[Currawong]

I can only hope the dialogue for ‘more RAM’ can be put to rest.
Educate the consumer.
(It was a terrific post explaining why FiiO went with the modules they did. There reasons were nearly five vs one (the one being ‘bragging rights’).

More RAM is a burden more than a blessing (if ‘more power draw’, then shorter battery life, more heat under load as well…) The company investment in the ‘larger boards’ (by config, not necessarily ‘size’), with the longer ‘lead times’/ ‘wait times’ to get them, AND on such large ‘minimum order quantity’; ‘the liquidity caught up in those parts that may, or may not, sell quickly (needed to recoup outlay), means that the actual costs passed on to the consumer, with markup at retail, in the HiFi industry being a significant ‘bump’ (yes, I know ALL OF THIS WAS IN THE FIIO POST) for zero actual benefit, would be a ‘very poor business decision. (unless intending to sell to people who are not interested in absolute audio quality and usable features, but would rather ‘bragging rights’- remember extra RAM will not help in ANY WAY if it is not used. I cannot think of a single way to use ‘so much RAM’ in music player, so the negatives, again, which FiiO redressed.. they did the math and realised consumers could buy ‘practically ZERO markup (micro) SD cards’, and the huge amount saved on what FiiO would have had to sticker price the units at retail, means that end users essential get a free 512GB or possibly larger, amount of storage space. (that has the benefit of being removable as well)

There isn’t any part of their reasoning about this that is questionable.
In fact it had the effect on me, to realise that FiiO are ‘all about the sound and usability, and passing on ‘best bang for buck’ and know how to prioritise aspects that matter. (and are willing to ‘take the hit’ by not being on the radar of ‘spec sheet warriors (brave!)’.

The problem with buying the ‘extra RAM and storage capacity’ part, would have mostly destroyed their position in the market with these players, leading to slower sales rate, quite likely, and delaying the time it would take to have FiiO have their large outlay returned. (and take hits on interest or related issues of what happens when your money is tied up)
As they stated, the larger boards may have been usable in ‘other products’ so if the M17 DIDN’T sell quickly , they would have had less options in front of them to recover.
Of course the added operation costs lost to ineffective parts selection (in this case ‘significant’) would have meant a greater return on investment would be needed in the next part (whether that be budget segment or otherwise.. )

This is why good business decisions, carefully weighed up, benefit companies, and then ‘we don’t know what we are doing so we will sell ‘flavour of the moment fads’ (ie 6GB of RAM in a DAP with a storage amount that basically is inconsequential to any user.. (64 vs 128)..

I imagine the savings allowing a half tearbyte of reusable storage (removable) is more useful than having ‘super fast storage’ (designed for 4K video recordings) that in no way benefits a music player. (unless you move large amounts of data, regularly, only to internal storage… (most would rather external storage that can be removed and use their laptops much faster card bay for data copying.. )(and again, having 512GB-1GB of data, at ‘local office supplies stores prices’, given for ‘free’ (vs the price hiked unit due to bad engineering/‘parts selection’)

I know I am repeating myself, but it seems that many in the threads are repeating how useless the 6GB is /‘quoting the FiiO explaination that lists MANY LEGITIMATE REASONS for them making ‘the best decision’.


TL: DR
4GB of RAM is OVERKILL for a DAP.
Android devices didn’t really need 512MB of RAM, and each variant that ‘uses more RAM‘ is generally preloading all your ‘potential apps’ (in a phone that does everything, and runs bloaty phone platform software). A DAP could be streamlined down to ‘bugger all’ (embedded systems would have been happy with 4MB, ‘once upon a time’; literally having four thousand times more than a music player needs is SAFE OVERHEAD in case a user wants to experience Apple Music etc.

Once Android moved to 1GB for the Operating System ‘working pool’ there was ‘so much overhead’ and redundancy, having 2GB of RAM in a music player, I would consider ‘more than needed (by a long shot).
I gather some power social media die hards, using their DAP ‘like a phone’ might have several consumption platforms open, and even if running lots of bloated apps, are still in ‘over kill territory with 2GB, but 3GB is ‘future proofing the device’.

What happens if you don’t have enough RAM to have 300+ webpages open, and lots of apps etc?.. the device simply drops the least used webpage from RAM, and you have to reload it again if you reselect that tab (from the 300 that you have open).
The problem is with european privacy laws, most platform holders get around ‘don’t send user data’ with “reload everything all the time” (we get their user data by the nature of them requesting things); so for a few years now, most platforms barely keep ten webpages in active memory, and ’all those gigabytes sit around useless’ (happily draining power, so not useless; they keep our devices ‘warm’ for us!)

