Actually, this BJT circuit, for dead time adjustment (Sander, 'nano al.' only for it no more.), will have temperature drift more then 1ns/70C, fortunately, Lars no need such accuracy at all, because totem pole rise/fall >>>>>1ns. I guess, Lars hard to say how much exactly dead time in the his module, it's no instrumental matter, rather terminology. E.g. how much (in A, or mJ?) shoot through directly point to dead time = zero? Furthermore, Bruno's UcD400 drivers do little magic around shoot through, so, can anybody say how much d.t. here?
SSassen said:
I'm sure taking into consideration the type of controller and the size of the board, I don't think the auto-nano alignment could have been implemented. But the idea is cool. Don't really know what the benefits are and whether they are audible or make performance more reliable. 😕 Perhaps you want to elaborate on that more specifically about temporature issues on active versus passive setting, controller speed requirements, recurring and non-recurring cost, manufacturing tolorances and effects on performance, component reliability and failure modes effect, etc.?
I think, if it does not require a multi-turn precision trimmer, maybe it's not that sensitive to variation.
Well, if that's the case then 'nano-aligment' might just be a hollow phrase and not a genuine innovation perhaps? How much 'alignment' does one need really. Looking at Hypex' products it is safe to say their solution works beautifully, even taking manufacturing, component, etc. variations into account.
It is also clear, to those that have designed/built class-D amplifier from scratch, that if you indeed set the treshold for cross-conduction too tight, ie. time the mosfets in such a way that a temperature or component variation might cause them to cross-conduct, you're in for some headaches. In the end, due to the high switching frequency the difference between 8 or 10ns of delay is not going to reduce distortion tenfold, you maybe gain a few dB, but lose that again when temperature rises.
Ie. on paper it looks like a nice feature, in practice, across 100s or 1000s of boards shipped it isn't. That would require them to all be preset and run-in at the factory, or component variations will ruin any nano-alignment you try to accomplish. From the user end perspective you don't have much of a chance to set this nano alignment properly yourself, you'll need at least a 250MHz scope to be able to see a ns pulse reliably, and those don't usually sit in the bargain bin at Radio Shack.
Best regards,
Sander Sassen
http://www.hardwareanalysis.com
yeah this 1Mhz thing I find curious. I doubt it'll happen with this level of idle current, my finger tells me it won't do it, see what the scope say's soon enough.
Well I guess that explains you ignoring that question totally when I first asked it.
Now about the others... no need for concern, I'll figure it all out, piece by piece.
Ivan, you raise excellent points.
There's also the usual timing error in the control loop for a clocked based amp... this is self oscillating though so it's inherently corrected, the deadtime therefore seems to be just an idle current adjustment, where the higher the better in terms of THD.
It's funny but, they could easily be "nano aligned" to be 10us appart. Word games...
Well I guess that explains you ignoring that question totally when I first asked it.
Now about the others... no need for concern, I'll figure it all out, piece by piece.
Ivan, you raise excellent points.
There's also the usual timing error in the control loop for a clocked based amp... this is self oscillating though so it's inherently corrected, the deadtime therefore seems to be just an idle current adjustment, where the higher the better in terms of THD.
It's funny but, they could easily be "nano aligned" to be 10us appart. Word games...
classd4sure said:
Right On!
We didn't make a description of the trimmers, as we actually recommend people not tampering with the trimmers.
SSassen said:
Well i think we have established exatly what nano alignment is, both in word, and with help from IVX a few pages back, also with a schematic of somebody else's implementation of the 'nano alignment' function.
novec said:
Thank You for this post.
Well in the future i might have access to better ferrite materials, (an area with progressive development) so the wide frequency range gives the possibility to upgrade the modules by that time.
Also some people are very interested to play with the switching freq vs. sound. And depending on some external factors, you can now do that.
soongsc said:
Soongsc: Totally cool with an auto nano alignment. It would require some special circuitry though, if you wanted to avoid quantization error. I guess with a superfast gate array you could have the chance. The risk of drift is not in the trimmers as pointed out by Ssassen, nor is it in the BJT's as claimed by IVX. The biggest contributors to drift are the 47pF capacitors.
