Blind testing amplifiers.

[Mulder]

If I want to blind test two different amplifiers with the same pair of speakers and the same preamp/DAC, what do you recommend I use to switch between the amplifiers.

 

[DVDdoug]

You can find switch boxes. Usually the same switch can be used whether switching speakers or amps. Or if they both have banana plugs (or some kind of connector) whomever is doing the switching can switch manually. With an ABX test it's pretty common to switch the connections manually, and to unplug every time even if they are re-connecting the same thing.

Assuming you are doing an ABX test (or similar) you'll need to make sure there is no "click" or other give-away when A or B is connected. And if you get a switch, it's best to get a 3-way switch, or a switch with an off-position, so it can be turned-off when you "switch" from A to A.

 

[sergeauckland]

I personally prefer to do AA, AB, BA, BB testing, where the switching is random, and the test is to be able to distinguish A from B with statistical validity. I find that ABX testing where X has to be identified as either A or B doesn't work for me, as by the time I've heard X, I've forgotten what A and B sound like, so end up guessing. The AB, BA, AA BB test, with rapid switching back and forth is exquisitely sensitive to small differences. It also just requires the answer 'same' or 'different'.

S

 

[Sokel]

Depending on the amps you might not want to get one who leaves the amp's output open.
Some don't like it.
On the other hand it goes without saying what you need a good quality one or else you might end up with a truckload of recycling materials.

 

[kemmler3D]

Whichever switch box you get, don't forget you need to match levels very precisely for the test to be valid. The commonly cited number is 0.1dB or so, at the speaker terminals... you will need to use a multimeter for this.

 

[Blumlein 88]

Might be helpful to know which two amps and which speakers you'll be using.

 

[Mulder]

Might be helpful to know which two amps and which speakers you'll be using.

I am going to compare a Purifi-Class D with an EAR Tube amp.

 

[Blumlein 88]

I am going to compare a Purifi-Class D with an EAR Tube amp.

You'll want to place a 220 ohm 2 watt resistor across the outputs of the tube amp. That way when you switch away from the speakers it won't go into oscillation from having no load. Otherwise you could cause damage to the tube amp. That was one of the reasons I thought it good to ask.

 

[mcdn]

I personally prefer to do AA, AB, BA, BB testing, where the switching is random, and the test is to be able to distinguish A from B with statistical validity. I find that ABX testing where X has to be identified as either A or B doesn't work for me, as by the time I've heard X, I've forgotten what A and B sound like, so end up guessing. The AB, BA, AA BB test, with rapid switching back and forth is exquisitely sensitive to small differences. It also just requires the answer 'same' or 'different'.

S

ABX tests done correctly allow you to switch between A, B and X as often and early as you like. Which makes them tricky for amp testing.

 

[Blumlein 88]

Maybe some other folks have better ideas. I think using 2AFC might be more practical for this test. It is often used in food testing. You choose a preference. The one you like better. 2 alternative forced choice. The forced choice means you have to pick a preference even if you think they sound the same. Believe it or not you can sometimes pick reliably one as different even if you think they are sounding the same.

The reason it is more practical than an ABX test is you know each time you switch you are listening to different amps. You just don't know which one is A and which is B. So with an amp switcher doesn't matter you always will be swapping from one to the other and not from one to the same one. You still need to randomize which is A or B first for each trial. But for each trial you can switch back and forth knowing for certain it is a different amp until you think you have a preference. The statistics are simpler to. If you achieve a consistent preference 75% or more of the time you heard a difference at only 5% or less chance of this being a random result. I'd say you need to do at least 20 trials.

Two other things, be sure you check for polarity inversion between both amps. Also check channel to channel balance by checking signal level on both channels of both amps.

 

[Roland68]

If I want to blind test two different amplifiers with the same pair of speakers and the same preamp/DAC, what do you recommend I use to switch between the amplifiers.

The most important thing is that the ground is separate and is switched in the same way as the + line.
Most finished devices are equipped with very simple switches and very thin cables, which is why I built my box myself.
A high-quality switch costs around €10 - €25. With the right connections, everything can be crimped and screwed, no soldering required.

You can also set it up as a switch with a completely insulated cable breakout, without any housing.

 

[Audiolab800]

If I want to blind test two different amplifiers with the same pair of speakers and the same preamp/DAC, what do you recommend I use to switch between the amplifiers.

I did that recently with a friend. What we did was the following (not ideal but as good as we could). Streamer WiiM Ultra -> Optical -> Marantz HD-DAC1 -> Fixed out -> Amp1 -> Variable out Amp2 | Both speaker out connections to a speaker switch A | B.
Volume matching with REW and a calibrated Mic feeded by Pink noise...
Scenario 6 sons each playing for 1min 30 sec in the sequent
song1 -> Amp A | switch | Amp B
song2 -> Amp B | switch | Amp A
...
The listener does not know the relation between the switch and the connected Amp, he is taking notes we discussed before showing which amp is connected to the switch in position A and B
Between the Amps is a price difference of factor 20, the result was stunning....

