Review and Measurements of Hypex NC400 DIY Amp

[JohnPM]

I knew the ex-Bosch engineer really well from F1 but can not think of his name at the moment.

Udo (Zucker). Lives in Portugal these days.

 

[JohnPM]

I read that was a Douglas Self design. Was that correct?

Yes, Doug did the TMA amp designs.

 

[Frank Dernie]

Udo (Zucker). Lives in Portugal these days.

Of course, thanks John.
I know he was a Hi-Fi enthusiast but knew him best for the electronic controllers he did for Bosch and TAG-McLaren electronics.

 

[March Audio]

So far, class D amplification is problematic with tweeters, at least the old class D.

When I was n-reviewing the specifications and measurements of the IcePower 1200AS2 module PDF manual, the only graph I found different was that of the phase. That it varied so much was an alarm signal in my brain.

This would explain why unaltered voices (you know, without Autotune, vade retro satana) and acoustic instruments sound bad with this technology.

Many audiophiles reneged on the NC400, which is why I supposed that he also had a big problem with the phase. They have the problem but it is not that huge. Maybe the NC500, with custom buffers mask that problem.

With the current music, so badly recorded and manipulated should not be problematic. This would explain why many do not dislike sound. Let is face it, what percentage of people listen to (high) quality recordings with acoustic instrumentation and unaltered voices via software? And with DR > 11dB. The percentage must be ridiculous.

I think there are a few sweeping statements there.
Re phase, just have a bit of a think about what your speaker does to phase. It ain't pretty.

 

[Armand]

I'm also keen to see the transient overload characteristics of these modules. Alan is going to do some toneburst testing on his at some stage so hopefully we may get to see what happens in gross overload and/or on the edge of clipping.

I too have lost tweeters in a split second when a high powered amplifier had an out-of band-oscillatory burst.

Back in the day, I inspected many tweeters (performed autopsies) to determine whether they were valid warranty claims or not. In my experience, around 40% had vaporised lead-in wires and undamaged voice-coils, with no evidence of long term heat damage. They were the audiophiles with their dynamic CD sources and high powered amplifiers that in my opinion were bursting into oscillation at transient overload. Often they would have repeat failures. Changing amplifiers sometimes solved the problem.

The heat damaged ones were the AC/DC guys who played loud all night after drinking a bottle of bourbon.

It was after inspecting many expensive dead berrylium tweeters, I decided to repair a few just for fun by unwinding one turn of the VC and resoldering to the lead-in wire. Many of those repairs never came back and never cost their owners a cent. (NS-1000 tweeters were around AU$200 at the time).

In fact, one of the tweeters in the NS-1000Ms (Japanese import 10th Anniversary model (1984)) at my father's place is still going 20+ years after I did the same repair. We do have a brand new one in case it ever fails however.

I have done some burst tests on a Hypex NC500 that drove it into overload condition. It was powered from Hypex SMPS1200 with idle rails @ 85V.
On the picture below the signal is 100Hz into a 4 Ohm load (684W). Peak voltage on the first cycle is 74V on the two next cycles we can see the artifact on the negative side. A small square pulse with a duration of 0.5ms. Further abuse did not create any more dramatic events.

 

[DonH56]

I have not really followed, too many other things going on... A quick comment on phase: there will be latency through the amplifier (group delay) in addition to a phase shift due to the output filter. Group delay is the negative of the change in phase over frequency. If it is constant, phase shift will not change the timing (pulse response) of the signal thus is benign. This complicates the whole phase issue -- you want to know the constant delay that is independent of frequency and separately the phase shift over frequency at the output. Have to be able to do phase separation, natch. Or just plot group delay instead.

Please forgive if this has already been said or does not make sense, only skimmed the last few pages.

 

[Deleted member 2944]

So far, class D amplification is problematic with tweeters, at least the old class D.

When I was n-reviewing the specifications and measurements of the IcePower 1200AS2 module PDF manual, the only graph I found different was that of the phase. That it varied so much was an alarm signal in my brain.

