R/E/P Community

R/E/P => R/E/P Archives => Brad Blackwood => Topic started by: Ed Littman on July 29, 2005, 07:20:55 PM

Title: 88.2 & 96k playback problem
Post by: Ed Littman on July 29, 2005, 07:20:55 PM
I use two rme cards. 1) pad with aes out & 2) the pst with spidf out on an aes cable(made the adapter rca to xlr)
My main set up is to play the original file with the pad & capture with the pst after processing through analog.

I usually capture to 44.1,
This time I did a test, &
my pst card does not get a lock when at 88.2 or 96k but does at 44.1. the card settings show correctly at any sample rate & the file plays in the DAW. when I replace the pst with the pad the pad plays with no problem.
What am I missing? if the card is screwing up i would assume 44.1 would be effected as well.
any ideas?
Ed
Title: Re: 88.2 & 96k playback problem
Post by: bblackwood on July 29, 2005, 07:30:40 PM
Unless I'm missing something, it sure sounds like your AES/SPDIF cable/connector is giving you issues. Do you have an active AES/SPIDF converter handy?
Title: Re: 88.2 & 96k playback problem
Post by: Ed Littman on July 29, 2005, 08:06:23 PM
Yes! that was it. with an active converter it works.

The wiring on the adapter is done very cleanly. why would a spdif to aes conversion loose the clock for higher sample rates?
Ed
Title: Re: 88.2 & 96k playback problem
Post by: bblackwood on July 29, 2005, 08:13:13 PM
Ed Littman wrote on Fri, 29 July 2005 19:06

The wiring on the adapter is done very cleanly. why would a spdif to aes conversion loose the clock for higher sample rates?

It's not converting the impedance - AES is 110 Ohms, SPDIF is 75 Ohms - this alone can cause issues with more sensitive receivers...
Title: Re: 88.2 & 96k playback problem
Post by: Ronny on July 29, 2005, 09:46:24 PM
bblackwood wrote on Fri, 29 July 2005 20:13

Ed Littman wrote on Fri, 29 July 2005 19:06

The wiring on the adapter is done very cleanly. why would a spdif to aes conversion loose the clock for higher sample rates?

It's not converting the impedance - AES is 110 Ohms, SPDIF is 75 Ohms - this alone can cause issues with more sensitive receivers...




Yes, but the ohms are the same with the 44.1k send aren't they and it works without the active converter. I suspect that there is more to it than ohms. I've been interfacing AES and SPdif with just the straight cable mod since early 90's and have had no problems with short cable runs. Never tried it with 88.2 or 96k though.  
Title: Re: 88.2 & 96k playback problem
Post by: bblackwood on July 29, 2005, 10:00:18 PM
Ronny wrote on Fri, 29 July 2005 20:46

bblackwood wrote on Fri, 29 July 2005 20:13

Ed Littman wrote on Fri, 29 July 2005 19:06

The wiring on the adapter is done very cleanly. why would a spdif to aes conversion loose the clock for higher sample rates?

It's not converting the impedance - AES is 110 Ohms, SPDIF is 75 Ohms - this alone can cause issues with more sensitive receivers...

Yes, but the ohms are the same with the 44.1k send aren't they and it works without the active converter. I suspect that there is more to it than ohms.

Errr, the 'impedance' does have an impact when the data is transmitted at 2x or greater speeds...

What else do you suspect it could be aside from "ohms"?
Title: Re: 88.2 & 96k playback problem
Post by: Ronny on July 29, 2005, 10:30:54 PM
bblackwood wrote on Fri, 29 July 2005 22:00

Ronny wrote on Fri, 29 July 2005 20:46

bblackwood wrote on Fri, 29 July 2005 20:13

Ed Littman wrote on Fri, 29 July 2005 19:06

The wiring on the adapter is done very cleanly. why would a spdif to aes conversion loose the clock for higher sample rates?

It's not converting the impedance - AES is 110 Ohms, SPDIF is 75 Ohms - this alone can cause issues with more sensitive receivers...

Yes, but the ohms are the same with the 44.1k send aren't they and it works without the active converter. I suspect that there is more to it than ohms.

Errr, the 'impedance' does have an impact when the data is transmitted at 2x or greater speeds...

What else do you suspect it could be aside from "ohms"?



