[Thorsten Loesch]

Considering I have like 4 ifi usb gadgets in my chain, is it *possible* for one to be faulty and corrupt the data inegrity? And if so, id assume the other 3 would just fix it? XDD

No, that's not how USB works.

Let's separate data (which is digital) and signal (which is analogue).

The source sends an analogue waveform in which digital data is encoded.

This waveform passes through the cable and at the sink end an analogue circuit decides if it has a "transition", that is, if a lower Level changed to a higher level or a higher to a lower and the "decision circuit" works on absolute values.

All of this works best if the "edges" are clear (square waves). At 480Mbps square waves become truncated triangles (actually at much lower frequencies).

We get something called an "eye pattern" because it kinda looks like the eye of Ra.

The more "open" the eye, the easier the detection of 0 or 1.

It is of course possible to make circuits detect logic levels reliably under difficult conditions and you get some that, while passing USB Standard Compliance Testing do terrible in the real world.

This problem is analogue, not digital. The usual USB hardware on CPU chips (programmed to become USB to audio bridges) are functional. Not much more.

When you take a hub or repeater, it will only detect 0 or 1 and pass that on. Usually there is not intelligence there, so it doesn't care if the data is right or not. It just repeats what it gets.

As it so happens, Hub Chip makers tend to make better USB analogue Hardware than CPU Chip makers.

So back to our USB gizmo (in series). Let's take my earlier 25m multiple USB Repeater active USB cable.

After the first 5m cable (let's use a cheap one) the signal is already at the limits of us standards.

Or beyond, if the cable was made cheaply in ch!na or by some artisan in a garage workshop using nine nines pure silver in Teflon sleeving hand twisted and without sound quality damaging screen.

If we choose a "good" hub / repeater / receiver chip, it will reliable detect the data embedded in a badly degraded signal (poor signal integrity but as system high data integrity). And it will pass this signal into the next cable. And so on.

Note, nothing tries to actually make sense of the signal or data, all it does is to repeat (and route one to many ports) the received signal.

If we have a CPU Chip made by the lowest bidder fab in mainland china (who perhaps generously provide the USB IP for free) we may have problems even with signals that are just barely beyond standard (good or adequate signal integrity but due to poor chip design poor data integrity).

If we have such a CPU chip as downstream USB device (all the common ones are like this) then given all else in USB audio Class (2) we have high risk problems of data integrity problems.

Here, a USB gizmo can help by re-producing a clean, wide open eye eye pattern with minimal calling in-between to make it worth again.

And daisy-chainging 5 x 5m super cheap cabled made in ch!na or by some artisan in a garage workshop using nine nines pure silver in Teflon sleeving hand twisted and without sound quality damaging screen with 5 repeaters will make sure that at the input pons of our USB 2 AUDIO bridge IC the eye pattern is near perfect, even if after every 5m it was so bad with just source cable and sink you would already get dropouts.

I know that doesn't quite address your question but hopefully answers to perfectly reasonable thinking behind it.

Thor

 

[Assationater]

No, that's not how USB works.

Let's separate data (which is digital) and signal (which is analogue).

The source sends an analogue waveform in which digital data is encoded.

This waveform passes through the cable and at the sink end an analogue circuit decides if it has a "transition", that is, if a lower Level changed to a higher level or a higher to a lower and the "decision circuit" works on absolute values.

All of this works best if the "edges" are clear (square waves). At 480Mbps square waves become truncated triangles (actually at much lower frequencies).



We get something called an "eye pattern" because it kinda looks like the eye of Ra.

The more "open" the eye, the easier the detection of 0 or 1.

It is of course possible to make circuits detect logic levels reliably under difficult conditions and you get some that, while passing USB Standard Compliance Testing do terrible in the real world.

This problem is analogue, not digital. The usual USB hardware on CPU chips (programmed to become USB to audio bridges) are functional. Not much more.

When you take a hub or repeater, it will only detect 0 or 1 and pass that on. Usually there is not intelligence there, so it doesn't care if the data is right or not. It just repeats what it gets.

As it so happens, Hub Chip makers tend to make better USB analogue Hardware than CPU Chip makers.

So back to our USB gizmo (in series). Let's take my earlier 25m multiple USB Repeater active USB cable.

After the first 5m cable (let's use a cheap one) the signal is already at the limits of us standards.

Or beyond, if the cable was made cheaply in ch!na or by some artisan in a garage workshop using nine nines pure silver in Teflon sleeving hand twisted and without sound quality damaging screen.

If we choose a "good" hub / repeater / receiver chip, it will reliable detect the data embedded in a badly degraded signal (poor signal integrity but as system high data integrity). And it will pass this signal into the next cable. And so on.

Note, nothing tries to actually make sense of the signal or data, all it does is to repeat (and route one to many ports) the received signal.

If we have a CPU Chip made by the lowest bidder fab in mainland china (who perhaps generously provide the USB IP for free) we may have problems even with signals that are just barely beyond standard (good or adequate signal integrity but due to poor chip design poor data integrity).

