Fosi vs NCore/Purifi: When would there be an audible difference?

[mfaughn]

I'm weighing options for which new amplifier(s) to purchase. The budget option is Fosi v3 or ZA3. The more expensive option would be an Ncore/NcoreX/Purifi amp (from Buckeye). Obviously the latter would have more power and is technically superior. It seems that a v3 of a couple of ZA3 should still deliver plenty of power for my typical use case (moderate volumes @ 3 meters with a variety of speakers). Are there any good reasons to believe that the Ncore/Purifi options would be audibly superior?

 

[MaxwellsEq]

I'm weighing options for which new amplifier(s) to purchase. The budget option is Fosi v3 or ZA3. The more expensive option would be an Ncore/NcoreX/Purifi amp (from Buckeye). Obviously the latter would have more power and is technically superior. It seems that a v3 of a couple of ZA3 should still deliver plenty of power for my typical use case (moderate volumes @ 3 meters with a variety of speakers). Are there any good reasons to believe that the Ncore/Purifi options would be audibly superior?

One of the areas of weakness (if you can call it that) of the current crop of chip amps is that they are less load-tolerant than the Purifi/ NCoreX equivalents. This is critical for speakers with complex and reactive aspects to their impedance curves.

 

[Zapper]

One of the areas of weakness (if you can call it that) of the current crop of chip amps is that they are less load-tolerant than the Purifi/ NCoreX equivalents. This is critical for speakers with complex and reactive aspects to their impedance curves.

Other than the +/- 1dB fr variation at 20kHz (due to the LC lowpass filters on outputs), are there measurements that show load dependency at lower frequencies?

 

[Blumlein 88]

Other than the +/- 1dB fr variation at 20kHz (due to the LC lowpass filters on outputs), are there measurements that show load dependency at lower frequencies?

That was with 4 ohm vs 8 ohm resistance. You can be pretty sure with varying impedance with frequency and reactance you'll get some variations that enter the audible range. Maybe not hugely audible (though maybe so with some difficult speakers), but definitely not an audible match to Purifi/Hypex.

https://www.audiosciencereview.com/forum/index.php?attachments/fosi-audio-v3-stereo-amplifier-budget-frequency-response-measurement-png.293699/

 

[boXem]

It'not only FR. The output impedance being variable with frequency generates tons of IMD.

 

[Ron Texas]

I Don't know if Buckeye is any better, but I found the speaker wire binding posts on the Fosi V3 to be useless unless you have banana plugs.

 

[Zapper]

That was with 4 ohm vs 8 ohm resistance. You can be pretty sure with varying impedance with frequency and reactance you'll get some variations that enter the audible range. Maybe not hugely audible (though maybe so with some difficult speakers), but definitely not an audible match to Purifi/Hypex.

https://www.audiosciencereview.com/forum/index.php?attachments/fosi-audio-v3-stereo-amplifier-budget-frequency-response-measurement-png.293699/

Yes, I looked at the TI TPA3255 reference design. It has two 10uH inductors in series with the speakers (and two 1uF caps in parallel for the 50kHz lowpass). That's enough to cause ~1dB amplitude variations in speakers with tweeter crossover resonances in the 2-4 kHz range. Like this from Erin's R5 Meta review:

 

[Guddu]

It'not only FR. The output impedance being variable with frequency generates tons of IMD.

Is it also applicable in case of PFFB based chip amplifier deployments?

 

[welwynnick]

It'not only FR. The output impedance being variable with frequency generates tons of IMD.

Now that's really interesting. Can you measure that? Amir does power measurements into reactive loads, but not distortion measurements. These days, when sources and amplifiers measure so well, it would be great to get a handle on something that does make a difference and explains the differences that we hear.

 

[Zapper]

It'not only FR. The output impedance being variable with frequency generates tons of IMD.

How exactly? A passive LC filter in series between the amp feedback loop and the speaker is in principle no worse than the LCR circuits in a passive speaker's crossover. It's not the impedance that causes IMD but the nonlinearities in the L and C. The powdered iron core inductors that TI specs might not be the best in that regard.

The TPA3255 chip amps have more IMD in Amir's tests than Hypex or Purifi, but I think "tons of IMD" is an exaggeration.

 

[boXem]

Is it also applicable in case of PFFB based chip amplifier deployments?

The PFFB is only 6dB loop gain if memory serves. It won't suppress IMD.

Now that's really interesting. Can you measure that? Amir does power measurements into reactive loads, but not distortion measurements. These days, when sources and amplifiers measure so well, it would be great to get a handle on something that does make a difference and explains the differences that we hear.

@pma measured it on an Ayima A07 some time ago. Bed time here, I let you search for it.

It's not the impedance that causes IMD but the nonlinearities in the L and C.

You right, sorry, as just written, bed time

 

[Zapper]

Now that's really interesting. Can you measure that? Amir does power measurements into reactive loads, but not distortion measurements. These days, when sources and amplifiers measure so well, it would be great to get a handle on something that does make a difference and explains the differences that we hear.

Amir does measure IMD. Any effect that the LC output filter has is included there. The TPA3255 amps are OK, not terrible, not great.

 

[welwynnick]

Yes he does measure IMD, but I believe it's into a resistive load.

 

[Zapper]

Yes he does measure IMD, but I believe it's into a resistive load.

True, but the output filter is still there and still has variable impedance.

 

[Greenman]

moderate volumes @ 3 meters with a variety of speakers)

This is key here. What volumes and what speakers. There’s a huge difference in the demands of different speakers especially when you approach high ish volumes. Efficient speakers that dont fall below 4ohms and are listened to at 80db could easily be driven by the Fosi ZA3 option. Difficult speakers at 90db+ (with transients above say 100db) is a completely different ballpark. Many small speakers in particular can be pretty inefficient.

 

[Streamc]

One of the areas of weakness (if you can call it that) of the current crop of chip amps is that they are less load-tolerant than the Purifi/ NCoreX equivalents. This is critical for speakers with complex and reactive aspects to their impedance curves.

So we need simply measurements for different impedances?

 

[MaxwellsEq]

So we need simply measurements for different impedances?

No, if there is load-dependant frequency amplitude inaccuracy, it should be obvious even with 4 and 8 Ohm resistance-only tests. The graphs should perfectly overlay. Have a look at a few of Amir's reviews, you will see some that have variance at high frequencies when an 8 Vs 4 Ohm load is used . If there is a variance real world reactive speaker loads could emphasize it more.

Ideally something like the Powercube can be used to show the impact of capacitive or inductive reactance combined with low resistance, at 1kHz.

 

[mfaughn]

So, am I hearing that I'm unlikely to notice an audible difference unless I am using speakers that present a difficult load?

 

[boXem]

So, am I hearing that I'm unlikely to notice an audible difference unless I am using speakers that present a difficult load?

Not really, the easier (high impedance) the load will be at high frequencies, the bumpier will be the response from amplifiers without post filter feedback.

 

[charlielaub]

It'not only FR. The output impedance being variable with frequency generates tons of IMD.

Got a reference or data to support this claim?