I could hear the difference. But why.

[voodooless]

To be clear, you'll need about .5% accuracy to get it within .1 dB

 

[Sokel]

To be clear, you'll need about .5% accuracy to get it within .1 dB

Yep.
-/+ 1% (or worst as we go higher in freq) is way off.
I can short of trust the particular one cause the results are constantly repeatable normally but I wouldn't for a serious application.

Cheaper ones are a nightmare,they jump all around the place and its not only because of the circuitry but because of their horrible leads too.
Some need a vice to give readings,I have measured as much as 4 Ohm resistance to a 20 euro one across the leads.

 

[voodooless]

I can short of trust the particular one cause the results are constantly repeatable normally but I wouldn't for a serious application.

Yeah, so yours is in the top right segment Sadly, usually, it's not specified how the inaccuracy manifest itself, so better be safe than sorry.

 

[SIY]

To be clear, you'll need about .5% accuracy to get it within .1 dB

No. You need 0.5% precision, which again is trivial- even the $10 Harbor Freight DMMs will give you that.

Remember that for a level match, you just don't care whether the actual level is (say) 0.800V or 0.810V, all you care is that both are the same.

 

[Blumlein 88]

I don't know why. OVER and OVER and OVER again a huge majority of people jump straight to level matching with their phone. Then they'll be obstinate and argue far too long that it is good enough against evidence to the contrary.

Maybe we should come up with a reasonable way to do it that way. @preload did claim that (for headphones) and I did test it and it did work. However the details of that are also NOT going to be not done by most people. Maybe if we come up with a peculiar, but easy to use test signal and made it available for level matching.

What preload did with headphones was place an SLM very close to the headphone in stable position, and keep very quiet. I did get .1 db consistency using REW RTA readout with 128k FFT size or larger as that averaged the result over 3 seconds. I even managed to get that with speakers. The SLM or Umik had to be placed and left alone. You need a long FFT to average over time or an SLM with long averaging. You also could not move yourself anywhere in the room during the test. Obviously all of this could work if you had no other options. Just using a voltmeter is far easier with fewer pitfalls.

Anyone have other ideas for a stable easy to use signal you just read with your phone?

 

[Blumlein 88]

No. You need 0.5% precision, which again is trivial- even the $10 Harbor Freight DMMs will give you that.

Remember that for a level match, you just don't care whether the actual level is (say) 0.800V or 0.810V, all you care is that both are the same.

Yes, where one might run into trouble is making allowances for different levels. Like one device has far higher output, but say it is a DAC with variable output. You measure it with one voltage and measure another at a different voltage and adjust levels without checking again. The meter may vary between say .8 volts and 1.6 volts enough to throw you off. Not common, but really cheap meters can be this bad. So you measure DAC 1 at 1.6 volts and reduce it 6 db expecting it to match the .8 volts of DAC 2. Easy solution is after the fine tuning check it one more time at the same voltage for both sources.

 

[Mnyb]

with a cheap meter i would also chose to use a fairly low frequency <400Hz, 50Hz or 60hz maybe ? that's often what they are built to measure ?

 

[voodooless]

No. You need 0.5% precision, which again is trivial- even the $10 Harbor Freight DMMs will give you that.

Yes, I agree, but precision isn't generally something that's in any spec sheet of a DMM, so accuracy as a proxy will do.

 

[SIY]

Yes, I agree, but precision isn't generally something that's in any spec sheet of a DMM.

No, because it's basically built in. If you have a five digit meter, the precision will inherently be better than 0.1%.

 

[SIY]

with a cheap meter i would also chose to use a fairly low frequency <400Hz, 50Hz or 60hz maybe ? that's often what they are built to measure ?

Normally, anything up to a kHz or two is fine, remembering that even if the meter rolls off and gives an inaccurate number, the repeatability will still be more than good enough.

Again: the actual amplitude is unimportant. Only the relative number matters.