 

[MaxwellsEq]

Volume matching with REW and a calibrated Mic feeded by Pink noise...

That's a totally inadequate way to level match. You MUST use a voltmeter and tone and get levels equal to within 0.01V. any other method is invalid.

 

[Audiolab800]

That's a totally inadequate way to level match. You MUST use a voltmeter and tone and get levels equal to within 0.01V. any other method is invalid.

Thanks for the technical correction but… There is nothing I MUST do, we did what we liked and we didn*t notice any difference in level between the two. Besides you should not move your head while listening, the slightest movement will give other results… it was good enough for us

 

[pma]

There is nothing I MUST do, we did what we liked

You surely are not obliged to do something. But if you do not follow the recommendations, your test will be pointless and you do not need to share results of improperly done test. Of course, your choice only.

 

[JiiPee]

I 'd say it depends on what You want to accomplish with the test. If the goal is to perform objectively valid sound quality test, then it is actually fairly demanding, as can be concluded from the comments on this thread. On the other hand, if You just want to find out if there is a difference in sound quality that is big enough for You to clearly notice it, and pick the one you prefer, things become much easier. You still want to be blindfolded, and have somebody to do the switching for You, but You can overlook the absolute accuracy of level matching, head movements, the time it takes to perform the switching etc... finer details. Yes - they all do matter when You are searching for truth, but not so much, if You just want to find out if You clearly favor one above the other in practice.

 

[DVDdoug]

Besides you should not move your head while listening, the slightest movement will give other results…

No, you don't have to hold your head still. Head movement makes a difference but it will make a difference with both A and B, and of course you move your head in "real life" listening.

If you are doing an ABX test you can walk around the room, or listen to A, B, and X for weeks, as long as you don't know which one you are listening to whenever you switch to X. But, we usually don't want to listen for long durations because our audio memory is short. And since ABX tests have to be repeated and statistically reliable, it's not practical to listen for weeks between switching.

And if for some inconceivable reason, A sounds different when you move your head, and B doesn't (or if A & B behave differently when you move your head) that's a valid difference. That can happen with speakers, but since speakers sound different anyway, you can always tell which is A and which is B and only possible to conceal the make/model/price, etc.

...Besides level mis-matching, there can be other "give aways" that invalidate the test such as a click whenever you switch to A but not to B, or a click when you actually switch but not when you don't switch, etc.

 

[pma]

Yeah. Nothing is easy and proper knowledge and experience is needed. Audio testing is not a task for lay persons.

 

[sfogg]

I personally prefer to do AA, AB, BA, BB testing, where the switching is random, and the test is to be able to distinguish A from B with statistical validity. I find that ABX testing where X has to be identified as either A or B doesn't work for me, as by the time I've heard X, I've forgotten what A and B sound like, so end up guessing. The AB, BA, AA BB test, with rapid switching back and forth is exquisitely sensitive to small differences. It also just requires the answer 'same' or 'different'.

If you have an ABX testing box you get to choose if you are listening to A, B or X for each trial in a test. You can switch between them as often and as quickly as you like. When I start a new test I typically just jump back and forth between A and B to determine what I think is different between the two DUT. Once I think I have identified a difference I just jump between A and X . If I hear them as the same I then ID X as A, if you hear them as different you identify X as B and then move onto the next trial in the test which re-randomizes the identify of X. At that point I just keep going between A and X and get through 10 or 15 trials in the test.

 

[Audiolab800]

I 'd say it depends on what You want to accomplish with the test. If the goal is to perform objectively valid sound quality test, then it is actually fairly demanding, as can be concluded from the comments on this thread. On the other hand, if You just want to find out if there is a difference in sound quality that is big enough for You to clearly notice it, and pick the one you prefer, things become much easier. You still want to be blindfolded, and have somebody to do the switching for You, but You can overlook the absolute accuracy of level matching, head movements, the time it takes to perform the switching etc... finer details. Yes - they all do matter when You are searching for truth, but not so much, if You just want to find out if You clearly favor one above the other in practice.

I think it was way better than knowing which amp was playing and then have some flavor for the most expensive. Even if the test was not perfect by the book it's way beter than thinking that the most expensive gear must be better than the cheaper one...
Anyhow I'm vary happy with outcome, the Fosi V3's are very good amplifiers, would I by the Accuphase again, wel I don't know for sure..
Just my 2 cents