Register a SA Forums Account here!
JOINING THE SA FORUMS WILL REMOVE THIS BIG AD, THE ANNOYING UNDERLINED ADS, AND STUPID INTERSTITIAL ADS!!!

You can: log in, read the tech support FAQ, or request your lost password. This dumb message (and those ads) will appear on every screen until you register! Get rid of this crap by registering your own SA Forums Account and joining roughly 150,000 Goons, for the one-time price of $9.95! We charge money because it costs us money per month for bills, and since we don't believe in showing ads to our users, we try to make the money back through forum registrations.
 
  • Locked thread
qirex
Feb 15, 2001

I saw this on an audiophile website and decided more people needed to see it

quote:

As with any passive network, cables contain both resistive and reactive components. This creates resonances and anti-resonances in the cable. A series resonance is when the reactive components cancel each other. At the resonant frequency the complex impedance will be quite low. This series resonance doesn’t impede the signal flow in a cable. An anti-resonance, however, is formed when the reactive components add together to form a highly complex impedance. This “parallel resonance” does impede signal flow in the cable.

It’s generally assumed that the electrical bandwidth of an audio system should be ten times greater than the audio bandwidth. That is, the electronic components should operate out to at least 200kHz. So, what are the first issues that cause distortion when a cable doesn’t work well within that band of frequencies? Cables suffer from a parasitic series resonance at frequencies below about 1.5kHz and from parallel resonances at higher frequencies, determined by the values of the inductance and capacitance. The cable doesn’t function as an ideal inductor. All audio products act as low-pass filters. Cables without networked terminations function as a lossy low-pass filter because of this parasitic capacitance as well as shunt capacitance. The vector seen at the input terminals of an audio signal-carrying cable should be an inductive vector at all frequencies and at all power levels.

We can correct for the parasitic and shunt capacitance by adding reactive components in the network that will offset these effects.

A conventional [non-networked] cable will also operate in a “bi-stable” state. State analysis shows that a system can work in three states—stable, astable, and bi-stable. A cable carrying a low signal level will function in a bi-stable state because of the parasitic capacitance within and between the individual conductors, which when twisted or coiled together form the inductor of the cable. The low-level signal must overcome this parasitic capacitance before it can pass current (the audio signal). So the cable shifts between a capacitive element and an inductive element many times per second because of the audio signal’s varying amplitude. The cable must carry sufficient current to overcome the parasitic capacitance. That makes the cable bi-stable.

During the time it takes for the cable to shift from being astable to stable, the low-level signal carried by the cable is turned into noise. We call this “analog jitter.” Removing analog jitter is one of the reasons why MIT cables have such a black background and have such good low-level detail. The result is proper timbre, transparency, soundstage size, and point-point location of images. No jitter equals no noise component.

Think of a cable carrying two tones of the same frequency, but one is very high in level and the other very low in level. With MIT cable the low-level signals remain intact and are not converted into noise, and are sent in-phase with the high-level signal. Also, the harmonics of the low-level tone are transported in time within the complex tone’s envelope.

All of this describes the technologies I’ve developed over the years: “2C3D,” “JFA” and “JFA-2” (“Jiiter-Free Analog”), and “SIT” (“Stable Image Technology”). SIT means that the image of the instrument or voice won’t move within the soundstage.

The circuit elements in our networks are “time invariant.” That means the relationship between the input and output signals doesn’t change over time. The system should not respond differently to the same input signal at different times.

We have a whole range of impedance analyzers that we’ve bought over the years. We can increase or decrease the applied voltage and measure the impedance with varying power levels at any frequency. This allows us to fully characterize any capacitive or inductive component we use in any given cable. MIT is the only cable manufacturer that quantifies the performance of the products.
this interview was conducted to promote an $80,000 cable

Adbot
ADBOT LOVES YOU

silence_kit
Jul 14, 2011

by the sex ghost
Oh man that's pretty bad, and misuses terms describing real effects in cables and circuits.

However, when you are going at very high speeds, with signals at frequencies much much higher than audio frequencies, you no longer get good functionality from cabling and interconnections for free and the cable and connector construction & design actually matters.

Really nice coaxial cabling which is durable, flexible, and has good phase stability for speeds of 10's of GHz can be pretty expensive, and can set you back on the order of a couple grand per cable assembly. High-quality coaxial connectors which work well up to 67 GHz are $50 a connector.

silence_kit fucked around with this message at 12:52 on Aug 18, 2017

Corla Plankun
May 8, 2007

improve the lives of everyone
i used to work at a company that tried to break into that market with zero understanding of the context

they had the wherewithal to buy an s-mparameter machine but ran out of steam before anyone figured out how to use it (but after they had already ordered a shitload of fancy coax and connectors, of course)

Sagebrush
Feb 26, 2012

qirex posted:

I saw this on an audiophile website and decided more people needed to see it
this interview was conducted to promote an $80,000 cable

ok, i guess i'll be the $80,000 that you need to spend on a cable to mitigate noise up to 200khz

please ignore that 48MHz dollar-store USB 1.1 in the corner

Poopernickel
Oct 28, 2005

electricity bad
Fun Shoe

qirex posted:

I saw this on an audiophile website and decided more people needed to see it
this interview was conducted to promote an $80,000 cable

i guess at that price you only need to sell one a month to book 1M in revenue

FlapYoJacks
Feb 12, 2009
Device tree is so easy to learn and good that if you don't understand it in a hour of looking at other device trees you are bad and shouldn't be modifying kernel code.

hobbesmaster
Jan 28, 2008

ratbert90 posted:

