One of the improvements claimed of the Ncore topology over the original Ucd self-oscillating class-d amplifier is that the modulator transfer function is more linear. I created a couple of simulations to check this out and, indeed, the Ncore transfer function is very linear until just under saturation whereas the original UcD design has a more tube-like, rounded transfer function with a bit of an s-curve. To reproduce these curves, the x-axis must be changed from "time" to "V(a)" by hovering the mouse pointer over it until the ruler icon appears, then click and edit the axis parameter.
The Ncore also has about 20dB more loop-gain across the audio band (to see this, switch to an ac analysis for both simulations and click on the "fb" node).
The Ncore also has about 20dB more loop-gain across the audio band (to see this, switch to an ac analysis for both simulations and click on the "fb" node).
Attachments
Some astute readers may have noticed that the self-oscillation frequency predicted by the ac simulation is close to, but higher than the actual operating frequency in the transient simulation.
As Bruno Putzeys pointed out, the UcD is not a sine wave oscillator, so the additional phase shift of harmonics influences phase crossover for the composite waveform. Oscillation occurs at the frequency where if the loop were broken (and driven as before the break), the signal on both sides of the break would be the same shape and phase relationship.
At quiescence, at the input to the power comparator the output square wave has been heavily low passed, so it appears as a slightly distorted sine wave, so the influence of the harmonics is very small and the ac analysis is quite close. However, when the output level is near the rails, the output is a narrow pulse rich in harmonics that must be taken into account.
In the ac simulations (which are open loop), I set the ac source to a value so that when traversing the feedback loop, zero crossing of magnitude would occur at the same point as phase crosses zero. I have done this simply to make the results appear more intuitive by mimicking what the closed loop would do naturally through feedback.
As Bruno Putzeys pointed out, the UcD is not a sine wave oscillator, so the additional phase shift of harmonics influences phase crossover for the composite waveform. Oscillation occurs at the frequency where if the loop were broken (and driven as before the break), the signal on both sides of the break would be the same shape and phase relationship.
At quiescence, at the input to the power comparator the output square wave has been heavily low passed, so it appears as a slightly distorted sine wave, so the influence of the harmonics is very small and the ac analysis is quite close. However, when the output level is near the rails, the output is a narrow pulse rich in harmonics that must be taken into account.
In the ac simulations (which are open loop), I set the ac source to a value so that when traversing the feedback loop, zero crossing of magnitude would occur at the same point as phase crosses zero. I have done this simply to make the results appear more intuitive by mimicking what the closed loop would do naturally through feedback.
As feedback is taken after the LP filtering, more filtering probably wouldn't do much to the HF peak other than reduce the feedback there (and possibly degrade phase margin)
Yes. Of course there could be added filtering after that output feedback point that is not in the loop.
I doubt it would make it sound better. I would welcome HF boost in that high range although I can't hear quite that high now anyway.
boB
Perhaps if I have the time I will attempt to create a simulation with a linear version of the N-core with the same response as the switching version (except for the switching, of course). Why? Then I can run the same audio .wav file input into both simulated amplifiers and difference their outputs, saving the result as a .wav file.
If the responses for moderate signal levels are truly identical the difference will yield no output (at least at low to moderate input levels). However, with the original UcD, for signals with outputs that exceeded about half of full power, the difference was no longer zero. This difference is the coloration added by the UcD switching process. I believe these .wav files should still be available here on the diyAudio website. They are interesting to hear although I don't claim to know what they mean.
Is anyone interested in this exercise for the N-core, assuming it can be done? What other simulations with the N-core might be interesting?
If the responses for moderate signal levels are truly identical the difference will yield no output (at least at low to moderate input levels). However, with the original UcD, for signals with outputs that exceeded about half of full power, the difference was no longer zero. This difference is the coloration added by the UcD switching process. I believe these .wav files should still be available here on the diyAudio website. They are interesting to hear although I don't claim to know what they mean.
Is anyone interested in this exercise for the N-core, assuming it can be done? What other simulations with the N-core might be interesting?
Last edited:
Maybe I am too late and the thread is dead or its in the nick of time. I just created a thread dedicated to my own project, a NAD M22 special edition. While I will not spend time going over what it is, you can just look that up, I am curious about one thing. Does someone present in this thread have enough knowledge to isolate the Feed-Forward Error correction circuit used in the NC400/M22 ?
While I do have access to all I need, I can't simply do that isolation on my own due to limited knowledge and as far as I know, there are multiple ways to do feed-forward feedback.
Many thanks.
(https://www.diyaudio.com/forums/class-d/335418-nad-m22-special-edition.html)
While I do have access to all I need, I can't simply do that isolation on my own due to limited knowledge and as far as I know, there are multiple ways to do feed-forward feedback.
Many thanks.
(https://www.diyaudio.com/forums/class-d/335418-nad-m22-special-edition.html)
How quick these threads just seem to randomly die, makes you wonder. This site is about DIY, not DIY to learn new things from everyone else and then simply bugger off and profit with your own commercial wares... that said I wouldn't be surprised anyone with a successful nCore reproduction would end up on the wrong side of a certain someone, at some point Hypex wont even have a product to promote that is superior to a modern IcePower if that is how they play games. I am fortunate good people like Soren/Soekris still sticks around here despite coming up with an amazing product and originally debuting it here, and not having to throw a gross premium on it just because. (nc400)
Last edited:
My favourite is when they began to sell nc400 kits for more than what the nc500 can be had for at a good OEM such as Nord. Essentially the only major difference between the 400 and 500 was the working voltage. IMO the nc400 was artificially limited to not detract from the ridiculous OP NC1200 that was unobtainium except the privileged few that made sure to flaunt that they were able to get nc1200's, posting them all over this site, despite nothing being DIY about it! EDIT: rant over, noticed all the ncore threads were locked a long time ago, strange that, but great that an OEM-only commercial product who's creator is long gone on this forum gets no free promo here now. Hypex had no intentions beyond DIY ncore other than their crippled nc400 beta test platform. You can see straight through it.
Last edited:
The best example is how the nc500 better suits 8 ohm speakers at higher outputs, the nc400 would have stepped on the nc1200's toes had it had the extra 8 ohm headroom! To get the extra headroom with the nc400 you needed to bridge them. 2x$$$$$$
Given how the nc500 and nc1200 both operate at the same working voltage you know the nc400 was voltage limited, no reasoning to derate nc400 other than intentionally limit performance to market the nc1200 to OEMs. Crappy business model despite owning nc400 and I will voice my displeasure. 500mA of DC offset on the *inputs*, after a few years of use.
Given how the nc500 and nc1200 both operate at the same working voltage you know the nc400 was voltage limited, no reasoning to derate nc400 other than intentionally limit performance to market the nc1200 to OEMs. Crappy business model despite owning nc400 and I will voice my displeasure. 500mA of DC offset on the *inputs*, after a few years of use.
Last edited:
Have you taken this issue up directly with Hypex? Unlike the designer of your dams, I have always found the response from Hypex to be prompt, courteous, and very helpful. In fact, they have bent over backwards to help out.
I would hardly call the nc400 a beta release- unlike the dam 1021, it delivered exactly what was advertised.
It better have delivered what it advertised for what it cost, with the many years of UcD behind them all things considered. I think everyone can agree Soren originally oversold the dam1021, however I think the latest models have very much lived up to the expectations and he doesn't overcharge for OEM parts and sells to everyone even with the top end board. Nobody would have an issue if Hypex did the same.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.