The "Elsinore Project" Thread

This is a purifi based class d amp with custom linestage and psu, and as such will face comparisons with many other amps. Its main pull is the price - I wouldn’t bother dyi-ing my own if I wanted to go down that route. Its main downside is the lack of integrated streamer option, as many of its purifi-based competitors do have that, and it makes for a neat and cost-effective single-box solution.
 
Was just looking at this exact amp or even it's bigger brother the p422. Not sure if the extra power is needed...

Because the Elsinores are basically 6 Ohm, then we are looking at around 200W per channel. The Elsinores can go pretty loud with a 20 Watt amplifier. How much power do you want?

What I do like is that the P262 is still able to drive 2 Ohm and 400W plus. I doubt very much that I will be using speakers with such a low Z, but it could be interesting for testing purposes in the future (I have something in mind).

I simply bought if from the website for AUD $1899.99 and the process added $71.50 to send from Western Australia to Sydney. I suppose for export, the 10% GST/VAT comes off and then add postage to anywhere in the world. It is only sold direct and I think the price is very reasonable indeed. I have had a checkered history with Class D amplifiers, even built hybrid tube versions for myself, I have owned Tripath 6W (I still have one), Hypex UcD 180 and 400, and NCore 400.

But this amp is not just the best Class D amplifier I have heard, it is the best sounded solid state amp I have heard. Keep in mind, the kind of tube amplifier I like is not the fat and bloaty kind, I definitely want transparency and hearing into the music - 'cocktail' tube amplifiers is not for me.
 
When the power amplifiers (eg Hypex with custom buffers) have reached this superior level of quality and power, the sonic differences are almost undetectable. My Orchard Audio Starkrimson Ultra (GaN) was compared to one of the markets absolute best power amps (last Hypex generation). No difference was heard by the non-biased third part audience. (tube) Preamps and speakers dominate when it comes to coloring the sound IMHO. A matter of taste of course, how do you like your rum, with/without Coke or lime?
 
Yes, I can attest to that, the input/buffer stage in Class D is very important. But not if the Class D engine is not up to it. The best I heard was my own circuit based on zero feedback "Diamond Transistor" stage into (UcD400); this also a circuit, sans EQ, that I use in my JLTi Stand-alone Phono Stage. I also did one with the venerable 6DJ8/6922 tube and that sounded fine as well. The NCore 400 used its own input buffer stage and to me was a bit of a disappointment. But I have not heard the NCore 500 with a custom buffer. But all I have heard has been superseded in performance with this P262.
 
Hello All,

All this test lead stuff has been sitting open on the bench for a couple of days.

I did find that with and without the 8R test resistor in parallel with the driver there was no measurable difference in current distortion through the driver's sensing resistor.

APx555, APX1701 at 2.83 volts output. Edit 5 second frequency sweep

See the attached plot:

1) current distortion through 8R test load resistor + 0.1R Current Sensing resistor.

2) Current distortion through P17WJ-00-08 + 0.1R Current Sensing resistor only.

3) Current distortion through P17WJ-00-08 + 0.1R Current Sensing resistor With 8R test load resistor in parallel.

Test Plots 2) and 3) show that the parallel 8R test load resistor had no measurable effect on on the P17WJ-00-08 Driver Current Distortion.

Thanks DT

Current Sense Resistor THD Ratio -_ Smooth.PNG


Edit: Distortion tests with a microphone will be more interesting to me.
 
Last edited:
In #5511 I really ment electrically as in measuring the fundamentals on the input connectors.

No, not much in the acoustics FR /dist. But now we se above that there are really no difference using an 8 ohm resistor in parallel. To crown it, one have to do a strict blind test I suppose. Most probably it will not be detectable and if it was, there is something to be learned (I wouldn't hold my breath).

//
 
DualTriode said "the driver's sensing resistor" so it's in series with the driver, and the pair is in parallel with the extra 8 ohm resistor so it measures the driver current, not the amplifier current. So following, if the driver current is the same then the driver Voltage is the same and the measured response will also be the same.
 
  • Like
Reactions: DualTriode
1. That quick 8R resistor across the speaker terminals with global results and no need to do any design work. The results may vary, but in most cases you will end up with an 4 Ohm load and better sounding speakers.

2. Use the resistor across the driver terminals, this will have a more local and targeted effect. But this must be integrated into the overall design - and this is already done in some commercially available speakers and best to leave well alone.

3. The full discrete current EQ, this is also targeted but not much heat wasted. The other two above will produce will produce quite a lot of heat. My approach does not, not even close.

But there are heaps more benefits to EQ the current this way. Such as total insensitive to the amplifier's impedance (you can even use a current-drive). It also locks in the Crossover and makes it far more dynamically stable.

But all three just improves the sound and the question is why?

I have developed some ideas about that, that I believe can account for the better sound. It seems that there is an interaction between the speaker and no measurements yet. We need to measure what the amplifier does on the current side when the amplifier is a voltage source.

So I am working on a measurement that will. Wish me luck.
Now, would be the time.

