Bottlehead Forum

Bottlehead Kits => Crack => Topic started by: hwaitung on August 14, 2014, 10:24:23 PM

Title: Channel matching measurements?
Post by: hwaitung on August 14, 2014, 10:24:23 PM
Suppose I have multiple combinations of 6080 & 12au7. Will I be able to measure which combination give me the best channel level matching?

Will something like this work?
1. generate artificial sound files of various frequencies
2. play those files through my dac & measure the signals at TRS (AC voltage?)

Any suggestions are welcome!
Title: Re: Channel matching measurements?
Post by: Grainger49 on August 15, 2014, 01:09:04 AM
Start with a mono source and Y-Adapt it to each channel.  Inject a sine wave, 500 Hz is good, and measure the voltage at each output.

You will need to either put in a headphone plug with no cable or flip the amp to measure the voltage.
Title: Re: Channel matching measurements?
Post by: hwaitung on August 16, 2014, 02:28:30 PM
Thanks! I think I need to make a dummy headphone cable then. Flipping the hole thing over will be quite a challenge for taller tubes.
Title: Re: Channel matching measurements?
Post by: hwaitung on October 12, 2014, 01:58:03 AM
Just a followup question. Is there a formula to convert voltage measurement to volume level in dB?
Title: Re: Channel matching measurements?
Post by: Paul Birkeland on October 12, 2014, 06:37:45 AM
20Log(V1/V2) will give you a dB difference between two voltages.

-PB
Title: Re: Channel matching measurements?
Post by: Paul Joppa on October 12, 2014, 12:18:54 PM
... and if V2 is 0.775 volts, then you get dBu, which is often used in pro audio.

The value of 0.775 volts is a historical leftover from telephone engineering, which technology drifted into radio and eventually recording studios. Telephones must use matched impedance lines to prevent echoes, and long ago standardized on 600 ohms; 0.775 volts into 600 ohms is 1.000 milliwatt. The same telephone technology is at the historical root of balanced lines for audio also.
Title: Re: Channel matching measurements?
Post by: hwaitung on October 12, 2014, 02:50:11 PM
Thanks! The worst set of tubes I have yield a measurements of 0.145V/0.136V which translate a 1.28dB difference. That actually confirms what I hear. Luckily, I do find a set that match almost perfectly ;D.
Title: Re: Channel matching measurements?
Post by: Paul Birkeland on October 12, 2014, 03:17:27 PM
That's just under 0.6dB, which is quite difficult to hear.

Gain mismatch will be compounded through both the 6080 and 12AU7, but dominated by the contribution from the 12AU7.

-PB
Title: Re: Channel matching measurements?
Post by: hwaitung on October 13, 2014, 03:59:43 AM
I may well be imagining things up  ;D but it is a very interesting exercise to measure the outputs. I tried a total of 8 combinations using two 6080 and four 12au7. Some combinations actually indicate that the imbalance in 6080 and 12au7 cancel out each other.

BTW, the bad one is a Psvane 12au7.
Title: Re: Channel matching measurements?
Post by: mcandmar on October 13, 2014, 05:53:04 AM
The value of 0.775 volts is a historical leftover from telephone engineering, which technology drifted into radio and eventually recording studios. Telephones must use matched impedance lines to prevent echoes, and long ago standardized on 600 ohms; 0.775 volts into 600 ohms is 1.000 milliwatt. The same telephone technology is at the historical root of balanced lines for audio also.

That is an interesting bit of history.  I used to work with telecomunications and it always fascinated me how technology evolved to allow the humble phone line to transmit data at ever increasing rates.  They used to say 2400 baud was the limit of a twisted pair phone line, then some bright spark figured out how to divide up the signal to give an effective 4800 baud, then echo cancellation raised that to 9600 and 14.4k, then shortly after came 33.6k, and 56k with compression, to current multi channel digital signalling in the megahertz range allowing DSL connections up to 100Mbits down the same bit of old twisted pair copper phone line. Not bad for a system designed to transmit voice ~2000hz :)