If we have such a CPU chip as downstream USB device (all the common ones are like this) then given all else in USB audio Class (2) we have high risk problems of data integrity problems.

Here, a USB gizmo can help by re-producing a clean, wide open eye eye pattern with minimal calling in-between to make it worth again.

And daisy-chainging 5 x 5m super cheap cabled made in ch!na or by some artisan in a garage workshop using nine nines pure silver in Teflon sleeving hand twisted and without sound quality damaging screen with 5 repeaters will make sure that at the input pons of our USB 2 AUDIO bridge IC the eye pattern is near perfect, even if after every 5m it was so bad with just source cable and sink you would already get dropouts.

I know that doesn't quite address your question but hopefully answers to perfectly reasonable thinking behind it.

Thor

Im grateful for the plathora of info that is so beyond me, i tried using my all 10 of my braincells to decipher that...
So at the worst if one of my usb gizomos 'malfunction' it would simply make it so the signal is 'blurrier' and harder for the next gizmo to detect, but wouldnt be able to 'damage' the signal at all?

 

[Thorsten Loesch]

So at the worst if one of my usb gizomos 'malfunction' it would simply make it so the signal is 'blurrier' and harder for the next gizmo to detect, but wouldnt be able to 'damage' the signal at all?

If it "malfunctions" no signal will pass.

Remember, "active" USB gizmos will decode the "blurry" signal and output a new clean signal.

Unless you have long or severely non usb standard compliant USB cables between individual "gizmos" there should be no way for the signal to degrade enough to cause problems.

Passive (filter) type Gizmos only filter out noise, if designed correctly, the should not impact signal integrity.

Connecting multiple Gizmo in series unnecessarily will increase delay in the chain without offering any benefits. Too much delay, breaks the USB transmission.

One or three iPurifier directly plugged into each other have the same output signal as a single one, just more delay.

In a "tricked out" system you might place a single iPurifier directly plugged into the DAC (NO ADAPTERS) to get the best possible signal integrity at the DAC's USB socket. That's why it's designed as "dongle".

You might add an "iSilencer" at the source end attach clean powersupply. Or you can use an iUSB instead.

From a strictly technical standpoint there is no benefit adding more gizmos in this chain, unless you need to use multiple long cables to bridge the distance between source and DAC, like source is a touchscreen all in one placed at the listening position.

In this case using a iPurifier (adapters allowed) or iUSB 3.0 (or a cheap bus powered USB Hub) between each two cables as "repeater" will allow a long cable.

Note that using more gizmos in this case doesn't make things "better", but rather it allows to overcome fundamental limitations to USB cable length.

Also, strictly speaking the "repeaters" do not need to be "audio grade", only the last device in the Chain might be improving sound quality if it is.

Thor

 

[Assationater]

Man i really shouldve stopped at one ipurifer, but every single one changed the sound so i kept going lol.
Have you seen the new ifi releases? or do you not care at all haha

 

[Thorsten Loesch]

Man i really shouldve stopped at one ipurifer, but every single one changed the sound so i kept going lol.

Hmmm, it's possible, but I'm stumped for a reason. I would need to analyse the overall system to hazard a guess why.

Among other reasons, each IP will lower noise on the USB power line. So if for example the power source is noisy and the DAC uses USB power, it's possible that filtering more noise produces changes in sound.

Have you seen the new ifi releases? or do you not care at all haha

I do notice at times, why?

I see nothing new so far. The only things without my fingerprint all over the blueprints are the Uno and Network streamers.

Even though the streamers integrate building blocks form if gizmos I designed, and use powersupplies based on my designs.

I suspect a Raspberry Pi with a good power supply and a cheap powered hub and volumio will do as well for USB.

Thor

 

[Assationater]

Hmmm, it's possible, but I'm stumped for a reason.

I wish cables and such didnt make a differnce, it would save me alot of money, but welp. And ill be the first one to say that cables and whatnot terrible 'price to performance' usually.

I see nothing new so far.

Idk, i just think the new stuff is kinda interesting. Feels like audios been stagnating for the past few years on the dac and amp side. Diablo 2 looks...cool. Idk,

 

[Thorsten Loesch]

I wish cables and such didnt make a differnce, it would save me alot of money, but welp. And ill be the first one to say that cables and whatnot terrible 'price to performance' usually.

The key is to understand the underlying principles. There are no "magic" cables. Understand the problem and solve it. A correctly designed cable need not be expensive.

For example, for USB I like CAT6e trunk cable with solid copper conductors, one pair USB, 3 pairs for power/ground, shield just shield with a "soft" connection at one side.

I also like CAT6e for line level or even microphone level signals. I also like CAT6e in more complex installed audio systems as "multicore", as each CAT6e cable has four separate shielded pairs that each can carry either AES/EBU or ADAT digital or a single balanced audio channel.

Amazing stuff. Buy a reel, it makes great signal cables for almost any application.

Idk, i just think the new stuff is kinda interesting.

What new stuff?

Feels like audios been stagnating for the past few years on the dac and amp side.

Yup.

Diablo 2 looks...cool. Idk,

Looks are a matter of taste.