 

[voodooless]

Anyone have other ideas for a stable easy to use signal you just read with your phone?

Your phone must be super stable for this to work. A deviation of 2.5 cm from the source would already be the .1 dB difference in gain that is needed for level matching. And even then you'll need some long averaging, and you then have no way of knowing if the differences are from some room source, or a gain difference. it would need several runs at random times to get this done. Just use the f*ching DMM and save you all the trouble

This brings us to another point: Equally for the blind test to work, you yourself cannot move more than 2.5cm as well while performing the test, otherwise the gains won't match either. Never mind different reflections and other room effects that can be highly localized. All of these things just show how silly it is to use speakers to try to perceive differences that are so tiny.

No, because it's basically built in. If you have a five digit meter, the precision will inherently be better than 0.1%.

That's the whole point, isn't it

 

[pinger]

I think if you did one of these blind tests and never changed the source the listener would still think he heard differences.

 

[Sokel]

No. You need 0.5% precision, which again is trivial- even the $10 Harbor Freight DMMs will give you that.

Remember that for a level match, you just don't care whether the actual level is (say) 0.800V or 0.810V, all you care is that both are the same.

I have a cheap one that has this deviation (and larger) measuring the same thing after 2 seconds.
It also deviates while applying different pressure at the leads

Now imagine that I have to measure 3-4 freq points at dubious gear and the same at the other one.
0.5 % precision looks like utopia to me with the shorts of that one.

You engineers are spoiled by your nice gear

 

[pinger]

Won't that always be the case? Should hearing be any different than our other senses? Just like two people will experience a different flavor profile when drinking, for example, from the same bottle of fine wine. I'm definitely a "believer" when it comes to ASR and the need for objective measurement, but I'm not blind to the reality that all sound must be processed through our inherently subjective human brains. That's why even someone like Amir, with his wealth of experience and knowledge, still subjectively skews listening results without an objective measurement of same. If anything, we're too prone in ASR (guilty here) of eschewing subjectivity as irrelevant rather too quickly.

The issue here though is comparing two different bottles of wine not different people drinking the same bottle

 

[Sokel]

Your phone must be super stable for this to work. A deviation of 2.5 cm from the source would already be the .1 dB difference in gain that is needed for level matching. And even then you'll need some long averaging, and you then have no way of knowing if the differences are from some room source, or a gain difference. it would need several runs at random times to get this done. Just use the f*ching DMM and save you all the trouble

This brings us to another point: Equally for the blind test to work, you yourself cannot move more than 2.5cm as well while performing the test, otherwise the gains won't match either. Never mind different reflections and other room effects that can be highly localized. All of these things just show how silly it is to use speakers to try to perceive differences that are so tiny.


That's the whole point, isn't it

I wonder about the precision by measuring with my interface and REW or Multitone,which is what I do for convenience (I mean electrically,calibrated against the DMM which is also calibrated))
That one is stable to .001V (averaging) all the time but I don't know if its because of accuracy and if it's right.

 

[voodooless]

I wonder about the precision by measuring with my interface and REW or Multitone

Do you mean using your interface instead of the DDM? That should be perfectly fine as long as you don't tough any analog or digital settings

 

[SIY]

You engineers are spoiled by your nice gear

I have some nice meters, but for stuff like this, I use a $10 special from Harbor Freight.

 

[morillon]

just a correct true-rms is generaly enought ...
(very common now and affordable)
;-)

 

[Blumlein 88]

I have a cheap one that has this deviation (and larger) measuring the same thing after 2 seconds.
It also deviates while applying different pressure at the leads

Now imagine that I have to measure 3-4 freq points at dubious gear and the same at the other one.
0.5 % precision looks like utopia to me with the shorts of that one.

You engineers are spoiled by your nice gear

While obviously we cannot know what someone will use, I'd throw such a meter in the trash bin.

 

[Salt]

Back on topic and the question above: can the HDMI stream be captured and compared via Deltawave?