• 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!

Crown XLS2502 Stereo Amplifier Review

anmpr1

Major Contributor
Forum Donor
Joined
Oct 11, 2018
Messages
3,741
Likes
6,464
Wish they would have updated the FC spec for multi-channel but the manufacturers fought hard against it.
I'm not a home movie fan, but if I was, I'd think that these Crown amps (or others like it) would be the bees knees. I mean, when it's exploding all around, and you're trapped in the jungle with a Predator after you, and you're blasting away willy nilly with a mini-gun, well... you not only don't have time to bleed, but you're not worrying about power bandwidth and THD.

ventura.jpg
 

gattaca

Member
Joined
Nov 29, 2019
Messages
97
Likes
87
I have no problem with testing different classes of equipment with different standards and test. The write-up was very clear that this is not intended or recommended to be on somebody's hifi rack connected to a turntable. This is meant to power PA speakers and and would also do a great job with a subwoofer or could fill the role in a home theater for movies. It was not designed to have a great SINAD. I'm judging it for what it is.

... I would personally love to see some QSC amps get tested as Crown's quality control has gone really downhill. I've replaced four of their amps all less than six years old in the last year. None were abused. Half were hardly ever used. Every professional I've talked to have move to purchasing the QSC product lines. I'd send one to Amir if I didn't need them being used. Might still if an opportunity arises.

Please continue to tell me real world pertinent information on the products and tailor your tests to what product you are testing so long as it's clear in the write-up what the product is intended for and what it might be recommended for.

Very well stated! The XLS2502 should be a viable, powerful SUB-class amp with a good power envelope as long as we pay attention to it's boarder limits. That what Amir's testing found and he conveyed early in the report. Now, when we want to take that motor and run it differently from his recommendations - that's a different USE CASE. We need to be careful not to boil-the-ocean with specs and use-cases.

It is akin to engine manufacturer saying I have a "great engine for $500" And then you buy it for your built-to-order auto b/c you figure the engine will fit into your custom Honda but then you decide you'd rather use it in your custom hot-rod Mustang. Those are quite a different use cases. Could you find an poweramp for both probably if that's what you state you need.

I am also a bit cautious with watt-efficient, cost-optimized Class D's as the years have taught me well "we get what we pay for (99.9999%) most of the time." But sometimes you don't know until you try either and that's part of what Amir's is helping us decide. Crown, QSC, Crest and others have AMPS designed for many price points. Each version is compromise in cost, performance, reliability and serviceability.

Do you spend $350 on a used Crown XLS2502 and use it for 5-6 years and then trade it or sunset it or pitch it when it dies or do you spend $1500-$2000 for another AMP that might well last for 30 years? Many times, this is a personal call. How many times to do you want to tear out your AV setup and/or deal with warranty issues? Our time is more valuable than we often remember.

WRT Q/C of the Crown, the QSC RMX2450 units are often in the top 5 recommendations of "good" sub AMP too. In this case b/c of it's build (I read it can go down to 5Hz) but there's a host of others too: Crown K2 (no longer made), anything Powersoft, , Crown XLI3500, plus others.

I spent probably 40+ hours scouring the web for recommendations and info is starting to blur. WRT the Crown line, there are varying views and surprisingly the XLi3500 surfaces quite often for a good SUB amp. But as long as I say I'm looking for an amp which can drive from 5Hz - 200Hz motor loads (pistons, not speakers) @ 4 ohm from 400 (min) to 1000W (max) for mostly HT use - that's *@(@* specific use case. Now if I say I want to listen to classical 2 channel in the absolute "cleanest" environment I can afford, that's a totally different zebra / use case.

I've got some AMPs which are easily 25 years old. I have them re-capped every 15 or so years and they are GTG another cycle. I'm seriously considering just picking up some older amps and doing that for these loads b/c outside of a very small slice of AMPs, most cannot deal well with my "piston" use-case.

BTW, someone mentioned the Damping Factor maybe in this thread. I ran across an 25 year old AMP which stated it had a damping factor of 600.... nothing like that I have seen anywhere today. Have a great NYE and 2020!
 
