DDF
100+ Head-Fier
- Joined
- Aug 11, 2006
- Posts
- 238
- Likes
- 12
Quote:
Hi Skylab,
Thanks for clarifying that your tests used the same headphone, which is very important. For the uninitiated: different headphones can give the same SPL in free space (i.e. not on a head) but very different perceived SPL when used. Headphones produce different SPLs into different acoustic loads (seals), and each headphone is often uniquely sensitive to differences in the seal. So, lets say the gain of one amp is adjusted to measure 80dB in free space on one headphone with an RS meter, then repeated with another set of headphones; the perceived levels will very likely be slightly (1 to 2dB) to moderately (3 to 4dB) different between the two headphones (even if the RS meter gave the same SPL). This isn't a "major" level of difference but its more than enough to change a quality perception: for a careful listener, even 1 dB louder will sound better, all else equal. What this means is that to begin to really get a fair assessment of the small differences in amps, all the comparisons need to occur on one set of headphones. If multiple headphones are used, a good break should be taken between cans.
When all amps are used on the same headphone (sounds like they were), then its more accurate a test if the voltage level into the phones is measured, than if SPL is measured (the voltage to SPL will always be the same on the same phones). Even quite small differences in SPL meter placement will result in notable (easily 1, maybe 2 or 3 dB) reading differences out of the headphones. So even if the meter reads the same in different measures, you could be a couple dB out due to even small differences in placement. With a voltmeter and a careful hand on the amp volume control, you can easily nail this down to 0.1 dB matching. In order to get around any differences caused by amp o/p impedance, I measure the voltage using the actual headphones as the load. Take a y-adapter, cut off one connector, and measure the voltage on the stripped cables.
Skylab, I hope you don’t take this as a nock on your results. Your tests are very valuable, and it sounds like you were pretty careful. It’s just that by using a voltmeter instead of an SPL meter, you can helpfully further lessen some of the uncertainty, which may make a difference when comparing amps that are close to each other in SQ.
The other benefit beyond accuracy is that it fast to use a voltmeter to adjust the amps. If you don’t have a voltmeter, a PC sound card will work well. The amp o/p is fed into the sound card i/p and monitored. A great deal of freeware exists that will give a fast level measure (e.g. RMAA).
Dave
PS for context: I used to work in audio and at one time spent months in the lab designing/measuring/specing headphones, on everything from head and torso simulators, to telecom pinna to measuring headphones with Etymotic probe mics in ears of real subjects (more like victims; they didn’t take kindly to all the scotch tape running up their neck
). Measuring headphones present some “unique and difficult” (i.e. total PITA) challenges.
Originally Posted by Skylab /img/forum/go_quote.gif Also, DDF, w/r/t level-matching, I'm not sure I understand your point. I used a pink-noise track to match the levels of the amps, using the same headphones, to within 1db. Why wouldn't this work? Genuinely curious here. It seems that if all the amps were delivering exactly 80db at the headphones on the same pink noise track, then they're pretty well level matched... |
Hi Skylab,
Thanks for clarifying that your tests used the same headphone, which is very important. For the uninitiated: different headphones can give the same SPL in free space (i.e. not on a head) but very different perceived SPL when used. Headphones produce different SPLs into different acoustic loads (seals), and each headphone is often uniquely sensitive to differences in the seal. So, lets say the gain of one amp is adjusted to measure 80dB in free space on one headphone with an RS meter, then repeated with another set of headphones; the perceived levels will very likely be slightly (1 to 2dB) to moderately (3 to 4dB) different between the two headphones (even if the RS meter gave the same SPL). This isn't a "major" level of difference but its more than enough to change a quality perception: for a careful listener, even 1 dB louder will sound better, all else equal. What this means is that to begin to really get a fair assessment of the small differences in amps, all the comparisons need to occur on one set of headphones. If multiple headphones are used, a good break should be taken between cans.
When all amps are used on the same headphone (sounds like they were), then its more accurate a test if the voltage level into the phones is measured, than if SPL is measured (the voltage to SPL will always be the same on the same phones). Even quite small differences in SPL meter placement will result in notable (easily 1, maybe 2 or 3 dB) reading differences out of the headphones. So even if the meter reads the same in different measures, you could be a couple dB out due to even small differences in placement. With a voltmeter and a careful hand on the amp volume control, you can easily nail this down to 0.1 dB matching. In order to get around any differences caused by amp o/p impedance, I measure the voltage using the actual headphones as the load. Take a y-adapter, cut off one connector, and measure the voltage on the stripped cables.
Skylab, I hope you don’t take this as a nock on your results. Your tests are very valuable, and it sounds like you were pretty careful. It’s just that by using a voltmeter instead of an SPL meter, you can helpfully further lessen some of the uncertainty, which may make a difference when comparing amps that are close to each other in SQ.
The other benefit beyond accuracy is that it fast to use a voltmeter to adjust the amps. If you don’t have a voltmeter, a PC sound card will work well. The amp o/p is fed into the sound card i/p and monitored. A great deal of freeware exists that will give a fast level measure (e.g. RMAA).
Dave
PS for context: I used to work in audio and at one time spent months in the lab designing/measuring/specing headphones, on everything from head and torso simulators, to telecom pinna to measuring headphones with Etymotic probe mics in ears of real subjects (more like victims; they didn’t take kindly to all the scotch tape running up their neck