• 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). Come here to have fun, be ready to be teased and not take online life too seriously. We now measure and review equipment for free! Click here for details.

Review and Measurements of Neurochrome Modulus-286 Amp

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
22,046
Likes
32,097
Location
Seattle Area
#1
This is a review and detailed measurements of the Neurochrome Modulus-286 Power (speaker) amplifier Kit. It is on kind loan from company designer and member, @tomchr. It is to ship by end of this month and costs $1,299. See: https://www.neurochrome.com/product/modulus-286-kit-2/

As you may recall from my review of Tom's headphone amplifier, Neurochrome designs strive for very low distortion and good, objective performance. So I have been quite excited and anxious to measure one of Tom's power amplifiers.

From the outside, the Modulus-286 is rather small but very chunky look with massive heatsinks (for its size) on each side of the unit:

Neurochrome Modulus-286 Power Amplifier Review.jpg

The heart of Modulus-286 and other solid state amplifiers Tom designs are the integrated LM3886 IC from Texas Instruments. I suspect this is from the National Semiconductor team though which was acquired by TI.

The LM3886 is a pretty powerful little IC capable of producing about 70 watts at 4 ohm. The Modulus-286 uses a pair of them to get over 100 watts of power at the same rating for each channel. Typical of mass market amplifier ICs though, the distortion rating is nothing to write home about at 0.02% (SINAD of 71 dB). In Tom's design, an active feedback loop (i.e. uses an op-amp) brings this distortion way down as you will see in the measurements.

The LM3886 is a class AB amplifier unlike other products we have tested that use much more efficient class D. So while the LM3886 by itself is small, it requires pretty large heatsinks to cool it which is the reason for them being included in Modulus-286. But in addition to the amplifier circuit, another source of size, weight and some heat is the power supply. Here, Modulus-286 deviates from classic class AB designs by using a switching power supply to gain its size and weight advantage. You can see it in the center of the Modulus-286:

Neurochrome Modulus-286 Power Amplifier teardown.jpg


It is a commercial switching supply which Tom says has been customized for his use.

The amplifier modules are actually hard to see on each side of the central power supply. I tried to take a picture of them but it was hard:

Neurochrome Modulus-286 Power Amplifier Amplifier Module.jpg


You can see the two LM3886 ICs mounted to the heatsink at the bottom. I am sure there are better pictures at Tom's website.

So let's get into the measurements and see how she does.

Measurements
As usual, I start with my dashboard view with 5 watts of output and in this case, XLR balanced inputs:
Neurochrome Modulus-286 Power Amplifier Measurements.png


SINAD (signal over noise and distortion) breaks the 100 dB barrier which is nice. You can see that the distortion of 0.0009% is orders of magnitude better than the specified level by TI (although that number is likely at higher powers).

I noticed a lot more mains/power supply noise in one channel than the other. Nothing I did would impact its level. I could also feel the "tingling" sensation from the case which indicates mains leakage. This is legal and safe (allowed in safety standards) but does explain the elevated levels there. Of course at -110 dB is not audible since our hearing is so poor in low frequencies.

Next, I measured the all important power versus distortion+noise:

Neurochrome Modulus-286 Power Amplifier Power vs Distortion at 4 Ohm Measurements.png


Compared to our best measured so far, Hypex NC400, the Modulus-286 nearly matches it in noise level. On distortion, one channel also tracks the NC400 until its max power of 95 watts. This was with 0.975 millivolts of input. Attempting to feed it strong signal would cause it to go into protection so I could not quite get to the specified power rating.

One channel deviates some though as you see in light red. the better channel goes as low at 0.0007% of THD+N or SINAD of 103. Seeing how this is at max power, it is very nice figure to arrive at. Performance is basically noise dominated with no distortion until the shut down.

The Hypex DIY NC400 has more power but its distortion starts to rise from where Modulus-286 stops. It produces 226 watts at the point I have marked in blue. So it has more headroom.

