25dB cleaner than the Cambridge as 1kHz SINAD would suggest. The Topping is actual less clean than the Cambridge by a significant margin, despite 1kHz SINAD results that would suggest the opposite. The Aegir is slightly worse, again contrary to 1kHz SINAD results suggesting the opposite. The difficulty with 1kHz SINAD is that it is can turn into something of a noise ranking. 1kHz also does not test amplifier linearity all that well. In other words, it doesn't really tell you which amplifier is likely to have more distortion overall, both harmonic and intermodulation. At 15kHz (or ideally 20kHz) the feedback is doing its job less and less well, and the amplifier is struggling harder. If the amplifier continues to do extremely well, it is highly likely it will have no problems. And so it is, surprisingly, for the Cambridge.

Last but not least, Hypex NCx500:

Look at this! It's less clean at full power than the Cambridge at 15kHz! Eh, sort of, but not really. As power increases, the feedback loop does lose a bit of control. But not nearly as much as it looks like. We're starting from a shockingly low baseline of -90dB up to 20W at 15kHz. Nothing else ever even gets that low except the Benchmark. It's a clean amplifier, but on the 15kHZ full power metric, it's technically worse than the Cambridge. So, one number does not tell the whole story. You still need to read the whole chart, since the Hypex is cleaner, but it has nothing to do with it being cleaner on 1kHz SINAD. The Hypex is cleaner on the multitone, and it's cleaner on the CCIF twin tone than the Benchmark. And yes, it also "cheats" with low gain a little bit, but so does the Benchmark. Odds are, it would do that on other tests too since it is so incredibly clean. No 1kHz component on the multitone, and almost no sidebands on the twin tone. The old ones weren't nearly this good. The new NCx and Purifi amps are really quite amazing. But, so is the Cambridge for a really simple design with a linear supply. At some point, this story could continue with the new Topping amps, which in turn are able to absolutely decimate the Hypex amps. Just in case you were thinking the Hypex results had anything to do with it being Class D. They don't. Hypex does this despite being a Class D, not because it is.

Cambridge's merit here is that it does this all in an integrated amplifier, with a brand name, a real warranty, and a nice package. And once you get past the supply noise, it measures up really well. And you can see that from simple chart plotting SINAD at 15kHz.

Moral: You need to learn to read charts to understand what they mean, and what they show the amplifier is doing if you want to have a chance of sorting this all out. If your goal is finding an amp that is clean and the least likely to "misbehave" you cannot rely on any single number, particularly any single number taken at 1kHz. Usually, you can rely on 20kHz much more confidently (or 15kHz, if that's what you've got). But even then, it still take sorting and verifying.

Thank you. Not an amplifier I would buy, but it really didn't deserve the casually derogatory review by Amir and the pile-on it got in this thread.

Its absolutely deserving of a "competent" award and not at all to be derided for its target market: a nice update to a good mainstream integrated amp aimed at being paired with a decent set of mid-price speakers for the average UK consumer in a small/medium listening space. In most of the places this amp is deployed it will do a very nice job. For the technically-minded you could likely do better, but most people don't care enough about the minor differences to make it worth the hassle.

 

[NoMoFoNo]

I've been torn about this amp and the reviews of it here. First, I GREATLY value and appreciate this community and Amir's work. ASR has become my go-to audio forum. Second, I own this amp so it's impossible to be completely unbiased about it.

IMO the tone of his verbiage in the first iteration of the CXA81 (first version) review came across as aggressively harsh, almost personal. His follow up that described interaction with Cambridge reps was really useful, but I have to believe that something about Cambridge Audio the company, or perhaps the tenor of interactions with them surrounding these reviews, colored his remarks about these amps. There is something about the way he writes about them that almost seems like he is gritting his teeth and struggling not to slam them and I don't understand it.

One thing to consider is that these amps can easily be found for far less than MSRP direct from Cambridge, new and with warranty, on their ebay store. I paid $800 for a new open box with new warranty. That's the same price that the Yamaha A-S701 and A-S801 models (I also own an A-S801) can be found for and I see the CXA81 as a competitor to those models. I like them both very much. The Yamahas are better value IMO and offer marginally more power and tone controls but otherwise are similar, but from what I've seen the Yamahas don't hold value very well because they are so ubiquitous.

