Audio-gd Digital Interface

[upstateguy]

Quote:

upstateguy, Yes, the usbface is not asynchronous while hiface is. Although the concept of asynchronous usb audio transfer has been around for years, only a handful of manufacturers use it because of the difficulty in writing the software.

It will be interesting so see if his non-async is going to be better than the HiFace async........

 
Quote:

heh. might as well call it PookieFace or something like that. But in all seriousness, it's outside design really doesn't have anything fun about it and would probably warrant a more serious name like DigitalConvert/Transit and some of the other stuff that has already been suggested.
 
I love MonkeyFace, but the design just doesn't suit the name

I don't care for the word "monkey".  Doesn't seem to have the greatest connotations .....  as in throwing in a monkey wrench,  grease monkey, monkey business, monkey around. or spank the monkey......  I'd prefer a more serious name....my 2 cents. 
 
I also don't care for the use of "Face" for the same reasons......  better to use the whole word, interface.

 

[joseph k]

Jkeny,
 
Quote:

This is because the pin of a male 50ohm BNC is thicker than that of a 75ohm one.

 
Unfortunately, Bruno is wrong here. It's simply not true, 75ohm BNC has exactly the same geometry as 50ohm BNC, center pin diameter included.
This very same topic had been already discussed also at diyhifi.org, and clarified. BNC 50/75 ohm connectors are all interchangeable. I'm doing it daily. No damages.
 
http://www.nexteklightning.com/resources/articles/Lightning%20protection%20of%2075O%20Impedance%20Lines.pdf
 
ciao, George
 
http://www.diyhifi.org/forums/viewtopic.php?p=42500#p42500

 

[mmerrill99]

Yes, Joseph K, I thought Bruno was wrong on a number of counts but thank you for this clarification - it seems that the centre pins of 50 & 75ohm BNC connectors were once of a different size but have been the same size since the 70's or 80's according to this discussion that I was given a link to http://www.radiobanter.com/showthread.php?t=151715&page=3
 
Well, if I hadn't been disabled without notification from Diyhifi.org I might be able to follow your link to Diyhifi - I wouldn't recommend this forum based on my experience & on the personal abuse that was allowed to be posted about me by Jocko & others!

 

[mmerrill99]

Joseph k,
One thing that has me confused is that on the Huber & Suhner 75 ohm attenuators that I have, the female end has white ring of insulation around the inside of the ground shield. This, I thought was a visual sign that the connector was of the 50 ohm variety, the 75 ohm not having any insulation on this ground shield? My 75 ohm minicircuits ones do not have this insulation. Can you clarify this in any way?
 
This is the attenuator i have with 75 ohm on it's label. This same image is also used in the datasheet for the 75 ohm & 50 ohm part - doesn't give great confidence really.

 

[joseph k]

Nice find..
And You are right, it's right there also on mine... (I have the same Huber &Suhner)
And it's not 75ohm, that end of the connector, it's for sure. It's a 50ohm part in there.
What to say, I don't think that they would not know what they are doing..
Rather they have decided to take the risk, for some economy reasons?
 
I had taken a look around in the lab - most of my other 75ohm parts  are OK,  insulator removed everywhere - BNC tee, male/ female connectors etc.

 

[joseph k]

I can try to measure this - it will be not simple, to demonstrate it, even with my Tek TDR.
Because the shortness of the "deviant " part of the signal path, one needs a really high resolution system, very short rise time excitaton pulse, and very high speed sampling. The Tek stops at ~2.3 GHz bandwith, 150ps.
150psec timing resolution gives ~3cm; or ~1" spatial resolution. Even if in practice one can go a bit below this, here we are talking about 5mm signal path length, which deviates downward from 75ohm to ~60-65ohm?, because the plastic will be partially interrupted anyway in the female connector, IF a real 75ohm female is connected.
 
So: this "feature" of this attenuator will give, for 1/6th of 150psec, that is for a ~25psec  period, a ~20% impedance "glitch".
To have a clear resolution of this, one should go for a <30psec rise time system. 35psec would be 10 GHz.
The attenuator guaranteed to work from DC to 1GHz.
I think this is the reason why Suhner do not care about it..
 
I hope the above deduction also gives some hint about how much "paranoid" we should go..
It's valid also for the Hiface..  which should work up to 200MHz.
 
Ciao, George

 

[joseph k]

BUT, the above reasoning should not permit us to deduct that connectors do not count at all. IF we apply full 50 ohm connectors in a 75ohm system everywhere, that means the signal path will deviate for the full length of a mated connector pair, ~3cm, 1", from 75 to 50 ohm. And that starts to be visible even at 2nsec edges! (And clearly obvious on a TDR)
 
Ciao, George

 

[mmerrill99]

Joseph K,
Another thing you might have some idea about is the datasheet error on DIT4192 - I have the latest one SBOS229B – DECEMBER 2001 – REVISED JUNE 2003 (It's also the same on the DIT4096 datasheet )
 
On page 4 the setting of CSS are given as Not Connected mode=0 software mode; mode=1 hardware mode 
On page 8 & subsequent pages the settings are CSS = 0 hardware mode; CSS = 1 software mode.

I have written to TI for clarification.
 
PS can you connect a number of these attenuators together to make it easier for testing?

 

[joseph k]

 
Quote:

PS can you connect a number of these attenuators together to make it easier for testing?

 
An attenuator can be divided into three parts: male input connector-attenuating body-female output connector.
I suppose and ~~see, that the "body" is right impedance. The female is right as well, the only short glitch part is the male connector. The "defective" part is 5mm, the full body is 50mm.
If I would connect them in series, it would be like this: glitch~25psec -- OK for ~225psec - glitch 25psec - OK for 225psec   etc etc..
 
So no, it does not help. I had tried, for now I cannot point out a real difference, with this 150psec instrument.
 
CSS - I did not work with this part yet, but:
 
CSS is a "mode control" pin, used only in hardware mode. In software mode, it's not needed / do not exist, because:
 
 
Quote:

 In Software mode, the channel status data buffer is accessed through the serial control port
 

 
So for me the data sheet is consequent in this.
 
Ciao, George

 

[mmerrill99]

Looking up this evaluation board "SBAU103 – August 2004" seems to indicate that CS connected to ground puts it in hardware mode (if the datasheet on that board is correct )

 

[joseph k]

I think You are quite confused here? The Mode pin (pin 28) is controlling hard / software mode. It's fixed to "1", that is, hardware mode in that EV board.

 

[mmerrill99]

I'm not a bit confused at all, I think you might be however - the CSS pin setting determines how the channel status data is loaded either via hardware settings on pins 2, 3, 22, 23 or via software ie, sent as serial data. Totally separate from the mode setting of pin 28. 

 

[joseph k]

Ok, now I start to understand Your original problem.
 
BOTH the DIT as a whole AND the CSS pin itself can function in two modes: hardware/software.
 
On page 4 of datasheet they refer to the MODE pin settings, so they refer to the  DIT  hard/soft mode.
If MODE pin= 0  then it's software mode, but the whole DIT is in software mode, not the CSS. CSS does not exist in this mode.
 
If MODE pin=1, then DIT is in hardware mode. Here we can start to look at CSS. But nobody said anything on page 4 about the meaning of CSS. It's defined only on page 8, and it's in hardware mode when CSS= 0. And so it is consistent with the EV board.

 

[mmerrill99]

Ah, yes I see that on pg 4 hardware/software refers to the setting of pin 28 MODE pin.
 
And it's only in Hardware mode that the CSS pin settings are relevant - you are correct, george, I apologise! 

 

[shogo33]

Any updates from Kingwa re: usbface?