Review and Measurements of Hypex NC400 DIY Amp

[bluesaint]

I have the new NAD M22 Version 2. it's a custom NC400 board/layout made exclusively for NAD using NC500 output. It's an incredible amp.

 

[phoenixdogfan]

I thought I'd join the conversation after reading in guest mode for so long. I first want to thank everyone for such a great forum and community.

I built a Hypex 2xNC400 + SMPS1200A400 amp for daily use and have a few thoughts that might help the conversation.

The benefits of SMPS600 include a slightly higher rail voltage (65 vs 63 V nominally) but the difference in maximum output at the speaker should be imperceptible from a practical perspective. The SMPS600 also has separate diodes in the bridge rectifier and automatic mains voltage switching. Neither seemed to offer any functional benefit but I've yet to see any EMI or noise comparisons for the SMPS1200 vs SMPS600. My amp is dead quiet so that bodes well.

The SMPS600 provides a 16V Vdr which allows the NC400 internal driver regulator to remain off, dissipating 5W less per NC400 than when using the SMPS1200 (Vdr = 15.6V). My amp modules don't seem to generate much heat so this doesn't seem to be a cause for concern in practice. They were cool enough to the touch at idle or when driven hard that I haven't felt any need to monitor the amp modules' heat sink temperatures. They're mounted directly to the base plate of an all aluminum chassis with very thin thermal paste. I chose aluminum as it is much more thermally conductivity than carbon steel.

I occasionally listen to compressed music at high levels including hip hop with drawn out bass, or classical with long extended bass lines. Most speakers will have an impedance minimum an octave or so above woofer resonance so the long duty cycle bass lines and this speaker load can make for a tough combination for an amp and its power supply. The added headroom of the 1200W supply seemed to provide welcome insurance and this amp does play very loud and clean. The much lower cost of the 1200W supply vs 2 600W supplies sealed the deal.

The power supply heat and reliability is something to consider. After a year or so of use I had one channel start to drop out when played near clipping with sustained bass lines. Hypex performed a quasi RCA from afar and looked at hi res pictures of the internals. They replaced the SMPS1200 under warranty, sending a unit that I noticed uses caps from different manufacturers than the original supply. So far, all is working as expected. Hypex's customer support has been fantastic.

The case around the power supply does get modestly warm to the touch when it's idle, so I connect it to a switched power bar and shut it down when not in use, boosting the odds of a long life.

It has allot of power. To me it's not clinical, bright or glary at all and I don't detect it adding or subtracting anything. I can't say that about most other amps I've heard (rightly or wrongly). I think it's a wonderful amp. If it proves reliable, it'll probably be my last amp for a very long time.

I'm about to pull the trigger on an NCore build, and will be using the SMPS1200A400 with 2 NC400s and the NC400 x2 connection module. I've already bought the 300mm wide Ghent audio NC400 case which has all the IEC connection wiring.

I don't think this will be very difficult to assemble at all, but I do want to ask if Hypex supplied all the necessary nuts and bolts, or did you have to purchase them from Parts Express? Also have you had any issues with thermal shutdown? And what speakers are you driving, and what kind of load do they present?

I'll be using my completed unit for LS 50s, so I think there will be minimal problems with overheating. Even if I blow a power supply, we're still only talking 180 Euros to replace it. The complete all in cost for this amp will be around $1150 USD, so I see this as a stone cold bargain even with the added reliability risk in the SMPS.

 

[phoenixdogfan]

This will be one of four amps for a 4 way active system with DSP and a certain amount of FIR corrections.

Sounds like Lx 521s.

 

[phoenixdogfan]

My goodness Dave, I can't seem to get through to you can I?

You enjoy your 'Hypex Watts' and I'll continue to test and confirm specifications the way it has been done (and regulated) for 40+ years.

Seem to remember the US government used to requirere amplifier manufacturers to rate their amps in RMS wpc from 20 to 20khz, all channels driven continuously at l.t.1% THD, what happened to that standard?

 

[DDF]

I'm about to pull the trigger on an NCore build, and will be using the SMPS1200A400 with 2 NC400s and the NC400 x2 connection module. I've already bought the 300mm wide Ghent audio NC400 case which has all the IEC connection wiring.

I don't think this will be very difficult to assemble at all, but I do want to ask if Hypex supplied all the necessary nuts and bolts, or did you have to purchase them from Parts Express? Also have you had any issues with thermal shutdown? And what speakers are you driving, and what kind of load do they present?