Having said all of that, I wish I understood what appears to be a burr under the saddle of Amir when it comes to the CXA81 and why his reviews seem different from so many other reviews here. We'll probably never know but my interpretation is that there is something personal going on there, something like a harsh or nasty interaction he had with their reps that soured him on the company, perhaps well deserved. If Cambridge poisoned the well somehow I guess it's on them, but the CXA81 has served well in my home.

 

[Doodski]

I've been torn about this amp and the reviews of it here. First, I GREATLY value and appreciate this community and Amir's work. ASR has become my go-to audio forum. Second, I own this amp so it's impossible to be completely unbiased about it.

IMO the tone of his verbiage in the first iteration of the CXA81 (first version) review came across as aggressively harsh, almost personal. His follow up that described interaction with Cambridge reps was really useful, but I have to believe that something about Cambridge Audio the company, or perhaps the tenor of interactions with them surrounding these reviews, colored his remarks about these amps. There is something about the way he writes about them that almost seems like he is gritting his teeth and struggling not to slam them and I don't understand it.

One thing to consider is that these amps can easily be found for far less than MSRP direct from Cambridge, new and with warranty, on their ebay store. I paid $800 for a new open box with new warranty. That's the same price that the Yamaha A-S701 and A-S801 models (I also own an A-S801) can be found for and I see the CXA81 as a competitor to those models. I like them both very much. The Yamahas are better value IMO and offer marginally more power and tone controls but otherwise are similar, but from what I've seen the Yamahas don't hold value very well because they are so ubiquitous.

Having said all of that, I wish I understood what appears to be a burr under the saddle of Amir when it comes to the CXA81 and why his reviews seem different from so many other reviews here. We'll probably never know but my interpretation is that there is something personal going on there, something like a harsh or nasty interaction he had with their reps that soured him on the company, perhaps well deserved. If Cambridge poisoned the well somehow I guess it's on them, but the CXA81 has served well in my home.

Can you quote some of this stuff you feel so strongly about? Otherwise it's all rubbish.

 

[Yevhen]

I'm going to return to this amplifier for just a bit, because it stands out on a chart I'm building, and its a relatively inexpensive name brand product. I said before that it was linear. This is also a good spot for this until I have the time to do a more detailed post about amplifier linearity. For those who are interested in this sort of thing, it hopefully it will be helpful and we can learn a thing or two from each other.

For the CXA81 Mk II, I don't think I fully appreciated just how linear it is. So far, I have only found an absolutely tiny handful that best it: Benchmark AHB2 and Topping B100 (and the other similar discreet Toppings), and Soncoz SGP1. Stereophile has reviewed and measured a few in the multiple kilobuck range, but out of the 20 or so I've catalogued from here, that's it. Hypex is a unique case I'll address at the end.

The chart that shows it best is this one, here:

Look how clean that is. The noise floor is somewhat higher because of a linear supply, but for most of us with normal speakers, that noise floor is not a relevant issue. The noise is not correlated to the signal, and will generally not result in audible problems. What matters is what happens once the distortion starts to come out it, and we can see how the amplifier circuitry itself is performing. If there is something audible on normal speakers (emphasis on the "if") this is one primary area you will find, particularly in a traditional AB amplifier. At 1kHz, distortion is somewhere below the noise. Likely no issues there. Distortion at 15kHz never rises above -77dB. Oddly, it even gets lower as power increases. I can only assume that's because as the signal continues to increase, the distortion is so low that the noise plays an ever lower role. It's relatively unusual. As frequency increases, distortion usually will increase as the feedback loses control of the amplifier's output, since there is ever less feedback at higher and higher frequencies. Generally, the same will happen as power increases if there is any nonlinearity affecting the output. It doesn't do that here. It flattens out, rises a bit, and then drops as power increases. The 5kHz curve is a typical curve for a well-controlled amplifier. But, let's just call this under .015% at any power level or frequency up to 15kHz. Not that impressive right? That's like a -77 SINAD! Average at best. WRONG! It's actually very good.

Let's compare this to a couple of old favorites, most of which are "better" in the rankings. First, Topping PA5 for a Class D chip amp. Remember this one for later when we look at a Hypex.

There is less noise, overall, but there is a significantly larger amount of distortion. By 5W, the amplifier has already lost the plot compared to the Cambridge. At full power, which we'll call 80W, we're at -63 or .07%. To be clear, that's still good.

