Hypex Nilai 500 DIY Stereo Amp Kit Review

[peng]

The four mono's that I use (icepower, 1200as2 and 300a2) are dead silent as well but that's with my rather inefficient speakers.
But I wouldn't be confident to suggest them to someone with 105dB sens horns for example (the most sens I have tested them is about 100db and no pop) .

I like measuring and watching the chart as I power on gear, the only one I have seen jumping with a large spike was one that I had its PSU floating by mistake
The horror!

In my recent experience with my Hypex and Purifi amps, pops at powering up is a little like ground loop, sometimes hard to find the root cause. Case in point, my Purifi (VTV) amp out of the box would pop almost every time but after I opened it up and sand/tighten the chassis ground connection, it stopped, but then when I used with my DIY BMR bookshelf monitors, it would do such pop about 1 out of 10 times, most of the time not very loud so I have no concern, but used with my other speakers I have never heard such pop, though I should add, yet... My Hypex amp is from buckeye, same deal, after ensuring the chassis ground was good, it never pops any more. I do have the standby switch feature, if I didn't it would likely pop even if I always switch on the preamp first but I wasn't going to do such experiments when the standby feature is there to use.

 

[Sokel]

In my recent experience with my Hypex and Purifi amps, pops at powering up is a little like ground loop, sometimes hard to find the root cause. Case in point, my Purifi (VTV) amp out of the box would pop almost every time but after I opened it up and sand/tighten the chassis ground connection, it stopped, but then when I used with my DIY BMR bookshelf monitors, it would do such pop about 1 out of 10 times, most of the time not very loud so I have no concern, but used with my other speakers I have never heard such pop, though I should add, yet... My Hypex amp is from buckeye, same deal, after ensuring the chassis ground was good, it never pops any more. I do have the standby switch feature, if I didn't it would likely pop even if I always switch on the preamp first but I wasn't going to do such experiments when the standby feature is there to use.

Most amp modules like Hypex, Purifi, icepower, etc have their ground path through their mounting plates, so it helps if you identify the specific screw (most have little circuits around it, be it diodes,or small value resistors ending to Vclamp, etc) and sand it as well if the case is anodized or painted (or even securely tighten the screw but not much, their fragile, it's only aluminum! )

Same goes even with PC nice cases and painted screws, a lot of people having problems just because of that. Sanding the point where the PSU screws to the case make a nicer low resistance path to ground.

 

[Mort]

Sanding the point where the PSU screws to the case make a nicer low resistance path to ground.

My wife could hear the difference from the kitchen.

 

[Sokel]

My wife could hear the difference from the kitchen.

If by doing it was eliminating or reducing CPU/GPU/Fan/Hard disc noises going through the DAC as it sometimes happens I would hear it from my kitchen as well

 

[peng]

Most amp modules like Hypex, Purifi, icepower, etc have their ground path through their mounting plates, so it helps if you identify the specific screw (most have little circuits around it, be it diodes,or small value resistors ending to Vclamp, etc) and sand it as well if the case is anodized or painted (or even securely tighten the screw but not much, their fragile, it's only aluminum! )

Same goes even with PC nice cases and painted screws, a lot of people having problems just because of that. Sanding the point where the PSU screws to the case make a nicer low resistance path to ground.

That's exactly what I found with those two amps, buckeye did a better connection than the more expensive VTV amps (bad to the point the screw wasn't even tight enough). Not hard to fix, except if I had time I would have traced down the occasional pop with the VTV/BMR speaker combo, that I suspect has to do with some sort of capacitive coupling somewhere, since it would only pop 1 out of 10 even with the BMRs (and only 1 channel every time, could be either one though) and just a milder kind so I would likely probably leave it alone. I have now set it up with my KEF R900 so unless it evenly pop there too, I definitely wouldn't waste time chasing.

 

[onununo]

The reason to add the second power supply is to double the current capacity, which roughly doubles the output power at 2 and 4 ohm loads. For an 8 ohm load the amplifier is voltage limited, so adding the second power supply does not accomplish much except waste money.

My speakers are 4 ohm, and I added the extra power supply. I heard no difference. But, it may come in useful down the road if I ever really need 525W/ch.

You are over-simplifying. The output impedance of the amp, seen at its output terminals is what counts. You can think of it as two impedances in series. The first one is determined by the PSU (I'll stop at the PSU, although all the elctric power supply chain ahead of it also plays a role, though smaller), and the second one is determined by the amp (its output stage devices, its feedback factor, etc). This is still a simplification, because the two impedances in series, and especially the one determined by the amp, are not constant, but change with the dynamic of the signal. To add a second power supply (or if you prefer, to double the current capacity of the PSU), only changes the first impedance component. As you'll know, generally speaking speakers' impedance is not simply "8 ohm", "4 ohm", but instead varies a lot with frequency, and can have strong reactive components that also depend with frequency. The impedance seen at the amp output terminals, composed of the 2 said components (PSU and amp) gets in series with the impedance of the speakers and can badly (sometimes VERY badly) affect the sound (meaning distorsion), depending on how good the PSU and amp are (they both need to have the smallest possible equivalent impedance, and this must vary as little as possible with the dynamic of the signal). We'll never know how an amp with its own PSU will sound with "difficult" loads (and even with not-so difficult ones, because the effect is proportional, it's not "black or white") if we don't make proper measurements, as for example, the ones I have suggested in my post. This was the reason I wrote the post, and it wasn't for asking why you may or may not want to add a stronger PSU to an amp.
Presumably, if one decides to try doubling the PSU, it is because he hopes to get better sound (unless he's got free money to spend), so I begin to doubt on the quality of the amp, or the PSU, or both.

