“Let our rigorous testing and reviews be your guidelines to A/V equipment – not marketing slogans”
Facebook Youtube Twitter instagram pinterest

Cherry Audio MEGAschino Mk2 Amplifier Measurements & Conclusion

By

Cherry MEGA MK2 #20531 Test Results

Measuring the Cherry MEGAschino amplifier turned out to be a tough proposition. In theory, the amplifier is capable of close to 1000 watts per channel into 4 ohms RMS and sustained output well past 600 watts. Yet it also produces distortion levels, at more reasonable power levels, that test the limit of my equipment. The combined need for low noise and distortion with very high power handling made testing difficult, as the gear I have just isn’t good for that. Add that it is a Class D amp and without the Class D measurement filter, distortion and noise rise as a result of the effects of ultrasonics feeding back into the audio band and potentially slewing the input opamps.  I’ve included a factory measurement of the exact amplifier under review showing a sweep into both 4 and 8 ohms of THD+N.  Since the max power that this produced exceeded my measurement limits at the time of this review, the factory measurements were confirmed at lower levels and included here for reference.

The main test rig consists of a Quantum Asylum QA401 & QA451 with external resistive load build using ARCOS LPR100 resistors. The test rig is really designed to test no more than around 175 watts into 8 ohms and 300 watts into 4 ohms, but with additional attenuation, can measure any power level necessary, with the negative side effect of increasing noise floor with added attenuation. As such, a decision was made to focus on the frequency response and noise distortion at set power levels within the linear limits of the testing apparatus. In addition, the manufacturer supplied and repeated tests of the same amplifier using their own Audio Precision based test system, which both validated my own measurements and allowed for more accurate measurements beyond the limits of my measurements.

CheryMega_R8_1KFFT_1W_Balanced

Cherry Audio MEGAschino Amplifier FFT @ 1 watt

First, looking at the 1 watt 1Khz FFT (64K with 10 averages), we see a few interesting properties. First, the basic noise floor is around -110 to -120dB with an overall noise plus distortion noise floor of -76 dBV. THD+N came in at .0056% (-86dBV) and THD came in at a much lower .0005% (-105.5 dBv). Why such a difference? Where is the noise coming from? Well, that became more obvious throughout all testing and shows that one of the upgrades for the Cherry amplifier is probably worth buying. Power supply noise was significant relative to the noise floor, with the highest being the harmonics between 120hz and 800hz, but actually extending well past 1khz. Still, at 1 watt the spurs all fell below -90dBV, so maybe I am being too critical. I didn’t hear any obvious hum in use, but I would have liked to see less power supply noise in the measurement. Moving onto the S/N at 1 watt into 8 ohms, we see it comes in at 85.9 dB. That is decent performance for such an amplifier, not the best in the business, but certainly better than you see with lots of lesser amplifiers. I believe this spec would improve substantially with the optional power supply upgrade which was not tested here.

 Cherry Megachina Frequency Response

Cherry Audio MEGAschino Frequency Response - red (8 ohms); blue (4 ohms)

The frequency response of a Class-D amplifier varies with load, especially on lesser quality amps. This is a problem many of us thought we had long left in the past as we moved beyond Tube amplifiers, but with the advent of Class-D, their reactive nature brought back the problem, and in some cases, it can degrade the linearity of the amp so significantly that the change in response is very audible. As such, one of the most important measurements to perform on a Class D amplifier is a 4 and 8-ohm frequency response test. I did this and found the response did vary with the load but only slightly. With a reference level of 5.5 dBV at 1khz, the response is down -.5dBV at 10Hz for either 4 or 8 ohms, and this is very likely the LF response of my measurement system. Cherry claims a frequency response down to 0hz with no phase shift. On the other end of the spectrum, the 8-ohm load is down .5dBV at 30khz and 20Khz at 4 ohms.  This was tested using an output filter, but the output filter has little effect on the response at these low frequencies. The measurement filter is 6th order with a corner frequency of -3dB at 65khz. In future reviews, I will be building test cables that will allow measurement of the frequency response without the filter, but that was not possible during this review. Since the response change is so substantial between 4 and 8 ohms, it suggests the difference is the amp, not the filter. Now, having called it substantial, let’s be clear, the substantial difference is at or above the limits of hearing, there is no meaningful attenuation measured between 20hz and 20khz, and as such, I will consider this amp to be well behaved.

Mk2 high power testing 

What about high power testing you ask? Well, as noted, I couldn’t test the amplifier at high power levels, so instead, I tested the amplifier at various power levels until I hit a point where I was certain I was overloading the input of the analyzer. Again, changes are being made to the test rig to allow higher margins, but for now, these are the best I can do. For high power testing, I switched to 4 ohms, as this gave greater margins. Distortion goes up into 4 ohms with this amp, as with most amplifiers, so consider that into 8 ohms, it will achieve at least half these power levels with measurably lower distortion (and Cherry’s own AP results confirm this). At 10 watts, THD+N is .0028% (-91dB) and THD is .0012% (-99dB). At 20 watts into 4 ohms, we see THD+N of .004% and THD of .002%. SNR at 10 watts was a good but not great 92.6 dB, again strongly impacted by power supply noise, with the 120hz power supply spur at -100dB. At 200 watts into 4 ohms, THD+N and THD both increased to around .007%. This means noise is no longer dominating the spec and THD starting to emerge as the dominant factor. At 270 watts distortion remains about the same, increasing to .008%. At the time of this test, the way the Q451 was setup also reflected the point at which the QA451’s own distortion rises to a similar level, and I lose confidence in any power results above this.

I did measure up to 350 watts where distortion remained below .05%, but this reflects more the rising non-linearity of the load than it does of the amplifier. Sadly, after putting together a high power test load with sufficient attenuation for full-power testing, I inadvertently unplugged and re-plugged a cable while live quickly destroying the amplifier. For those concerned that this is a sign that the Cherry protection is weak, keep in mind that what I did is something you should never do with any amplifier and would cause the failure of most amps. A very large 1khz test signal was going through the amplifier when the load was set up for a much lower power level than it would see. When replugging it, either the 1khz tone or, more likely, a huge 60hz signal suddenly burst through an amplifier capable of well over 1000 watts and instantly destroyed a sense resistor in the load. This led to a chain reaction that destroyed parts inside the QA451 which led to the destruction of the output transisters in the amplifier. All in a matter of milliseconds. I can’t fault the Cherry MEGAschino; this was all me. This is why I can’t have nice things.

Cherry Dac Dac 64K_32Bit 

I wasn’t able to take a lot more measurements but will note that rising power supply noise was noted as higher power measurements were taken. This increase is a clear sign of insufficient power supply capacitance, though at all times the power supply noise remained still modest. Many receivers would measure as bad or worse. My main concern is that this is not a receiver but a very expensive and high-end amplifier. I do wish this amplifier came standard with the capacitor upgrade.

Summary of Measurements

How do we sum up these measurements? Are they good? Great? State of the Art? Well, no I don’t think they are state-of-the-art with respect to the highest-performing amplifiers out there, but I do think they are very good. The amplifier frequency response changes far less than we see with most pro audio class D amplifiers, along with many past cheaper options. While I don’t have a graph to share here, I have measured ICEpower modules and found their highest power modules to have far greater changes in the response between 4 and 8 ohms. The noise floor of the amp was dominated by the power supply spurs and as such were not state of the art. I was concerned with this and asked Cherry to measure amplifiers they had to see if he could replicate the results. It became clear that better power supplies dramatically lowered the noise to the point that I think this could be a state of the art amplifier in terms of noise if the power supply were better. As an example, we saw the noise floor of the FFT to move from around -110dBV to -140dBV when moving from the basic power supply to a lab supply (Keep in mind, these lab supplies cost 10’s of thousands of dollars often). Distortion was very low out to very high power levels, and it seems clear this amplifier has more power than the new Purifi 1ET400A. So while its noise and distortion were not quite as good, it is also putting out a lot more power.

Further, can we really complain about distortion levels of .004% or less over most of the power range? Is this really audible? In fact, looking at the manufacturer’s own AP results at 1khz, we see that into 8 ohms the THD+N levels were mostly below .002% from 1 watt to over 100 watts with clipping becoming evident at 270 watts at .004% THD+N. Compared to the results of some of the best new amplifiers like the earlier mentioned Purifi module, I can’t call the Cherry amplifier state of the art. On the other hand, both power output and performance is still better than a lot of other amplifiers on the market. Further, the price of the MEGAschino is substantial enough that I can’t call it a bargain for such measured performance, but I think the price is justified for other reasons.

Cherry Dac Dac 64K_32Bit

Cherry Audio Dac Dac FFT

One measurement I want to add is of the Dac Dac used in the review.  This DAC was so clean that I had a hard time measuring it, so the only measurement I am including here is a 1khz FFT taken with my gear and a QA480 notch filter.  This shows a device with a S/N ratio of in excess of 121dB (this is not A weighted, expect a few dB improvement if A weighted) and THD+N of -112dB.  It is about as good as one can expect.  I really found nothing to complain about with the Dac Dac from these measurements.    

