• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required as is 20 years of participation in forums (not all true). There are daily reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Cambridge Audio CXA81 Review (Sample 2)

KSTR

Major Contributor
Joined
Sep 6, 2018
Messages
2,153
Likes
4,729
Location
Berlin, Germany
@amirm,
Thanks for this follow-up review and nice to see you unmistakingly nailed the core issue of this product: The device quite likely pollutes its own audio signal from suboptimal internal layout and cabling.

Some of the balancing current that enters at USB (and other digital connectors) shares a ground path with the audio GND, flowing right across the main analog PCB instead of being diverted away from the analog section and its ground reference.

Internal view suggests it's even simpler and the DAC chip is on the digital input PCB (the three SO8-chips -- DualOpAmps -- seem to indicate this), creating an unbalanced local output signal which they tried to bring down to the main PCB with a short cable connector. If the GND of the cable carries any other current than signal return then this current impresses the error voltage.

cambridge-audio-cxa81-innen.jpg


It's very easy to get Signal integrity wrong in a complex product like we have here.
 

DMill

Active Member
Forum Donor
Joined
Jan 9, 2022
Messages
295
Likes
344
Like many current integrated amps, I think Cambridge has tried to boil the ocean by adding too many features into a $1500 box. It looks good to many consumers because it’s a Swiss Army knife marketed to the mid fi market and for many it is good enough. I have many friends who ONLY stream music with devices like this and it leaves me scratching my head why they pay for things they don’t use. They are very marketable to the uniformed because most don’t realize there are compromises to do it all solutions at a fixed price.
 

Talisman

Active Member
Forum Donor
Joined
Mar 27, 2022
Messages
123
Likes
241
Like many current integrated amps, I think Cambridge has tried to boil the ocean by adding too many features into a $1500 box. It looks good to many consumers because it’s a Swiss Army knife marketed to the mid fi market and for many it is good enough. I have many friends who ONLY stream music with devices like this and it leaves me scratching my head why they pay for things they don’t use. They are very marketable to the uniformed because most don’t realize there are compromises to do it all solutions at a fixed price.
But maybe many people understand very well and prefer to have a single complete instrument, with many inputs, many outputs, an integrated bluetooth, a beautiful aesthetic, a single remote control etc. etc. instead of having 22 boxes connected by 400 different cables, which maybe do better on the paper in the measurements but then little or no difference is heard when simply listening to music.
It doesn't seem so absurd to me, everyone will have his own tastes and preferences, I don't see why denigrate them
 

Cambridge Audio

New Member
Manufacturer
Joined
May 19, 2022
Messages
1
Likes
40
Dear ASR community

Firstly, thank you to Amir for listening to us and agreeing to do a re-test of our CXA81.

When the original test was published a couple of months ago we were part devastated and part confused. When his first review landed Amir didn’t know us as a company, and his findings and suggestions made Cambridge Audio come across as a "quick and dirty" company that knocks out products designed by people who don’t know what they are doing. Nothing could be further from the truth. The engineers are responsible for every PCB, line of software code and product design detail in our CXA81 are all UK-based and pride themselves on their craft, taking personal responsibility for their work, individually and as a team. They all work on products across our range, including several that have won great praise in previous tests by Amir and others in the ASR community, and they are fastidious – some would say fanatical – about what they do.

If we were a bigger, more “corporate” business it would be normal to pass the job of handling this off to someone in a comms or marketing department, however Cambridge Audio, and I, aren’t like that. My team and I have a deep connection with our business, our brand and the products we design. Our products take years to create and we take great pride in what we do. They are not churned out in a cookie-cutter process that lacks heart, we consider every detail end to end and spend many months hand picking components, comparing to competitor products, tuning, fine tuning and listening. So we were naturally upset to read Amir’s original review, knowing that something was very wrong.

Of course we’re only human, so let’s not pretend we’re perfect and don’t drop the odd ball, but with CXA81 I’ll do my best today to reassure you we haven’t. When it comes to our product, we really do try hard to make it as good as we possibly can, and whatever you may have thought about us reading Amir’s previous assessment, please give us a second chance and get to know us better by reading both this and Amir’s second test.