Last edited:

gattaca

Member
Joined
Nov 29, 2019
Messages
97
Likes
87
Nice! 500W continuous! Yikes! As soon as I opened the page, I see custom engineered (made-in-house) torrid and huge caps leaped off the page! All high quality. It's a class AB, just like some of the other more-economical recommendations I've encountered for a sub-amps: the Crown XLi3500, QSC RMX2450. My gut says the Rotel is in the more extreme-end of the engineering, performance compared to these two priors. A common thread I see is these AB AMPS have huge/heavy torrid transformers, large caps to carry the load during high demand voltage rail sags or peaks. They have tons of headroom and higher than most damping factors.

This is one of those compromises designers make on the Cost, Performance, Reliability and Serviceability intersection. This is also why, from what I've read, Powersoft Amps are specifically suited for sub-duty. I'm guessing their engineer/designer optimized that end of the frequency spectrum and has much less concern about the upper frequency range. It's their compromise. Of course, that's also a more expensive amp too.

The Crown XLS2502 is quite an engineering feat. Just let's all understand the compromises they made and what that AMP is ans is not good for... When I open them I don't see a huge torrid or large bank of caps inside. ;) Peace.
 
Last edited:

Doodski

Grand Contributor
Forum Donor
Joined
Dec 9, 2019
Messages
21,801
Likes
22,063
Location
Canada
Nice! 500W continuous! Yikes! As soon as I opened the page, I see custom engineered (made-in-house) torrid and huge caps leaped off the page! All high quality. It's a class AB, just like some of the other more-economical recommendations I've encountered for a sub-amps: the Crown XLi3500, QSC RMX2450. My gut says the Rotel is in the more extreme-end of the engineering, performance compared to these two priors. A common thread I see is these AB AMPS have huge/heavy torrid transformers, large caps to carry the load during high demand voltage rail sags or peaks. They have tons of headroom and higher than most damping factors.

This is one of those compromises designers make on the Cost, Performance, Reliability and Serviceability intersection. This is also why, from what I've read, Powersoft Amps are specifically suited for sub-duty. I'm guessing their engineer/designer optimized that end of the frequency spectrum and has much less concern about the upper frequency range. It's their compromise. Of course, that's also a more expensive amp too.

The Crown XLS2502 is quite an engineering feat. Just let's all understand the compromises they made and what that AMP is ans is not good for... When I open them I don't see a huge torrid or large bank of caps inside. ;) Peace.
The large toroidal transformer and capacitors are for the low frequency 60 Hz power transfer. As the frequency goes up to say 50-60KHz the filter caps get smaller and the toroidal transformer gets smaller too. They are inversely proportional. For the switching power supplies.
 

RayDunzl

Grand Contributor
Central Scrutinizer
Joined
Mar 9, 2016
Messages
13,257
Likes
17,249
Location
Riverview FL
It's really a no-brainer. I'd personally rather use the money for something more useful like speakers. Or a coilover suspension for my car. Or a vacation.

View attachment 44145

I suppose the spectrum supplied was created with an amplifier with flat frequency response.

Therefore, wouldn't a rolled off amp roll that more?

I don't see how the amp with the big rolloff would retain the spectrum as shown, without some EQ applied to bring up the highs..
 

digicidal

Major Contributor
Joined
Jul 6, 2019
Messages
1,985
Likes
4,846
Location
Sin City, NV
This stuff will make you a sexual tyrannosaurus!
I bet it would bring the percentage of "working every time" up to at least 80-90% over it's competitor:
 

Francis Vaughan

Addicted to Fun and Learning
Forum Donor
Joined
Dec 6, 2018
Messages
933
Likes
4,700
Location
Adelaide Australia
Nobody is rolling the frequency response off. See post 96
The question is about the bandwidth used to measure distortion products. Power delivery remains solid and flat 20-20k. But if the amp cannot be sensibly measured beyond 20kHz it makes no sense to look for harmonic distortion products above 20kHz.

