Class D Bass amp?
- Thread starter Gus
- Start date
Help Support GroupDIY Audio Forum:
bcarso
Well-known member
I haven't played with this design, although from a cursory glance at the performance claimed I would probably favor one of the larger chips in the NXP series, which I have used and gotten decent results.
While I haven't any first hand experience with either.. the NXP approach strikes me a damn clever, using the lag in the output filter to dither the PWM... Is that a Bruno design? http://www.nxp.com/acrobat_download/usermanuals/UM10155_2.pdf
While I am a little outside my comfort zone, I believe there are some first order reductions of PS problems with the bridged output drive (Nat uses). While the NXP take it's negative feedback after the output coil, so PS issues get reduced by NFB.
I have been meaning to do some bench work on a low performance class D for improved battery life in my little tuner... but the complexity so far has made it not worth the effort. I don't need anything as fancy as either approach. NXP has an interesting little 2w chip with 70%+ efficiency at 4 Ohm. Better at 8 ohm.
JR
While I am a little outside my comfort zone, I believe there are some first order reductions of PS problems with the bridged output drive (Nat uses). While the NXP take it's negative feedback after the output coil, so PS issues get reduced by NFB.
I have been meaning to do some bench work on a low performance class D for improved battery life in my little tuner... but the complexity so far has made it not worth the effort. I don't need anything as fancy as either approach. NXP has an interesting little 2w chip with 70%+ efficiency at 4 Ohm. Better at 8 ohm.
JR
bcarso
Well-known member
JohnRoberts said:
For reasons that I cannot fathom, NXP appears to have made no attempt to port Putzeys' wonderful UcD design onto a chip. Although Bruno has continued to develop the approach for one of his affiliated companies, Hypex, and for many good reasons doesn't tend to share the details anymore that I know of, the reference you link is the same-o same-o from the earliest disclosure by Bruno. Maybe someone at NXP is working on chips as we post.
At least one person in the organization is well aware of its advantages and has done higher-power versions; in keeping with what must be some sort of characteristic sense of humor over there I think the one guy called the last one "D Best"* :
The 892X series amps that I recommend with qualification over the NS chipset are conventional fixed-frequency PWM designs. They maintain good control over deadtime and the body diode reverse recovery noise, making emissions easier to control. Their specs are not stellar overall compared to some other more elaborate constructions but I think they represent excellent value, and are also amenable to synchronization, something which UcD does not lend itself, being an inherently variable frequency self-oscillating machine.Hypex is hypersensitive about this last issue and have insisted that you will not have a problem with heterodyne artifacts using their amplifiers. I haven't tried to use a group of them in a system yet, and they may well be right. I am deep into an alternative topology right now, and despite two channels on a board with sometimes different self-oscillating frequencies I have only been able to detect heterodyning under very contrived conditions---so maybe the issue is easier to deal with than I had thought. I will have to give Hypex the benefit of the doubt until it is proven otherwise.
*This following another big thigh-slapper, "State of D Art" (groan)
I'm not sure that syncing one of the two major transitions in a PWM switcher buys you all the much.. the other edge will always be modulated by the audio, or load in a PS switcher... The free running PWM just means both edges are varying.
I repeat I haven't had one of these puppies on the bench, but I recall going through the mental exercise from syncing a switching PS in a DSP platform, and finally determined it wasn't worth the trouble.
JR
I repeat I haven't had one of these puppies on the bench, but I recall going through the mental exercise from syncing a switching PS in a DSP platform, and finally determined it wasn't worth the trouble.
JR
bcarso
Well-known member
It's not an issue with the performance of a single channel so much. With a fixed frequency PWM amp with synchronized multiple channels (i.e., same reference oscillator), you don't get heterodyne artifacts at the difference frequencies (well, that is, the difference freq. is zero, a.k.a. d.c., so there can be offset voltage generation)---that's one of the appeals of such.
With a multiple-channel variable-frequency amp system you can get audible artifacts, and they manifest usually as whistlers or birdies at ~zero signal. If a given pair of amps is offset at the zero-signal frequency by a few kilohertz (region of high aural acuity), it can be heard deep into the noise floor. So even if it measures well, it can be audible if you are listening for it. And potentially embarrassing.
Now most of the time the music overwhelms this sort of thing, so if you are synchronized at zero signal there is not often much of an audible problem. The other approach is to intentionally offset each amp from the other by, say, thirty-forty kilohertz. With that strategy at least you won't hear birdies at quiescence.
And if you isolate power supplies, and lowpass-filter inputs adequately, and use input devices with low rectification propensities (like JFETs), you may not have much of anything to worry about.
With a multiple-channel variable-frequency amp system you can get audible artifacts, and they manifest usually as whistlers or birdies at ~zero signal. If a given pair of amps is offset at the zero-signal frequency by a few kilohertz (region of high aural acuity), it can be heard deep into the noise floor. So even if it measures well, it can be audible if you are listening for it. And potentially embarrassing.
Now most of the time the music overwhelms this sort of thing, so if you are synchronized at zero signal there is not often much of an audible problem. The other approach is to intentionally offset each amp from the other by, say, thirty-forty kilohertz. With that strategy at least you won't hear birdies at quiescence.
And if you isolate power supplies, and lowpass-filter inputs adequately, and use input devices with low rectification propensities (like JFETs), you may not have much of anything to worry about.
sahib
Well-known member
Last night I was having my quality time in front of tele, flicking through Elektor 300 series circuit books and I came across this (I have been an Elektor reader since 1977). It is an older version of the chip but from the same family. I thought you might be interested if you are considering building an instrument amp. If you need high res scans pm me your e-mail address.
Attachments
sahib
Well-known member
sahib
Well-known member
sahib
Well-known member
Junction
Well-known member
These guys do some cost effective Class D modules or even up to 9000W in a 1 ru rack box and only 8kgs.
Michael
http://www.cadaudio.dk/cadpwmmodules.htm
Michael
http://www.cadaudio.dk/cadpwmmodules.htm
Similar threads