So how did this go so wrong the first time round, and why haven’t we got involved in the discussion here sooner?

Firstly, we do not believe there was anything at fault with the first unit on test. The distortion issues reported in the original test are clearly down to one main reason.
  • Noise introduced by external devices in the test setup (we have tested over 20 units ourselves, and have replicated how this could have happened, with data to back it up). To elaborate on this in more detail, we have found that some computers create a noise differential that conducts noise between USB port grounds, the computer upstream connected to CXA81 via USB input, the APx balanced outputs upstream connected to CXA81 balanced input via XLR cables, and the APx connected to the computer via USB, a copper conducted audio noise loop is created and can add approximately 20db of low frequency noise.

CXA81_computer_noise_connections_09-06-2022.png


Only a few types of products with such a feature set can be connected in such a way (which eliminates most standalone DAC products). This particular low frequency noise is only added to the audio due to the addition of the test/measurement equipment that would never be seen by any normal user set up.

Also, after extensive testing on our 20+ different CXA81’s, we can confidently say that the extra noise is added through the test set up and not created by the product. It is not possible for the CXA81 to create the low frequency noise shape internally. Even if there was the potential for there to be burst noise from a faulty component in the CXA81 (very unlikely in any case), the low frequency noise shape would be very different between the left and right channels. Also, there is no capacitance or mutual inductance large enough in the CXA81 or the setup for the low frequency noise to couple into the CXA81. The low frequency noise can only be conducted through cables in the setup.

  • We also noticed that the correct USB Cambridge Audio driver was not being used and no effort was made to improve the performance by changing the grounding switch on the rear panel. Although, this had no bearing on the addition of noise.
It serves no purpose to repeat Amir’s own words on this, which I believe largely vindicate the above.

Another area of the first review relates to distortion measurements taken when a 2V signal is taken from the pre-out unbalanced RCA connectors. 2V is Amir’s test standard and may be the line-level standard, but it is not a level that is typical of that used when such pre-out sockets are connected to one of our power amplifiers, or another in a similar class. Amir as we all know is effectively testing CXA81 as a stand-alone DAC, whereas we all know it is designed to be used as an integrated amp which includes an attenuation stage in its signal path. The result is that Amir pushed the level from these jacks up to one higher than it was designed for. When, at our request, he subsequently tested output at 1.5V (a perfectly adequate and ideal level to drive one of our power amps, or those by other brands) the THD is at a level you'd expect from a quality brand like Cambridge Audio.

Those with a CXA81 themselves, or who may considering buying a CXA81 or anything Cambridge Audio, should be reassured that in a normal hi-fi system or set up, whatever showed on those original graphs will not be manifesting when playing music.

We know some of you who commented in the original review post were dismissive of the positive magazine reviews we have received as a result of Amir's original "findings". All I can say is that those posting those negative comments did so on the basis of what they read and not from first-hand experience listening and testing a CXA81 in front of them. Hopefully Amir's own words, his re-test and my message here convinced those now that this product is not the duffer you were led to believe. Jon Atkinson’s Stereophile review also backs this up.

Amir was also critical about some of the things we did with user experience, as well as some of the words used on our website. Small stuff, but we sweat the small stuff, so let’s cover those off too, and not just sweep them under the carpet..

The UX comments surprised us as no one before has given us such feedback (and we’ve enjoyed a lot of reviews). But we respect his opinion and are never too big or proud to take on board feedback, to learn how we can do better. We consciously haven’t put a display on the front of CXA81 because we simply don’t see a need, or a real customer benefit given this is a high-performance value-orientated integrated amp with an onboard DAC, not a standalone DAC. We also wanted to keep button count down, thus dual-purposing a couple of input buttons. Roon labelling follows Roon’s rules which we have little/no influence over.