Device tree is so easy to learn and good that if you don't understand it in a hour of looking at other device trees you are bad and shouldn't be modifying kernel code.

so i guess i won't be seeing device tree in any bsps soon

Sapozhnik
Jan 2, 2005

Nap Ghost
good news, i'm not modifying kernel code! i just want this loving thing to boot in a usable state!

bindings docs are poo poo so yes i guess you do have to go code spleunking in practice to figure out what the actual gently caress is going on

like, why the poo poo doesn't this uart work? gpio pinmux? lack of a clock signal to the functional block?? who the gently caress knows???

the raspberry pi 3 is the only embedded platform i've seen where it is far easier to get a framebuffer console than it is to get a uart console and that's just perversely impressive somehow

hobbesmaster
Jan 28, 2008

the raspberry pi project exists to trick you into thinking Broadcom is actually good

Sapozhnik
Jan 2, 2005

Nap Ghost
oh i'm well aware it's just a broadcom marketing exercise

but hmm do i buy a 1GHz 1GB quad core compute module which has a well known retail price of $40 or do i speak to some other vendor's loving salesprick who wants to see how far into the three figures he can push his luck

hobbesmaster
Jan 28, 2008

the thing I don't understand about the rpi is projects go down like this:
start up: woah cool this prototype based on a rpi works great! now we can just get someone to lay it all out on one board and we'll have a product! what's a 30 board pilot going to run?
Broadcom: lol no
startup: huh?
Broadcom: 3k minimum *spam filters startup's email*
startup: hey disty what the gently caress how do I buy Broadcom stuff?
disty: can I interest you in some nice ti processors?

Sagebrush
Feb 26, 2012

that's what they made the rpi compute module for :shepface:

Bloody
Mar 3, 2013

hobbesmaster posted:

the thing I don't understand about the rpi is projects go down like this:
start up: woah cool this prototype based on a rpi works great! now we can just get someone to lay it all out on one board and we'll have a product! what's a 30 board pilot going to run?
Broadcom: lol no
startup: huh?
Broadcom: 3k minimum *spam filters startup's email*
startup: hey disty what the gently caress how do I buy Broadcom stuff?
disty: can I interest you in some nice ti processors?

its easy. you buy rpis and desolder the processors and

eschaton
Mar 7, 2007

Don't you just hate when you wind up in a store with people who are in a socioeconomic class that is pretty obviously about two levels lower than your own?

Bloody posted:

its easy. you buy rpis and desolder the processors and

you buy RPi Compute Modules and SO-DIMM sockets

hobbesmaster
Jan 28, 2008

eschaton posted:

you buy RPi Compute Modules and SO-DIMM sockets

only $40 on the bom!

Sapozhnik
Jan 2, 2005

Nap Ghost
yeah and if you can meet a 3000 unit moq then you can get an actual application soc and put it right on your design, pi cm isn't intended for large volume applications

that and the pi cm doesn't particularly serve the goals of broadcom's marketing department so they actually put a fairly juicy profit margin on it

hobbesmaster
Jan 28, 2008

let's say you did a beagle board project and want to do that pilot run. processor is $20 in quantity 1

but $20 is a rip off you say. well yes digikey charges a ton for actually stocking stuff. TI says <$10 in 1k quantities. it's probably cheaper if you actually have a distributor quote everything on your bom for a few kilounits

FlapYoJacks
Feb 12, 2009

hobbesmaster posted:

so i guess i won't be seeing device tree in any bsps soon

Tons of device trees are provided by Bsps.

hobbesmaster
Jan 28, 2008

how many sign offs do i need for schematics on jokes?

Cocoa Crispies
Jul 20, 2001

Vehicular Manslaughter!

Pillbug
i have no idea to tell why my stupid complicated teensy 3 project doesn't appear to work lol

Sagebrush
Feb 26, 2012

The teensy 3 is real and strong and my friend. What is going wrong?

Cocoa Crispies
Jul 20, 2001

Vehicular Manslaughter!

Pillbug

Sagebrush posted:

The teensy 3 is real and strong and my friend. What is going wrong?

well ya boi got too ambitious trying to be the first to get https://crystal-lang.org/ compiled and running on arm-none-eabi https://github.com/bkerley/os-crystal

i can actually compile everything into an ELF (this took a couple weeks of free time), link it out to an ihex that teensy_cli_loader accepts, but it doesn't blink the LED like i want it to https://github.com/bkerley/os-crystal/blob/master/src/main.cr#L8

next time i feel like looking at it i'm probably gonna make one of the C files that initializes the chip do something with the LED so i can bisect where it's failing

Schadenboner
Aug 15, 2011

by Shine

:flaccid:

Sapozhnik
Jan 2, 2005

Nap Ghost
the kinetis mcu found on the teensy 3 is, shall we say, the dick dastardly of microcontrollers

it is fast but it is also bad

Sagebrush
Feb 26, 2012

it sounds like you people are trying to use something other than god's own programming language, C++

JawnV6
Jul 4, 2004

So hot ...
just attach with a debugger, whats the big deal

Cocoa Crispies
Jul 20, 2001

Vehicular Manslaughter!

Pillbug

Sagebrush posted:

it sounds like you people are trying to use something other than god's own programming language, C++

c++ is trash though, the safety of c combined with the readability of perl

Adbot
ADBOT LOVES YOU

Schadenboner
Aug 15, 2011

by Shine

JawnV6 posted:

just attach with a debugger, whats the big deal

I'm pretty sure that conversion therapy doesn't work and even if it did, you should accept people for who and what they are not your opinion of who/what they should be. Hth.

  • Locked thread