[Daniel Johnston]

Rekkr into DCA stealth load of 23 ohms yields about 700mW max; similarly, Diana V2 42 ohms, 380mW max; both cases should yield >110dB peaks (assuming HP drivers capable of this extreme output/ no compression/ etc.)

but kinda expensive / inefficient use of 8x 6N6P tubes for a preamp-only application.

perhaps @Jason Stoddard might comment on using 2 or 4 6N6P tubes for this use case? or he might recommend a Valhalla 2 instead?

My primary headphone I use with FV is the Meze Liric. Low gain and impedance multiplier off. It's way better sounding than it should be.

I have a Valhalla 2 as well. I was mainly concerned with the ability of the Rekkr to handle high impedance loads.

 

[AndreYew]

Intuitively:
- Volume control in the digital domain drops bits (The least significant ones). This is called truncation. Think 0011 becoming 00xx (the last two bits are dropped in this example).
- Premise: If one is to truncate anyway, can we find a nearby value that is close enough to the desired target volume, but minimizes information loss? Paper shows a cute way of doing so. This is the tradeoff They mention. Instead of bringing you from a specific point down to, say -100db, the control may bring you to -100.2db because that’s where the loss was minimal and the distance to target was “good enough”. For illustrative purposes, imagine the task is to drop two bits (least significant, rightmost):

-100.2db -> 001100010010011110100001101101110000
-100.1db -> 001100010010011110100001101101110010
-100.0db -> 001100010010011110100001101101110011

Dropping the two 0s at -100.2causes the least amount of information loss, so the technique will “pin” you there instead of exactly taking you to -100.0db.

Sorry to be That Guy (tm) but I don't think that's what they're doing. What they're doing is finding the smallest precision volume coefficient that when multiplied against a sample gets close enough to the desired attenuation level.

Some background: when computers do integer multiplication of two numbers, say one with A bits and another with B bits, you need A+B bits to hold the product without any loss of information. If you multiply a 16-bit number with an 8-bit number, you need 16+8=24 bits to hold all of the resulting bits. You can show this with just simple long multiplication you learned in grade school.

What they're doing is trying to find a volume coefficient with a low enough number of bits so that when multiplied against a sample on a recording will make a product that's smaller than the precision of the DAC.

And therein lies the limitation with their technique: if you're playing back stuff with say 20-bit info, and you have a 24-bit DAC, you only have 4 bits to use for your volume control before losing precision or in their case causing truncation distortion. A recording with real 20-bit information is very rare, but possible, and a 24-bit DAC does not exist. (For common line levels, its noise floor would be equivalent to the thermal noise of a 50 Ohm resistor which means it has to look like a big power amp because all of your equivalent resistances must be 50 Ohms or lower, which in turn causes more heat, which makes you have to use more extreme cooling measures, etc.).

So in effect, their volume control might work OK with 16-bit recordings but probably not for higher bit ones. Considering that most if not all DACs are probably sub-20-bit when listened to by the human auditory system (the amount you can hear into the noise floor depends on the critical bands in your hearing system as well as a bunch of other physiological effects), the system is probably not very practical in the real world.

The other thing I found striking is how quickly they dismissed dithering. Any time you truncate bits, you need dithering: you can see the effects of not dithering in their figure of a 0dBFS 1.5 kHz sine wave with all its harmonic products. Even worse, those extra spikes are anharmonic, meaning they are not related in a simple way to the main signal like analog distortion, and have also aliased, so they will be more audible. This is why dithering seemingly has big effects even though it works on the smallest values: the distortion products due to truncation can occur all over the frequency range due to aliasing.

TL;DR: it's an interesting idea, but probably not practical.

 

[RickB]

One thing about Apple that naysayers do not acknowledge is that Apple, for whatever reason, often gets to a place before other companies realize such a place even exists. Maybe, those places aren't worth going to, either; otoh, how does one know without trying?
Some examples off the top of my head:

• Apple Desktop Bus (universal serial bus in 1986 before USB some 20y later?)
• Ethernet on their machines (who needs 5Mb/s networking in the 80s that aren't mainframes?)
• Bitmap display (only in research labs when Apple "popularized" them)
• personal digital assistants (aforementioned Newton)
• touchscreen cell-phones.
• lots more examples.