Inside there seems little real change from the 2017 designed "Red Label" (which got released in fire-engine red instead of Ferrari red and called Diablo) except adding Bluetooth and bumping the price from $ 899 (which was very poor VFM) to $1299.

The "Phantom" is just merging iCAN Pro & iESL Pro into one box and removing the ability to use a speaker Amp to drive the iESL.

The "Gryphon" is just copy-paste merging xCAN and xDSD and not even upgrading the HP Amp to newer (and better) designs I did for iFi.

From where I am looking there is nothing "new", just tired old and outdated designs without even fixing the known problems where the tech to fix the problems is available to iFi, in new boxes with looks/features, shall we say, that are questionable.

Thor

 

[Assationater]

The key is to understand the underlying principles. There are no "magic" cables. Understand the problem and solve it

Everywhere I listen, there's so much that changes the sound, and it truly hurts my brain trying to think about the physics behind it. Especially when most people are so dead set on nothing but headphones make a difference to audio and there seems to be so little thorough studies on the niche sides of cable, EMI cloth, anti vibration, power filtering and file quality nonsense. Too much for me man...

On the new ifi products though, I really do hope that they sound 'better' but I don't exactly have the money to just throw at em.

 

[Thorsten Loesch]

Everywhere I listen, there's so much that changes the sound, and it truly hurts my brain trying to think about the physics behind it.

Some people are more sensitive than others.

Plus a lot of electronic design these days seems "copy & paste" from a datasheet, often not great system integration, so a lot of potential for small changes in sound exist.

I tend to channel my old self as "six sigma evangelist" and focus on where the least effort can yield the biggest improvements first. Usually I find once fundamentally the root causes for issues are sufficiently well understood we can solve the problem. "

And doing so often stops many "tweaks" that had an impact before from doing anything, because these tweaks were randomly tinkering on symptoms without resolving the underlying problem, resolve the problem and the tweaks are "defeated".

Especially when most people are so dead set on nothing but headphones make a difference to audio and there seems to be so little thorough studies on the niche sides of cable, EMI cloth, anti vibration, power filtering and file quality nonsense. Too much for me man...

Same for me, hence my six sigma approach, take the big wins and don't loose sleep on small fry.

Thor

 

[Assationater]

Same for me, hence my six sigma approach, take the big wins and don't loose sleep on small fry.

Really is good advice, no point if your not enjoying yourself, good music will sound good even on the most sub-optimal hardware if you're in the mood.

 

[Assationater]

@Thorsten Loesch I hope i'm not bothering you, but as ifi support is useless....
if ifi's usb cables are 90ohm, are the ipurifer's 90ohm too, or does it even matter at all?

 

[Thorsten Loesch]

@Thorsten Loesch I hope i'm not bothering you, but as ifi support is useless....
if ifi's usb cables are 90ohm, are the ipurifer's 90ohm too, or does it even matter at all?

All USB 2.0 cabling and devices are mandated as 90 Ohm +/-15%.

Thor

 

[Assationater]

All USB 2.0 cabling and devices are mandated as 90 Ohm +/-15%.

Thor

Thank you! I can rest easy knowing theres a chance my ipurifers are 80ohm and 100ohm XD

 

[Thorsten Loesch]

Thank you! I can rest easy knowing theres a chance my ipurifers are 80ohm and 100ohm XD

No, I doubt it.

The original designs were quite precisely for 90 Ohm and unless there has been some specification creep (different PCB materials, different PCB stackup etc.) should be on spec.

I did not quite get from your question that you asking about the expected tolerances on iFi products.

Modern PCB manufacturing is very precise, if the original design is correct and it is manufacturered according to design individual variation between items should be less than 1%.

The Repeater IC's datasheet is silent on the precision of the build in termination resistors, but 1% is trivial.

Lastly, the actual length of the signal connections in the iPurifier are VERY short, this means they are likely far too short for transmission line effects (which is where this impedance comes in) to become noticable.

This is why iPurifier plugs directly into the DAC's USB port. If the accumulated USB signal path length is below 12" (almost certain) even significant impedance mismatches will have little effect.

So I think you worry about things not relevant in context.

Thor

 

[Assationater]

No, I doubt it.

The original designs were quite precisely for 90 Ohm and unless there has been some specification creep (different PCB materials, different PCB stackup etc.) should be on spec.

I did not quite get from your question that you asking about the expected tolerances on iFi products.

Modern PCB manufacturing is very precise, if the original design is correct and it is manufacturered according to design individual variation between items should be less than 1%.

The Repeater IC's datasheet is silent on the precision of the build in termination resistors, but 1% is trivial.

Lastly, the actual length of the signal connections in the iPurifier are VERY short, this means they are likely far too short for transmission line effects (which is where this impedance comes in) to become noticable.

This is why iPurifier plugs directly into the DAC's USB port. If the accumulated USB signal path length is below 12" (almost certain) even significant impedance mismatches will have little effect.

So I think you worry about things not relevant in context.

Thor

I was mostly joking about the tolerances, thank you for the info though!