/
 
Am I correct in noticing in 5512 there is -other than some minimal measuring differences- no measured difference in both SPL, H2 and H3 with and without the 8 Ohms parallel resistor?
The extra stuff you see in the plot posted in 5512 I believe is noise. The test plot used a test microphone that has its own noise plus there is additional noise from the analyzer and test amplifier cooling fans.

Thanks DT
 
DT, you realise that I am not surprised, I can do the same measurement and up to H5 at that and more. As we discussed earlier are you proceeding or just resting here? I made sure you got Esa's article and what are you going to do next? Harmonic distortion test like that is not going to show up much, you will need multi-tone IMD and showing products that go way beyond H3. Agreed?

I can assure you that those who have heard that 8R trick are not deluding themselves. There are additional elements involved when you are looking at crossovers in the usual 2-3KHz region. There is a Mk2 version of a compact 2-Way speaker system selling around ten grand around the world and I know the designers personally (for decades) and the Mk2 version has been hailed a great improvement over Mk1. The reason for this improvement? You guess... and reviewers just listened and made the comparison? The drivers are the same, they are just told that the crossover is improved. I am not sure, but I think the Mk1 might get returned and the 'improvement' fitted for a fee. This is a premium speaker using top-line Scan-Speak drivers.

Also, as you know, there is a potential 20dB distortion differences at mid-frequencies when using current source in the midrange. This is not speculation.

1682158374153.png


See that forward midrange driver in the Kii Three speakers, that is current-driven by what I believe is a modified NCore Class D amplifier. I have heard an Ncore amplifier that Menno Vanderveen demonstrated at ETF17 with an 18 Ohm output Z. I am sure he could have made it even higher than that as it was switchable to three different values, 0, 3 and 18 Ohm. The 18 Ohm sounded better and pretty much all agreed to that.

1682158619851.png


This is by SGR in Melbourne. See that dome midrange? These are also active speakers and that dome midrange is current-driven.

My point is, that using current-drive on the critical midrange, you are preventing distortion, you are not reducing distortion. If you hear that extra 20dB of distortion, then it is because it was prevented, not reduced.

Take for example the Purifi ULD drivers. Where it matters, which is "impedance modulations" and these drivers are relative free of that, then current-drive would have far less reduction in distortion. They get you the low distortion because they don't modulate the current of the amplifier anywhere near as much as just about any other driver out there.

Lars, if you read this, your driver shows the way to get the right current from the amplifier.

The parallel resistor trick needs to be tested with signals that are spread well apart, something that combines LF and midrange stimuli. Harmonic distortion or THD is not likely to show up much at all.

There is something else to add. The 8R trick works well in bog-standard 2-Way designs that have a peak Z in the midrange and that gets flattened out by the 8R parallel resistor. Even better is an LCR that is tuned to that peak and bring the Z as low as possible.

The crossover can make things worse, or they can make things better.

My 5:1 ratio, agreed upon by both Esa and myself in conversation ten years ago, apply that to the driver, add the highest value inductor possible - good drivers have lower inductance that is also more stable with motion/excursions.

Inductor-Drive.gif


The current sense resistor is only for testing purposes, so please ignore as it won't be used in a crossover. But the above example should be around -6dB circa 3KHz. It will give you the benefit of current drive around the crossover and the immediate octaves below. It should give you 15-20dB reduction (actually preventing) distortion that goes beyond the rate of roll-off. Now create the high-pass to match, may I suggest you look closely at the Elsinore high-pass design as a guide.

Voltage-drive can still do a great job. But you need the best drivers and thoughtful crossovers.

The perfect driver would work equally well with voltage-drive and current-drive - by preventing distortion.

Now you guys know why after suggesting a bunch of different drivers, very nice Sartori drivers and one would have worked very well, but examining the Purifi ULD driver "W" version, I said if you must, it had to be this driver.

I hope Lars reads this. I was suggested many different drivers and said no. It was those "impedance modulations" that made me say yes. I knew what I was looking for.

Late here, goodnight.

PS: I have been asked what the main difference between the SB17MFC driver and the Purifi ULD "W" driver? Both are great, the SB is a tad warmer, but what sets the ULD apart sound wise? Purity!
 
DT, you realise that I am not surprised, I can do the same measurement and up to H5 at that and more. As we discussed earlier are you proceeding or just resting here? I made sure you got Esa's article and what are you going to do next? Harmonic distortion test like that is not going to show up much, you will need multi-tone IMD and showing products that go way beyond H3. Agreed?
There may be some agreement, I am not sure to what extent.

If you start adding Harmonics they rapidly fall outside the usual bandwidth we usually use for audio measurements. Do not be discourage about the common practice of using H2 and H3. I put FR and H2 plus H3 on the same plot, there was a request for Frequency response. FR does not show up in the current sensing resistor only tests.

If you look at the previous post (5508) Total Harmonic Current Distortion was measured across the driver's current sensing resistor.

If you want Two-Tone tests here they are.

The plot with the added parallel 8R resistor seem to have a little less noise, otherwise they are pretty much the same.


Thanks DT

With Parallel 8R resistorFFT Spectrum.PNG
Without Parallel 8R resistor FFT Spectrum.png