Now, did they invent all that stuff? Probably not. Did they invent some of it? Probably.

One thing I like about Schiit is that it has some of that DNA. Do they invent everything they put into a product? Probably not.
Did they take a high end audio tech, R2R, and bring it to the masses. Schiit yeah!
Did they invent some of that tech. Darn tootin!

Fortunately, Schiit (for whatever altruistic or diabolical reason you can imagine) does not take after Apple in pricing their products at whatever the max is they can get away with.

 

[JanMc]

You may think you're disagreeing with me, but you're actually not.

Excellent. Just wanted to make sure no one was being misled.

 

[tincanear]

My primary headphone I use with FV is the Meze Liric. Low gain and impedance multiplier off. It's way better sounding than it should be.

I have a Valhalla 2 as well. I was mainly concerned with the ability of the Rekkr to handle high impedance loads.

Should be fine, just proportionally lower max power with increasing impedance

 

[theeclone]

Excellent. Just wanted to make sure no one was being misled.

Yes, certainly don't want to mislead anybody. Just think the principles behind how those amps work is interesting. Perhaps the marketing of them as "digital" is unfortunate, but such is marketing. I've never actually heard one of them so I can't comment on how they sound.

What I'm really curious about is how that design could work if the amp could be fed DSD and directly convert it to PWM, or perhaps staying in PDM. As far as I know, a class D amp can use PDM instead of PWM. Since DSD is already so close to the representation of the signal within a class D amp maybe it wouldn't actually sound too assy!

 

[AstronomerXI]

Me too! I have less rarified stuff than you guys, but a club member has a 130 Starfire that he brings out for public star parties (albeit using it only for EAA), and the two local places we do our public star parties have a TEC 200FL and an AP 180 they let the public look through. Currently, I'm really enjoying the Analog Sky Magic, which is a 3D printed binocular telescope that you can build at home. It's a lot easier to bring that thing out and set it up than the 150mm refractor, its mount (a Rowan AZ100), and the tripod. I have an Agema 120 on order, but who knows when they'll get that done. I'm also a Televue fan with mostly Delos, but also the 31T5 and 21 Ethos. Other favorite eyepieces include the Nikon Nav-HW series.

Very nice, and pleased to virtually meet you! This makes at least 4 Schiity astronomers on the forum. Nice to hear that you do public outreach events. There's nothing like a person's first-ever view of Saturn or detailed lunar craters. I've had seniors get emotional and seen many young child's eyes go saucer-wide on their first view through a 'scope. Nothing beats that.

 

[sam6550a]

Yes, certainly don't want to mislead anybody. Just think the principles behind how those amps work is interesting. Perhaps the marketing of them as "digital" is unfortunate, but such is marketing. I've never actually heard one of them so I can't comment on how they sound.

What I'm really curious about is how that design could work if the amp could be fed DSD and directly convert it to PWM, or perhaps staying in PDM. As far as I know, a class D amp can use PDM instead of PWM. Since DSD is already so close to the representation of the signal within a class D amp maybe it wouldn't actually sound too assy!

Then the burden of D to A conversion just moves farther down the road, which puts the burden on the power amp. Very interesting, but I am not able to predict how it would sound.

 

[Mr Rick]

Sorry to be That Guy (tm) but I don't think that's what they're doing. What they're doing is finding the smallest precision volume coefficient that when multiplied against a sample gets close enough to the desired attenuation level.

Some background: when computers do integer multiplication of two numbers, say one with A bits and another with B bits, you need A+B bits to hold the product without any loss of information. If you multiply a 16-bit number with an 8-bit number, you need 16+8=24 bits to hold all of the resulting bits. You can show this with just simple long multiplication you learned in grade school.

What they're doing is trying to find a volume coefficient with a low enough number of bits so that when multiplied against a sample on a recording will make a product that's smaller than the precision of the DAC.

