Kyma Forum
  Kyma Support
  White noise at startup

Post New Topic  Post A Reply
profile | register | preferences | faq | search

next newest topic | next oldest topic
Author Topic:   White noise at startup
tuscland
Member
posted 28 April 2005 08:42         Edit/Delete Message   Reply w/Quote
Hi !

My Capybara outputs a (sort of) white noise when powered on. The level is - 2 dBu ; it is very loud if no attention is given when powering on the devices.

I would like to know if it is a bug or a feature ;-)
By the way, I saw that there will be a firmware update in order to run ASIO or CoreAudio drivers with the Capybara. In the case that this is not an intended feature, is there a way to take this opportunity and tell the Capybara to output nothing when powered up (before Kyma connects to it)?

Last thing: there is a tiny high pitched noise while operating. I don't find it annoying but I would like to know if others noticed it and if yes, if they measured it. The noise is not present on digital outputs.


Best Regards,
Camille

IP: Logged

SSC
Administrator
posted 28 April 2005 10:08         Edit/Delete Message   Reply w/Quote
Unfortunately, the D/A converters output noise until they receive a proper sample rate clock. The clock is started when the first program is loaded into the Capybara, during the start up of Kyma. I think the only way to work around this "feature" is to put the levels down on your mixer until Kyma has finished starting up.

IP: Logged

tuscland
Member
posted 28 April 2005 10:45         Edit/Delete Message   Reply w/Quote
OK !

:-)

IP: Logged

KX
Member
posted 28 April 2005 12:43         Edit/Delete Message   Reply w/Quote
quote:
Originally posted by tuscland:

Last thing: there is a tiny high pitched noise while operating. I don't find it annoying but I would like to know if others noticed it and if yes, if they measured it. The noise is not present on digital outputs.


Best Regards,
Camille


My Capy is VERY quiet; if I want to hear a hiss I have to crank the gain to max and monitor level to max!


IP: Logged

SSC
Administrator
posted 28 April 2005 13:21         Edit/Delete Message   Reply w/Quote
Is this noise present in the analog outputs or could it be coming from the Capybara itself? Can you record 5-10 seconds of this and email the recording to symsound@symbolicsound.com?

[This message has been edited by SSC (edited 28 April 2005).]

IP: Logged

pete
Member
posted 29 April 2005 12:03         Edit/Delete Message   Reply w/Quote
Hi Cam

I just wondered if you were connecting to your mixer as balanced or unbalanced ?


IP: Logged

tuscland
Member
posted 29 April 2005 12:11         Edit/Delete Message   Reply w/Quote
quote:
Originally posted by pete:
Hi Cam

I just wondered if you were connecting to your mixer as balanced or unbalanced ?



Hi Pete,

My Capybara is connected balanced with 'XLR male / XLR female' connectors, but I plan to buy 'XLR / balanced jack' cables and connect the Capybara on the Line jack input because the gain is too hot for the mic input - however the mic input of my small Mackie has lots of headroom, but still I need to lower the Capybara's output gain.


Cheers,
Cam

IP: Logged

rafe
Member
posted 29 April 2005 12:42         Edit/Delete Message   Reply w/Quote
hi cam

there was a posting a while back that the capy analogue outs are actually phase reversed so if your inputs on the mixer are not truly balanced (some manufactures use balanced connecters but the internal wiring is such that the neg side of the signal and the ground get wired together) if this is the case then the combination can be sending noise from the ground line into your audio chain. That noise may not always sound like 60hz buzz it can sometimes just be a hiss. Perhaps someone else on the forum can verify about the analogue i/o as the posting was a fair while ago.

Also the under the kyma > preferences menu there is a setting for the output level of the capy.

r

IP: Logged

tuscland
Member
posted 29 April 2005 13:09         Edit/Delete Message   Reply w/Quote
quote:
Originally posted by rafe:
hi cam

there was a posting a while back that the capy analogue outs are actually phase reversed so if your inputs on the mixer are not truly balanced (some manufactures use balanced connecters but the internal wiring is such that the neg side of the signal and the ground get wired together) if this is the case then the combination can be sending noise from the ground line into your audio chain. That noise may not always sound like 60hz buzz it can sometimes just be a hiss. Perhaps someone else on the forum can verify about the analogue i/o as the posting was a fair while ago.


The hiss is around 8 kHz, here a screenshot of an analysis I made under Wavelab. I can live with it, but it is a bit annoying, see my answer to your next question.


quote:
Also the under the kyma > preferences menu there is a setting for the output level of the capy.

Yeah, as the Capy's output are very hot for the mic inputs of the mixer, I used this setting to lower the output. But this leads to two VERY annoying side-effects :

1. The DSP status window doesn't show a normalized digital output, but the output that is actually sent to the D/A converters. I mean if a Sound outputs at 0dB, the DSP status window will show a signal at -15dB. I find it annoying because I have less readability of the actual headroom I can dispose, and I don't really know if the Sound is clipping or not. (I hope this is clear enough, english is not my mother tongue ;-))

2. If I digitally lower the output of the Capy by 15 dB, I also loose 15 dB of dynamics. For instance, I measured the hiss to have a peak at 8 kHz, - 96 dB (the measure is not precise at all). Having the Capy outputting 15 dB lower, makes the hiss by relation 15 dB higher : - 81 dB.


