• 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 daily reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Hypex Nilai500DIY Amplifier Review

Rate this amplifier:

  • 1. Poor (headless panther)

    Votes: 4 1.1%
  • 2. Not terrible (postman panther)

    Votes: 14 3.8%
  • 3. Fine (happy panther)

    Votes: 74 20.1%
  • 4. Great (golfing panther)

    Votes: 276 75.0%

  • Total voters
    368

Dunlavyphile

Member
Forum Donor
Joined
Feb 5, 2023
Messages
28
Likes
43
Location
Seattle
As someone who presently has a pair of Nilai mono block kits enroute, I would appreciate knowing why I should care there are perhaps comparable finished amps available in the same price range. Assembling the kits sounds like fun, and shouldn't take long. Is there something about their design as kits that will diminish the finished product, in terms of performance or reliability?
 

Redwine

Member
Joined
Sep 22, 2022
Messages
97
Likes
31
As someone who presently has a pair of Nilai mono block kits enroute, I would appreciate knowing why I should care there are perhaps comparable finished amps available in the same price range. Assembling the kits sounds like fun, and shouldn't take long. Is there something about their design as kits that will diminish the finished product, in terms of performance or reliability?
From Nords’ homepage:
“Using Nord REV D Input Buffer Boards with high-end discrete Sparkos Labs Voltage Regulators and a choice of Sparkos Labs SS2590 or the Sonic Imagery 990Enh discrete Op-Amps, both run in full Class A. And offers a richer, warmer, tonally dense musical sound with greater soundstage.”
Whether it makes a difference soundwise to Hypex Nilai’s standard, I don’t know.
 
OP
amirm

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,154
Likes
231,579
Location
Seattle Area
Hi amirm, thanks for all the work put into making this review and your other reviews.

I wanted to point out a couple of things that stood out to me after trying to make some comparisons to the Benchmark AHB2:

1) The two (lower) charts for displaying power vs distortion over bandwidth are not easy to compare at all. It would be much more helpful to stick with one type or the other.

2) The Nilai chart displaying power vs distortion over bandwidth is labeled as the “Low Gain” setting, but when compared to the top chart of power vs distortion w/ 4ohm load it looks to me like you measured the Nilai in “Medium Gain.” This further complicates a fair comparison…

Although I got a lot of useful information from this review, I’m left questioning your data portraying the Nilai’s “ballooning of distortion.” It’s also unclear if this data is with the lid on or off.
HI there. Coming late into this discussion. Benchmark AHB2 was measured in very early days of me measuring amplifiers. I have since refined and standardized the measurements across a large sample size. I don't have an AHB2 to go back and recreate the new measurements on. A member has offered one so perhaps I should do that.

As noted, the THD+N vs frequency is over 45 kHz bandwidth and with class D amps, this reduces their measured performance fair bit. The 45 kHz bandwidth came from much testing and arguments. If set much higher than this, it completely screws up class D amp due to increased noise above audible band. To set it lower as Hypex/Purifi do, would be wrong as well. So I have settled at 45 kHz where we can see the differentiation with class AB designs well.

I have also eliminated showing the line for 20 kHz (even though I measure it) since it is definitely misleading (it routinely shows less distortion than 15 kHz for class D which is wrong).

I could modify the test to only measure THD and not noise. Then I can expand the bandwidth. But then we won't see noise levels. Is this desirable?

On gain settings, I think they are right.
 

Julf

Major Contributor
Forum Donor
Joined
Mar 1, 2016
Messages
2,985
Likes
3,976
Location
Amsterdam, The Netherlands
From Nords’ homepage:
“Using Nord REV D Input Buffer Boards with high-end discrete Sparkos Labs Voltage Regulators and a choice of Sparkos Labs SS2590 or the Sonic Imagery 990Enh discrete Op-Amps, both run in full Class A. And offers a richer, warmer, tonally dense musical sound with greater soundstage.”
Whether it makes a difference soundwise to Hypex Nilai’s standard, I don’t know.
By definition, if you change the sound fo a neutral and transparent (accurate) amplifier to "warmer" and more "tonally dense" (whatever that is), you are coloring/distorting the signal.
 

capslock

Senior Member
Joined
Jun 19, 2020
Messages
305
Likes
144
I just compared to the Audiophonics HPA-S1ET400 at low gain (7.2 dB). The 1ET400 with no buffer has similar balooning, part of which is hidden in its higher noise floor. So the Nilai is actually better at every frequency and power level.
HI there. Coming late into this discussion. Benchmark AHB2 was measured in very early days of me measuring amplifiers. I have since refined and standardized the measurements across a large sample size. I don't have an AHB2 to go back and recreate the new measurements on. A member has offered one so perhaps I should do that.