This would explain why unaltered voices (you know, without Autotune, vade retro satana) and acoustic instruments sound bad with this technology.

Many audiophiles reneged on the NC400, which is why I supposed that he also had a big problem with the phase. They have the problem but it is not that huge. Maybe the NC500, with custom buffers mask that problem.

With the current music, so badly recorded and manipulated should not be problematic. This would explain why many do not dislike sound. Let is face it, what percentage of people listen to (high) quality recordings with acoustic instrumentation and unaltered voices via software? And with DR > 11dB. The percentage must be ridiculous.

I hope the speaker systems in use here are single-driver crossover-less types that have no phase distortion.

Regarding the phase response measurements, sometimes it's best to just break out the dual-trace oscilloscope and take a look. No fancy AP measuring scheme, no AES 17 filtering, or anything else. The scope measurements posting my March are a nice example.

BTW, the custom buffers in use with the NC500 modules have no effect on the phase response of Ncore systems. Those are there strictly to provide the extra 8db (or so) voltage gain necessary and provide an opportunity for manufacturers to utilize a preferred op-amp type that might required by the audiophile types and/or apply their own "flair" to the design.

Dave.

 

[DonH56]

The problem with just using a DSO is that, unless you sweep and plot the phase (time difference converted to phase), you don't distinguish just the delay through the amp (component) from the phase shift caused by e.g. the output stage (filter, feedback, etc.) If you want to look at phase a better tool is a VNA.

 

[Deleted member 2944]

I have not really followed, too many other things going on... A quick comment on phase: there will be latency through the amplifier (group delay) in addition to a phase shift due to the output filter. Group delay is the negative of the change in phase over frequency. If it is constant, phase shift will not change the timing (pulse response) of the signal thus is benign. This complicates the whole phase issue -- you want to know the constant delay that is independent of frequency and separately the phase shift over frequency at the output. Have to be able to do phase separation, natch. Or just plot group delay instead.

Please forgive if this has already been said or does not make sense, only skimmed the last few pages.

Yes, a factor as well. I mentioned in Post 182 above.
This type of phase response measurement is actually more complicated than it might seem. You're correct regarding latency altering a simple DSO phase response measurement. Any frequency-independent delay would need to be characterized.
This reminds me of reviewers measuring the Sunfire amps years ago and not understanding what they were seeing regarding phase response.

Dave.

 

[maty]

@dreite, my second audio system works with my tweaked and modded KEF Q100, 5.25" coaxial speakers, with two first-order filters.

Like almost dynamic speakers, the phase went progressively / smoothly in a range: 0º - 45º with which the human ear will hardly appreciate that variation.

The problem is an amplifier whose output impedance is not basically resistive. Specially if there is a large variation in its phase that affects the whole, with which the phase of the system will no longer be in that range of 0 -45º !!!

custom buffer -> custom discrete op-amp.

There are discrete op-amps with terrible harmonic profile and noise but with great audiophile press. Fortunately nobody uses them with these amplifiers. I have a post about op-amps measures, with graphs. How can someone recommend that garbage?

 

[andreasmaaan]

@dreite, my second audio system works with my tweaked and modded KEF Q100, 5.25" coaxial speakers, with two first-order filters.

Like almost dynamic speakers, the phase went progressively / smoothly in a range: 0º - 45º with which the human ear will hardly appreciate that variation.

Do you have measurements to validate this please?

The KEF Q900, while using an electrical first order crossover, gives an acoustical second order crossover between the tweeter and midrange, with resultant phase shift of (approx.) 180 degrees around the crossover point.

Measurement here (see Figure 8).

I’d be surprised if they went for a completely different filter topology on the Q100, especially given that the same coaxial drive unit is used in both speakers.

 

[maty]

I have a lot of docs about KEF Q100 and others coaxial speakers (KEF or not), in open and closed Internet (with much more info in a forums where I write). But it is not the thread, and most likely my comments are deleted again.

If I am not wrong, ALL Qx00 series uses first order filters.