Conclusion

Cherry AmpThe bottom line for the Cherry Audio MEGAschino Mk2 is it is a very good amp, but that is as one would expect from its $6k+ cost. It doesn’t quite match the technical performance of the very best amps such as the Class-D amps from Hypex or Purifi or the Class-H amps from Benchmark in terms of THD, but its shortcomings relative to those models are still well below audible thresholds. Furthermore, it can come much closer to those extremes of performance with the optional upgrades to the power transformer and capacitors, although those upgrades do carry a hefty premium on the cost. What the MEGAschino does deliver that those amps do not is a lot more power. It’s a tall order to have both monster quantities of Class-D power with fantastically low distortion and noise, but it can be had in the Cherry MEGAschino, and it is built almost entirely in the USA, unlike those other brands. I enjoyed my time with the MEGAschino and was heartbroken to have accidentally wrecked my review unit, but my understanding is that the damage I did can mostly be repaired. I can easily recommend the Cherry MEGAschino, and were I shopping for high-powered amps in its price point, it would be at the top of my list.

The Score Card

The scoring below is based on each piece of equipment doing the duty it is designed for. The numbers are weighed heavily with respect to the individual cost of each unit, thus giving a rating roughly equal to:

Performance × Price Factor/Value = Rating

Audioholics.com note: The ratings indicated below are based on subjective listening and objective testing of the product in question. The rating scale is based on performance/value ratio. If you notice better performing products in future reviews that have lower numbers in certain areas, be aware that the value factor is most likely the culprit. Other Audioholics reviewers may rate products solely based on performance, and each reviewer has his/her own system for ratings.

Audioholics Rating Scale

  • StarStarStarStarStar — Excellent
  • StarStarStarStar — Very Good
  • StarStarStar — Good
  • StarStar — Fair
  • Star — Poor
MetricRating
Frequency Response LinearityStarStarStarStarStar
SNRStarStarStarStarStar
Measured Power (8-ohms)StarStarStarStar
Measured Power (4-ohms)StarStarStarStar
Two-channel Audio PerformanceStarStarStarStarStar
Build QualityStarStarStarStar
Fit and FinishStarStarStarStarStar
Ergonomics & UsabilityStarStarStarStarStar
FeaturesStarStarStarStar
PerformanceStarStarStarStarhalf-star
ValueStarStarStar
About the author:
author portrait

Matthew has spent the better part of the last two decades studying acoustics and good sound reproduction. He provides down to earth explanations of complex scientific topics related to audio reproduction.

View full profile

Confused about what AV Gear to buy or how to set it up? Join our Exclusive Audioholics E-Book Membership Program!

Recent Forum Posts:

PENG posts on June 15, 2021 13:59
Matthew J Poes, post: 1489476, member: 85392
But what SPL are we talking about?

I was talking about 105 dB peak SPL.

Your REW test wouldn't be able to test for clipping properly. Clipping is something that happens during dynamics. The most you could do, and it would still be a problematic method of testing, would be to test at the very high level at which clipping should be taking place. However, the energy contained in a sweep wouldn't induce that because its such a tiny narrow band of energy, that the amplifier would be putting out way more power than it could with a musical or full range signal. You would need white noise to do that but then you cant extract distortion. So a sweep test wouldn't be proving anything useful. Thats the wrong test.

I never use REW to tell me what my peak spls are anyway. I just make sure I have enough “power” on hand to achieve 105 dB, by calculations, and then to size my amp I would allow for about 6 dB of extra headroom, on top. That isn't hard for me to achieve as I rarely listen to more than 80 dB average, 70 dB from 10 to 11ft most of the time is about it as loud as I can tolerate and enjoy. So I know my 4 ohm rated 400 W A21 and 500 W 4B SST will never clip for my needs.

What Bob did and what I am planning to do is a much better way to address this. The multimeter and sweep tests don't accurately reflect what is really going on.

Understood. Multimeters are not the right tool for sure, though my Fluke 87V's 250 micro second peak capture is pretty good.

I think we do have to keep things in perspective though, that for a lot of people who listen to no more than 95 dB peak SPL from just one speaker that has sensitivity 87 to 90 dB/2.83V/m, nominal impedance 4 to 8 Ohm, sitting less than 4 meters away, their 200 W/300 W into 8/4 ohms amp is not going to clip. So much is hinged how how loud one listen to, distance and speaker sensitivity.