Not sure, Brad, it just doesn't feel like ohms to me, it feels more like a clock thing. AESid is 75 ohms and is closer to SPdif than it is with AES/EBU. If the cables are short the difference between 110 and 75 ohms hasn't proven to be a problem at 44.1 and 48k, that's all that I know. How is the 110 ohm side going to change or the 75 ohm side going to change, if the sample rate is different? Are you saying that a 75 ohm SPdif output changes ohms when it's outputting 96k versus 44.1k?
Title: Re: 88.2 & 96k playback problem
Post by: bblackwood on July 29, 2005, 10:38:16 PM
Ronny wrote on Fri, 29 July 2005 21:30

Not sure, Brad, it just doesn't feel like ohms to me,

Let's call it 'impedance'...

Quote:

it feels more like a clock thing.

OK, then explain why it works when he swaps cards (same clock) or when he switches to an active format converter (same clock)...
Title: Re: 88.2 & 96k playback problem
Post by: Ed Littman on July 29, 2005, 10:43:17 PM
bblackwood wrote on Fri, 29 July 2005 22:00


Errr, the 'impedance' does have an impact when the data is transmitted at 2x or greater speeds...

What else do you suspect it could be aside from "ohms"?


Ronny may be onto something,  but I don't know what.
The input of my pst card has the same type of adapter/connector as the output.
I was able to send the program through my adc set to 88.2/96k respectfully into the pst & capture with no problem. If it was strictly an impedance issue wouldn't there be a problem going in not just coming out?
Ed
Title: Re: 88.2 & 96k playback problem
Post by: bblackwood on July 29, 2005, 10:51:23 PM
Ed Littman wrote on Fri, 29 July 2005 21:43

bblackwood wrote on Fri, 29 July 2005 22:00


Errr, the 'impedance' does have an impact when the data is transmitted at 2x or greater speeds...

What else do you suspect it could be aside from "ohms"?


Ronny may be onto something,  but I don't know what.
The input of my pst card has the same type of adapter/connector as the output.
I was able to send the program through my adc set to 88.2/96k respectfully into the pst & capture with no problem. If it was strictly an impedance issue wouldn't there be a problem going in not just coming out?

Not necessarily, it could be one of several things:
1- Bad card
2- Proper input impedance but improper impedance presented to output of card (unlikely ime with RME cards)
3- Clocking, but only if you are clocking the output from the card while clocking the input from the incoming AES/SPDIF signal...
Title: Re: 88.2 & 96k playback problem
Post by: Ronny on July 30, 2005, 01:05:09 AM
bblackwood wrote on Fri, 29 July 2005 22:38

Ronny wrote on Fri, 29 July 2005 21:30

Not sure, Brad, it just doesn't feel like ohms to me,

Let's call it 'impedance'...

Quote:

it feels more like a clock thing.

OK, then explain why it works when he swaps cards (same clock) or when he switches to an active format converter (same clock)...




You snipped my question and answered it with another question, but I'll answer yours. Because the ohms aren't changing from 44.1k, but the sample rate on the clock is. Same clock, different sample rates. That's where I'd look for the problem, not impedance. The main differences between AES3 and SPdif as far as the data is concerned is the status bit, the audio data is exactly the same, the newer receiver chips don't typically have a problem interfacing AES3 and SPdif with a straight cable if it's a short length these days. The only big difference I can see is the sample rates on the clock have changed.      
Title: Re: 88.2 & 96k playback problem
Post by: bblackwood on July 30, 2005, 07:01:03 AM
Ronny wrote on Sat, 30 July 2005 00:05

You snipped my question and answered it with another question, but I'll answer yours. Because the ohms aren't changing from 44.1k, but the sample rate on the clock is. Same clock, different sample rates. That's where I'd look for the problem, not impedance.

Then how does a simple active format converter fix the issue?

The reason the impedance matters as at higher fs, you're dealing with a 'faster' signal - if the impedance is incorrect, the receiver in the card may still be receiving enough data at the 'slower' rate of 44.1kHz to read it, but be unable to at higher fs...
Title: Re: 88.2 & 96k playback problem
Post by: Ronny on July 30, 2005, 08:04:09 AM
bblackwood wrote on Sat, 30 July 2005 07:01

Ronny wrote on Sat, 30 July 2005 00:05

You snipped my question and answered it with another question, but I'll answer yours. Because the ohms aren't changing from 44.1k, but the sample rate on the clock is. Same clock, different sample rates. That's where I'd look for the problem, not impedance.