More RAM than we need doesn’t speed anything up. This has been an IT fact for many decades. A reasonable excess is nice (futureproof).. 4GB is vastly more than a DAP needs, based on a range of reasons. But never let the truth get in the way of a good story.
Most phones built for the Chinese market need to have ‘bragging rights amounts of RAM’ (that serves no purpose).
Consumers must buy based on spec sheet warfare, and that shows FiiO has a backbone in ‘such a silly market place’

I commend FiiO for doing this, as we know it will cost them ‘some’ sales in their home market, but out of duty to end users (pro ‘consumer’), they take the hit.

 

[Currawong]

The following criteria apply before I tube roll...

1. I must be well rested.
2. Completely sober.
3. Consume 1 shot of espresso for heightened alertness.
4. Glasses must be cleaned for optimal vision ensuring pin-to-socket alignment.
5. Change into loose fitting clothes.
6. Loosen up by doing stretches and toe-touches to ensure adequate tube-rolling mobility.

My process is as follows:

1. Power off my amplifier.
2. Power Freya N off at the switch, power conditioner, and circuit breaker.
3. Remove power cord.
4. Wait 15 minutes for current tubes to cool off.
5. Remove tubes carefully while wearing clean cotton gloves.
6. Carefully place them in their appropriately labeled boxed with care to ensure the correct tube goes into the correct box as the boxes have the measurements on them.
7. Insert the new tubes carefully while wearing clean cotton gloves.
8. Go to the breaker panel and flip the breaker ON
9. Re-install the power cord.
10. Switch on the power conditioner.
11. Power on Freya.
12. Ensure it is in tube buffer mode.
13. Wait 60 seconds for tubes to reach optimal temp.
14. Select a well-recorded CD or stream album via Roon
15. Wonder why there's no sound.
16. Double-check input selection.
17. Make sure MUTE is not engaged.
18. Turn volume up and down.
19. Remove the tubes and re-insert.
20. Repeat steps 15-18.
21. Remove tubes and check them on my B&K tester. (they always test OK)
22. Place tubes back in and try again.
23. Repeat steps 15-18.
24. Place the previous tubes back in.
25. Repeat steps 15-18.
26. Remove original tubes and replace with the tubes previously removed.
27. Repeat steps 15-18.
29. Shout expletives!
30. Dejected and looking down, realize that the amplifier is still powered off.
31. Repeat step 29.
32. Power on the amplifier.
33. Confirm that there is finally sound.
34. Repeat step 29.
35. Power everything off and go to bed as I am now too tired and too pissed off to listen to music.

 

[Currawong]

Useful posts about amplifiers and power.

It is exactly this "feeling" that makes reliable, comparable measurements important.

More, power is very misleading, we need the maximum output in Volt at a given load. And we need the normalised headphone sensitivity in dB/V.

And ignore the volume control Position, it is not an indicator of "power". It is strictly "gain". Labels like turbo are also arbitrary.

The Gryphon at 1% THD should manage around 5.8V into 32 Ohm BAL or 3V into 16 Ohm SE. For higher impedances (say 300 Ohm and up) we get around 7V into 300 Ohm BAL or 3.5V SE. Measurements here:

https://www.l7audiolab.com/f/ifi-gryphon/

Based on Goldensounds Measurements the maximum output with 1% THD from Diablo 2 into 32 Ohm BAL was 7.2V.

That is a 1.8dB difference. Meaning the difference in maximum loudness is a lot less than 3dB, the limit of what can be reliably detected by listening to music.

Based on Goldensounds Measurements the maximum output with 1% THD from Diablo 2 into 32 Ohm SE was 8.7V.

That is a 3.5dB difference. Meaning the difference in maximum loudness is more than 3dB, but just above the limit of what can be reliably detected by listening to music.

So realistically, for 32 Ohm or lower impedance headphones Gryphon and Diablo deliver around the same maximum SPL from a given headphone within around 3dB or less.

What we have not looked at however is how the Diablo 2 does into 300 Ohm. There it can (well the original could) output 20V into 300 Ohm at 1% THD.

Compared to 7.6V on the Gryphon that is a massive 8 dB difference and it is maintained for SE respectively at 10V and 3.8V.

Now, translating this whole thing into power looks like this

Gryphon: 1.05W 32R BAL
Gryphon: 0.62W 32R SE
Gryphon: 0.16W 300R BAL
Gryphon: 0.04W 300R SE

Diablo: 1.62W 32R BAL
Diablo: 2.36W 32R SE
Diablo: 1.33W 300R BAL
Diablo: 0.33W 300R SE

So, when driving low impedance headphones in BAL, there is not a lot between the two in terms of power.

But for high impedance headphones the ratio between power from Gryphon and Diablo (2) is nearly 1:10 in favour of Diablo.

But the output voltages already told this story.

Why are not headphones rated in dB/1V (or V/90dB) and headphone amp's in how many volts they provide (continuous and peak according to AES2-2012) in to a set of representative loads AND they should be rated in "dB/V".