Frequency response shows somewhat early roll off:

Neurochrome Modulus-286 Power Amplifier Frequency Response Measurements.png


We are down 0.4 dB at 20 kHz. Would have been nice to see this flatter seeing how this is a traditional class AB design. Likely it is part of the computation to keep the amplifier stable with different loads. Of course not an audible concern for most of us who can't even hear 20 kHz let alone 0.4 dB of drop in there. :)

Signal to noise ratio is quite good:

Neurochrome Modulus-286 Power Amplifier SNR Measurements.png


Broadband spectrum analysis to 1 MHz shows very little of concern:
Neurochrome Modulus-286 Power Amplifier FFT Spectrum Measurements.png


We see some indication of the switching frequency of the power supply around 160 kHz but its level is quite low at nearly -100 dB. With class D amplifiers we see levels as high as -30 dB! Note that since this is a class AB amplifier, I did NOT use my AES-17 filter with any of the tests.

I tested THD+N versus frequency at 6 different power levels as you see on the graph:
Neurochrome Modulus-286 Power Amplifier THD vs Frequency Measurements.png


As expected, distortion rises at higher frequencies (feedback gain is lowest there). At 97 watts, the Modulus-286 could produce frequencies above 1 kHz but below that, it would enter into protection mode as indicated by vertical lines in orange.

At the lowest power level (0.25 watt) we are dominated by noise and hence the elevated line in red for that. That is true of any power amplifier.

Finally here is the crosstalk for one channel (the other is better):
Neurochrome Modulus-286 Power Amplifier Crosstalk Measurements.png


Please take this as worst case response. I am testing the amp using a bunch of wires everywhere which can impact crosstalk (and noise levels at low amplification). I need to build a dedicated module with tight wiring for tests like this. Regardless, numbers like this are amply good enough as far as audibility is concerned.

Conclusions
The Neurochrome delivers good power with very low distortion using a combination of active feedback and off-the-shelf amplifier IC. Use of a switching power supply keeps costs, size and weight down. It produces nearly 100 watts of very clean power into 4 ohm load. Personally I like to have much more power than that. :) But if you have subwoofers, or don't listen as loud and/or have more efficient loudspeakers, you should be fine.

I am not sure what value judgement to put on the amplifier. Certainly at $1,300, the Modulus-286 amplifier is "cheap" compared to any high-end amplification. At the other externe, the DIY Hypex NC400 I tested produced more power at I think $1,500. Purchasing the Neurochrome Modulus-286 would let you support a smart, Canadian designer and entrepreneur. I let you decide how valuable that is.

Overall, I can recommend the Neurochrome Modulus-286 as a well-engineered product with no design faults.

------------
As always, questions, comments, recommendations, etc. are welcome.

They say you should save money for a rainy day. It rains for about 6 months here so I need plenty of money!!! Please consider donating funds using:
Patreon: https://www.patreon.com/audiosciencereview), or
upgrading your membership here though Paypal (https://audiosciencereview.com/foru...eview-and-measurements.2164/page-3#post-59054).
 

tomchr

Technical Expert
Technical Expert
Manufacturer
Joined
Nov 5, 2018
Messages
271
Likes
779
Location
Calgary, Canada
#2
Amir: First off thank you very much for your review. I really appreciate it. Thank you for taking the time to measure the amp. I hope you found time to hook it up to speakers and take a listen as well.

One question: In your comparison of the MOD286 and NC400 on THD+N vs output power, did you use the same output filter? At least I'm assuming that you used a "Class D filter" (AP AUX-0025, AUX-0040, or equivalent) when measuring the NC400.

The difference in mains hum between channels is due to inductive coupling between the SMPS and the amplifier. The rectifier in the SMPS will couple into sensitive nodes. I oriented the SMPS to minimize this coupling, but one channel will always be closer to the rectifier than the other. That's the price you pay in a compact design like that. Like you, I decided not to worry too much about the hum, as it is well below audible.