Lets unwind what is going on. The issue is noise shaping. There is no difference in how the amp is behaving and being measured than any part of a digital recording chain that uses noise shaping. Which is all of them. The underlying mathematics is intrinsic to the nature of the universe we live in and is quite beautiful. The bottom line comes from Shannon and information theory via Fourier theory. With a given bandwidth channel and a given signal to noise in that channel, there is a fixed amount of information you can transfer. This is true in the digital domain or the analog domain, indeed they are just specific implementations of the general case. We have a target channel, the human auditory system. The bandwidth is at best 20-20kHz, and the signal to noise (aka dynamic range) varies across the channel, being much worse at the ends than the middle. If we can keep our distortion products outside of the range of the auditory channel, they will be impossible to detect. Simple. With conventional (continuous, unquantised - aka analog) system we try hard to keep distortion products down in amplitude, so they drop below the threshold of detectability. Harmonics resulting from higher frequencies escape our notice by falling outside the bandwidth of the channel. Which makes worrying about harmonic distortion products of anything about 10kHz meaningless.

But we can do better than this. Much better.
What if our amplifier is operating well past the bandwidth of the human auditory channel? Well any distortion products we make that can be placed outside that channel are not detectable. This leads to the notion of noise shaping. Here noise is any unwanted signal, including distortion products.
It is possible to design a PWM (aka class D) system so that distortion (noise) is shaped in its spectral distribution such that it is placed outside of the human channel's frequency range of detection. The well known idea of shaped dithering in the audio chain is one example of this. It moves distortion (quantisation noise in this case) in the mid-bands where the ear is very sensitive, say 105db or more, to the higher frequencies where the ear is much less sensitive. No free lunch, but all that space in the high frequencies where we have poor sensitivity is just going to waste, so we use it. In a class D amplifier, there is space above 20kHz that the distortion (noise) can be shaped into. So that is done. And done to great effect.

But it now makes no sense to go looking for harmonics of in-band signals above 20kHz. Why? Because the amplifier has deliberately filled that space with all the shaped noise that allowed it to get such good in-band performance. This is a technique unavailable to the designer of a conventional (ie class A/AB) amplifier. There going looking above 20kHz can tell you something interesting about in-band operation. But once the designer is able to exploit noise shaping, looking for harmonics of in-band signal out of band is simply meaningless. What you see out there tells you very different things about the in-band performance.

Bottom line. Limiting the bandwidth of the measurements (not the amplifier implementation) of harmonic distortion products is done simply because not doing so is mathematically invalid.

It was just a matter of luck that making measurements of harmonic products above 20kHz was useful with conventional amplifiers. And we got used to the idea. It provided insights into in-band operation, and thus was useful. But demanding that all amplifiers behave this way is mathematically indefensible. There is bandwidth to be exploited to make the operation where it matters better, so we use it.
 
Last edited:

restorer-john

Grand Contributor
Joined
Mar 1, 2018
Messages
12,822
Likes
39,338
Location
Gold Coast, Queensland, Australia
Bottom line. Limiting the bandwidth of the measurements (not the amplifier implementation) of harmonic distortion products is done simply because not doing so is mathematically invalid.

It was just a matter of luck that making measurements of harmonic products above 20kHz was useful with conventional amplifiers. And we got used to the idea. It provided insights into in-band operation, and thus was useful. But demanding that all amplifiers behave this way is mathematically indefensible. There is bandwidth to be exploited to make the operation where it matters better, so we use it.

Not a matter of luck, but a matter of deliberate design and method based in science. Do you think THD analyzers and meters settled on 80-100KHz bandwidths through sheer luck?

Perfectly defensible to demand harmonic distortion metrics for all frequencies up to and including the uppermost specified frequency at the claimed continuous power output of ANY amplifier, regardless of topology. Indeed, it is a requirement. To suggest otherwise is farcical. High power inter-modulation 19/20KHz only inspected within the audio bandwidth leaves all manner of variable sins out of band, to wreak havoc.

Burying noise and switching products out of audible band is one thing, expecting (or demanding) the mess be ignored altogether, is another.
 

Francis Vaughan

Addicted to Fun and Learning
Forum Donor
Joined
Dec 6, 2018
Messages
933
Likes
4,700
Location
Adelaide Australia
Perfectly defensible to demand harmonic distortion metrics for all frequencies up to and including the uppermost specified frequency at the claimed continuous power output of ANY amplifier, regardless of topology.

It is mathematically useless to make this demand. There is no justification for demanding that harmonics that fall outside of the audible band when analysing the output of a power amplifier be explicitly singled out. These are not metrics of audible phenomena, and that must be the bottom line.

Indeed, it is a requirement.
Says who? Cite?

