|
@kawzx7
Any Purifi module isn't junk! - they are all excellent.
The problem was the mix and match approach that spawned this old thread had the expected result. If you want a class D amp to sound good, you have to give it a respectable power supply like you would with any other amplifier. If that happens to be a switch mode supply, to really make it work it should be designed for the application rather than something off the shelf, even if that off the shelf source is excellent.
The input buffer can have a big effect too. Its a sad state of affairs that so many class D amps get shot down on this account.
Any bugs in either of these departments and the whole thing goes down the loo.
If these issues are properly controlled you find that a class D amp can be just as good as the best class A amps price no object.
|
Seems unlikely that this would be the case in both channels (assuming
that's the case). Since this is a stereo amp, perhaps an issue with the
power supply? I agree. I don't know how the Purifi module works, but if its anything like ours there is a low voltage side of the module where the comparitor and triangle wave processor reside. If the power supply has a regulator for all that, it could be that something with it is amiss- for example, the regulator is oscillating. At any rate it seems a good idea to get it checked out. |
Sounds sort of metallic, "buzz-y", which is why I want to call it a distorted sound. That can be caused by an offset problem at the input of the comparitor- if its something like that I would regard it as a malfunction. |
I can't agree with this statement Atmasphere, it really depends on the
power supply, how it was designed, and circuit limitations.
Most
switch mode power supplies for power amps I see for audio are not power
factor corrected. They tend to be single stage LLC, with a simple bulk
capacitor arrangement on the front end. Poor design choice in my mind,
but is what it Is.
The problem I was getting at is for the most part you can't use an off-the-shelf SMPS and get top drawer results. I think you could if you had a supply custom built but IME that's a very pricey thing to do- and not something that most 'high end audio' manufacturers can pull off due to the quantities involved. Obviously storing electrons at a higher voltage allows a lower capacitance, since voltage dominates the energy formula; I think the problem might come in when that conversion to the actual voltage being used has to occur and occur quickly. We were able to use SMPSs for prototyping, but continually ran into current limit shut down issues and the like. After a while (and also once we had sorted RFI issues) it got much easier to simply use a conventional power transformer. We still had to be careful about the power supply since it had to have a number of protection circuits built in. |
Take it up with the likes of D’Agostino, Curl, ect ect manufacturers of
the greatest amps that are recognized as being able to drive speakers
with loadings that make Class-D’s run for the hills, they will tell you
the same thing. Actually they won't. They will all tell you the same thing- the limit of any solid state amp, class D or no, will be current the output section can withstand, the ability of the heatsinks to adequately cool the devices and how much current is available from the power supply. I will note one thing though- I've seen a number of SMPSs (Switch Mode Power Supplies) used in class D amps. I think its a bad practice. They work, sort of; the problem is that class D amps go from almost no current draw at idle to quite a lot at full power. SMPSs don't have a lot of capacitance at their output and so can be drained faster than they can replenish. A lot of them will shut down if you put a lot of capacitance at their output. So if you're not running an amp with a regular power transformer and a lot of capacitance available for the class D amp to feed on, you're probably not hearing what it can do. |
Show us just one that’s not, and can almost double it’s rms measured
wattage from 4ohm to 2ohm. The true indication of not current limiting,
like many good A/B bjt amp. George, seriously, you would do yourself well to educate yourself on how amps work, at a basic level. Your metric seems to be "at full power". That is actually causing you to have confusion; its a misleading and lousy metric. What is important is that the amp behave as a voltage source, and by that I mean that it will make constant voltage regardless of the load impedance, at music power levels. If you exceed the power the amp makes into whatever impedance, it will clip (overload). I'm stating stuff that should be obvious but because you trot this trope so often, I have to assume that everything I'm telling you here is new for you. With any amplifier the designer is faced with limitations; often verses cost. For example, our module has GaN output devices that are rated at 35 amps. This means that we should not exceed 17 amps in the field if we want the output section to survive; employing 50% margins is a good idea with any kind of semiconductor! In addition the heatsink has to be effective, and yes you do need them for GaNFETs if you plan to run the amp at higher power levels, and especially if that involves lower impedances. So far all traditional solid state amps all have this factor in common too. Finally you have to look at how much you want to spend on the power supply. Given you know the limitations of the output devices there's an argument for setting the power supply current to a certain maximum so that current will not be exceeded. So really what this comes down to is what I've been harping to you all along: the current capacity of the output section, its heatsinks and the current available to support the output power of the amp are the variables and these are no different than traditional solid state. What **is** different is that the output impedance of any class D amp employing GaNFETs is several times lower than any traditional solid state amp. That is why they can act as a voltage source to a much lower impedance than traditional amplifiers. So your metric of 2 ohms is not harder to attain than it is for traditional solid state should the designer feel its important. |
Take that as I would as colouration distortions, countering what they
say earlier in the same sentence as "The NC500 is practically devoid of
any sonic signature ” Actually the two comments sound entirely in agreement with each other. |
No they are the output buffers in the rear of the MSB discrete R2R dac,
which can swing 7.5v and do it with ease into 2kohm, with no temp rise
or voltage drop or heat sink needed into the raw NC500 module
Did you test to see if they actually do that in this application? There are a lot of variables here, not the least of which is how surprising 1 watt of power actually is and how distortion can affect your perception of sound pressure, to say the least. The issue is performance, and without a proper low impedance buffer, at
the input, not 6 feet of cable, with capacitance and inductance, the
performance suffers in my testing (with actual test equipment).