Here's a demo/video, you can see the big blue meters show around 1 W average, rarely peaked pass the 12 W mark. Those are real watt meters, though I doubt they probably won't always capture the highest peaks. But then the truly power hungry 800 D3 are not the kind of speakers that a lot of people have.

At request: B&W 800 D3 playing Norah Jones - YouTube
Matthew J Poes posts on June 15, 2021 13:01
PENG, post: 1488402, member: 6097
Thank you again Matthew for this post in which you have covered a lot of ground on this topic. I thought I had read just about every credible (in my opinion only obviously) article Dr. Google could find but I might have missed Cordell's 2007 show one.

There is nothing (could have missed something though as I was speed read..) I would disagree based on my own understanding, and experience on the topping of “clipping”.

That's why when people come to AH and ask at least once a week the questions related to “Do I need an external amp?, Does my amp have enough juice for my 4 Ohm speaker etc., kind of question.., I would typically ask them to first try and figure out how much power they actually may need for their applications/systems and report back, and go from there. If I think a simple question might suffice, I would high ball it based on their speaker's specs that include manufacturer's recommended amp power, maximum power handling number etc. and state my assumptions.

Some would come back with a number, even with a screenshot of the calculator they used, making it easy for my follow up response, after checking to see if they had input the data correctly, such as speaker sensitivity (important as most calculator are based on a fixed impedance of 8 Ohm), number of speakers and room gain allowance. I don't mean those things are exact, but if one or more are way out of whack then the final results could be almost meaningless. Once a number is reached, then I would typically bump the calculated number up by at least 3 dB, example: if I think they need 100 W, I would suggest a 200 W amp and that is exactly because I am concerned about clipping during peaks, not knowing what their favorite source contents are.

In my case, amps typically output between 0.1 and 0.5 W on average so I know any of my half a dozen power amps will never, or almost never clip.

All in all, clipping or not is not that hard to know as they (or was it I..) said, its the distance, speaker sensitivity, impedance, phase angles, and SPL stupid!!

I posted multiple times my experience in comparing my AVR-X3400H (that I tried for about a week) with my separates including the 4B SST and Halo A21, that I couldn't hear any difference whatsoever no matter how hard to try. During that time, I also use REW to check and see if the tiny AVR altered the FR and it didn't. So I am certain it's not THD, FR, DF, TIM of modern well made amps including AVRs, but much more about voltage and current capability such that as long as the DUTs are not pushed close to their limits all would be good. Any time I see someone reporting night and day difference, especially when heard even at ”low volume", I would say there's something wrong in their comparison listening setup, or they just say things that's likely wouldn't apply to everyone else.

In your linked Cordell article:

“At the end of the session the attendees were polled in a very informal and unscientific way as to which amplifier they thought was the tube design. They could answer one amplifier or the other, or they could answer that they honestly could not tell which was which.

There were no ”night and day“ results. Indeed, for many attendees the differences were difficult to hear. Moreover, those who perceived a difference were just as often wrong in selecting which amplifier they thought was the tube amplifier. This shocked all of us.”


So yes amps likely would clip more often than people think, but is the “all about SPL stupid” (my simplified version of the truth) that would determine whether one's amp would clip and how often.

In this case it was a comparison between a 35 W DIY tube amp and a 250 W Denon power amp. It just showed how silly people get thinking just by replacing their amp with a different brand, or simply adding an external 125 W SS amp to their 120 W rated AVR for two channel listening, they would get better sound quality like opening up their speakers, hearing details they would never hear before.
But what SPL are we talking about? You've said that before. If someone listens on average to a level at around 85dB that isn't the SPL level you use to calculate the amp needed. For movies, you then need to account for 20dB of headroom for dynamic peaks. Straight calculations of that necessitates often unrealistically high power levels at reasonable listening distances of 3-4 meters.

Most speakers are way over-rated for their sensitivity figure so using that as a baseline can be misleading. The saving grace being that the in-room sensitivity increases. As an extreme example, a lot of typical consumer speakers have a sensitivity between 85dB 1w/1m and 87dB 1w/1m. A straight SPL calculation of that show that for a listening distance of 3 meters, 10 watts would cover the average SPL fine. However, 1000 watts is needed to handle the 20dB peaks that the movies may contain. I know, its unlikely anyone is really trying to use such low sensitivity speakers for such high output, but the point remains, if anyone is really trying to do that, they are absolutely clipping their receiver, no way they aren't. And I don't believe its inaudible. Numerous studies have looked at that, its quite audible. People claiming they don't hear it either are listening at much more benign levels (most music has very little dynamic range) or simply don't know what they are looking for. I stand by that is a likely reality.

https://www.aes.org/e-lib/browse.cfm?elib=3793
as one example where this was explored. Worth getting for members.