As noted, the THD+N vs frequency is over 45 kHz bandwidth and with class D amps, this reduces their measured performance fair bit. The 45 kHz bandwidth came from much testing and arguments. If set much higher than this, it completely screws up class D amp due to increased noise above audible band. To set it lower as Hypex/Purifi do, would be wrong as well. So I have settled at 45 kHz where we can see the differentiation with class AB designs well.

I have also eliminated showing the line for 20 kHz (even though I measure it) since it is definitely misleading (it routinely shows less distortion than 15 kHz for class D which is wrong).

I could modify the test to only measure THD and not noise. Then I can expand the bandwidth. But then we won't see noise levels. Is this desirable?

On gain settings, I think they are right.
If there's one thing I would add, it is 18 + 19 kHz IMD or something very much like this (e.g. 18.5 + 19.5). This is apparently most revealing of how HF feedback factor was sacrificed.
 
Last edited:

Sokel

Master Contributor
Joined
Sep 8, 2021
Messages
5,533
Likes
5,450
HI there. Coming late into this discussion. Benchmark AHB2 was measured in very early days of me measuring amplifiers. I have since refined and standardized the measurements across a large sample size. I don't have an AHB2 to go back and recreate the new measurements on. A member has offered one so perhaps I should do that.

As noted, the THD+N vs frequency is over 45 kHz bandwidth and with class D amps, this reduces their measured performance fair bit. The 45 kHz bandwidth came from much testing and arguments. If set much higher than this, it completely screws up class D amp due to increased noise above audible band. To set it lower as Hypex/Purifi do, would be wrong as well. So I have settled at 45 kHz where we can see the differentiation with class AB designs well.

I have also eliminated showing the line for 20 kHz (even though I measure it) since it is definitely misleading (it routinely shows less distortion than 15 kHz for class D which is wrong).

I could modify the test to only measure THD and not noise. Then I can expand the bandwidth. But then we won't see noise levels. Is this desirable?

On gain settings, I think they are right.
As other friends here say,it would be nice to know what causes the rise in each occasion,noise or distortion.
Thanks for considering this!
 

pma

Major Contributor
Joined
Feb 23, 2019
Messages
4,513
Likes
10,529
Location
Prague
I could modify the test to only measure THD and not noise. Then I can expand the bandwidth. But then we won't see noise levels. Is this desirable?
It would be desirable if we could see both THD+N plots (vs. level) and pure THD plots (vs. level), both at the current set of frequencies. It would tell much more about amp nonlinear behaviour than the cumulative THD+N plots only, which tend to mask origins of measured behaviour.
 

Mnyb

Major Contributor
Forum Donor
Joined
Aug 14, 2019
Messages
2,396
Likes
3,265
Location
Sweden, Västerås
From Nords’ homepage:
“Using Nord REV D Input Buffer Boards with high-end discrete Sparkos Labs Voltage Regulators and a choice of Sparkos Labs SS2590 or the Sonic Imagery 990Enh discrete Op-Amps, both run in full Class A. And offers a richer, warmer, tonally dense musical sound with greater soundstage.”
Whether it makes a difference soundwise to Hypex Nilai’s standard, I don’t know.
By definition, if you change the sound fo a neutral and transparent (accurate) amplifier to "warmer" and more "tonally dense" (whatever that is), you are coloring/distorting the signal.

Yes i prefer to select an amp builder whos not into this buffer board FUD and smoke and mirror show :)

BoXem or March or Buckyeye (as examples ) makes one proper design and stick with it . There is simply no need for a consumer choice here only for features as selectable gain auto on or triggers etc .
 
OP
amirm

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,154
Likes
231,579
Location
Seattle Area
If there's one thing I would add, it is 18 + 19 kHz IMD or something very much like this (e.g. 11.5 + 19.5). This is apparently most revealing of how HF feedback factor was sacrificed.
As soon as I add this, then folks ask where that measurement is for older amps. :)

Here is the thing folks: the idea behind my tests is to have enough tests to quantify the overall performance of the unit. It is poor, good, very good, excellent or great? The tests are NOT meant to reverse engineer the device, find every bit of performance difference, etc. Each test takes time to run, document, post, and defend. So unless we are failing in the mission as I just stated, I am unwilling to add more tests.
 

ivo.f.doma

Active Member
Joined
Jan 21, 2021
Messages
217
Likes
95
Location
Slovakia
As soon as I add this, then folks ask where that measurement is for older amps. :)

Here is the thing folks: the idea behind my tests is to have enough tests to quantify the overall performance of the unit. It is poor, good, very good, excellent or great? The tests are NOT meant to reverse engineer the device, find every bit of performance difference, etc. Each test takes time to run, document, post, and defend. So unless we are failing in the mission as I just stated, I am unwilling to add more tests.
PMA: "It would be desirable if we could see both THD+N plots (vs. level) and pure THD plots (vs. level), both at the current set of frequencies. It would tell much more about amp nonlinear behaviour than the cumulative THD+N plots only, which tend to mask origins of measured behaviour."
 