Then how does a simple active format converter fix the issue?

The reason the impedance matters as at higher fs, you're dealing with a 'faster' signal - if the impedance is incorrect, the receiver in the card may still be receiving enough data at the 'slower' rate of 44.1kHz to read it, but be unable to at higher fs...




Higher fs does not mean that the signal travels twice as fast as the halved sample rate, only that it's being sampled twice "as much", all frequencies travel at the same rate, although not the same energy. If freq's traveled at different speeds all your ears would hear would be beat frequencies and harmonic structure would be nigh impossilbe to capture and reproduce. I guess we'll just have to disagree on this one, Brad.
Title: Re: 88.2 & 96k playback problem
Post by: bblackwood on July 30, 2005, 08:21:55 AM
Ronny wrote on Sat, 30 July 2005 07:04

Higher fs does not mean that the signal travels twice as fast as the halved sample rate, only that it's being sampled twice "as much", all frequencies travel at the same rate, although not the same energy. If freq's traveled at different speeds all your ears would hear would be beat frequencies and harmonic structure would be nigh impossilbe to capture and reproduce. I guess we'll just have to disagree on this one, Brad.

No, you simply don't understand what I'm saying.

It's a fact that higher fs means that data is crossing the zero point at a faster rate in any given period of time. That's what sampling frequency means, Ronny - that in one second period of time, you have 96k sampled versus 44.1k samples. The greater number of samples means that you have more instances of zero crossings in any given time period - hence the term 'faster'.

No one is suggesting that the actual signal travels faster through the wire, but higher sampling frequencies means that the signal appears 'faster' to the receiver chip (not faster from point to point, but the number of zero crossings/given time period), and the impedance can definitely have an impact on higher freqs...
Title: Re: 88.2 & 96k playback problem
Post by: bblackwood on July 30, 2005, 09:15:27 AM
Ed Littman wrote on Fri, 29 July 2005 21:43

If it was strictly an impedance issue wouldn't there be a problem going in not just coming out?

Not necessarily - that would depend on the device the card is feeding. The receiver chip is only on the input of a device.

Regardless, without being there, there's little one can do except help you trouble shoot. Have you got it working?

For this very reason, the few devices I have the utilize SPDIF have active format converters and I have had zero issues...
Title: Re: 88.2 & 96k playback problem
Post by: Ed Littman on July 30, 2005, 09:59:20 AM
yes, it's working with the format converter.
Just curious at this point...

I did think that spdif can can run over a aes/ebu cable with no problems.....humm

My aes/ebu cable length is about 25 ft.

no faulty cards, the same issue happend on a third rme card in another computer.

using the clock from my adc or internal respectfully gets the same result.

no problems when it's spif all the way.

thanks for the input,
Ed
Title: Re: 88.2 & 96k playback problem
Post by: Chris Cavell on July 30, 2005, 10:06:45 AM
The change in clocking frequency can have an impact on the load requirements (which you need to meet to prevent transmission line reflections).  The change is usually in how strictly you have to adhere to the requirement, and less often a change in value. There are too many factors that come into play to say why one person's setup actually works in this situation...but suffice it to say that the effects of reflections on a working setup are small enough that they don't have an impact large enough to completely lose sync (you might however be encountering an undue amount of jitter and simply not realize it).

Everything from a tiny cold spot in a solder joint to a minimal amount of oxidation at the connection can have pretty drastic influences on the reflective property of the connection.  It may be that Ed's setup is encountering these very difficult to narrow down elements, or that his system is simply more sensitive to transmission line reflections.

