ALK Jr type . done

[kg4guy]

Another pair done.tap 2-5 are on the small terminal.

[kg4guy]

Another single done I am sending it to a member for a few weeks for him to try out to see if he wants to switch his stock AA's out for this type of network so it will be available after that for someone else if they want to try it out.I will also add that I am not selling these networks to make money I am lending this out for free however I will build this for anybody for parts cost.

[kwingylee]

kg4 guys, I am about to embark on building some universals, where did you get the transformers? Can I use the ones from my AA network?

[Al Klappenberger]

Mark,

It is entirely possible that if you can't measure the difference between to things and you can hear a difference, you are simply measuring the wrong thing! One of the beauties of instrument measurements is that they are extremely focused on one single parameter. The usual reason though is that it is EXTREMELY easy to fool ones self with listening tests. The human brain will very often hear what it expects to hear. No one is immune to this. Even though I am aware of it, I often fool myself as well. This is one of the main reasons I seldom express my opinion about how something sounds. I believe that only a blind A / B switch test where someone else flips the switch is an actual valid test. Even then it's only one set of ears.

Al K.

[sfogg]

"I believe that only a blind A / B switch test where someone else flips the switch is an actual valid test."

Actually that isn't a valid test. Single blind can be influenced by the expectations of the person throwing the switch if they know the identity of what is being switched.

listen to this.... NOW.... LISTEN TO THIS!!!

You want double blind to avoid this. Person throwing the switch doesn't know what they are switching to and can't influence the test.

It is also very possible to do these sorts of listening tests yourself and avoiding bias. That is what A/B/X tests are all about. You know the identity of A and B and X is randomly either A or B. Your job is to identify is X is A or B. You do this for a specific number of trials and if you answer correctly enough times (95% confidence level) you 'pass' the test.

If you can find one I think you would really like an A/B/X Comparator. QSC made one awhile ago, I have one and it is a GREAT tool. For each trial it randomizes the identity of X. Using a remote control you can switch between A, B or X as often as you like. When you have decided the identity of X you tell it your answer and it moves on to the next trial again randomizing X. It will keep score for up to 25 trials and will automatically tell you the probability when you are done. You can of course extend the trials out over days or weeks if desired but I have never felt the need to do that. It lets you know if you are hearing what you think you are, or if you are fooling yourself. Manual here.

Shawn

[Al Klappenberger]

Shawn,

YES! That is the last word in subjective testing. I doubt it's ever going to be done to help resolve the age-old capacitor contrivance though. It would require a very fancy switch to swap out every cap in a crossover network at once. Evan to swap out two complete networks built with two different brands of caps would require a 4 pole two position relay, one for each channel! It could be done, but I bet it never will be! That's too bad too. I would really like to see it.

AL K.

[sfogg]

Al,

It would be easy enough to build an add on to the ABX box to let you switch out complete networks. Just use the existing speaker level relays in the ABX box to switch 5v or 12v DC to external relay banks to switch in/out different networks. You would use single pole relays for each side, not double pole. Otherwise you wouldn't be switching the same way on each change and the sound of the relay switching could give away the identity of X.

Shawn

[Al Klappenberger]

Shawn,

I'm not following you on the switching. Of course single pole switches can be riged to select two differnt things, one is off while the other is on, then swap them, but that still functioning like a single pole 2 postion switch. You would need one set of these for each driver and one to switch the amp between netwroks. That works out to a 4 pole 2 postion switch requiring 8 single pole ON / Off switches for each channelin a 3-way speaker. Am I missing something?

Al K.

[Al Klappenberger]

Shawn,

Here's what i think would be required. The switch would be run by the A/B/X controller. The two networks would be as Identical as they could be made but using two different brands of caps and matched left for right for value as close is possible.

Al k.

[sfogg]

Al,

That would switch it but that would also be giving away the identity of X when you make the switch due to the physical sound of the relays switching, or not switching as the case would be.

In your wiring say for A you are using the NO side of the relay and for B you are using the NC side. For this trial X = B.

The ABX breaks before making when you switch from any input to another.