And therein lies the limitation with their technique: if you're playing back stuff with say 20-bit info, and you have a 24-bit DAC, you only have 4 bits to use for your volume control before losing precision or in their case causing truncation distortion. A recording with real 20-bit information is very rare, but possible, and a 24-bit DAC does not exist. (For common line levels, its noise floor would be equivalent to the thermal noise of a 50 Ohm resistor which means it has to look like a big power amp because all of your equivalent resistances must be 50 Ohms or lower, which in turn causes more heat, which makes you have to use more extreme cooling measures, etc.).

So in effect, their volume control might work OK with 16-bit recordings but probably not for higher bit ones. Considering that most if not all DACs are probably sub-20-bit when listened to by the human auditory system (the amount you can hear into the noise floor depends on the critical bands in your hearing system as well as a bunch of other physiological effects), the system is probably not very practical in the real world.

The other thing I found striking is how quickly they dismissed dithering. Any time you truncate bits, you need dithering: you can see the effects of not dithering in their figure of a 0dBFS 1.5 kHz sine wave with all its harmonic products. Even worse, those extra spikes are anharmonic, meaning they are not related in a simple way to the main signal like analog distortion, and have also aliased, so they will be more audible. This is why dithering seemingly has big effects even though it works on the smallest values: the distortion products due to truncation can occur all over the frequency range due to aliasing.

TL;DR: it's an interesting idea, but probably not practical.

Sorry, but I have to say it. I didn't understand a "bit" of that.

 

[cschock]

One thing about Apple that naysayers do not acknowledge is that Apple, for whatever reason, often gets to a place before other companies realize such a place even exists. Maybe, those places aren't worth going to, either; otoh, how does one know without trying?
Some examples off the top of my head:

• Apple Desktop Bus (universal serial bus in 1986 before USB some 20y later?)
• Ethernet on their machines (who needs 5Mb/s networking in the 80s that aren't mainframes?)
• Bitmap display (only in research labs when Apple "popularized" them)
• personal digital assistants (aforementioned Newton)
• touchscreen cell-phones.
• lots more examples.

Now, did they invent all that stuff? Probably not. Did they invent some of it? Probably.

One thing I like about Schiit is that it has some of that DNA. Do they invent everything they put into a product? Probably not.
Did they take a high end audio tech, R2R, and bring it to the masses. Schiit yeah!
Did they invent some of that tech. Darn tootin!

Apple did not invent most of that. For example, the bitmap display was first developed for the Alto computer in 1973. The first commercially available personal computer system to use one was the Xerox Star workstation (I used this in my job in the mid 80's, gorgeous graphics on huge monitors, but oh god so slow.) I have heard it said by people who might know, that Steve Jobs saw the Xerox Star and decided that is what Apple should do for a display type. They purloined the mouse from the same source.

IBM created the worlds first "smart" phone with a screen, and it was even a touchscreen, with the Simon, released in 1994, 13 years before the iPhone! Was it a winner? No it was not, definitely a technology not quite ready for prime time at that point.

As for ethernet, another technology developed for the Alto computer, it was standardized in 1980 originally and implemented in mostly minicomputers initially. Apple did make it available pretty early in the desktop arena, to their credit, but by that time I believe there were add-on cards for IBM-compatible machines though it is absolutely true that other topologies were far more frequently used in PC LANs of that time. We used ethernet enabled Compaq II portables in our office in 1986.

Apple has done a lot of good things, don't get me wrong. But research was never their forte and so inventing things wasn't, often, something they did. They have often applied technologies in interesting and unexpected ways and/or popularized things in the public consciousness however, no small feat!

 

[theeclone]

Then the burden of D to A conversion just moves farther down the road, which puts the burden on the power amp. Very interesting, but I am not able to predict how it would sound.

Exactly. There is another thread on here claiming sonic nirvana with a design like that, but hard to know what to believe until I can hear one myself. I do think that at least in the case of powered speakers class D can sound great.

And in case it isn't clear by now, I'm not making any claims about what any audio company should be pursuing . Again, just interesting to contemplate!

