Jump to content
  • 0
IGNORED

What is the speed of audio signal ?


STC

Question


I am trying to calculate the speed of audio signal through the speaker wires and wonder if 15 feet difference In the speaker cable length between left and right would introduce a delay of more than 10 microseconds?

 

I thought the speed is about 80% of light but it looks like it is much slower than that. Just trying to figure where the 10 microseconds delay is coming from?

 

thanks. 

Link to comment

18 answers to this question

Recommended Posts

  • 1
8 hours ago, pkane2001 said:

Speed in copper should be in the vicinity of 0.8-0.9x the speed of light. A 15 foot cable would add less than 20ns delay, so look for the 10 microseconds delay elsewhere :)

That figure is a bit on the high side for generic cabling. For most audio cables the signal propagation speed is around 200 million m/s. That gives a delay of roughly 5 ns per metre of cable. For typical cable lengths, we're still talking about tens of nanoseconds of delay, of course.

 

8 hours ago, STC said:

Sitting here thinking whether to replace the cable or add 15 feet because I have given up trying find where the delay is originating from.

Are you seeing a delay between left and right channels? Where in the system have you measured? I'd start by checking if the delay is present already at the output of the DAC.

Link to comment
  • 0
11 minutes ago, pkane2001 said:

Speed in copper should be in the vicinity of 0.8-0.9x the speed of light. A 15 foot cable would add less than 20ns delay, so look for the 10 microseconds delay elsewhere :)

 


Thank you so much. Sitting here thinking whether to replace the cable or add 15 feet because I have given up trying find where the delay is originating from. 
 

Once again thank you. Now I can start to finish wiring the other speakers. :)  

Link to comment
  • 0
2 hours ago, mansr said:

That figure is a bit on the high side for generic cabling. For most audio cables the signal propagation speed is around 200 million m/s. That gives a delay of roughly 5 ns per metre of cable. For typical cable lengths, we're still talking about tens of nanoseconds of delay, of course.

 

Are you seeing a delay between left and right channels? Where in the system have you measured? I'd start by checking if the delay is present already at the output of the DAC.

 

Right. I’ve seen numbers from about 0.6c to 0.95c.

Link to comment
  • 0
10 hours ago, mansr said:

That figure is a bit on the high side for generic cabling. For most audio cables the signal propagation speed is around 200 million m/s. That gives a delay of roughly 5 ns per metre of cable. For typical cable lengths, we're still talking about tens of nanoseconds of delay, of course.

 

Are you seeing a delay between left and right channels? Where in the system have you measured? I'd start by checking if the delay is present already at the output of the DAC.


Actually, I worked with50% of the speed of the sound.  And it is still still in nanos. 
 

I suspected the delay was from the serial data streaming. At 96KHz sampling the delay of 11 microsecond is explainable but you clarified in another thread that the DAC buffers and account for the timing difference. 
 

Now I am thinking how would the DAC treat intentional delay based on each channel. If it buffers for the serial data stream by adding one sample delay then the 11 microsecond is explainable but not sure how I can solve them. 
 

In theory, I wanted 77microseconds delay but a setting of 88 gives the correct result. So I thought the place to look for was the cable length difference. but now, it looks like serial data streaming and DAC buffer is the the most plausible answer to me. 
 

I hope you understand what I wrote just now. I am pretty sure I won’t understand this if I were to look again at this post a couple of months. :)  

Link to comment
  • 0
17 minutes ago, STC said:

Actually, I worked with50% of the speed of the sound.  And it is still still in nanos. 
 

I suspected the delay was from the serial data streaming. At 96KHz sampling the delay of 11 microsecond is explainable but you clarified in another thread that the DAC buffers and account for the timing difference. 
 

Now I am thinking how would the DAC treat intentional delay based on each channel. If it buffers for the serial data stream by adding one sample delay then the 11 microsecond is explainable but not sure how I can solve them. 
 

In theory, I wanted 77microseconds delay but a setting of 88 gives the correct result. So I thought the place to look for was the cable length difference. but now, it looks like serial data streaming and DAC buffer is the the most plausible answer to me. 
 

I hope you understand what I wrote just now. I am pretty sure I won’t understand this if I were to look again at this post a couple of months. :)  

What DAC are you using?

Link to comment
  • 0
15 minutes ago, mansr said:

What DAC are you using?


So many different DACs. They all showed the same behavior. I have replaced everything in my system and it still existed. It is not a deal breaker and I may not even know the difference if the original position of the instrument is not known. 
 

The data streaming is serial from the PC to the DAC, correct?  And when I record digitally will the delay still be compensated by the digital recorder?  But now the reprocessed filed involves an intentional delay which will be corrected by the DAC by one sample, correct?

Link to comment
  • 0
23 minutes ago, STC said:

So many different DACs. They all showed the same behavior. I have replaced everything in my system and it still existed. It is not a deal breaker and I may not even know the difference if the original position of the instrument is not known. 