And because I need another sort of decent AB amplifier (and not that many have been reviewed outside of AVRs and the like), the Schiit Aegir:

Yup, it is quieter than the Cambridge. Props where they are due, the Schiit boys got the linear supply in this thing worked out. It is exemplary. Not a lot of linear supply amps where the power supply is this clean, particularly for this cheap. But then watch what happens as frequency and output rise. By 5 watts, the amp is at .02% THD, or -77dB. And it keeps going up. At higher powers, it's maybe -73if we take an average. Still not bad, really, "by high end standards". But it's 15W.... This is interesting, though, because without that low noise, this would look better.

And finally, since we want to be mean to the Cambridge as well, the Benchmark AHB2:

Now, that's clean. But, all is not quite as it would appear. The measurements were taken in "low gain" mode. Why does this matter? It matters because it means that the amplifier has to do a lot less work. It effectively gets the benefit of having a $20,000 test rig do the hard work of voltage amplification for it. Let's see if we can fix that. Fortunately, this amp has been profiled a lot. According to measurements taken by the legendary Bascom King (https://www.soundstagenetwork.com/i...er&catid=97:amplifier-measurements&Itemid=154), the AHB2 clocks in at .002% THD+N into 4 ohms at 120W. That's minus -94dB. Same, but Bascom didn't specify the gain. Stereophile? (https://www.stereophile.com/content/benchmark-media-systems-ahb2-power-amplifier-measurements). Gain not specified. But, about .0025 at 60W. So -92. Amir's old review? Low gain specified there, too. So, you just compared 3 data sources, apples to apples, and lined them up. And we cannot confirm that any of the measurements were high gain mode, so we do not know whether this Benchmark is an apples to apples comparison. Now, somewhere I actually do have high gain results but cannot find them at the moment. The result is around -90, so it drops perhaps 3dB. Still very respectable.

Now that we know all that, let's compare:

Cambridge: -77 (or -80 at full power)
Topping: -63
Aegir: -73
Benchmark: -90

So, the Benchmark is more linear than the others. But it is not 25dB cleaner than the Cambridge as 1kHz SINAD would suggest. The Topping is actual less clean than the Cambridge by a significant margin, despite 1kHz SINAD results that would suggest the opposite. The Aegir is slightly worse, again contrary to 1kHz SINAD results suggesting the opposite. The difficulty with 1kHz SINAD is that it is can turn into something of a noise ranking. 1kHz also does not test amplifier linearity all that well. In other words, it doesn't really tell you which amplifier is likely to have more distortion overall, both harmonic and intermodulation. At 15kHz (or ideally 20kHz) the feedback is doing its job less and less well, and the amplifier is struggling harder. If the amplifier continues to do extremely well, it is highly likely it will have no problems. And so it is, surprisingly, for the Cambridge.

Last but not least, Hypex NCx500:

Look at this! It's less clean at full power than the Cambridge at 15kHz! Eh, sort of, but not really. As power increases, the feedback loop does lose a bit of control. But not nearly as much as it looks like. We're starting from a shockingly low baseline of -90dB up to 20W at 15kHz. Nothing else ever even gets that low except the Benchmark. It's a clean amplifier, but on the 15kHZ full power metric, it's technically worse than the Cambridge. So, one number does not tell the whole story. You still need to read the whole chart, since the Hypex is cleaner, but it has nothing to do with it being cleaner on 1kHz SINAD. The Hypex is cleaner on the multitone, and it's cleaner on the CCIF twin tone than the Benchmark. And yes, it also "cheats" with low gain a little bit, but so does the Benchmark. Odds are, it would do that on other tests too since it is so incredibly clean. No 1kHz component on the multitone, and almost no sidebands on the twin tone. The old ones weren't nearly this good. The new NCx and Purifi amps are really quite amazing. But, so is the Cambridge for a really simple design with a linear supply. At some point, this story could continue with the new Topping amps, which in turn are able to absolutely decimate the Hypex amps. Just in case you were thinking the Hypex results had anything to do with it being Class D. They don't. Hypex does this despite being a Class D, not because it is.

Cambridge's merit here is that it does this all in an integrated amplifier, with a brand name, a real warranty, and a nice package. And once you get past the supply noise, it measures up really well. And you can see that from simple chart plotting SINAD at 15kHz.