I'm a bit more concerned that you "only" get 97 W out. I measured 105 W (both channels driven) and 115 W (one channel driven). I wonder if it's because the supply voltage on the SMPS is a little low. There is an adjustment pot that would allow you to raise the voltage if it is low. Granted the 0.3 dB difference is mostly academic, but my conscientiousness requires me that I try to determine the cause of that 0.3 dB difference. :)

Your mention of headroom makes we wonder: How do you (or anyone else here) define that? The reason I'm asking is that I occasionally get asked, "how much headroom does your amplifier have?" and I honestly struggle to answer that question. To me, the term 'headroom' is meaningless in that context. A 100 W amplifier providing 100 W will have no headroom. If the same amp provides 10 W, it'll have 10 dB of headroom. If it provides no signal, it'll have infinite headroom.

Thanks,

Tom
 
Last edited:

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
22,046
Likes
32,097
Location
Seattle Area
#3
One question: In your comparison of the MOD286 and NC400 on THD+N vs output power, did you use the same output filter? At least I'm assuming that you used a "Class D filter" (AP AUX-0025, AUX-0040, or equivalent) when measuring the NC400.
I did not use a filter with Modulus-286 in this final run. I recall using it before but didn't make a difference. The AES-17 filter I have is only down a few dB at 200 kHz so there is not much for it to filter in your design. But I will check again when I get a chance.
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
22,046
Likes
32,097
Location
Seattle Area
#4
I am not sure what value judgement to put on the amplifier. Certainly at $1,300, the Modulus-286 amplifier is "cheap" compared to any high-end amplification. At the other externe, the DIY Hypex NC400 I tested produced more power at I think $1,500. Purchasing the Neurochrome Modulus-286 would let you support a smart, Canadian designer and entrepreneur. I let you decide how valuable that is.
I am using "headroom" in a true sense here of pure continuous power. Here, the Hypex has more power so when large low frequency peaks arrive, it is able to reproduce them better.
 

tomchr

Technical Expert
Technical Expert
Manufacturer
Joined
Nov 5, 2018
Messages
271
Likes
779
Location
Calgary, Canada
#5
I did not use a filter with Modulus-286 in this final run. I recall using it before but didn't make a difference. The AES-17 filter I have is only down a few dB at 200 kHz so there is not much for it to filter in your design. But I will check again when I get a chance.
Isn't AES-17 20 kHz bandwidth? Or are there different AES-17 filters?

I'm just a little puzzled by the difference in THD+N between the MOD286 and NC400. I couldn't get the NC400 below 0.0008% THD+N (4 Ω, 1 kHz, 20 kHz BW) in my THD+N vs output power sweep of that amp.

Tom
 

tomchr

Technical Expert
Technical Expert
Manufacturer
Joined
Nov 5, 2018
Messages
271
Likes
779
Location
Calgary, Canada
#7
In that case I'm puzzled the NC400 measures as well as it does. I seem to recall its switching frequency being above the audio band, but not by that much. Now, it's been a few years since I measured it, so I could be entirely wrong on that. Anyway. As long as the two amps are compared under the same conditions, life is grand.

Tom
 

jackenhack

Active Member
Patreon Donor
Joined
Oct 25, 2018
Messages
190
Likes
419
Location
Stockholm, Sweden
#8
Amir, just curious, why don't you test it at 1 watt? Five watts probably gives a better indication of the performance, but it's fun to compare to the specifications from the manufacturer, which usually shows one watt.
 

Blumlein 88

Major Contributor
Joined
Feb 23, 2016
Messages
7,047
Likes
7,600
#9
In that case I'm puzzled the NC400 measures as well as it does. I seem to recall its switching frequency being above the audio band, but not by that much. Now, it's been a few years since I measured it, so I could be entirely wrong on that. Anyway. As long as the two amps are compared under the same conditions, life is grand.

