Piano World Home Page Forums Our Most Popular Forums Piano Tuner-Technicians Forum Calibrating Tuning Software and Microphones

NIST and the equivelent organization in other countries provide audio frequency standards via radio, phone, and internet. Even with atmospheric errors, jitter and the like, the accuracy exceeds the precision of the ETD being calibrated.


Edit: I wouldn't use a voip phone or the internet for frequency calibration. The NIST audio signal is digitized and then converted by your computer back to audio. Whatever errors occurred in the digitization process will be in the resultant signal.

Analogue Radio reception of the audio will give you extreme accuracy. (I don't actually know what level of accuracy - probably 10^-8 or better.)

I now also understand why there are special crystals for high end audio (eg in my EMU0404 external DAC) that reduce the stability down to say 10ppm.

>NIST and the equivelent organization in other countries provide audio frequency standards via radio, phone, and internet. Even with atmospheric errors, jitter and the like, the accuracy exceeds the precision of the ETD being calibrated.

Interesting.


Phone - well your provider most likely transfers the signals digitally and then you are back to some - prabably dirt cheap - DAC

Internet - digital only ? How can you transfer a frequency over that?

Analog radio yes this sounds like the best alternative. What's the frequency on which they transmit these test tones?

DAB would not work.

Check out the NIST website. It gives you all the data you need. I usually use 5, 10, or 15MHz, depending on time of day and atmospheric conditions. They use 500 and 600 Hz amplitude modulated tones with the same accuracy of the carrier.

An high quality outboard DAC/ADC is the best solution for recording, but any cheap internal computer sound card/chip will work for tuning, if properly calibrated.

Actually the DACs (and audio chips) used in phones are extremely high quality. I should know - me and some of my colleagues have been involved in their pre-silicon verification. The transmission medium for digital makes no difference - it's the final DAC and reconstruction filter that matters. So long as you have an accurate and low-jitter clock, all is usually good. I did a test on my devices and got them all to play a digital 440Hz test tone - there was no audible beating between any of them over a time period of a minute or so. This means the frequency accuracy of the clock supplying the DACs was excellent.

Paul

I dont believe that DACs in phones are high quality. Seems to me it would be a waste of money as nobody is expecting very high quality here. There may be many reasons to not hear beating. I suppose the beating you get with a standard 100ppm crystal is very hard to hear anyway, as the beating would be in the order of 1 cycle per 23 seconds worst case.

You might not believe it, but it is true. Cost is related to volume and the volume is exceedingly high for the audio processing IPs that are used in the SOCs on phones. The DACs are usually 1-bit coupled to class-D amplifiers. Just take a look at the specs of the chips Cirrus Logic sells for some examples of what is currently in production. I and my colleagues have worked on both Cirrus Logic and Wolfson Audio audio IPs that end up in phones and the quality is so good that you would be pressed to hear any distortions unless you have an exceptional listening environment and transducers.

If 100ppm tolerance gives 1 beat in 23 seconds at 440Hz, then, to all intents and purposes, it's bang-on and any kind of calibration will likely not improve this any further. No-one can tune to that level of accuracy anyway.

Paul.

I'll check that!

>If 100ppm tolerance gives 1 beat in 23 seconds at 440Hz, then, to all intents and purposes, it's bang-on and any kind of calibration will likely not improve this any further.

Apparently not so 'bang on', this 23sec beats is exactly the +0.2 cents of difference that people were talking about above, and the need for calibrating that.

In my research on this topic, I found that the biggest need for accurate clocks is maintaining the sync of audio, independently recorded, with the video.

It doesn't take long to get perceptibly out of sync with a 100ppm error.

But for tuning, who cares, or can hear 1 beat in 23 seconds? If you can tune your P5s with that kind of accuracy, my hat is off to you.

That's what we do in movie and TV production -- the cameras, sound recorder, and slates all have crystal oscillators. We jam sync -- plug a master into each of them to set them to the same time -- twice a day, before the first shot, and again after lunch.

Still, that's not the most accurate readily available clock. It turns out that GPS requires far greater accuracy, and provides it on each of the satellites. Some studio equipment grabs GPS time, but crystals are good enough for field production.