Your REW test wouldn't be able to test for clipping properly. Clipping is something that happens during dynamics. The most you could do, and it would still be a problematic method of testing, would be to test at the very high level at which clipping should be taking place. However, the energy contained in a sweep wouldn't induce that because its such a tiny narrow band of energy, that the amplifier would be putting out way more power than it could with a musical or full range signal. You would need white noise to do that but then you cant extract distortion. So a sweep test wouldn't be proving anything useful. Thats the wrong test.

What Bob did and what I am planning to do is a much better way to address this. The multimeter and sweep tests don't accurately reflect what is really going on.
Matthew J Poes posts on June 15, 2021 12:45
AcuDefTechGuy, post: 1488539, member: 26997
Or their abilities to tweak or work the system just to get those extra decimal points?

Does the best THD+N and SINAD guarantee that every single thing inside the component is absolutely the best quality?

Or do designers/engineers just have to tweak their components so that they will get the best THD+N and SINAD if they know exactly what the reviewers are looking for?

For example, maybe components like the Marantz AVP and Anthem AVP had great SNR, Crosstalk, and THD numbers on S&V magazine for years because their engineers knew exactly what S&V and Stereophile were looking for in their measurements. But these same AVP didn’t seem to have great THD+N and SINAD from ASR because Amir doesn’t measure the same as S&V and Stereophile.

Now that S&V and Stereophile no longer do measurements, and Amir now does more measurements than anybody, Marantz AVP and Anthem AVP engineers may change their approach and tweak their components to get the best SINAD since now they know exactly what Amir is looking for.

Just a thought.
I actually think that the SNR numbers obtained by the two/three sources were about the same, it was more about how the reviewer perceived those. Most receivers have never had outstanding SNR or Crosstalk, but they were deemed good enough. Amir holds things to a much higher standard and it skewed peoples perception.

Amir also requires things to perform at a higher output level than others measured to. While 2 volt was the standard output, it wasn't uncommon for others to measure at 1 or 1.2 volt or to measure at whatever was the max clean output and not really making a big deal about it.

Amir's standards far exceed audibility, its just about good engineering practices. He just wants to see things perform to what is the current state of the art, regardless of how audible that is. The vast majority of products on the market today (of any decent quality) are audibly transparent. The main difference is noise floor, and cost isn't an indicator of that necessarily. Lots of very good receivers have very low noise floor.
lovinthehd posts on June 09, 2021 20:39
buckchester, post: 1488473, member: 78944
I don't think my logic is skewed.

I assume that if you asked most people in this hobby what their primary purpose would be in purchasing new gear it would be to try and improve the sound quality of their system. Now, I did not suggest getting rid of the measurements. I believe they do provide value. They can help people make reasoned assumptions as to whether or not a piece of gear will yield an audible improvement. However, they still do often leave a degree of uncertainty for many people. Blind testing is a way to get us closer to that level of certainty.

Yeah that improving sound quality gets deep into the significant return category and that can be tough these days with so much gear with very good performance at very good price points….
AcuDefTechGuy posts on June 09, 2021 19:41
ryanosaur, post: 1488471, member: 86393
…many of us here do tend to agree that those extra decimal places may be irrelevant, but it speaks volumes to the care and attention to detail of the designer and their capabilities…
Or their abilities to tweak or work the system just to get those extra decimal points?

Does the best THD+N and SINAD guarantee that every single thing inside the component is absolutely the best quality?

Or do designers/engineers just have to tweak their components so that they will get the best THD+N and SINAD if they know exactly what the reviewers are looking for?

For example, maybe components like the Marantz AVP and Anthem AVP had great SNR, Crosstalk, and THD numbers on S&V magazine for years because their engineers knew exactly what S&V and Stereophile were looking for in their measurements. But these same AVP didn’t seem to have great THD+N and SINAD from ASR because Amir doesn’t measure the same as S&V and Stereophile.

Now that S&V and Stereophile no longer do measurements, and Amir now does more measurements than anybody, Marantz AVP and Anthem AVP engineers may change their approach and tweak their components to get the best SINAD since now they know exactly what Amir is looking for.

Just a thought.
Post Reply