Amir picked up on the statement that claimed CXA81 was “designed and specified without compromise”, of course at this price level there has to be some element of compromise, so we’ve amended these words.

On the latter point (why didn't we get involved in the conversation on the previous test comments/forum), we took an initial decision to see how things panned out whilst also fully investigating and verifying internally what we believed had gone wrong. I reached out to Amir and over a number of days we saw a meeting of minds and a mutual respect. But the whole process took quite some days and by that time what was being said on the forum was hard to respond to and frankly toxic at times, making it hard for us to respond without Amir or a third party test company verifying our findings. Amir agreed to listen to us and re-test, and myself and the engineers who designed CXA81 had a Zoom call with Amir to talk things over and get to the root of the problems experienced by Amir.

I hope this helps, and shows you the care and attention we at Cambridge Audio put into making genuine high quality audio products that are designed to offer value and performance and delight our listeners all over the world.


Best regards

James Johnson-Flint
CEO
Cambridge Audio
 

Somafunk

Addicted to Fun and Learning
Joined
Mar 1, 2021
Messages
588
Likes
1,236
Location
Scotland
I hope this helps, and shows you the care and attention we at Cambridge Audio put into making genuine high quality audio products that are designed to offer value and performance and delight our listeners all over the world.

Well said and thanks for taking the time to respond.

(I’m somewhat biased as a very satisfied user of cxn v2)
 

Ajax

Member
Forum Donor
Joined
Aug 14, 2018
Messages
92
Likes
233
Location
Byron Bay, Australia
I must say i never understood this hypex hype. I've never heard a nice (not good), nice hyper amp by it's own. They all sound like crap even if they porvide twice as much power for the same price. Unless u put a nice (maybe tube) pre/power amp , which again will add cost. Hence why you would see most expensive Hyper implementations doing so. So if i had to choose between twice more powerfull SOTA lifeless crap , or less powerful nice sounding OK amp, guess what ? ;)
So SOTA transparent power amplifiers, designed to amplify a signal without adding distortion or noise, have their own sound?
 

jasonhanjk

Active Member
Joined
Mar 24, 2019
Messages
186
Likes
144
Also, after extensive testing on our 20+ different CXA81’s, we can confidently say that the extra noise is added through the test set up and not created by the product.
That's not a correct statement.
If uses a PC that's grounded and an external amplifier that's also grounded; you would see the correct measurement data as what Amir sees.

This product is good when using it's internal amp and bad when using it's analog out.

index.php
 

JSmith

Major Contributor
Joined
Feb 8, 2021
Messages
2,657
Likes
5,814
Location
Algol Perseus
Some of the balancing current that enters at USB (and other digital connectors) shares a ground path with the audio GND, flowing right across the main analog PCB instead of being diverted away from the analog section and its ground reference.

Internal view suggests it's even simpler and the DAC chip is on the digital input PCB (the three SO8-chips -- DualOpAmps -- seem to indicate this), creating an unbalanced local output signal which they tried to bring down to the main PCB with a short cable connector. If the GND of the cable carries any other current than signal return then this current impresses the error voltage.
I would suggest that every class I audio product with USB input should have the USB link galvanically isolated.
Also, after extensive testing on our 20+ different CXA81’s, we can confidently say that the extra noise is added through the test set up and not created by the product.
Thanks for posting here. I'm interested in your comments in relation to the above quotes.
When his first review landed Amir didn’t know us as a company, and his findings and suggestions made Cambridge Audio come across as a "quick and dirty" company that knocks out products designed by people who don’t know what they are doing.
FWIR Amir had already tested some of your companies other products, which were fairly positive. I didn't find the tone of the original review as you describe, moreso disappointment that a good company had an possible issue with a popular product;
I have had a very positive impression of Cambridge Audio products up to now.


JSmith
 
OP
amirm

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
38,643
Likes
168,954
Location
Seattle Area
To avoid damaging the Cambridge's power amplifier stage with high-level digital signals, I performed all the measurements of the digital inputs' performance at the headphone output; inserting the headphone plug into the front-panel jack mutes the preamplifier and loudspeaker outputs.