  • Like
Reactions: pma

pma

Major Contributor
Joined
Feb 23, 2019
Messages
4,513
Likes
10,529
Location
Prague
PMA: "It would be desirable if we could see both THD+N plots (vs. level) and pure THD plots (vs. level), both at the current set of frequencies. It would tell much more about amp nonlinear behaviour than the cumulative THD+N plots only, which tend to mask origins of measured behaviour."
Yes.
 

capslock

Senior Member
Joined
Jun 19, 2020
Messages
305
Likes
144
As soon as I add this, then folks ask where that measurement is for older amps. :)

Here is the thing folks: the idea behind my tests is to have enough tests to quantify the overall performance of the unit. It is poor, good, very good, excellent or great? The tests are NOT meant to reverse engineer the device, find every bit of performance difference, etc. Each test takes time to run, document, post, and defend. So unless we are failing in the mission as I just stated, I am unwilling to add more tests.
That is hard to avoid but should not stand in the way of continous improvement. If you could do that for the two Hypex amps, that would be a good base to compare anything to that gets measured after that.
 

Ralf Stocker

Active Member
Joined
Jan 22, 2019
Messages
146
Likes
90
It is known from switched mode PSUs that leakage currents are always present at the outputs, especially at ground. How does that work with these amplifiers?
 

Julf

Major Contributor
Forum Donor
Joined
Mar 1, 2016
Messages
2,985
Likes
3,976
Location
Amsterdam, The Netherlands
It is known from switched mode PSUs that leakage currents are always present at the outputs, especially at ground. How does that work with these amplifiers?
What does "leakage current at ground" mean?
 

pma

Major Contributor
Joined
Feb 23, 2019
Messages
4,513
Likes
10,529
Location
Prague
What does "leakage current at ground" mean?
It seems to be improperly stated only. It would be the usual capacitive current trough the transformer winding, its path is closed according the real circuit configuration, grounding scheme etc. It of course exists, always.
 

Julf

Major Contributor
Forum Donor
Joined
Mar 1, 2016
Messages
2,985
Likes
3,976
Location
Amsterdam, The Netherlands
It seems to be improperly stated only. It would be the usual capacitive current trough the transformer winding, its path is closed according the real circuit configuration, grounding scheme etc. It of course exists, always.
Sure, but a clarification would be useful.
 

pma

Major Contributor
Joined
Feb 23, 2019
Messages
4,513
Likes
10,529
Location
Prague
Sure, but a clarification would be useful.
You know, it is quite important. Amir tests the power amplifier under ideal, clinically clean, unrealistic conditions. Not a single user here will have it connected in such way. Talking about stray impedances and leak ground currents, just have a look what kind of difference it makes if the NC252MP is supplied directly from mains or through the isolation transformer with only 120pF stray capacitance.

PMA-NC252MP mains without isolation trafo, load 4ohm.png


PMA-NC252MP mains with isolation trafo, load 4ohm, BW22kHz.png


And, the plots shown in Hypex datasheets are completely idealized, moreover with strictly limited BW to 20kHz and THD+N plots shown up to 20kHz thus pointless. It is a kind of marketing, and almost everyone eats it as served.
 

Rob Fens

Member
Joined
Feb 8, 2023
Messages
39
Likes
50
Location
Haarlem Holland
@ PMA;
What Amir is doing seems to me to be valid as long as he uses the same procedure for all comparable devices under test. If you want realistic tests, you have to take into account the average home situation and that will be different around the globe. Tests are performed following a procedure and a procedure is almost always an idealised or controlled situation. If this is not the case, the test will be poluted by unknown factors and the result will have no value. Furthermore; I may be wrong, but in your graphs i only see a difference in noise level between the 2 sitiuations. I guess this will be the case with almost any poweramp in those situations and has nothing to do with the Nilai.
This thread is not about Hypex in general or even the NC 252MP. What about the real subject of this thread the Nilai, is it based on your thoughts comparable with its competitors (such as Purifi) and what are the differences? Thats the info that i want to know.
 
