Darn, after some intereting pages, I jumped to the last one just to add on the wishlist . Only to see Bruno had left the ship. I also found why.... 
Oh, whatever, If I'm here I might as well add to the wishlist.
IF Ncore's arrive with lesser Power spec, consider using the same supply rail voltages , or allow them to overlap between different models.
For those few building HT amps, it makes lifes easier if the PSU's can be the same, whilst the output needs for the surround channels doesn't warrant the top-of-the line ncore modules.
... for the sleeping crowd ...
/E
Oh, whatever, If I'm here I might as well add to the wishlist.
IF Ncore's arrive with lesser Power spec, consider using the same supply rail voltages , or allow them to overlap between different models.
For those few building HT amps, it makes lifes easier if the PSU's can be the same, whilst the output needs for the surround channels doesn't warrant the top-of-the line ncore modules.
... for the sleeping crowd ...
/E
To Noah Katz
Sorry for not answering earlier but I was absent.
Class-d does indeed not know the same type of crossover distortion as an AB amp does.
The idling current through the output inductor mentioned resembles a triangular wave quite closely. It is caused by the switching of the output devices and has therefore also the same frequency.
Its peak-to-peak amplitude rises for a given switching frequncy when the output coil's inductance is lowered. For a given inductor it rises also when the switching frequency is reduced.
This idling current helps to shorten the switching phase and minimises the distortion caused by deadtime.
As soon as the output current exceeds the peak value of this triangular current the "helping effect" is lost and THD increases.
In the case of the nCore the distortion rises from extremly low to very low at this point.
It would be better in fact not to call this effect crossover distortion at all. The reason why this term is used is most probably the lack of a better one. Maybe we should start a poll for this subject. ;-)
I would say that other effects dominate the THD behaviour of the IR amp such that this crossover effect can't be seen at all or at least not very clearly (keep in mind that the difference in THD between the ncore and the IR is almost an order of magnitude). The difference between the channels is simply manufacturing torerance IMO.
Regards
Charles
Sorry for not answering earlier but I was absent.
Class-d does indeed not know the same type of crossover distortion as an AB amp does.
The idling current through the output inductor mentioned resembles a triangular wave quite closely. It is caused by the switching of the output devices and has therefore also the same frequency.
Its peak-to-peak amplitude rises for a given switching frequncy when the output coil's inductance is lowered. For a given inductor it rises also when the switching frequency is reduced.
This idling current helps to shorten the switching phase and minimises the distortion caused by deadtime.
As soon as the output current exceeds the peak value of this triangular current the "helping effect" is lost and THD increases.
In the case of the nCore the distortion rises from extremly low to very low at this point.
It would be better in fact not to call this effect crossover distortion at all. The reason why this term is used is most probably the lack of a better one. Maybe we should start a poll for this subject. ;-)
I would say that other effects dominate the THD behaviour of the IR amp such that this crossover effect can't be seen at all or at least not very clearly (keep in mind that the difference in THD between the ncore and the IR is almost an order of magnitude). The difference between the channels is simply manufacturing torerance IMO.
Regards
Charles
Last edited:
phase_accurate
You have a good way of explaning things.
I just want to add that the "help" actually continues to exist but non symmetrically. When generating a positive output current the "help" actually becomes stronger and stronger with current when the PWM node is going negative but it becomes tougher and tougher for the output stage to pull the PWM node back up leading to longer dead time. So at idle you have a symmetrical positive/negative slope but as you increase the output current one slope becomes faster and one becomes slower.
You have a good way of explaning things.
I just want to add that the "help" actually continues to exist but non symmetrically. When generating a positive output current the "help" actually becomes stronger and stronger with current when the PWM node is going negative but it becomes tougher and tougher for the output stage to pull the PWM node back up leading to longer dead time. So at idle you have a symmetrical positive/negative slope but as you increase the output current one slope becomes faster and one becomes slower.
Hi Noah
A good drawing sometimes helps a lot. Pages 18 to 24 of this presentation by Bruno are a good description of what is going on.
http://www.hypex.nl/docs/papers/AES124BP.pdf
Regards
Charles
A good drawing sometimes helps a lot. Pages 18 to 24 of this presentation by Bruno are a good description of what is going on.
http://www.hypex.nl/docs/papers/AES124BP.pdf
Regards
Charles
& 
Another quick update
The amp module has been debugged, we're still sifting through some minor issues with the SMPS which keep me from completing the test report. The hand-stuffed v1.0 board (same as the actual first production run) certainly doesn't look bad. Also the fact that the 6kHz plot (red) is worse than the 1kHz plot will get some attention before I can release the product.
(back into hiding)
The amp module has been debugged, we're still sifting through some minor issues with the SMPS which keep me from completing the test report. The hand-stuffed v1.0 board (same as the actual first production run) certainly doesn't look bad. Also the fact that the 6kHz plot (red) is worse than the 1kHz plot will get some attention before I can release the product.
(back into hiding)
Attachments
Great work from the hiding place Bruno!
The board looks really neat and well thought through regarding the connections too.
Wonderfully it seems that the DIY nCore promise even lower distortion up to about 300W/4ohm than the otherwise more powerful nc1200 which is in itself already offers stunningly low distortion figures.
When you are able to find the time, it would be great with some update on the final design regarding e.g. the specifications on the input stage power supply needs and other interesting details which may have been discussed through this thread but would be nice to know
BTW its a nice looking hot speaker wire connection that golden screw like thing on the board, if i guess correctly.
cheers,
The board looks really neat and well thought through regarding the connections too.
Wonderfully it seems that the DIY nCore promise even lower distortion up to about 300W/4ohm than the otherwise more powerful nc1200 which is in itself already offers stunningly low distortion figures.
When you are able to find the time, it would be great with some update on the final design regarding e.g. the specifications on the input stage power supply needs and other interesting details which may have been discussed through this thread but would be nice to know
BTW its a nice looking hot speaker wire connection that golden screw like thing on the board, if i guess correctly.
cheers,
I noticed it too but have a certainly stupid question: why has the hot connection such a "good and strong' connection, whereas the cold one is not apparent on the plate and hence probably done via a smaller connector ??
I think I can see where the pictures might lack clarity. There are two screw tags. On the front view, the cold connection is hidden by the hot connection (it's directly behind it) and on the top view only the cold connection is visible, the hot one being underneath the image. If you look closely though you can see both of them in the side view.
Nice Modules ;-)
Do You have some Information about delivery?
I asked Hypex and got this response from Mr. Jan-Peter van Amerongen:
The NCORE modules are only available for selected customers. Next year we'll plan new modules and these will be more widely available.
The plans for any new NCORE modules are not yet defined.
Do You have some Information about delivery?
I asked Hypex and got this response from Mr. Jan-Peter van Amerongen:
The NCORE modules are only available for selected customers. Next year we'll plan new modules and these will be more widely available.
The plans for any new NCORE modules are not yet defined.
- Status
- Not open for further replies.