 

[terryforsythe]

You are over-simplifying. The output impedance of the amp, seen at its output terminals is what counts. You can think of it as two impedances in series. The first one is determined by the PSU (I'll stop at the PSU, although all the elctric power supply chain ahead of it also plays a role, though smaller), and the second one is determined by the amp (its output stage devices, its feedback factor, etc). This is still a simplification, because the two impedances in series, and especially the one determined by the amp, are not constant, but change with the dynamic of the signal. To add a second power supply (or if you prefer, to double the current capacity of the PSU), only changes the first impedance component. As you'll know, generally speaking speakers' impedance is not simply "8 ohm", "4 ohm", but instead varies a lot with frequency, and can have strong reactive components that also depend with frequency. The impedance seen at the amp output terminals, composed of the 2 said components (PSU and amp) gets in series with the impedance of the speakers and can badly (sometimes VERY badly) affect the sound (meaning distorsion), depending on how good the PSU and amp are (they both need to have the smallest possible equivalent impedance, and this must vary as little as possible with the dynamic of the signal).

Yes, I am over-simplifying. Yes, the output impedance of the amplifier is a very important parameter. Oftentimes damping factor is specified in lieu of output impedance, but this is a misnomer. The true damping factor of each driver in the system depends not only on the amplfier's output impedance, but also on the driver's impedance and the impedance of all components between the amplifier and the driver (e.g., the crossover in passive speakers). This too often is overlooked.

In your previous post to which I responded you stated:

There's not enough evidence as to how "muscular" the amplifier should sound, in the measurements.

I'm not sure what you mean by "muscular". Given the context of your most recent post, perhaps you mean low output impedance, which makes sense. For active speakers, that would fit. But, as noted above, for passive speakers damping factor not only is a function of the individual drivers (the woofer is the primary concern) and the amplifier's output impedance, but also the crossover. The passive filters in the crossover can have a significant negative effect on the damping factor. Below are charts of the damping factor for the woofer in my speakers both with and without the crossover. I modeled them in quc-s some time ago:

Note: In an active speaker in which long speaker cables are used (like in my family room setup), the impedance of the speaker cables will have some effect if the amplifier's output impedance is really low, but it is negligible for any speaker using a crossover that includes an inductor in series with the woofer or for any amplifiers with high output impedance. Thus, I included the speaker cable in the no-crossover simulation, but not in the simulation using the crossover since the impedance of the cables was so low in comparison to the inductors.

so I begin to doubt on the quality of the amp, or the PSU, or both.

You are free to doubt all you want. Nonetheless, the Nilai has very low output impedance, which makes sense since it uses a lot of negative feedback according an interview with a Hypex engineer I saw a few years ago. He also stated that the output impedance is less than 50 microhoms if I remember correctly. It would be nice to get independent confirmation of this, but I have not seen amplifier reviews where they actually measure the output impedance.

Also, FYI, I have seen numerous highly regarded amplifiers using Purifi amplifier modules with Hypex power supplies. Hypex power supplies generally are regarded as being very good for use in class D amplifiers.

 

[Julf]

Nonetheless, the Nilai has very low output impedance, which makes sence since it uses a lot of negative feedback according an interview with a Hypex engineer I saw a few years ago.

Pretty much all modern amps use a lot of negative feedback. One of the things Bruno Puzeys pioneered was the inclusion of the output filter (inductor) of a class D amp in the feedback loop.

Worth reading: Putzeys - the F word

 

[Sokel]

Pretty much all modern amps use a lot of negative feedback. One of the things Bruno Puzeys pioneered was the inclusion of the output filter (inductor) of a class D amp in the feedback loop.

Worth reading: Putzeys - the F word

No one is immune to marketing though.
Look at the lengths Purifi states they have gone:

It almost sounds like snake oil since even the most proper speaker cable will destroy that at its birth.

 

[terryforsythe]

Pretty much all modern amps use a lot of negative feedback.