I'll be using my completed unit for LS 50s, so I think there will be minimal problems with overheating. Even if I blow a power supply, we're still only talking 180 Euros to replace it. The complete all in cost for this amp will be around $1150 USD, so I see this as a stone cold bargain even with the added reliability risk in the SMPS.

95% of the work for mine was creating the harness and drilling the chassis. Since that's already done, this will be an easy job for you.

The NC400s come with a mount set (https://www.diyclassd.com/), can't recall the smps (if its a concern, I've found their tech support responsive). Screws are standard metric and should be readily available (Home Depot etc).

With the SMPS1200A400, you'll have enough current to drive just about any impedance. I design my own speakers and that was incentive for this supply, to not be current constrained when trying new designs.

It runs warmest at idle with no audio on the input and cools noticeably when driving it with any normal audio signal. I connect it to a power bar and turn it off when not used. There's no difference in sound when cold or warm and I think amir's tests showed no measured difference. This is a well engineered product and doesn't seem to have any boutique drifts over temperature.

Enjoy the amp!

 

[pos]

It runs warmest at idle with no audio on the input and cools noticeably when driving it with any normal audio signal.

Hi,
This is surprising and in contradiction with the power loss vs power curves published for both the power supply and amp module.
Are you sure this is the case?

 

[Roen]

Is this still the best measuring speaker amp tested so far?

 

[DDF]

Hi,
This is surprising and in contradiction with the power loss vs power curves published for both the power supply and amp module.
Are you sure this is the case?

Yes indeed.

 

[restorer-john]

It runs warmest at idle with no audio on the input and cools noticeably when driving it with any normal audio signal.

Can you post some idle consumption figures (one of those little killawatt things maybe)?

 

[Samoyed]

Reddi?

 

[6speed]

@restorer-john All Killawatt measurements, but take them with a grain of salt. The lower the power draw and the more inductive the load, the greater the inaccuracy. Googling for threads on computer forums where people compare with Flukes or other gear suggests Killawatt over-estimates as much as 100%.

 

[brodie_mc_choad]

View attachment 25097

Indeed, no other power amplification technology reviewed here at ASR has measured like the NC400 in mono - it's audibly transparent.

Nice to see that Bruno still wants to push the engineering limits further though.

I've had bad luck with NC502s...the power supply is very sensitive to surges, but more importantly dips in AC power - as surge protectors don't block the latter. Had two of them fail on me so far. :-( The NC500 amp I've got is connected to an isolation XFRMR so I've not had any problems with the SMPS3K700 PS that I'm using in that one.

 

[rajapruk]

No. Speaker output goes through the AES-17 filter. The feed to the amplifier is unmodified (flat).

@amirm On a Swedish hifi-forum (www.faktiskt.io) your measurements of power amplifiers have been questioned because of the AES-17 filter applied (questioned by an Swedish manufacturer of class-AB amps called Sybarite Audio).
Do you apply AES-17 filter only on class-d amps? Or on all amps?
Would it be possible to measure class-d without the AES-17 filter? Could you for example show Hypex Ncore without AES17 applied?
AES17-filter makes class-d amps to measure better than they should, compared to other non-class-d amps? The comparison is skewed?
Sorry if this has been elaborated on before, a search on AES17 on this site gave many many hits.

 

[DonH56]

It's an AES Standard filter to reject out-of-band noise. http://www.aes.org/publications/standards/search.cfm?docID=21 It targets digital circuits due to the significant extremely high frequency noise things like class-D amplifiers can generate. The frequency is high enough to be audibly irrelevant but will make class-D SNR/SINAD performance worse than AB amps due to the out-of-band noise. So naturally any class A, AB, G, H, etc. amplifier manufacturer will be quick to highlight its use as a selling point for their amps having higher SNR even though in-band SNR may be better or worse. The HF noise is real, but the use of it as a differentiator to say "class D is bad" is misleading IMO. Mostly by implying it can destroy tweeters, or be modulated back to the audio band, or that it corrupts ultrasonics so vital to the listening experience.

The debate is over words like "measure better than they should". The argument is whether the HF switching noise, far above the audible band and a by-product of the design, should be reflected in the measurements even if there are no audible consequences. A manufacturer with comparable or poorer in-band performance can use the wideband measurements of a class-D amplifier to claim superiority (e.g. superior SNR/SINAD/THD+N) based upon out-of-band noise. True, but whether it is relevant (matters to the listener) is left to the consumer. Do I buy a class-AB (or whatever) amp with worse in-band performance and requiring much more power, or a much more efficient class-D amp with better in-band performance but high out-of-band noise? That is up to the consumer to decide.