With Headphones rated dB/V (like sennheiser does) and headphone amp's in dB/V (dB over 1V) you just add the two numbers to get maximum peak SPL and numbers start making sense, rather than to cause confusion and mislead.

Thor

What is the volume and gain setting on Gryphon?

Another factor here is the difference between "continuous" power (aka "RMS") and "Peak"/"Music"/"Impulse"/"Dynamic"/"Programme" power. And there are more possible standards than standard bodies, due to some extinct ones showing up.

Goldensound showed that at 4.5W with 32Ohm a 1kHz signal can be reproduced without clipping for 4 cycles, or 4mS, before clipping starts and available power progressively reduces.



The Diablo can likely can produce even higher peaks with a shorter duration. Music is not a sinewave. it looks like this:



As long the amplifier is able not start to clip for the duration of the highest peak AND can support the "Average RMS" indefinitely, the actual acoustic output from the attached transducer will be the same as of an Amplifier with an RMS output equal to 1/2 the peak (a 100W RMS Amplifier produces AT LEAST 200W Peak).

The original Diablo design accounts for this, the power supply is current limited and will progressively collapse the power supply voltage until an equilibrium is reached between current demand and available power (depends on battery state and external power supply if used).

If running a "nominal" 3.7V battery at 50% charge state, and having the the power supply switching at the nominal 3.8A peak current with 90% duty cycle (the the limit where the supply throttles down), we can supply around 7W to the two amplifier channels, which at 70% efficiency gives around 5W across two channels or 2.5W per Channel, continous, which squares well with Goldensounds Measurements.

Diablo can produce a lot of voltage into high impedance headphones and can produce the same voltage for short Bursts into low impedance's, there the power supplies peak current capabilities and the charge stored in the power supply capacitors support the power supply voltage for short peaks.

So peaks in music sufficiently short, can pass undistorted. Here a typical.musical instrument note:



This is a piano, but most percussion etc. looks similar too, even a plucked acoustic guitar does. We can see that the intial transient raps up and peaks and then the signal falls of.

This means if the "holdup" of the power supply bridges the initial transient which exceeds continuously available power by a lot, power demand will fall off rapidly and the limited holdup will suffice for the signal to pass undistorted.

The AES Standard AES2-2012 attempted to solve this in 2012 for speakers and looked at actual music:



As a result AES2-2012 recommends a power rating based on testing with pink noise, bandwidth limited according to speaker requirements and a 12dB Peak to average ratio. No minimum peak duration is stated.

As a result a speaker that was rated 200W RMS under a traditional standards (say a 15" Electrovoice EVM15L rated 200W RMS and 400W "Program") would actually be rated 3,200W AES2-2012, as 12dB is 16 times the power.

So an Amplifier delivering 3,200 W ito 8 Ohm would be usable with this speaker as long as:

1) the thermal (RMS) Power is not greater than 200W
2) The safe cone excursion is not exceeded

I would suggest that making an Amplifier that drives 3200W RMS into 8 Ohm to drive a 200W Speaker makes no sense, so as next step we should re-rate amplifiers using a similar scheme.



The Gryphon has no excess gain.

This means with the volume maxed out, gain maxed and Bass-Boost off any full scale digital signal will drive the Amplifier to the onset of clipping with 50Ohm or higher impedance headphones.

The Diablo has a shedload of excess gain. Full gain to balanced in turbo is 30dB. This means a full scale signal will attempt to drive the amplifier output to 63.2V, but the output clips at 20V

If you turn the volume to maximum you are overdriving the amplifier severely, with as much 10dB over clipping power.

This is why I wrote that volume control position is not meaningful in relation to output power, only to gain. Having the volume control at (say) 50% doesn't mean you are at 50% power and turning it up to the limit is not equal to 100%.

As said, with Diablo on Turbo turned up all the way, it's equal to to 1000% maximum power, where only 100% can be delivered.



Excess gain on Diablo and the ability that the Diablo can probably produce a single 20V peak into 32 Ohm for a fraction of mS (which is 12.5W!) which may be enough to allow music to pass subjectively undistorted, while having "transient" power levels much greater than "continuous".

It would be equal in this specific case to an amplifier traditionally rated as 6.25W continuous with music, but without being able to deliver 6.25W continuously

The absence of an equivalent of AES2-2012 (which covers speakers only) for Amplifiers means we are lacking a valid and reality based standard to quantify this "dynamic power" behaviour of Amplifiers.

It is basically the illustration that in theory, there is no difference between theory and practice.

However, in absence of different standards, a manufacturer should always state technical data in accordance with established and standardised best practice (even if the numbers may look less impressive) and if using non-standard procedures to quantify for example the ability of an Amplifier to "overperform" with music, the test procedure should be disclosed and justified, referencing relevant literature.

Thor