You switch from B to A (NC to NO) and you hear the relays switch and the music switch. You switch from A to B (NO to NC) and you hear the relays switch again and the music switch again. Now you switch from B to X (NC to NC... in other words no voltage to the coils to no voltage to the coils) and there is no relay switching noise and no switch in the music. You just gave away the identity of X without actually listening to anything more then the relays.

Conversely if X = A you will always hear a relay switch when you go from B to X AND a DOUBLE switch. One from A to X then a switch again from NC to NO when ABX again makes the coil voltage connection. In that situation you are going from voltage to the coil, to no voltage to the coil then back to voltage to the coil again.

Wired in this manor I could identify X 100% of the time no matter what you were testing...even if A and B were identical the differences in switching are a giveaway to the identify of X.

To wire it to eliminate these giveaways you would need independent relays for each network. In each networks relay(s) the signal output of the relays would be on the NO side of the relay and wired to each driver. Now for any switch you will hear the same amount of relay switches no mater if you go from A to X or B to X. You will hear a set of relays open and a set of relays close again every switch. The relays no longer an automatic giveaway.

Shawn

[Al Klappenberger]

Shawn,

I see your point. I think that the sound from the speakers would cover the relay sounds if they were located in a box covered with an acoustic silencing materiel. Another idea would be to use a second or two of rapid relay switching which would stop at the setting dictated by the controller. It would just sound like a buzz. I am assuming that the controller would be a computer program running a random number generator to select the A, B or X condition. A transition period of buzzing shouldn't be hard to do.

I'm not sure if you figure the listener should know when the switching action takes place or not. I would think he should know when to take note of changes. If not, the buzz idea would be out the window!

BTW: I have a box here that has a bank of reed relays (SPST) that are operated by a GPIB buss from a computer. I use it to select channels when testing networks. The computer also runs the signal generator and the analyzer. That thing could easily be used to run the relays that ultimately swap the networks under computer control.

Al K.

[sfogg]

Al,

"'Im not sure if you figure the listener should know when the switching
action takes place or not. I would think he should know when to take
note of changes."

The
listener has complete control over when the switching occurs. It is
totally up to them when and how the switching occurs. For example when
I take ABX tests after I iniitally determine what to listen for I never
even bother listening to B. I just switch between A and X. If A = X
then I answer A, otherwise I answer B.

With the QSC ABX box you are the listener can be the only person involved
in the test since the box handles randomizing X and running the trials
for you as well as keeping score.

Shawn

[Al Klappenberger]

Shawn,

Yeah! That all sounds good. I have been thinking about that since my last post and I was going to say that I infinity want to know when the switch happens, but not what is selected. I also thought about the sound of the relays, if you hear them. You could definitely hear a clue about what was selected since a relay has a different sound on pull in then on release. With 4 poles though, it could be wired such that A is selected by two relays pulled in and two released. B would swap them. Two in and two out at the same time would make the same sound on A or B. I think the "buzz" would still be needed in case the same selection happened twice in a row.

Al K.

[sfogg]

Al,

You can switch between A,B or X whenever you want. With the QSC it is all done by remote control. You know the identity of A and B, you just don't know which X is. X stays constant for the entire trial so you have the ability to switch back and forth between all three as often as you like. When you have determined if X is A or B you enter that as the answer on the QSC and it moves you on to the next trial. At that point it randomly makes X A or B and then again keeps that constant for the entire trial.

"You could definitely hear a clue about what was selected since a relay has a different sound on pull in then on release."

That is why A and B have their own relay(s). On any switch you release the currently energized relay(s) then energize the chosen relay. You do this even if you are on A and select A.

"I think the "buzz" would still be needed in case the same selection happened twice in a row. "

That really is not needed. You just break then make on every selection and the switching is removed from the equation.

Shawn

[Al Klappenberger]

Shawn,

I suppose the next step is deciding to actually do a test like this. I would like to do it, but the cost of building four matched set of networks plus the switching box could get expensive. For me, knowing if I can or can't hear a difference in caps wouldn't matter. What matters is what my customers want. I still would have to use the expensive caps no matter if I can hear a difference or not. People expect it!

Al K.