Good catch! IMHO one of the reasons for the clarity of the GS is their independent suspension for the Mid/HF enclosure. When playing loudly you can't feel vibration from the LF drivers on the Mid/HF enclosure.
Also as was just pointed out, the systems are 24-bit DSP crossovers and eq to individual amps for each driver. Not cheap, but pretty well engineered and about to get better with the GS4s.
Cheers!
Howie
Interestingly enough, my perception of sonic problems with early class D was a "straining" sound to them, compared to a relative "ease" of linears back then. (sorry about the subjective terminology) They tested fine with sines and sweeps, but early Class Ds had a grunt or grit to the sound, especially when reproducing massed passages like full orchestras. I always attributed it to IM between PWM ripple and audio content. Non-linearity in the output inductors could encourage this phenomena.
I have to say I also heard this kind of distortion on Bob Carver's magnetic field amps as well...although I know these were just linears with "smart" power supplies. I auditioned a Carver M1.5T amp once and compared it to both an NAD 3020 and a Luxman L100. The L100 was flawless, the NAD extremely good but limited output and the Carver M1.5T sounded in comparison like someone straining to drop the kids off at the pool.
Since the power supplies are in series with the amps, lack of filtering of an SMPS could cause a situation similar to Class D issues...of course the Benchmark AHB2 amp uses a SMPS but it must be extremely well filtered; in my experience the amp performs very well and sounds like a high quality linear amp.
Just my observations, YMMV
Howie
Remember most early class D was open loop and not a low source impedance by accepted standards. Closed loop is much more common today.
I think this often gets forgotten on here.
I always admired the old kef reference Daleks with the pods on top for mid and treble. For domestic use, where ultimate POWA handling is not as important as in the studio if you stretch the mid down further you almost hit a minimonitor plus stereo subs arrangement which to me is the way to go at home
Happen to have an AHB2 right here. It was my main power amp until I A/B compared with with a modified Aragon 8008 Mk II. AHB2 sounds lower distortion, but strangely it loses some subtle detail information that the Aragon reproduces. I have an as yet unproven suspicion that there is some type of noise-modulation going on in AHB2 (aka 'noise floor modulation' so-named as it tends to appear on a typical FFT). It means noise is dynamically changing according the audio signal and masking small details (an effect, or symptom, also sometimes seen in dacs, but for different reasons). IME the effect can be hard recognize without careful A/B amplifier comparison. However, I do notice the same general perceptual effect from my laptop's built-in sound card as compared to the more detailed sound of a near SOA dac (not Benchmark DAC-3, better than that). The laptop dac has higher distortion too, but that's a different sounding type of effect.
Last edited:
That is interesting! Can you perform a measurement to quantify the effect you are hearing? I have noticed that the same piece of gear will perform differently depending on what system it is interconnected with, possibly due to stray capacitive coupling or ground currents...
Cheers!
Howie
Cheers!
Howie
Why magnetic hysteresis comes in discussions lately only when class D amps are mentioned? (A: because Bruno focused on it).
Well, nothing novel there, all magnetics have it to a degree depending on their B-H curve. (repeating myself here) The Black Hole......
George
Whatever that means. 😴 It could be level matched double blind or just subjective comparison. It's unknown without the details described.
Although Bruno doesn't say so, could be the hysteresis may be more of an issue when feedback is taken from after the output inductor. In other words, maybe its the response of the feedback loop to the distortion that creates some of the audible effect. Dunno.
Just curious '"when feedback is taken from after the output inductor" in a class D. Is it digitized and digitally subtracted from the digital word the amp is trying to reproduce, or is there some analog trick going on in the modulator that makes use of the signal directly?
Sorry I havent studied it myself to know "how it works"... Perhaps all_digital_right_to_the_Speaker_Terminal hasnt fully caught analog in all aspects (like "some subtle detail information") - yet. I'd think that's only a matter of time and once achieved (i.e. complete transparency) then comes all the emulations; "we can put back this tone, this or that distortion - make it sound however you like!"
I realize they do that already with guitar amps / speakers which is of course far more coarse than, say, emulation of various high end amplification sound. It will come, would be my point. A virtual "Bob Carver in a Hotel room" inside the DSP embedded in such an amp.
OK, maybe a little too far ;')
Sorry I havent studied it myself to know "how it works"... Perhaps all_digital_right_to_the_Speaker_Terminal hasnt fully caught analog in all aspects (like "some subtle detail information") - yet. I'd think that's only a matter of time and once achieved (i.e. complete transparency) then comes all the emulations; "we can put back this tone, this or that distortion - make it sound however you like!"
I realize they do that already with guitar amps / speakers which is of course far more coarse than, say, emulation of various high end amplification sound. It will come, would be my point. A virtual "Bob Carver in a Hotel room" inside the DSP embedded in such an amp.
OK, maybe a little too far ;')
Class D is switching, not quite the same as a R2R or delta sigma. They have an analog comparator at the center and the closed loop versions have feedback to that comparator. Its more of an analog than digital process. They are closer to a switching power supply than a DAC. I believe some German company did make a "power DAC" but its lost in the mists of time.
A normal class AB power amp can operate in class D, I think many amp designers will have done it by accident. You just need to get it to oscillate at low enough frequency that the oscillation reaches past the rails. This is basically self-exciting class D. With a positive input signal the amp will spend more time clipped at the positive rail.
So I don't see class D as strictly "digital". And the feedback is very analog, unless you deliberately add a microprocessor. For instance TPA3xxx chips do not have any processing in them AFAIK although it might be easy to imagine that inside the chip.
I thought there might even be a few amps designed to operate in both class AB and class D.
So I don't see class D as strictly "digital". And the feedback is very analog, unless you deliberately add a microprocessor. For instance TPA3xxx chips do not have any processing in them AFAIK although it might be easy to imagine that inside the chip.
I thought there might even be a few amps designed to operate in both class AB and class D.
Class D is indeed strictly analog. It operates on time, voltage, current, all analog quantities.
A digital amp would process information as a digital quantity, a set of ones and zeros like a digital audio signal itself.
The only example I know of a digitally feedback/controlled amplifier is those that use the Axign chip.
Jan
A digital amp would process information as a digital quantity, a set of ones and zeros like a digital audio signal itself.
The only example I know of a digitally feedback/controlled amplifier is those that use the Axign chip.
Jan
Last edited:
Yes, but OTOH applying the linear theory to Class D has it's unique set of challenges. Like, for example, applying negative feedback to a system (an open loop class D amplifier) with a "dead zone" transfer function (in the time domain).
Attachments
I have an off-topic question regarding the cap value after a DC umbilical cord. First the spec’s:
Wire: 23 ga twisted pair, Teflon jacket
Length; 4 ft
DCR: .088 Ohms
Impedance: 122 Ohms
Capacitance: 48 pF
Inductance: 719 nH
My question is what value Capacitor do I use at the end of the umbilical? If I use the DC resistance for a low pass RC, the capacitance is very large for 15 Hz cutoff. Or what is a practical low pass cutoff – i..e. is 15 Hz impractical in this case? Is the filter only for high frequencies? 100kHz, 1 MHz?
Wire: 23 ga twisted pair, Teflon jacket
Length; 4 ft
DCR: .088 Ohms
Impedance: 122 Ohms
Capacitance: 48 pF
Inductance: 719 nH
My question is what value Capacitor do I use at the end of the umbilical? If I use the DC resistance for a low pass RC, the capacitance is very large for 15 Hz cutoff. Or what is a practical low pass cutoff – i..e. is 15 Hz impractical in this case? Is the filter only for high frequencies? 100kHz, 1 MHz?