Tom
I think the NC400 runs at 480 khz switching rate. So that gentler filter would knock that down 50 db or more. Could you be thinking of some earlier class D designs that ran at 110 khz or so?
 

tomchr

Technical Expert
Technical Expert
Manufacturer
Joined
Nov 5, 2018
Messages
271
Likes
779
Location
Calgary, Canada
#10
I think the NC400 runs at 480 khz switching rate. So that gentler filter would knock that down 50 db or more. Could you be thinking of some earlier class D designs that ran at 110 khz or so?
Could be.

Amir, just curious, why don't you test it at 1 watt? Five watts probably gives a better indication of the performance, but it's fun to compare to the specifications from the manufacturer, which usually shows one watt.
I'd be interested in such a measurement as well. You'll probably have to run an FFT rather than just a THD+N measurement as I'd expect the THD+N to be dominated by the noise (+N) component at 1 W.

Tom
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
22,046
Likes
32,097
Location
Seattle Area
#11
Isn't AES-17 20 kHz bandwidth? Or are there different AES-17 filters?
AP sells two versions, one that has 20 kHz flat bandwidth and one that has 40 kHz flat bandwidth. AES states that you can use the latter but you need to mention its specs.

Note that both of those numbers are the guaranteed flat bandwidth. Both go well above that before the response starts to drop off. Here is the response of the one I have:

1548542084874.png
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
22,046
Likes
32,097
Location
Seattle Area
#12
Amir, just curious, why don't you test it at 1 watt? Five watts probably gives a better indication of the performance, but it's fun to compare to the specifications from the manufacturer, which usually shows one watt.
Yeh because it is not a situation anyone uses. The 1 watt spec was established years ago when watts were expensive. Measurements will likely be dominated by noise alone so it is not really distortion measurement.

I am actually tempted to go the other way around and standardizing at 10 watts. :)
 

Blumlein 88

Major Contributor
Joined
Feb 23, 2016
Messages
7,047
Likes
7,600
#13
AP sells two versions, one that has 20 kHz flat bandwidth and one that has 40 kHz flat bandwidth. AES states that you can use the latter but you need to mention its specs.

Note that both of those numbers are the guaranteed flat bandwidth. Both go well above that before the response starts to drop off. Here is the response of the one I have:

View attachment 20860
The AES 17 filters show -60 db at either 24 khz or 48 khz depending upon whether it is the 20 khz or 40 khz version. Both are said to be flat (- .1 db) at 20 and 40 khz respectively. So they do drop off pretty steeply just above the specified band.
 

tomchr

Technical Expert
Technical Expert
Manufacturer
Joined
Nov 5, 2018
Messages
271
Likes
779
Location
Calgary, Canada
#14
Yeh because it is not a situation anyone uses.
Except if you have efficient speakers. In that case you're likely to run less than 1 W in most cases. I know quite a few people who use speakers that have sensitivities around 95 dB/W*m.

Measurements will likely be dominated by noise alone so it is not really distortion measurement.
For a good amp with low distortion, that's certainly true. I suppose it could be true for a bad amp as well if the bad amp had a very high noise floor.

I am actually tempted to go the other way around and standardizing at 10 watts. :)
I'm fine with that as well. Then again, if you're setting up the FFT for 10 W, it would take only a couple more button clicks to get the 1 W measurements.

Tom

PS: Thanks for following up re. the AES17 filter.
 

tomchr

Technical Expert
Technical Expert
Manufacturer
Joined
Nov 5, 2018
Messages
271
Likes
779
Location
Calgary, Canada
#15
BTW: I should point out that the Modulus-286 Kit will be a Limited Edition. No, not limited to as many as I can sell... :) There will only be 28 of these made. Amir has #1/28.

With the compact design, I locked myself into one vendor for the power supply, which was an enormous risk to take, especially as my relationship with the supply vendor soured. Thankfully, I have all the supplies I need for the 28 amps so life is good, but I won't be getting more.