Moral: You need to learn to read charts to understand what they mean, and what they show the amplifier is doing if you want to have a chance of sorting this all out. If your goal is finding an amp that is clean and the least likely to "misbehave" you cannot rely on any single number, particularly any single number taken at 1kHz. Usually, you can rely on 20kHz much more confidently (or 15kHz, if that's what you've got). But even then, it still take sorting and verifying.

I think the Multitone Test is even more representative. It not only shows the impact of distortion at different frequencies, but also IMD.

 

[Yevhen]

I've been torn about this amp and the reviews of it here. First, I GREATLY value and appreciate this community and Amir's work. ASR has become my go-to audio forum. Second, I own this amp so it's impossible to be completely unbiased about it.

IMO the tone of his verbiage in the first iteration of the CXA81 (first version) review came across as aggressively harsh, almost personal. His follow up that described interaction with Cambridge reps was really useful, but I have to believe that something about Cambridge Audio the company, or perhaps the tenor of interactions with them surrounding these reviews, colored his remarks about these amps. There is something about the way he writes about them that almost seems like he is gritting his teeth and struggling not to slam them and I don't understand it.

One thing to consider is that these amps can easily be found for far less than MSRP direct from Cambridge, new and with warranty, on their ebay store. I paid $800 for a new open box with new warranty. That's the same price that the Yamaha A-S701 and A-S801 models (I also own an A-S801) can be found for and I see the CXA81 as a competitor to those models. I like them both very much. The Yamahas are better value IMO and offer marginally more power and tone controls but otherwise are similar, but from what I've seen the Yamahas don't hold value very well because they are so ubiquitous.

Having said all of that, I wish I understood what appears to be a burr under the saddle of Amir when it comes to the CXA81 and why his reviews seem different from so many other reviews here. We'll probably never know but my interpretation is that there is something personal going on there, something like a harsh or nasty interaction he had with their reps that soured him on the company, perhaps well deserved. If Cambridge poisoned the well somehow I guess it's on them, but the CXA81 has served well in my home.

Did you have chance to listen to Hypex or Purify amp? How does it compare to Cambridge? I like the measurements of CXA81 at high frequencies, but in real life the high freq content like cymbals sound much cleaner with Purifi.

 

[ryanmh1]

Can you quote some of this stuff you feel so strongly about? Otherwise it's all rubbish.

I believe everything he said was qualified by "appears to be", "if" "came across as", "have to believe that", "seems like" and "my interpretation is". The reality is that most people cannot interpret amplifier measurements well, so they rely on the subjective commentary and the little "panther position". Given that this thing has a 25% "poor" rating, and 57% "not terrible", 17% great, and 1% excellent (at least one of which was mine), I don't think his subjective impressions of Amir's subjective commentary about his measurements were probably all that off the mark or unfair. My aim is just to provide some basic insight into what the other amplifier measurements mean and how to interpret them, without getting too technical.

What I've pointed out, I hope, is that the performance of the Cambridge as an amplifier actually is excellent, when judged by the amplifier measurements on the whole, and the fair criteria by which (I believe) a linear supply amplifier must be judged (i.e. not turning this into a SINAD/SNR drag race which means nothing in 95% of cases once the music starts). I could go on talking about the 19+20kHz THD (measured at Stereophile) and the multitone (measured here), and what all of that most likely means, and it would only paint it in an ever more favorable light. If you've read Douglas Self's amplifier books, you'll quickly understand just how many of the tricks they had to use which most companies just... don't. This thing is objectively one of the more distortion free amplifiers ever tested here (once you lump all the Hypex and Purifi products into iterations of a single product, which they basically are). My impression was also that it got dumped on. I'm still at a loss to understand why, unless one is trying to judge it not as an integrated amplifier, but as a DAC or a preamp, neither of which it is, nor will it be used as. I think that's what happened.

The only other explanation is a lack of understanding by Amir of Class A/B amplifer topologies and distortion mechanisms, which I don't think is the case. To a degree, I think it was the case some years ago when he regularly belittled 20kHz and CCIF tests, but he often does those now, so I assume he has come to understand what they mean and why they must be done. As I noted in another thread, Richard Cabot of Audio Precision has ever written an AES paper about this using intentionally defective circuits which well illustrate why SINAD is not a useful test to characterize its overall performance. That is why other tests much be done. Amir now does them. I don't think as much important is placed on them as it ought to be, but it's understandable not to want to erase the SINAD chart and start over, even if it does border on being useless compared to other metrics.