 

[ricksome]

I came across these two vintage headphones today. They have very little resale value on eBay. I have not tested them. If you live in the lower 48 States and want them for free, I will pay for the shipping. First person to private message me, gets the goods.

I am doing this in the Spirit of Paladin79 who has given back so much to members of Head-Fi. You might want these headphones but I am not a craftsman like P79, so you don't want my woodwork.
Enjoy and Happy Holidays. / Rick

 

[Moondog55]

Apple did not invent most of that. For example, the bitmap display was first developed for the Alto computer in 1973. The first commercially available personal computer system to use one was the Xerox Star workstation (I used this in my job in the mid 80's, gorgeous graphics on huge monitors, but oh god so slow.) I have heard it said by people who might know, that Steve Jobs saw the Xerox Star and decided that is what Apple should do for a display type. They purloined the mouse from the same source.

IBM created the worlds first "smart" phone with a screen, and it was even a touchscreen, with the Simon, released in 1994, 13 years before the iPhone! Was it a winner? No it was not, definitely a technology not quite ready for prime time at that point.

As for ethernet, another technology developed for the Alto computer, it was standardized in 1980 originally and implemented in mostly minicomputers initially. Apple did make it available pretty early in the desktop arena, to their credit, but by that time I believe there were add-on cards for IBM-compatible machines though it is absolutely true that other topologies were far more frequently used in PC LANs of that time. We used ethernet enabled Compaq II portables in our office in 1986.

Apple has done a lot of good things, don't get me wrong. But research was never their forte and so inventing things wasn't, often, something they did. They have often applied technologies in interesting and unexpected ways and/or popularized things in the public consciousness however, no small feat!

Riding on coattails, standing on the shoulders of giants...technology moves (mostly) forward and Al Gore invented the........

 

[R3DFIV3]

In a world full of mega dicks doing mega dick things, it's nice to know that is not the kind of behavior in which Schiit partakes. I knew I was handing my money over to some skilled and genuinely nice peeps.

Amen, brother.

 

[adias]

Sorry to be That Guy (tm) but I don't think that's what they're doing. What they're doing is finding the smallest precision volume coefficient that when multiplied against a sample gets close enough to the desired attenuation level.

Some background: when computers do integer multiplication of two numbers, say one with A bits and another with B bits, you need A+B bits to hold the product without any loss of information. If you multiply a 16-bit number with an 8-bit number, you need 16+8=24 bits to hold all of the resulting bits. You can show this with just simple long multiplication you learned in grade school.

What they're doing is trying to find a volume coefficient with a low enough number of bits so that when multiplied against a sample on a recording will make a product that's smaller than the precision of the DAC.

And therein lies the limitation with their technique: if you're playing back stuff with say 20-bit info, and you have a 24-bit DAC, you only have 4 bits to use for your volume control before losing precision or in their case causing truncation distortion. A recording with real 20-bit information is very rare, but possible, and a 24-bit DAC does not exist. (For common line levels, its noise floor would be equivalent to the thermal noise of a 50 Ohm resistor which means it has to look like a big power amp because all of your equivalent resistances must be 50 Ohms or lower, which in turn causes more heat, which makes you have to use more extreme cooling measures, etc.).

So in effect, their volume control might work OK with 16-bit recordings but probably not for higher bit ones. Considering that most if not all DACs are probably sub-20-bit when listened to by the human auditory system (the amount you can hear into the noise floor depends on the critical bands in your hearing system as well as a bunch of other physiological effects), the system is probably not very practical in the real world.

The other thing I found striking is how quickly they dismissed dithering. Any time you truncate bits, you need dithering: you can see the effects of not dithering in their figure of a 0dBFS 1.5 kHz sine wave with all its harmonic products. Even worse, those extra spikes are anharmonic, meaning they are not related in a simple way to the main signal like analog distortion, and have also aliased, so they will be more audible. This is why dithering seemingly has big effects even though it works on the smallest values: the distortion products due to truncation can occur all over the frequency range due to aliasing.

TL;DR: it's an interesting idea, but probably not practical.

Digital volume control is never good.