Modern good amplfiers. There are a number of budget class D amplifiers that don't use enough negative feedback and/or have the feedback implemented sub-optimally. The following video discusses the impact of output impedance on speakers using various amplifiers:

 

[onununo]

Modern good amplfiers. There are a number of budget class D amplifiers that don't use enough negative feedback and/or have the feedback implemented sub-optimally. The following video discusses the impact of output impedance on speakers using various amplifiers:

Good point.
Let's not forget that the output impedance depends on frequency, and though I haven't made experiments to prove it, I am of the idea that it also strongly depends from the dynamic of the signal. It may even be very small if measured say at 1W in stationary conditions (such as with a sinusoidal signal of constant amplitude, be it swept or not in frequency), but I am pretty convinced that it varies hugely during a transitory, especially due to PSUs slow to recover from load transients. By load I am referring in this case to the one seen by the PSU, which is a function of the one seen at the amplifier output terminals AND the dynamic of the signal, and also by effects introduced by the amp (for instance slow thermal energy transfer among the amp devices and the environment compared to signal swing, especially during transitories, and amp stability issues in general).
An example: power a very capable amp (I mean capable in terms of the current it can provide to the load) with a weak (undersized transformer) unstabilised linear PSU with a large battery of capacitors. Load the amp with its intended impedance. Apply a step-like signal burst the amp, say a sinusoidal 100 Hz tone just to pick up one. What you'll get is very small output impedance at the beginning of the transitory (due to the capacitors) but such impedance will increase dramatically over time as a function of the energy drawn from the capacitors, leaning towards the one you would get without the capacitors in place.
We may make a similar example but this time with an ideal PSU and badly designed amp, measuring a "dynamic" output impedance affected by who-knows-what imperfections in the design and construction of that amp...
Ultimately, the output impedance of any amp and especially "how" it behaves during signal transients (and with different connected loads, because let's not forget that the load affects the stability of the amp), is one of the most important parameters for understanding its quality, along with the classical measurements made in stationary conditions such as frequency response, IMD, crossover distorsion - when applicable -, noise, SNR, the much beloved (and IMHO hugely overestimated) SINAD, etc..
I believe the importance of the behaviour of the amp under transitories and at different loads is way underestimated and some ways to some how evaluate it should be put in place.
A well-regarded Italian paper magazine used to publish their so-called TRITIMs for every amp they were testing. It is a graph showing the time-behaviour of the amp under various complex loads under transitory, a very interesting representation IMHO, along with a graph of the continuous and dynamic (or short-term, we may say) powers at various, decreasing resistive loads.
The only hint I can find in Amirm orherwise very detailed measurements, is sometimes the continuous maximum power measurement at various loads, but I think much more could (and should?) be done to investigate the aspects related to the behaviour of the amp under the point of view of its output impedance and especually during transitories, with (more) properly oriented measurements.

 

[onununo]

for passive speakers damping factor not only is a function of the individual drivers (the woofer is the primary concern) and the amplifier's output impedance, but also the crossover.

True, but you're now moving to the loudspeaker's terminals, which is outside the purpose of the analysis when we are trying to evaluate the quality of the amp as a piece of gear (i.e. from the plug into the mains socket to its speakers' output terminals, as that piece of gear is sold to the customer).
I mean to say, that the crossover network, if present, is a peculirity of the speaker set, and similarly long and undersized interconnecting cables are a peculiarity of... the cables (unless we're discussing about the cables INSIDE the amp).

 

[terryforsythe]

True, but you're now moving to the loudspeaker's terminals, which is outside the purpose of the analysis when we are trying to evaluate the quality of the amp as a piece of gear (i.e. from the plug into the mains socket to its speakers' output terminals, as that piece of gear is sold to the customer).
I mean to say, that the crossover network, if present, is a peculirity of the speaker set, and similarly long and undersized interconnecting cables are a peculiarity of... the cables (unless we're discussing about the cables INSIDE the amp).

I think we can sum it up with use a good amplifier with low output impedance. As Erin noted in the video to which I provided a link, if you have an amplifier with higher output impedance, you are rolling the dice on how it will work with your speakers. You can always use PEQ, I guess.

As far as cables go, if you have a passive crossover using inductors in series with drivers, the impact of the cables is negligible. I ran analyses, and cables only really matter when they are relatively long and there is not a passive crossover in the speaker using series inductors.

 

[onununo]

Erin has only measured the static output impedance. There's more to that.

 

[terryforsythe]

Erin has only measured the static output impedance. There's more to that.

Please explain how you would like the output impedance measured and what equipment would be used for that.

 

[onununo]

Please explain how you would like the output impedance measured and what equipment would be used for that.

I gave some hints in my previous posts. A possible way could be to measure max burst power vs max continuous power under challenging loads, such as 2, 4 and 8 ohm purely resistive, 2, 4 and 8 ohm 60° capacitive, and 2, 4 and 8 ohm 60° inductive. I see Amirm already measures max continuous power under such loads. Adding max burst power would be very revealing.

 

[Julf]

I gave some hints in my previous posts. A possible way could be to measure max burst power vs max continuous power under challenging loads, such as 2, 4 and 8 ohm purely resistive, 2, 4 and 8 ohm 60° capacitive, and 2, 4 and 8 ohm 60° inductive. I see Amirm already measures max continuous power under such loads. Adding max burst power would be very revealing.

Please give more than just hints. What should the test setup be?