A proper test
SoundStage! Measurements - Luxman L-509U Integrated Amplifier (2/2009) Following the introduction of our highly successful loudspeaker-testing program with Canada's National Research Council (NRC), in 2001 the SoundStage! Network began its amplifier-testing program. Originally, tests were done by BHK Labs, headed by Bascom King, well-known design engineer, consultant, and technical writer for such magazines as the former Audio. BHK Labs tested amplifiers using the Audio Precision System Two.



Food for thought
This was addressed in the last review thread. That is, no one will use the headphone output to power an external amp. The RCA out is what it is for. That output must be independent of the internal amp when it is not being used. Otherwise it makes no sense to have pre-out in the first place (if it is not to have a more powerful external amp).
 

diddley

Senior Member
Joined
Nov 4, 2020
Messages
401
Likes
690
Location
The Netherlands
I hope this helps, and shows you the care and attention we at Cambridge Audio put into making genuine high quality audio products that are designed to offer value and performance and delight our listeners all over the world.
It certainly does and thank you for your letter in this forum showing that you care about your products aswell of the community here.
Great job!
satisfied customer
 

pma

Major Contributor
Joined
Feb 23, 2019
Messages
3,069
Likes
6,210
Location
Prague
Please let me demonstrate the issue that is created by USB loop between 2 devices on the same PC and how it is fixed by USB isolator. It is a similar issue that was seen here in the Cambridge review and the issue that many users are not aware of if they connect class I PC to class I DAC via USB. In case you see any noise floor modulation and strange spectrum components, raise your eyebrows and ask how it was measured.
 

Attachments

  • Preamp_loop_with_usbiso.png
    Preamp_loop_with_usbiso.png
    57 KB · Views: 69
  • Preamp_loop_without_usbiso.png
    Preamp_loop_without_usbiso.png
    64.5 KB · Views: 68

KSTR

Major Contributor
Joined
Sep 6, 2018
Messages
2,153
Likes
4,729
Location
Berlin, Germany
Firstly, we do not believe there was anything at fault with the first unit on test. The distortion issues reported in the original test are clearly down to one main reason.
  • Noise introduced by external devices in the test setup (we have tested over 20 units ourselves, and have replicated how this could have happened, with data to back it up). To elaborate on this in more detail, we have found that some computers create a noise differential that conducts noise between USB port grounds, the computer upstream connected to CXA81 via USB input, the APx balanced outputs upstream connected to CXA81 balanced input via XLR cables, and the APx connected to the computer via USB, a copper conducted audio noise loop is created and can add approximately 20db of low frequency noise.

CXA81_computer_noise_connections_09-06-2022.png
Thank for registering and answering here, much appreciated.

Let me add a few comments on the above:

I would think this is a scenario that is very common in practice. A PE-grounded analog source and a PE-grounded digital source, with a noise potential between those PE-grounds (very easy to have when one of the PE-grounds is the USB-ground from a desktop computer. This will give rise to really significant "balancing" current flowing through the Amplifier as the noise source is low impedance.
Note that the problem should still exists even if the AP was not connected via USB because the XLR-Pin1 and Shell of its analog inputs is still at the same PE-potential as before, the AP's main connection, whereas the USB input ground comes from the computer's (usually lousy) PE-ground at an USB socket.

Even with no PE involved at all the cross current still could exist just in the way you describe it. An USB DAC on one port of a floating laptop and the amp's USB input connected to the same laptop at a different socket with a noise potential between the two ports. Less likely a scenario in practice but not impossible.

Finally, even just the USB connection alone is prone to develop the error as the balancing current is internally flowing to the amp's PE (where the USB "GND lift" may actually help when the noise source impedance is low enough that 100 Ohms in series reduce the current substantially). It will flow along a different path than in our case but it still could affect internal ground references enough to create visible changes in the noise spectrum).