I don't doubt he did it, just pointing out that it is not an optimum implementation. As you might know our MP-1 is a tube preamp but can drive this through 30 feet of balanced cable with no worries- but that is our module, not a Hypex or the like to which I have no exposure. I would think that due to the low impedances involved the interconnect issues wouldn't be an issue. But there is a lot to unpack here; IMO the whole thing can be taken with a grain of salt without more information. |
No not really, the output buffers have .7ohm output impedance, but use a 50ohm series resistor for protection @georgehifi Are you saying you built a preamp or buffer circuit using these ICs? |
@unsound Thanks. There's still a bit of missing information, but you might be able to 4-5 volts out of one of those DACs (balanced), which would certainly be enough to make our module play although you wouldn't be able to drive it to full power. I can't speak to the Hypex but based on the idea that its input is about 2K (and thus similar to our module in that regard) I would not be surprised that George did this.
|
No, it is not, especially with the Hypex amps. The input is meant to be driven by a very low impedance source. In order to make the comparitor work properly (no offsets, no oscillation) with all the comparitors we tried we did wind up with an input impedance of about 2K. Through a lot of our prototype process, we simply drove this with our MP-1 (or MP-3) preamp, which has no problem at all driving a load like this. Of course, it was obvious that the circuit needed more gain and an easier load to work with most preamps. But it worked fine since we were not including the input buffer in the feedback loop anyway. FWIW our MP-1 can drive 32 ohm headphones- its got a miniature tube power amp (with direct-coupled output) as its line stage. So it is possible that George was telling the truth, but that would mean that he had a pretty gutzy preamp on hand. Most preamps I can think of wouldn't drive a load like that, so I do think his comments should be held in abeyance until he can explain how he did it. Since you really don't need much gain from the input buffer, almost any opamp that can drive 600 ohms (and most can these days) can do the job since your gain only need be a value of 2 or 3. So it would not surprise me in the slightest that the kind of opamp used has an inaudible effect. A proper buffer will have more than just one opamp of course, especially if you are executing a balanced input. But they would not need much gain so the feedback on the opamp would be very high! IME its when you are asking more than about 20dB of gain that you start hearing differences between opamps. |
Using King Crimson's track to make a final evaluation ? Ridiculous. Formantera Lady, from King Crimson's Islands LP; very nicely recorded. I play it at audio shows quite often as it can really show off a system! |
Can you illuminate me on what types of differences, among equally rated
and types of fuses you have encountered? How precise did the DVM have
to be for you to measure it, and what qualities do you look for?
The voltage drop across the fuse varies. The better sounding fuses correspond with the least voltage drop ('better sounding' doesn't always correlate to 'boutique', just to be clear). I use a Fluke DVM that's 3 1/2 digits. Like I said, its not rocket science. I prefer to use Littlefuse or Bussman. |
Almost all tube manufacturers....they pick caps, etc. by ear....and they don't measure any different. This statement is false- they do indeed measure differently (whether some companies bother to measure the differences is a different matter). We use custom made resistors as well (made by Caddock); they sound better but they also measure better. Most equipment stand manufacturers......I mean, can you measure a better stand?
Certainly! Sound Anchors was measuring the differences in performance of their stands 30 years ago. Its not hard. More like calling you out, on your snake oil mods, that you don’t backup
with any measurements, just a fusers "trust me it works" Its easy enough to measure the effects of a fuse, and the different fuses. Its not rocket science either. All you need is a DVM. Having said that I'm personally not a fan of boutique fuses. At least one of them uses a Teflon tube to 'damp' the fuse element. The problem is the fuse actually has to blow in order to protect; this type of fuse can have the metal of the fuse element go liquid and finally to a plasma state simply because of that Teflon tubing in which its held. Such a fuse will not reliably fail at its rating, and thus we don't recommend them. |