I think the Multitone Test is even more representative. It not only shows the impact of distortion at different frequencies, but also IMD.

Sort of. It shows it, but you do not need to do it to know that an amplifier does not distort. Since this test is done at low power, and uses so many signals, none of them individually push the amplifier that hard. It actually is not ideal for catching all possible causes of distortion (and hence, aggregate amplifier distortion on a musical signal). Ideally, you would use fewer tones and a slightly different mix to make it easier to read and interpret, slightly harder on the amp, and then take it at three power levels to characterize overall distortion at a variety of power levels. But that is perhaps asking a bit much. Instead, you can interpolate from how well the DUT does at low power on a multitone, and then take an educated guess as to how it will do at higher powers based on the 15kHz high power test, if performed. (Again, 20kHz would be better.) 20kHz (and to a less degere 15kHz) really is the best way to assure yourself an amplifier is clean.

For the Cambridge, that's always going to be really good because 15kHz is really good. As mentioned above, on Stereophile's CCIF test, it knocked it out of the park, and this at 25W. That was expected. There is essentially no 1kHz difference product and the sidebands are suppressed much more than in the bulk of reviewed products. That said, they could be lower, and if high frequency THD was lower, they probably would be. After further consideration, I don't think Cambridge got into things like using two pole compensation or other advanced methods to improve feedback at high frequency. It's still just a hint too dirty at 20kHz (Stereophile) or 15kHz (ASR). I think it's just an example of the Doug Self "blameless" design done right, with amp and preamp sections more than sufficient for what the amplifier can do, without playing with complex feedback schemes.

It gets a bit technical, but I do recommend the Cabot paper if you want to understand how to correlate various measurements with potential amplifier design issues: https://www.researchgate.net/public...n_of_Nonlinear_Distortion_Measurement_Methods. Note that there are two sections labeled "Output Nonlinearity". The second appears to be an editorial error, and should be labeled "Input Nonlinearity". The takeaway is that 20kHz distortion is the most stringent test--even more stringent than a multitone or "FASTest". If an amplifier passes that, it's pretty unlikely it will fail anything else, lending the conclusion here that the Cambridge is exemplary, hindered only by its noise performance vis-a-vis some switching supply amplifiers. In real life, though, this is just SNR, and it arguably only matters if you have highly sensitive speakers where you can hear the noise floor at the listening position.

Finally, note again that much of the above is not directly applicable to Class D amplifiers. I suspect the THD20 test still is the most difficult for most of them, but the mechanisms of what causes those distortions is going to be entirely different in many or most cases. And passing a THD20 test does not necessarily give the same "seal of approval" that it does for an AB amplifier, if your concern is maximizing technical performance. There are yet more papers about how to properly test Class D amplifiers for issues. You have to do some other tests, some of which have been discussed here, and some of which @pma has done with alarming results.

 

[Yevhen]

Ideally, you would use fewer tones and a slightly different mix to make it easier to read and interpret, slightly harder on the amp, and then take it at three power levels to characterize overall distortion at a variety of power levels. But that is perhaps asking a bit much. Instead, you can interpolate from how well the DUT does at low power on a multitone, and then take an educated guess as to how it will do at higher powers based on the 15kHz high power test, if performed. (Again, 20kHz would be better.) 20kHz (and to a less degere 15kHz) really is the best way to assure yourself an amplifier is clean.

I like your way of thinking! Is there a possibility to play with the phases of these multi tones to get lower peak-to-average ratio and drain more current from the amp, as you say, push it harder?

 

[Yevhen]

passing a THD20 test does not necessarily give the same "seal of approval" that it does for an AB amplifier

Purifi and Hypex have better multitone at 15kHz and I can definitely hear they play much cleaner in the that range. But for the rest you are right, most of the Texas Instruments based cheap class D amps measured excellent at 1kHz here don't have that overall feedback loop and, to my ears, only good for the mid-bass applications

 

[Doodski]

IMO the tone of his verbiage in the first iteration of the CXA81 (first version) review came across as aggressively harsh, almost personal. His follow up that described interaction with Cambridge reps was really useful, but I have to believe that something about Cambridge Audio the company, or perhaps the tenor of interactions with them surrounding these reviews, colored his remarks about these amps. There is something about the way he writes about them that almost seems like he is gritting his teeth and struggling not to slam them and I don't understand it.