Where and how are you measuring the delay? With the exception of ancient Sony CD players, every DAC I've checked has had an interchannel delay in the nanoseconds range.

 

23 minutes ago, STC said:

The data streaming is serial from the PC to the DAC, correct?  And when I record digitally will the delay still be compensated by the digital recorder?  But now the reprocessed filed involves an intentional delay which will be corrected by the DAC by one sample, correct?

Forget about the serial data transmission. It is not relevant to this issue. If you were using a DAC with I2S input, I might suspect an inverted LR clock signal, but you say it happens with many different DACs.

Link to comment
  • 0
11 minutes ago, fas42 said:

Where are you actually getting this figure of 10 usecs from? IOW, how are you measuring something, that ends up producing this number?


Speaker distance difference between the left and right ears and the the timing difference. I am sure you can do the math. 

 

1 minute ago, mansr said:

Where and how are you measuring the delay? With the exception of ancient Sony CD players, every DAC I've checked has had an interchannel delay in the nanoseconds range.

 

Forget about the serial data transmission. It is not relevant to this issue. If you were using a DAC with I2S input, I might suspect an inverted LR clock signal, but you say it happens with many different DACs.


As explained above. The calculation gives exact numbers. But in reality you need additional one sample delay. 

Link to comment
  • 0
3 minutes ago, STC said:

Speaker distance difference between the left and right ears and the the timing difference. I am sure you can do the math. 


As explained above. The calculation gives exact numbers. But in reality you need additional one sample delay. 

I don't see an explanation. Please describe precisely what you are measuring and how.

Link to comment
  • 0
5 minutes ago, mansr said:

Still anything but clear. Are you measuring something with a microphone at the intended listening position?


Yes. With UMIK but that procedure was not suitable for my purpose and bit hazy right now about the values I got there. With Roland spaces from 14 to 20 cm, I am unable to find the files now so I couldnt post them here so you could see the difference. 
 

So from the calculation I know exactly when to launch to second cancellation signal but the value is always stuck with the difference if 10 us. 
 

Now back to my snake pit which is coming to a completion after 8 days.  Testing for noise.  Hehehe.....

 

6BE2D719-5FAB-4F9D-8629-417FB698B781.thumb.jpeg.132998419d057c8941fd9affe3f9126f.jpeg

 

Link to comment
  • 0
2 minutes ago, STC said:

Yes. With UMIK but that procedure was not suitable for my purpose and bit hazy right now about the values I got there. With Roland spaces from 14 to 20 cm, I am unable to find the files now so I couldnt post them here so you could see the difference. 

How could you possibly hope to get the positions of the speakers and microphone accurate to 1 mm or so? You need to measure the signal on the wires and see if there's an offset there.

Link to comment
  • 0
13 hours ago, mansr said:

How could you possibly hope to get the positions of the speakers and microphone accurate to 1 mm or so? You need to measure the signal on the wires and see if there's an offset there.


You are looking at this from engineering view point about audio signal. I am aware of that but your ear to ear distant is fixed and no matter how much you move it will be the same distance. The difference of sound arrival time between the two ears will be fixed according to the source direction. it is true that we can’t distinguish the delay of about 10ms but the difference between the two ears delay can be as low as 10 or even 5 microseconds. 

Link to comment
  • 0
Just now, STC said:

You are looking at this from engineering view point about audio signal. I am aware of that but your ear to ear distant is fixed and no matter how much you move it will be the same distance. The difference of sound arrival time between the two ears will be fixed according to the source direction. it is true that we can’t distinguish the delay of about 10ms but the difference between the two ears delay can be as low as 10 or even 5 microseconds. 

10 μs is about 3.4 mm at the speed of sound in air. The slightest misalignment in the placement of speakers and microphone will cause differences of that magnitude. Forget about ears and perception. Unless you're in a lab with some kind of rails for exact positioning, you won't be able to get useful measurements with such precision.

 

If you suspect a skew between the channels in the DAC or other electronics, you should measure the output directly without using a microphone. Then you can easily get nanosecond resolution.

Link to comment
  • 0
32 minutes ago, mansr said:

10 μs is about 3.4 mm at the speed of sound in air. The slightest misalignment in the placement of speakers and microphone will cause differences of that magnitude. Forget about ears and perception. Unless you're in a lab with some kind of rails for exact positioning, you won't be able to get useful measurements with such precision.

 

If you suspect a skew between the channels in the DAC or other electronics, you should measure the output directly without using a microphone. Then you can easily get nanosecond resolution.


Thanks Mansr. I have even moved the speaker a few cm. But you actually made me to rethink about the distance of the pinna. What if it is not what I measured? I know you are going to say it is still not relevant and I do see your point. But I  (we) have been working with putting in this value based on calculation. Now if I substitute the pinna distance it now makes sense. Why didn’t I think of this earlier???!  I feel really stupid for missing the obvious. 

Link to comment

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now



×
×
  • Create New...