There are a lot of insights that can be gleaned from looking at out of band harmonic distortion products for amplifiers. They can be indicative of specific problems in the design. Which is good reason to have wide bandwidth analysers. But that is as far as it goes. Modern Class D amplifiers operate in a fundamentally different manner, and the out of band energy contains stuff of a different nature. Energy that is found at a harmonic frequency could be there for a whole range of reasons, and there is no reason to explicitly single it out. There is a wide band of energy - just look at some of the plots above - and all of that energy is there for a reason, not just the stuff at the harmonics. And none of it is audible, because it is outside of the range of human perception. Singling out energy at the harmonics misses the point of how the amplifier operates. You cannot directly apply ideas derived from simple conventional amplifiers in this domain. Simulating the energy spectrum expected from the noise shaping used and checking that it matches the measured spectrum well might be a good design goal during development.

Sure if this was the output of a DAC, and was fed unfiltered into an amplifier with dubious stability or other design problems, we might want to worry. We would add appropriate band limiting (aka the reconstruction filter used on any DAC.) But we don't have this problem driving speakers, so we avoid any narrow band reconstruction filter. The energy out of band doesn't wreak havoc anywhere. You need to show how and why it would. Numbers would help.

Unless you can show how this out of band energy can result in unwanted in-band energy that is audible, there is no justification for making any demands upon it.
 

sergeauckland

Major Contributor
Forum Donor
Joined
Mar 16, 2016
Messages
3,472
Likes
9,209
Location
Suffolk UK
When measuring Distortion Factor using a conventional Distortion Factor Meter with a 100kHz bandwidth, any ultrasonic noise pretty much invalidates measurements as the meter can't distinguish distortion harmonics from noise as it nulls the fundamental, and measures the amplitude of everything else. In older AB amplifier, that results in reasonable readings as noise should be sufficiently low and distortion higher than the noise, at least at higher powers. With Class D amplifiers or indeed most DACs, ultrasonic noise makes my DFM useless so I end up using FFT and band-limiting the results to remove the effect of noise.

Neither is better, but we need to understand the differences, and which one is more appropriate for the circumstances.

S
 

gattaca

Member
Joined
Nov 29, 2019
Messages
97
Likes
87
I think I actually understand this discussion - maybe in laymen's terms. Two sides WRT the XLS/Class D/ ClassA/SB/ and general measurements:

a) "Because the amplifier (this XLS Class D (my insertion)) has deliberately filled that space with all the shaped noise that allowed it to get such good in-band performance." Is saying that the designers shifted the "Class D switching noise" from the human audible band, which we standardized as: 20Hz-20kHz into higher bands as part their compromise to clean up that range. So since understand they did that, we should expect to see all sorts of "switching artifacts" here so it is WAD with these compromises.

b) "What you see out there tells you very different things about the in-band performance." +... "Do you think THD analyzers and meters settled on 80-100KHz bandwidths through sheer luck? " Examining the wider human inaudible ranges may indicate some of the compromises designers made to cost, performance, reliability and serviceability in the human hearing critical 20Hz-20kHz range.

"This is a technique unavailable to the designer of a conventional (ie class A/AB) amplifier. There going looking above 20kHz can tell you something interesting about in-band operation. But once the designer is able to exploit noise shaping, looking for harmonics of in-band signal out of band is simply meaningless."

That's exactly what is being shown here (no Class D noise...) ..a fascinating read - seldom can one see the exact same amp in A and AB mode.. -> https://archimago.blogspot.com/2019/12/measurements-emotiva-xpa-1l-gen-1-class.html

archimago-emotiva-ab-result01.jpg
Peace.
 

DonH56

Master Contributor
Technical Expert
Forum Donor
Joined
Mar 15, 2016
Messages
7,941
Likes
16,779
Location
Monument, CO
Extra bandwidth, to several hundred kHz or so, is routinely required to meet in-band flatness requirements if you want to achieve 20 kHz within a fraction of a dB. If you spec 20 kHz at -3 dB then bandwidth requirements are greatly relaxed. But extra bandwidth is usually required to achieve low in-band distortion, and of course loop bandwidth must be quite broad. All of which leads to a desire (by me and John alone, apparently) to characterize the amplifier up to and beyond 20 kHz to at least some degree, such as low-power 1kHz (or whatever) square waves with fast edges and full-power tests across the stated power bandwidth. LF square waves can also reveal thermal tails and such (like long settlers from doublets) that may not be obvious from steady-state sine-wave testing.