In conclusion, I don't really like to lower the gain using this option, I prefer to lower the gain on the mixer.


All the best,
Camille

IP: Logged

tuscland
Member
posted 29 April 2005 13:10         Edit/Delete Message   Reply w/Quote

Capy-Spectrum-96kHz.zip

 
I forgot the attachement ...
This is a gif file.

IP: Logged

rafe
Member
posted 29 April 2005 14:35         Edit/Delete Message   Reply w/Quote
Try the line inputs not the mic pre's.

IP: Logged

tuscland
Member
posted 29 April 2005 14:49         Edit/Delete Message   Reply w/Quote
I definitively decided to switch to the line inputs.
However, the hiss is also present when the Capy is directly plugged into the monitors.
Without the hiss, the Capybara's analog outputs would be perfect.

IP: Logged

KX
Member
posted 29 April 2005 19:15         Edit/Delete Message   Reply w/Quote
-96db being the theorical absolute lowest value on a 16bit signal,
it seems you crank your speakers to very loud level...!

I mesured my Capy and as your graph showed,
mine has a slight bump at 8khz (and its harmonics) too.
I never noticed that by ear before (and I don't hear it now
as it is very weak and narrow).

IP: Logged

tuscland
Member
posted 30 April 2005 08:16         Edit/Delete Message   Reply w/Quote
You're right !!!
I realized that my message did not reflected what I wanted to express. I don't really complain about the quality of the outputs, I was just surprised to hear that "slight bump at 8 kHz". My firewire sound card is not very good on the implementation and has a lot of side effects, that's why I though this could be one of those.

Thank you all for your help!
Camille

IP: Logged

SSC
Administrator
posted 30 April 2005 14:47         Edit/Delete Message   Reply w/Quote
We are trying to verify this today using an Audio Precision Portable One Dual Domain audio analyzer and a loop-back test (Capybara outs to Capybara ins) in combination with the Kyma spectrum analyzer. We will let you know what we figure out.

[This message has been edited by SSC (edited 30 April 2005).]

IP: Logged

SSC
Administrator
posted 01 May 2005 20:33         Edit/Delete Message   Reply w/Quote
It turns out the tones were around 7 or 8 kHz and more than 96 dB down. They were due to the repetition rate of limit cycles in the Burr Brown oversampling D/A that would occur only when converting a constant stream of zeroes. By changing the resting value to a value slightly different from zero, we are able to eliminate the limit cycle tones. We will make this fix into a Preferences item in the next update of Kyma that we post on the tweaky. (BTW, the tones appeared in the analog outputs only, not in the digital outputs.)

It's interesting that, in standard SNR measurements, sine wave peaks don't contribute much at all to the overall noise measurement (integrated from 0 to 20 kHz). But the human auditory system is able to pick out even these very low level sine waves that are just above the white noise floor.

Thanks to Camille and Karl (for bringing the tones to our attention and helping us test a solution over the weekend).

IP: Logged

robertjarvis
Member
posted 03 May 2005 05:12         Edit/Delete Message   Reply w/Quote
quote:
Originally posted by SSC:
We will make this fix into a Preferences item in the next update of Kyma that we post on the tweaky.

Why integrate this into the Preferences? Are there some occasions where this anomaly might be useful?

Robert

IP: Logged

tuscland
Member
posted 03 May 2005 06:44         Edit/Delete Message   Reply w/Quote
I agree, I don't see a reason why this should be configurable. However, this anomaly looks like it is linked to a calibration problem, thus varying between different systems. If this is the case, then a tool could be written in order to find the sweet spot (good exercise for a beginner, I will try this evening ! ;-) )


Camille

IP: Logged

SSC
Administrator
posted 03 May 2005 10:11         Edit/Delete Message   Reply w/Quote
The reason for making it optional is that the same stream of samples goes to the digital *and* the analog output. The limit cycles are in the D/A chip so they appear in the analog output *only*. Someone using the digital outputs *only* may not want to bother making this change. (On second thought, maybe you *could* make an interesting texture by fully cranking up the volume, pressing your ear up to the speaker, and listening to the noise floor ;-)

We have tried out our solution on 20 D/As over the weekend and it works for all cases. So it is not a situation where you will have to "tune" it or find the sweet spot. Since it is a digital filter, there won't be variations in its response from chip to chip. Once we have incorporated into the code, it will just work automatically on all Capybara-320s.

Limit cycles are well known artifacts of all digital IIR filters which, for certain constant inputs can get into a repeating cycle due to a combination of quantization and feedback. Since this particular limit cycle occurs in a digital filter that is *inside* the Burr Brown chip, we cannot change the filter itself. But we *can* change the value that we feed it as the "resting" or silent value.

If you are interested in reading more about limit cycles, here are a couple of links:
http://www.bores.com/courses/intro/iir/5_quant.htm http://mathworld.wolfram.com/LimitCycle.html

IP: Logged

All times are CT (US)

next newest topic | next oldest topic

Administrative Options: Close Topic | Archive/Move | Delete Topic
Post New Topic  Post A Reply

Contact Us | Symbolic Sound Home

This forum is provided solely for the support and edification of the customers of Symbolic Sound Corporation.


Ultimate Bulletin Board 5.45c