Tom
 

restorer-john

Major Contributor
Joined
Mar 1, 2018
Messages
2,756
Likes
4,874
Location
Gold Coast, Queensland, Australia
#16
It's a very neat looking little package. Beautiful attention to detail on the power supply plate, cable routing, grommet placement etc. Working in such a small space is tricky. You are a tamper-proof Torx aficionado eh?

So what are we looking at at 8ohms Tom, about 60 watts a channel with even better THD? Those THD levels for any chip amp driving 4 ohms are truly excellent.

I can't help myself thinking '286'. What's next a 386, 386DX and then down the track a 486DX4? ;)
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
22,046
Likes
32,097
Location
Seattle Area
#17
The AES 17 filters show -60 db at either 24 khz or 48 khz depending upon whether it is the 20 khz or 40 khz version. Both are said to be flat (- .1 db) at 20 and 40 khz respectively. So they do drop off pretty steeply just above the specified band.
They do not though. You can see the response graph I posted from AP above. I have measured the same.
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
22,046
Likes
32,097
Location
Seattle Area
#18
PS: Thanks for following up re. the AES17 filter.
I tested with and without AES-17 and results are the same. AES-17 is actually a hair noisier which is expected due to extra wiring involved.
 

tomchr

Technical Expert
Technical Expert
Manufacturer
Joined
Nov 5, 2018
Messages
271
Likes
779
Location
Calgary, Canada
#19
It's a very neat looking little package. Beautiful attention to detail on the power supply plate, cable routing, grommet placement etc. Working in such a small space is tricky. You are a tamper-proof Torx aficionado eh?
They're not tamper-proof, actually. The tamper-proof Torx has a nib in the middle that requires a corresponding hole in the screwdriver.

On the MOD286 Kit, I use Torx and Torx-Plus. The Torx-Plus can be driven using a regular Torx bit. In the final version, I went with hex drive for the six screws that attach the Modulus-286 modules to the heat sinks as I figured more people would have hex keys than small Torx drivers.

I love the Torx/Torx-Plus screws. They're foolproof and look pretty. I despise slotted screws. It is the most useless screw ever invented. Not that I'm opinionated or anything... :)

So what are we looking at at 8ohms Tom, about 60 watts a channel with even better THD? Those THD levels for any chip amp driving 4 ohms are truly excellent.
65 W into 8 Ω. You can see my measurements on the Modulus-286 Kit page: https://www.neurochrome.com/modulus-286-kit/#PerformanceGraphs

Based on Amir's measurements, it looks like I might have to revisit the max spec for power in to 4 Ω. Apparently there's enough variability in the output current limit in the power supply to limit some amps to less than 100 W, whereas I had no trouble getting 115 W out of my prototype.

I can't help myself thinking '286'. What's next a 386, 386DX and then down the track a 486DX4? ;)
Heh... Perhaps. A couple of OEMs have expressed interest in a MOD486. We'll see if it materializes. I'm not planning any more additions to the Modulus line anytime soon. I think I need to let the credit cards cool down a bit first. :) The biggest advantage of a 486 is cost savings compared to my Modulus-686. You can run the Modulus-686 on ±27 V for 130 W into 8 Ω (200 W into 4 Ω) for a safe-n-sane build with specs similar to or slightly better than what a 486 would feature. You can read more about the Modulus-686 here: https://www.neurochrome.com/modulus-686/

Tom
 
Last edited:

jasonq997

Active Member
Patreon Donor
Joined
Nov 4, 2018
Messages
115
Likes
165
#20
This is the sort of thing that really makes this website worthwhile. This process of testing and in some ways collaborating with these objectivist oriented small companies is going to reshape the stereo "audiophile" industry in the right direction. Tom, if you can get to the point where you can mass produce this stuff and get the prices down you will have a really big customer base. Amir, thank you again for the excellent work you are doing.
 
Top Bottom