FWIWFM - Don

 

[Matias]

In my understanding, if the measurements consider the out of band noise of class D, it would be further away from the reality of what people listen to, also because tweeters do not respond to those noises anyway.

 

[somebodyelse]

Given that the AES17 filter used is "only down a few dB at 200 kHz" and that the measurements are usually band limited below this (SNR 22.5kHz BW, THD+N 90kHz BW) it shouldn't make any difference. Having said that, I can understand the desire for identical test conditions. Is there any reason not to use the filter for all amp tests?

 

[sergeauckland]

Given that the AES17 filter used is "only down a few dB at 200 kHz" and that the measurements are usually band limited below this (SNR 22.5kHz BW, THD+N 90kHz BW) it shouldn't make any difference. Having said that, I can understand the desire for identical test conditions. Is there any reason not to use the filter for all amp tests?

The best reason I can think of for not using it in all tests is that only Class D amps should have any appreciable HF hash that needs to be eliminated from measurements. All other amps shouldn't have any HF hash, so using the filter would eliminate any hash that could indicate instability or oscillation, and which needs to be seen if present. I measure noise with a 20kHz bandwidth filter to see if the amp is within spec, but only after making sure that there's nothing of concern looking wideband. I normally measure distortion with a 100khz bandwidth but get strange readings with digital products especially those with noise-shaping, so in those cases I use a 20kHz bandwidth, which means that I can't capture more than the third harmonic at 6.3kHz.

S.

 

[amirm]

@amirm On a Swedish hifi-forum (www.faktiskt.io) your measurements of power amplifiers have been questioned because of the AES-17 filter applied (questioned by an Swedish manufacturer of class-AB amps called Sybarite Audio).
Do you apply AES-17 filter only on class-d amps? Or on all amps?

I have by accident applied AES-17 filtering to an amp that I thought was class-d, only to do the FFT and find out it wasn't. Removing it made no difference.

Would it be possible to measure class-d without the AES-17 filter? Could you for example show Hypex Ncore without AES17 applied?

I don't have the NC400 used in this review anymore but have a NORD amp with NC500 under review which I can do this with.

AES17-filter makes class-d amps to measure better than they should, compared to other non-class-d amps? The comparison is skewed?
Sorry if this has been elaborated on before, a search on AES17 on this site gave many many hits.

AES-17 is kind of a confusing standard. It both applies to analog filtering with hard stop at 20/40 kHz or to type of filtering we use for amplifier testing. This latter type does NOT use brickwall filtering. The only requirement for it is to be flat up to 20 or 40 kHz. What it does after that is up to the manufacturer.

Worrying about the same concern they are raising, I decided to get the AP filter that is flat to 40 kHz, instead of the far more common 20 kHz. If we can preserve full spectrum up to 40 kHz, we are covering "high resolution audio" just as well.

The reason the filter is needed as already covered by others but to expand, if one is trying to measure a very small signal and there is tone of switching noise at high frequencies, the analyzer can get confused and use for example the wrong auto-scaling. The front-end of the audio analyzer is designed to have ultra low-noise and high dynamic range. As a result it has low "slew rate" (i.e. inability to deal with signal that change very fast. So the filter is deployed to deal with very high frequency components.

To add some data to the conversation, I just ran a couple of test with and without my AUX-0040 filter that I use with class-d amps. First, the dashboard view which is limited to 22.4 kHz:

This is the stellar performance of Audio Precision APx555 generating an analog tone and measuring it. This is what happens when I route the output through AUX-0040 filter first:

As we see the filter didn't make things worse or better for a traditional source (i.e. non-class D).

Here is the frequency response impact of the AUX-0040:

The hit is only 1 dB up to 77 kHz which is well past bulk of my tests which limit to 22.4 kHz. Even at 100 kHz we are down just 2 dB. So this is not going to do any class-d amps favors if they have junk up to 100 kHz. Indeed I have tested a number of class-d and "digital" amplifiers that did have trash below 100 kHz and despite using the AUX-0040 filter, that issue dominated the THD+N vs frequency measurements.

Many amplifiers have such roll off anyway. Few are flat to 100 khz.

Note that the AUX-0040 can add its own distortion:

So any help it gives, can be taken away under some circumstances. Indeed there are less wires involved in testing without AUX-0040.

All in all, they have a tiny point which I have tried to mitigate. The overall message and summary of any product tested this way is not impacted.

 

[rajapruk]

Thanks.
Just for the sake of it, please measure the amp with Hypex nc500 both with and without this filter (as you already suggested). Would be interesting to see if any actual difference turns out.

 

[rajapruk]

Please also try to test ”Eigentakt” when available (Bruno Putzeys new endevour).