FWIWFM - Don
 

LTig

Master Contributor
Forum Donor
Joined
Feb 27, 2019
Messages
5,883
Likes
9,668
Location
Europe
Extra bandwidth, to several hundred kHz or so, is routinely required to meet in-band flatness requirements if you want to achieve 20 kHz within a fraction of a dB. If you spec 20 kHz at -3 dB then bandwidth requirements are greatly relaxed. But extra bandwidth is usually required to achieve low in-band distortion, and of course loop bandwidth must be quite broad. All of which leads to a desire (by me and John alone, apparently) to characterize the amplifier up to and beyond 20 kHz to at least some degree, such as low-power 1kHz (or whatever) square waves with fast edges and full-power tests across the stated power bandwidth. LF square waves can also reveal thermal tails and such (like long settlers from doublets) that may not be obvious from steady-state sine-wave testing.

FWIWFM - Don
That's certainly correct for class A/AB amps which exhibit a kind of 1st order lowpass behaviour where you need much higher bandwidth to not have -3dB at 20 kHz.

In the case of the XLS2502 this is different. Looking at @amirm's FR plot we have some -0.2 dB at 20 kHz and above 20 kHz FR drops like a stone. This is certainly the behaviour of the ADC and its anti alias filter. Therefore it's impossible to gain any insights into the behaviour of the embedded power amp stage at frequencies above 20 kHz, regardless of which tests signals you use at all. For this you have to open the housing, cut the connection between DAC and power amp stage and insert test signals direct into the power amp stage.
 

Panelhead

Senior Member
Forum Donor
Joined
Apr 3, 2018
Messages
348
Likes
137
I get a headache sometimes when electronics emit high frequency noise. Cannot hear it, Just get a dull headache quickly.
I suspect some tweeters can output these inaudible tones.
Built a current amplifying phono stage one time. It was an ultra sonic oscillator. Instant headache. This was with amps rated -3 dB it 60khz.
 

Panelhead

Senior Member
Forum Donor
Joined
Apr 3, 2018
Messages
348
Likes
137
That's certainly correct for class A/AB amps which exhibit a kind of 1st order lowpass behaviour where you need much higher bandwidth to not have -3dB at 20 kHz.

In the case of the XLS2502 this is different. Looking at @amirm's FR plot we have some -0.2 dB at 20 kHz and above 20 kHz FR drops like a stone. This is certainly the behaviour of the ADC and its anti alias filter. Therefore it's impossible to gain any insights into the behaviour of the embedded power amp stage at frequencies above 20 kHz, regardless of which tests signals you use at all. For this you have to open the housing, cut the connection between DAC and power amp stage and insert test signals direct into the power amp stage.

Some of this “hash” is generated after the ADC. The output stage artifacts. The output filter is designed to remove these.
Some implementations are more successful than others.
 

Xulonn

Major Contributor
Forum Donor
Joined
Jun 27, 2018
Messages
1,828
Likes
6,316
Location
Boquete, Chiriqui, Panama
I see no reason not to test all amps consistently and thoroughly.

Perhaps after Amir catches up on his backlog :rolleyes:, he will devote more time to measurements that are intersting from an engineering perspective and possibly useful to pro audio extreme applications, but unrelated to real-world home audio performance. :cool:
 

LTig

Master Contributor
Forum Donor
Joined
Feb 27, 2019
Messages
5,883
Likes
9,668
Location
Europe
Some of this “hash” is generated after the ADC. The output stage artifacts. The output filter is designed to remove these.
Some implementations are more successful than others.
Yes of course, and I have never written that one should not measure above 20 kHz at the output of a power amp - far from it. It just makes no sense at all to feed test signals with content above 20 kHz because they will never reach the power amp stage.
 

stunta

Major Contributor
Forum Donor
Joined
Jan 1, 2018
Messages
1,156
Likes
1,407
Location
Boston, MA
Besides, your speakers will distort more than the amps will.

I hear this a lot. Does speaker distortion mask the amp distortion or is it additive?
 
Top Bottom