Last edited:
Joined
Feb 14, 2023
Messages
49
Likes
123
You know, it is quite important. Amir tests the power amplifier under ideal, clinically clean, unrealistic conditions. Not a single user here will have it connected in such way. Talking about stray impedances and leak ground currents, just have a look what kind of difference it makes if the NC252MP is supplied directly from mains or through the isolation transformer with only 120pF stray capacitance.

View attachment 266628

View attachment 266629

And, the plots shown in Hypex datasheets are completely idealized, moreover with strictly limited BW to 20kHz and THD+N plots shown up to 20kHz thus pointless. It is a kind of marketing, and almost everyone eats it as served.
The Nilai five warranty from an established leader in the field is part of the value calculation for me as a buyer of Nilai Mono 500 diy kit. I define value by the following equation: V = P/C. Performance divided by cost determines value. The performance here is more than just lost in the weeds numbers. The Nilai has good numbers and very high power in a small package, this works for me, perhaps not ideal, but nothing is ideal in this world except abstract math. The amps barely get warm and they seem invisible sonically. The warranty and large manufacturer backing of this product is part of the performance cost equation for me.

I could have bought from a very small one person company, but these have a hidden high costs. In single owner companies the personality of the owner has a giant impact on the company and this tends to dominate... sometimes for good but it is often not very stable. A single visionary is often overwhelmed, due to new visions just like an artist, and like an artist he or she can also stop and move on... and their company disappears. Hypex seems to be the leader in class D, and Purifi is a viable challenger. If there is a good alternative in very small companies as consumers that becomes part of the V = P/C calculus.

I bought the Nilai to get a solid reference point on class D. I could have spent less but it would have had risks like dealing with a very small business that probably will not last. Some small companies will make it. I started in a smallish company that made it, and I have worked with many good small companies building the small company I was at into a giant. I am skeptical of measures that we probably cannot hear, but if you are sure you can finance measures and blind listening tests. I would not ask skilled hard working folks to do more work when they do not work for you and they give you their work for free. Just seems a bit ungrateful when someone is being generous but also practical.

How many choices do we have for a measured 500 watts, with a 5 year warranty, with awesome specs, and from the leader in the industry? PMA can add more measurements, that's fine but they should come with some kind of double blind listening tests that prove something important. I am skeptical, but go all the way to the end on something that's important, but it can only be done on certain key questions, for example active versus passive speakers.

For me I would like to see the next generation Hypex Fusion amps tested at some point. The next step, I think, is to stop buying amps, and cables and all that stuff because the numbers and subjective tests seem to be showing that if we want the most value, that is instrument grade system or just the highest performance for a given price point then active seems like the way to go. Why spend good money on huge fancy caps if they make no difference? Why clutter your house with a 100 boxes if they make no difference? Why not just have awesome active speakers and a great front end and then spend time with software to tweak stuff, or flip a switch to play around with first order crossovers?

It would be interesting to measure distortion of active versus passive implementations of the same speakers and blind tests with expensive and cheap capacitors to figure out what really is the optimal path, is a hybrid of some active and some passive best? This might be worth going all out with a ton of testing? But I don't think testing fantasies is what this thread is about. It is about a single product and if it has good value. The calculation of costs, like ease of use, ease of build, warranty, do they have service department, all these are factors in using the V = P/C equation. This equation is unique to each buyer, some people want chrome.

Audio can be death by a thousand cuts- or a thousand black boxes, the placebo and Veblen effects are in play, and many companies die slowly when lies are exposed and some just explode. However genuine value that is registered by the average consumer endure, and ASR as I understand it is trying to emerge from the voodoo audiophile cave into the light of real value. I applaud Amir for trying to keep people away from the ripoffs, if anything will breath life into hifi it is this. In the end a lot of measures are necessary... not everything can be tested but a lot of BS should be cleaned up so real improvements can be celebrated.
 
Last edited:

Dunlavyphile

Member
Forum Donor
Joined
Feb 5, 2023
Messages
28
Likes
43
Location
Seattle
I define value by the following equation: V = P/C. Performance divided by cost determines value.
I'm new to this forum: does value even properly come into play for product reviews? I don't see how it can. After all, the greatest value in amplifiers is found in the cheapest ones. They take you from silence to music, at the least cost. And probably satisfy the needs of a majority of music lovers (with their smart phones and earbuds). At the other end of the spectrum are folks who spend $1,000/foot for speaker cable because they are not limited by price and are easily persuaded.

Personally, for a review I just want to know how the device measures compared to other products, and what effect those differences might have on how I hear music played through them. I can draw my own (nonlinear) efficiency curves.
 
Top Bottom