@Gedlee: yes it has become a rabbit hole - but glad that some folks are finding put about it.
@Pallas: rather than dismissing a $50 amp with "oh it's not hifi" because of "xxx" (insert any number of esoteric requirements) - why don't you just try one? Look at Figure 8 in the data sheet where it shows THD at different frequencies vs power at say 24v supply. If you don't want to waste $50 get the $13 Ybdz variant from Aliexpress and listen for yourself before summarily dismissing something that so many people think sounds great.
http://www.ti.com/lit/ds/slos708c/slos708c.pdf
@Pallas: rather than dismissing a $50 amp with "oh it's not hifi" because of "xxx" (insert any number of esoteric requirements) - why don't you just try one? Look at Figure 8 in the data sheet where it shows THD at different frequencies vs power at say 24v supply. If you don't want to waste $50 get the $13 Ybdz variant from Aliexpress and listen for yourself before summarily dismissing something that so many people think sounds great.
http://www.ti.com/lit/ds/slos708c/slos708c.pdf
Wait a minute: FLAT FREQUENCY RESPONSE WHEN DRIVING AN ACTUAL LOUDSPEAKER is an "esoteric requirement?"
As for "trying one," why would I waste my time doing that? I have perfectly good amps. Nothing can be better, and anything different will be worse because an amp can only sound different from mine if it materially degrades the signal.
Um, THD ≠ frequency response. I never wrote a word about THD. Who cares about THD when an amp's frequency response varies with the driven load?
Take measurements of speakers with varying impedance driven by that amp and then driven by a known good one, and you'll see what I mean. Basically a board like that is good value for a closed system, but otherwise of little merit.
Last edited:
Finally had a look at your polar mapping software, despite my general aversion to not-a-Mac software. Great stuff! Not at all surprised that the Q15 measured so well. I had a pair from 1996 until 2005 or so. Couldn't find anything that sounded good enough to be replacement, though I tried several times because they looked like cheap things and had awful build quality. The baffle was a plastic molded piece that just press-fit into the cabinet! That was the price for made in the UK, good, and cheap, I guess.
Your provided link goes to a cirrus datasheet and nothing to do with the ncore.
I am aware of the ncore datasheet, what you seemed to have missed was the point of my post.
I've made amplifiers that show the ncores 50 watt distortion performance but at only 1 watt.
What I am saying here, is that if the performance of the ncore is being limited by the noise floor at 1 watt (because the signal is being attenuated enough such that a measurement at 400 watts is possible). Then I would like to see measurements for the ncore where the set up has been fully optimised to provide the best measurements possible at only 1 watt. This would require reducing the effect of the resistor divider going into the AP such that the signal to noise ratio is optimised for a signal level of just 2.83vrms. Then and only then would you see exactly what the output looks like.
There are no small signal distortion results, nor small or large signal FFT plots showing just the amplification of a 1kHz tone for example, which would be great at showing what the harmonic spectra of the amp actually looks like. It's nice and all, seeing 18.5+19.5kHz IMD tests, but when the measurement bandwidth has been limited to, what appears to be, 20kHz, that's a very good way of hiding certain things.
TIs measurements of their TPA311x series of amplifiers fall foul of the same thing, where they limit the measurement bandwidth. This causes the distortion to start dropping like a stone beyond around 8k because the third harmonic vanishes, then drops at 10k because the 2nd harmonic now disappears. It looks better on paper, but if the bandwidth was 100k we'd be looking at the distortion continuing to rise with frequency, rather than it falling off.
The only reason to bandwidth limit is to reduce the apparent noise floor, where it is sometimes necessary to see very low distortion with smaller signals. Or to eliminate ultrasonic noise out of a system that would otherwise cloud the results.
Attached is the 2.83v performance of said amplifier.
From my perspective the first watt of an amplifier is the most important one because this is where I do 99% of my listening at.
Attachments
Why would anyone do anything so . . . silly . . . as that? The whole point is to put between amp and driver exactly what belongs between amp and driver . . . nothing but (short pieces of) wire. Crossover, phase adjustments (delay), response compensation and all that stuff should be done digitally (or analog at line level) where it belongs, not with big coils of wire and foil stuffed into some speaker box where it can only foul things up.
You'll be hard pressed to find anything better for 1 watt that's for sure.
The load was a 9.4 ohm dummy resistor.
And no, I cannot do that. Why?
This is why.
I don't have a large three way to hook it up to. The amplifiers are low power class A designs for active use.
In the past I have done measurements of an amplifier driving a dummy load and of an amplifier driving a two way loudspeaker just to see the difference and there wasn't much.
Besides why would you want to measure the amplifier when it's hooked up to a big three way? Amplifiers are all tested into purely resistive dummy loads to give us a nice ground for comparison. If one were to start hooking up loudspeakers the measurements would be completely meaningless, unless everyone did the same and with exactly the same loudspeaker...
5th Element and Dewardh: excuse my ignorance on this area but I only ask because Pallas suggests that this amp is no good because it doesn't have a flat freq response with a varying load. Not sure how one goes about showing that it is. I think it is a non issue but just thought I would ask. You are right, I have only seen amps tested on resistive dummy loads. Yes, at 1 watt I think this probably beats many big iron room heater class A amps.
Yes, bad copy/paste.
Correct link is http://www.hypex.nl/docs/NC400_datasheet.pdf , as you might have guessed.
That would be 110dB S/N at 1W and less than 8uV of residual noise.