Would be nice, but unfortunately a very far cry from reality.
In your terms we could end up with a Class A - Class D amplifier. That would be nice wouldn't it? 🙂
The reality is, that when you use large amounts of dead time for 'break-before-make' safety (25-50 nS), then the coil will discharge the capacitive energy in the just-switched-off MOSFET. If you align the pulses up right on the spot, the opposite MOSFET will have to discharge the capacitive energy. That is why the idle loss is a bit higher than usual.
But overdoing it will of course increase the idle current, because of shootthrough, but i can promise you, the THD will not improve! 😀
It is? I'm sure you can fully substantiate such claims? In this case it is put up or shut up.
Best regards,
Sander Sassen
No affiliation whatsoever
Better laquer those trimmers then, or else you might get some DOAs due to people experimenting with them. That's the downside of having trimmers on a kit, people will always experiment, esp. those that forego any sort of common sense and will do anything to 'improve' their listening experience.
Either way, I'm just janking your chain a little Lars, trying to get some clear answers and some definitive information about your new modules.
Best regards,
Sander Sassen
Sound improvement debunker
SSassen said:
Sorry, I was unclear there. I meant "you" as the circuit designer, not as the end user.
SSassen said:
Isn't that what he's hired Paris Hilton to do? 😉
I read in HIFI4allDK Forum Digital that all preorders except those with 15000,80V caps (?? why haven't we been told about that option??) are now packed and I suppose will be sent out on monday.Is that correct?
Hmmmmmmmm.
With the 63V 10 000uF caps on my NCD supply, you kind of have to sandwich the cover on the case. My T-networks squeezed in nicely as well, but 80V 15 000uF, going to need a bigger case guys.
With the 63V 10 000uF caps on my NCD supply, you kind of have to sandwich the cover on the case. My T-networks squeezed in nicely as well, but 80V 15 000uF, going to need a bigger case guys.
The After Series
I'm aware how the wiring of the supply looks, like some sort of a spaghetti incident. Trust me, there's a method to the madness, and it has reduced heterodyne to much more acceptable levels, that's to say for the most part, it's now imperceivable.
If you have a careful eye you'll note alot of little changes. XLR are implemented, the obvious T-networks, magically floating above the supply (how'd he do that??), snubbers snuck in there somehow too, along with bleeders, unused plugs/connectors removed and a little twisting here and there.
You can't see the black gate caps installed, they're on the bottom side. The caps you see added near the op amp are panasonic FM's, 25V, 220uF. A good addition. This is how they stay.
I also took the outputs and bent them over the connectors as you can see, the goal here is obvious I should think, get them away from the power rails as much as possible.
There's no doubt more I can do... but this to me is the bare bones basic stuff I feel required in order for it to stand next to my UCD amp at all. Sadly I'm not granting Plitron the sale of another overpriced hunk of metal. Dealing with them just once left a bad taste in my mouth.
I'm aware how the wiring of the supply looks, like some sort of a spaghetti incident. Trust me, there's a method to the madness, and it has reduced heterodyne to much more acceptable levels, that's to say for the most part, it's now imperceivable.
If you have a careful eye you'll note alot of little changes. XLR are implemented, the obvious T-networks, magically floating above the supply (how'd he do that??), snubbers snuck in there somehow too, along with bleeders, unused plugs/connectors removed and a little twisting here and there.
You can't see the black gate caps installed, they're on the bottom side. The caps you see added near the op amp are panasonic FM's, 25V, 220uF. A good addition. This is how they stay.
I also took the outputs and bent them over the connectors as you can see, the goal here is obvious I should think, get them away from the power rails as much as possible.
There's no doubt more I can do... but this to me is the bare bones basic stuff I feel required in order for it to stand next to my UCD amp at all. Sadly I'm not granting Plitron the sale of another overpriced hunk of metal. Dealing with them just once left a bad taste in my mouth.
Attachments
- Status
- Not open for further replies.