The highlighted part is the key point. Ideally, any device should be as immune as possible to these cross-currents by virtue of the internal grounding scheme and circuit design. The outside loop will still exist and may manifest itself if unbalanced cables are used as the shield impedance allows a voltage drop to be developed just in the same way the internal grounding scheme allowed that cross current to develop a GND noise voltage. But the loop path inside the device is made irrelevant, that's the only thing a manufacturer can do, anyway.

So what should have been done during development and testing of such a device to identify this Signal Integrity issue early?
Measure a set of Noise Transfer Impedances (very similar to what you do when measuring shield effectiveness of an unbalanced cable):
You inject a current between two points of the amplifier that are grounds which in some way are connected together, nominally. For example, from USB-Ground to Main Ground (the PE pin of the connector). Or, from one analog input ground to another analog input ground (RCA shells or XLR Pin1&Shell). Or, in this case here, from analog input ground to USB ground.
You sweep this (constant) current vs frequency and measure the voltage developed at any of the existing analog outputs (speaker output in this case), which means this is equivalent to an impedance. Ideally there should be no effect of cross currents at all. The moment you see something you know the internal grounding scheme is not perfect and voltage drops along the ground find its way into the signal.

Once this is working as intended, you also don't need that "GND lift" on USB input anymore.

@Amir and @pma: I'll be writing a detailed article about the technique which can implemented with an AP or with a versatile audio device like the RME ADI-2 Pro.

-----

Having said all this I would add that while the problem exists it is not a deal breaker in my view. Measured performance suffered a bit but quite likely this is completely inaudible in most use cases.
 
Last edited:

Koeitje

Major Contributor
Joined
Oct 10, 2019
Messages
1,784
Likes
2,868
Nice to see CA respond. I also agree with most here that these issues are probably not an issue in practice. But I do want to ask how much harder it is to make a device reach 2v on the pre-outs? Can't really be that hard to make it not distort right if you do need that 2v?
 

sarumbear

Major Contributor
Forum Donor
Joined
Aug 15, 2020
Messages
4,445
Likes
4,007
Location
Southampton, UK
Another area of the first review relates to distortion measurements taken when a 2V signal is taken from the pre-out unbalanced RCA connectors. 2V is Amir’s test standard and may be the line-level standard, but it is not a level that is typical of that used when such pre-out sockets are connected to one of our power amplifiers, or another in a similar class. Amir as we all know is effectively testing CXA81 as a stand-alone DAC, whereas we all know it is designed to be used as an integrated amp which includes an attenuation stage in its signal path.
Nice to see CA respond. I also agree with most here that these issues are probably not an issue in practice. But I do want to ask how much harder it is to make a device reach 2v on the pre-outs? Can't really be that hard to make it not distort right if you do need that 2v?
Especially when their first digital product, the 2-box CD player CD1 was capable of output more than 4V.

Why would you disagree that a pre-out should have half the signal level capacity of your flagship product back in the 80s. If anything signal levels have generally increased since then.
 

Snoopy

Addicted to Fun and Learning
Joined
Jul 19, 2021
Messages
663
Likes
338
It certainly does and thank you for your letter in this forum showing that you care about your products aswell of the community here.
Great job!
satisfied customer

I have no issues either with my CXA81 & CXN V2.
Works flawlessly with roon, the trigger input starts the amp and Iove the old school "clicking" sound when the thing turns on.

Their class D amp is a interesting all In one product as well.

Looks like the CXA81 works within specification and people are going nuts again over stuff that isn't even audible or could be easily fixed with usb isolator.

But the DAC has always been just a gimmick for me. It's nice to have if you ever need one temporary. Similar to these cheap phono pre-amps or headphone amps in integrated amps and receivers.
 

laudio

Active Member
Joined
Dec 30, 2019
Messages
245
Likes
221
The issue was clearly explained as having ground noise between 2 USB ports on some computers as part of the test setup. All of the other scenarios saying you need a USB isolator for normal use are not applicable. The test setup caused the problem, not the DUT.
 
Top Bottom