Maybe is the better aka flatter phase graph at HF than Hypex NCxxx, which is one of the three improvements that I was interested in Purifi 1ET400A. Only 1º at 20 kHz.
I don't know much about what to measure but can say the highs are as good as anything I've ever heard... maybe better. Musical passages that are dense with high frequency content are kept clearly delineated... They don't descend into a mash. This from JBL horns.
Purifi is onto something that is very musical.
Purifi is onto something that is very musical.
^
And thanks to Tom/Neurochrome, it is about to get a lot better, at least from a customization perspective + mono which allow the placement of the modules to be more optimized and potentially produce a better chn separation since one can physically place them further away from one another, use shielding etc
And thanks to Tom/Neurochrome, it is about to get a lot better, at least from a customization perspective + mono which allow the placement of the modules to be more optimized and potentially produce a better chn separation since one can physically place them further away from one another, use shielding etc
No, Bruno Putzeys explained it already. It is basically the materials chosen for the output inductor, so that there is less hysteresis. Dunno what, he did not explain it in details, and I am not going to disassemble the output inductors to chemically analyse the materials, but maybe he moved from iron to permalloy. The (linear) phase (component) has nothing to do with it, it is inaudible!!!
By the way, the hysteresis is something the folks at ASR will never be able to properly measure – and to accept, since they are so fixated on sums of sinusoids. However, if you have a strong impulse, then the whole signal after that strong impulse gets shifted a bit for some time, and during that time all signals will be reproduced with a (slowly decreasing) DC shift. This considerably increases distortion, which is usually perceived as some "granularity" in the sound - in this case of the treble (I perceive sound "coming out as if in cubes" when there is a lot of distortion, esp caused by input saturation – on other amps – so when Bruno P spoke of granularity, I related to that description immediately). However it seems now to me that it is there anyway, only I was not aware of this).
NC500 amplifiers are absolutely great, among the best sounding things out there, and they provide twice as much power over 8Ohms than the Purifi modules, 50% more power over 4Ohms – so they are also top of the line, but with a different set of compromises. The Purifi 1ET400A is better for me for what I make out of it.
Last edited:
Well, the channel separation of the EVAL1 is already nearly perfect. So you are not going to get any real improvement there, as in audible.
Of course it will be easier to build monoblocks. But at the end of the day I am not putting my monoblocks as close as possible to the speakers, I have excellent speaker cables that I can also run for 3 meters without stressing the amplifier (actually, for a dozen of meters as well, if I wanted). So I am returning to a stereo build.
Last edited:
What I was thinking of was more in the line of dual mono configuration inside the same chassis. Fully separated channels is always better than when they share components, just my 2 cents
Oh yeah, I know the common wisdom, and is based on considerable experience as well. But I will still use the Universal Buffer, which is one component with a single power supply for both channels, and I will continue to use common power supplies for both channels. I am not worried by the minimal loss in channel separation, if any.
IN FACT, a single powerful power supply will probably serve the last stage of the amplifier better than two small ones, because the voltage may sag less on strong power demands. Whereas having two smaller ones may be at a disadvantage in this regard, so you may have better (but inaudible) separation but audibly worse dynamics.
Google fails. There were two parts. Domain flipping as a transient effect and linearity of gain the modulator and how hysteresis in the core affects that. The answer rather than solution in both cases was more loop gain. There was no mention of esoteric materials.
Last edited:
Thank you. I remember the fact that he addressed hysteresis, had a wrong recollection of how. The important thing is that he dealt with it!
Ahhh... Because.
Because the moon was in the seventh house and Jupiter aligned with Mars?
DB in MN
Because the moon was in the seventh house and Jupiter aligned with Mars?
DB in MN
Ooop. Here you go. It was in my bookmarks.
Purifi Audio - A Straight Wire to the Soul of Music | audioXpress
Elsewhere he discusses gain linearity in the modulator with the key being that the slopes at the input only needs to be matched at the transition points. Slope is equivalent to gain. Hysteresis in the core messes with that. As suggested the solution is more gain Egor.
Yes, I brought up that report in another thread.
What I don't understand (but I am not a class d engineer) is why not take core hysteresis out of the equation.
I measure the series output coil of an NC500 amp to be some 11 uH (ten turns on an air gapped pot core). Together with the parallel capacitor making a low pass filter.
16 turns on a 35 mm round bobbin also make an 11 uH coreless coil....and air coils have neglectable hysteresis
DC resistance of that (just by experiment) air coil is around 0.04 ohm.
Not sure if that would impair damping factor a lot (the coil is within the feedback loop), but when necessary lower DCR coils are simple to make using thicker wire.
Are there other reasons why class d amps exclusively use pot core coils instead air coils?
Last edited:
The only issue that initially or immediately comes to mind as it is often discussed is that of stray fields. It has even been suggested that toroid inductors cause problems due to flux leakage at the wire entry/exit point where the winding is not closed. I think that was in a discussion about ICE power, it could have been physically gapped ferrite toroids rather than amorphous mixes. The cores had to be mounted gap upwards away from the board to minimise the issue. Either way you can imagine that stray flux is likely to be a serious pain in a low noise environment. UCD/Purifi use closed external leg RM cores, gap is in the centre leg, also screened by the winding.