I reread the review and I think you are being overly sensitive and making things up.

One thing to consider is that these amps can easily be found for far less than MSRP direct from Cambridge, new and with warranty, on their ebay store. I paid $800 for a new open box with new warranty. That's the same price that the Yamaha A-S701 and A-S801 models (I also own an A-S801) can be found for and I see the CXA81 as a competitor to those models. I like them both very much. The Yamahas are better value IMO and offer marginally more power and tone controls but otherwise are similar, but from what I've seen the Yamahas don't hold value very well because they are so ubiquitous.

So if you bought the Cambridge for $800 and they are available everyday all year long then it is worth maybe $500 on the used market if only a few months old.

Having said all of that, I wish I understood what appears to be a burr under the saddle of Amir when it comes to the CXA81 and why his reviews seem different from so many other reviews here. We'll probably never know but my interpretation is that there is something personal going on there, something like a harsh or nasty interaction he had with their reps that soured him on the company, perhaps well deserved. If Cambridge poisoned the well somehow I guess it's on them, but the CXA81 has served well in my home.

You come across like a fan boy who is hurt.

 

[ryanmh1]

I like your way of thinking! Is there a possibility to play with the phases of these multi tones to get lower peak-to-average ratio and drain more current from the amp, as you say, push it harder?

Possibly a good backstop. Unless you're trying to tease out what causes a particular distortion, in an AB amplifier you may as well just hit it with 20kHz (or 15kHz) to assure yourself there is nothing untoward going on. If it passes that, it probably passes everything else you can throw at it. Although, to be fair, one should also test 20Hz to stress the outputs and validate current delivery. (The AudioPrecision paper was about opamps, but the principles are largely cross-applicable.) ASR does a low end test now, so that base is covered.

Where this falls apart are two scenarios: 1) Class D (potentially); and 2) an amp that "fails" the 20kHz which most will if we set a bar of .03% for a passing score. The Cambridge is an exception, so that's going to be a lot of them. So, you might need a backup.

Once an amp barfs on the 20kHz, I can see some merit to a multitone with, say 30Hz, 3000kHz, 4000kHz and 19kHz at a 1:1:1:20 ratio, near full power. You would need to work out the frequencies better to avoid overlap, but those might work. 30Hz might not be necessary. I have zero evidence for it, but something tells me that would work pretty good for sorting them out without being as hard to interpret as a multitone where everything is buried in the dirt and you can hardly tell up from down and can only see "clean, not clean". You would have to play around with a bunch of amps with known issues and a spectrum analyzer to try to tease things out. There are a variety of papers out there with a whole bunch of theories about which multitone is best. I think Gene Czerwinski of Cerwin Vega and Alexander Voishvillo of Harman have a paper about this. Try this on for size if you have an afternoon to blow: https://www.researchgate.net/public...lts_and_Conclusions_Part_1_History_and_Theory.

 

[NoMoFoNo]

Can you quote some of this stuff you feel so strongly about? Otherwise it's all rubbish.

No I'm not going down that road here but my impressions remain. I respect Amir and this site greatly and if there is a conflict I will likely fall on the side of Amir in any event. You can call my impressions rubbish if you'd like. Bottom line to me is that the CXA81 at $800 is a pretty great value compared to similar products.

 

[NoMoFoNo]

I reread the review and I think you are being overly sensitive and making things up.

So if you bought the Cambridge for $800 and they are available everyday all year long then it is worth maybe $500 on the used market if only a few months old.

You come across like a fan boy who is hurt.

Please stop being rude. I'm not up for personal attacks from other posters here. Stop it NOW.

Edit: reported for personal attack.

 

[AdamG]

Please stop being rude. I'm not up for personal attacks from other posters here. Stop it NOW.

Edit: reported for personal attack.

You Sir started the personal attacks on Amir. Doodski responded in kind and finished the argument. Since you felt the need to announce publicly that you reported Doodski’s post to the Moderators. Allow me to provide an update on your report. Since its a public comment from you. Report Rejected for the reasons outlined above.

Have a great night Sir.

 

[Purité Audio]

Never felt the need to report a post before, you should be ashamed.
Keith