I found a little animated GIF that shows the effect reflections can have on your clocking...kind of anyway, but picture instead of a single pulse (as in the GIF) what can happen as additional pulses are sent WHILE the reflections are bouncing around:
http://courses.ncsu.edu:8020/ece480/common/htdocs/images/aa- xmission-trm.gif

Cheers,
Chris
Title: Re: 88.2 & 96k playback problem
Post by: Ronny on July 31, 2005, 04:14:59 PM
bblackwood wrote on Sat, 30 July 2005 08:21

Ronny wrote on Sat, 30 July 2005 07:04

Higher fs does not mean that the signal travels twice as fast as the halved sample rate, only that it's being sampled twice "as much", all frequencies travel at the same rate, although not the same energy. If freq's traveled at different speeds all your ears would hear would be beat frequencies and harmonic structure would be nigh impossilbe to capture and reproduce. I guess we'll just have to disagree on this one, Brad.

No, you simply don't understand what I'm saying.

It's a fact that higher fs means that data is crossing the zero point at a faster rate in any given period of time. That's what sampling frequency means, Ronny - that in one second period of time, you have 96k sampled versus 44.1k samples. The greater number of samples means that you have more instances of zero crossings in any given time period - hence the term 'faster'.

No one is suggesting that the actual signal travels faster through the wire, but higher sampling frequencies means that the signal appears 'faster' to the receiver chip (not faster from point to point, but the number of zero crossings/given time period), and the impedance can definitely have an impact on higher freqs...




The frequency determines the zero crossing, it doesn't matter what sample rate that you sample a frequency at, provided that you follow Nyquist and have enough samples per second to include the the whole wavelenghth. Below are two pictures of a 10k tone at -6dB one sampled at 44.1k and the other sampled at 96k. I chose a 10k tone because one complete wavelength both positive and negative sides of the frequency is 100 samples when sampled at 44.1k. As you can see there is still only one zero crossing at either sampling rate. The timeline has not been altered, the zero crossings have not been doubled, the two examples are exactly the same except that the 44.1 example has sampled the wave 100 times and the 96k example has been sampled 220 times. In fact the zero crossing falls exactly at the same spot on the timeline at .0034 seconds on both sample examples. The original frequency determines the amount of zero crossings per second, not the sampling rate.
Can you clarify what you mean by 96k having twice the zero crossings?  


index.php/fa/1386/0/
Title: Re: 88.2 & 96k playback problem
Post by: Ronny on July 31, 2005, 04:17:43 PM
Here's the 96k 10k tone at -6dB. The squares represent each sample in time.

index.php/fa/1387/0/
Title: Re: 88.2 & 96k playback problem
Post by: Chris Cavell on July 31, 2005, 04:38:55 PM
The clocking signal has twice the zero crossings...not the data transmitted along that clocking signal.
Title: Re: 88.2 & 96k playback problem
Post by: Ronny on July 31, 2005, 06:38:13 PM
Chris Cavell wrote on Sun, 31 July 2005 16:38

The clocking signal has twice the zero crossings...not the data transmitted along that clocking signal.



Ok, I think I understand what you are saying.
Title: Re: 88.2 & 96k playback problem
Post by: bblackwood on July 31, 2005, 09:03:17 PM
Chris Cavell wrote on Sun, 31 July 2005 15:38

The clocking signal has twice the zero crossings...not the data transmitted along that clocking signal.

Precisely.
Title: Re: 88.2 & 96k playback problem
Post by: dcollins on July 31, 2005, 11:28:07 PM
bblackwood wrote on Sat, 30 July 2005 04:01


The reason the impedance matters as at higher fs, you're dealing with a 'faster' signal - if the impedance is incorrect, the receiver in the card may still be receiving enough data at the 'slower' rate of 44.1kHz to read it, but be unable to at higher fs...



Isn't this the same as the "Toslink" thread that we slogged through recently?

Aka, "where's the clock?"

The interested can Google "RF transmission lines"

DC


Title: Re: 88.2 & 96k playback problem
Post by: bblackwood on August 01, 2005, 05:59:45 AM
dcollins wrote on Sun, 31 July 2005 22:28

Aka, "where's the clock?"

You have no idea how many times I've almost typed that exact question in this thread already...
Title: Re: 88.2 & 96k playback problem
Post by: Chris Cavell on August 01, 2005, 08:56:12 AM
bblackwood wrote on Mon, 01 August 2005 04:59

dcollins wrote on Sun, 31 July 2005 22:28

Aka, "where's the clock?"

You have no idea how many times I've almost typed that exact question in this thread already...



Uh oh... Shocked

(I wasn't around for that one...I can only imagine the confusion that led up to it...or came about because of it)