What you see in that measurement (fig 9.2) is the real performance of that amp, with 0.001% at 1W for all frequencies, translating to a 100dB unweighted S/N at 1W.
If that is too much noise for you (for a compression driver I suppose?) then you can still put an lpad or an autoformer at its output.
Last edited:
Pallas is talking about the frequency response here not the distortion. I was discussing the distortion.
The fact is that no amplifier is load invariant it's just a matter of degree. With non class D amplifiers this is governed by the amplifiers output impedance, the higher it is the worse the result. If the amplifier has a low output impedance, which most do, this is largely an irrelevance.
With class D amplifiers, that take the feedback point before the output inductor (some don't even use feedback, but we'll ignore those completely), the upper end frequency response is dependent entirely on the reactance of the inductor in combination with the loudspeakers impedance.
If the inductor remains fixed in value and the loudspeakers impedance changes, then so will the frequency response. In other words, if you've got a perfectly flat pair of loudspeakers that are 8 ohm and a perfectly flat pair of loudspeakers that are 4 ohm. If you hook each one up to the class D amplifier with feedback before the output inductor. Then you may end up with a flat frequency response with the 8 ohm loudspeaker, but it wont be flat if you then change to the 4 ohm one. It might not be flat with either loudspeaker.
A class D amplifier whose frequency isn't load invariant isn't useless providing you have compensated for its effect. The trouble is most people don't. That isn't to say it's entirely useless, it's just less precise in the top octave.
Which is why I am saying that it might actually be lower if the measurement was taken in a way that didn't allow the full amplifier voltage on top of the noise floor.
As it stands that measurement had the capability to measure the voltage swing for 400 watts into 4 ohms or something, which is going to significantly reduce the SnR for a 1 watt measurement.
This is the trouble with output power vs distortion measurements, they are great for showing certain things, but most certainly not for showing you the amplifiers true performance at much lower signal levels and this only get worse the higher the power of the amplifier.
I am trying to say that the ncore probably has better performance at 1 watt than that measurement shows, but we wouldn't know unless the measurements we retaken to maximise the SnR at smaller voltage swings.
It is low, but class D amplifiers require an inductor at their output to eliminate the switching carrier on which they work. The inductor is only very small, but in principle it works in exactly the same way as an inductor in a passive crossover. As a result the high frequency response of a class D amplifier is dictated by the impedance of the loudspeaker that is connected to it via the reactance of the amplifiers output inductor.
Some class D amplifiers include the output inductor within a feedback network as thus compensate for the effect. Most do not however.
I am waiting for the day when TI will include output inductor inclusive feedback, but as the trend seems to be moving closer towards output inductor-less class D amplifiers (possible at lower power levels with short cables typical in integrated electronic products such as TVs), I do not know if this will happen any time soon.
The TPA311x series will adhere to EMI regulations when using a simple ferrite bead and shunt cap on the output, if twisted and short cable runs (125cm) are used.
Some class D amplifiers include the output inductor within a feedback network as thus compensate for the effect. Most do not however.
I am waiting for the day when TI will include output inductor inclusive feedback, but as the trend seems to be moving closer towards output inductor-less class D amplifiers (possible at lower power levels with short cables typical in integrated electronic products such as TVs), I do not know if this will happen any time soon.
The TPA311x series will adhere to EMI regulations when using a simple ferrite bead and shunt cap on the output, if twisted and short cable runs (125cm) are used.
Under such closed-system conditions, the performance requirements for amps are considerably relaxed. Most Class D and tube amps are just fine. But for those of us who do use passive crossovers or hybrid active/passive setups, a better amp design is required.
Stereophile tests into a published "simulated loudspeaker" circuit designed by Ken Kantor. That's generally the only part of their amp reviews that's worth a damn. Which then again gives Stereophile's amp reviews one thing over basically every other published amp review anywhere else, because that test reveals flaws of amps with high source impedance or poorly designed output filters that are actually audible in real listening.
Last edited:
The only thing they test into the simulated load is the frequency response of the amplifier itself, ie really to see how amplifiers with higher than average output impedances affect the frequency response. They do all their other testing into purely resistive dummy loads.
It isn't hard to infer how an amplifier is going to perform into a horrible loudspeaker load (in terms of distortion) by looking at how the amplifier performs into lower ohm dummy loads.
Power response can be straight regardless of lobing, and lobing axis can be more directive than non-lobing axis, reducing weight of lobing responses all around the room. It's also common recommendation to have some acoustic treatment for 1st order (vertical/lobing) reflections - also with coaxials.
I don't hate coaxial and other coincident arrangements, but it's not really necessary to get excellent sound into the spot, imo.
It's very good speaker, and error in axial and power response (by horizontal meas) is mainly based on cross-over implementation, fixable with 18u external cap.
Responses are windowed 6 ms at 1 m; valid >300 Hz.
Blue=Power, Black=Axial, Red=DI.
Off-axis 0...180 deg hor, step 5.
An externally hosted image should be here but it was not working when we last tested it.
IMD 500 Hz + 13 kHz, 90+84 dB/1m
An externally hosted image should be here but it was not working when we last tested it.
Not bad, but possible without coax and with better tweeter.
Last edited:
Actually, no. It is probably closer to the truth to say that the choice of an inferior passive crossover forces you to use what is likely to be an inferior (but nevertheless probably very expensive) amplifier as well.
I don't think so: all these % and db figures are matching the 25uV residual noise measurement, and I guess this one is easy to make correctly.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.