[Picture, original XO] http://maty.galeon.com/WP-imagenes/kef-q100/KEF-Q100-crossover.jpg

[Picture] http://maty.galeon.com/WP-imagenes/kef-q100/KEF-Q100-impedance-phase.png

with resultant phase shift of (approx.) 180 degrees around the crossover point.

180 degrees... FIX, without variation with the frequency. Fix, constant,...

Open a new thread.

 

[andreasmaaan]

I have a lot of docs about KEF Q100 and others coaxial speakers (KEF or not), in open and closed Internet (with much more info in a forums where I write). But it is not the thread, and most likely my comments are deleted again.

If I am not wrong, ALL Qx00 series uses first order filters.

http://maty.galeon.com/WP-imagenes/kef-q100/KEF-Q100-impedance-phase.png

Open a new thread.

@maty please, I'm not trying to be argumentative, really

You just need to understand that every speaker has an impedance response and a frequency response. The two are (very indirectly) related, but are completely different things!

Let me demonstrate with some graphs. Here's the impedance response (including phase) of the Wilson WITT:

As you can see, the phase of the impedance response stays within roughly +/- 22.5°.

And here is the same speaker's acoustical phase:

Here we see a phase shift of approximately 360° which is due (I infer) to 2nd order acoustical crossovers being used between each of the 3 drivers.

This is because the phase of the impedance response and the phase´of the frequency response are not the same thing!

You made the same mistake with the three graphs you posted of class D amps (can't find the post now, it was yesterday or the day before). The first graph showed one amp's impedance response, while the other two graphs showed the phase of the other two amps' frequency response.

Please be careful not to confuse these two very different things

 

[maty]

Off course they are not the same, that is why I wrote to the manufacturers!!! Only IcePower answered, to say it is not the same but did not provide the measurement.

Also keep in mind that English is not my first language.

I suppose you studied systems, and how these can be modeled as a black box. The problem is now, the source (the amp) affect the speakers response because it is not only resistive. Changing the value of amp output resistance only affected the magnitude but not the shape of the curves, that is the question.

That variation, how much will it affect the sound, the timbre? I do not know. I have my suspicions.

The violin will continue to sound like a violin or will be more like a viola?

An electronic instrument or created by soft will not have that problem, because the listener does not have the sound memory of the real instrument.

https://en.wikipedia.org/wiki/Timbre

*** If you want to continue, open a new thread ***

 

[Ron Texas]

I like the Hypex distortion numbers, but that power supply probably will not last long.

 

[andreasmaaan]

Off course they are not the same, that is why I wrote to the manufacturers!!!

Then why do you continue to insist that because your KEF speakers have an impedance response with phase +/- 22.5°, they must have a frequency response with phase +/- 22.5°???

Your speakers almost certainly have a 180° phase shift in the frequency response.

*** If you want to continue, open a new thread ***

No thank you, I really don't think this is going anywhere.

 

[maty]

To finish,

https://en.wikipedia.org/wiki/Black_box

Absolutely phase != relative phase. Fix != variable. All this was explained in Diyaudio's thread.

 

[Deleted member 2944]

Maty,

I've noted you going on and on about this on various audio forums for months now. I think it's time to give it a rest.

Dave.

 

[RayDunzl]

Re phase, just have a bit of a think about what your speaker does to phase. It ain't pretty.

JBL LSR 308 nearfield (a meter or so) phase in REW:

Add two traces at the listening position (what I infer that room reflections do to phase), one wrapped, one unwrapped:

The electrostats (red) throw a pattern like the JBL does in nearfield, but all the way to the listening position, I presume, due to lower room interactions:

 

[amirm]

Well it's simple to confirm with a (filter less) scope measurement.

Here is that:

500 Hz:

There is almost no phase differential with zero crossing almost the same.

20 kHz:

There is 2.6 microseconds of phase shift now. If my math is right, that is about 20 degrees of phase shift.

Running AP without AES filter makes it struggle to measure phase but going along with it anyway, we get: