Proac Response 3.8 stereo imaging is not centered


I picked up a pristine pair of Proac Response 3.8 speakers recently. They sound lovely, however the sound on one side seemed slightly weaker than the other side with the stereo image or soundstage offset to the louder side. This has never been an issue with my other speakers. Speaker cables are the same length. Switching the speaker wires at the amp output doesn't change the asymmetry. Switching the speakers from left to right and vice versa does change the asymmetry. So my conclusion is that the one speaker is slightly louder than the other? The individual speaker drivers are performing fine and the musical presentation is otherwise fine. Could the crossover components be the problem? Any other ideas to troubleshoot this?

kb3

I think @yogiboy above, is referring to relative phase, not absolute phase. As in, one speaker is wired with the negative cable going to positive terminal, while the other is correctly. 

This will cause the image to be offset, or diffuse, and not focused. 

The test with an SPL meter is good. There are many decent ones you can download on your phone. 

OK, changing the phase had no discernable impact. HOWEVER, it appears the midbass driver is not making noise on the weak side DOH! It was moving with the music (maybe sympathetically given the sound waves inside the enclosure?) and the bass and tweeter were filling in around it but it is noticeably silent.

I checked the connections, cone, surround and spider. All appeared good. I pulled the crossover out and all solder connections were good and couldn't detect any loose parts. Unfortunately everything internal to the speaker from the driver connections to the board connections are soldered.

What is my next step? Unsolder everything and check component values? I can't check speaker resistance as long as it is soldered into the circuit. If everything is intact in the driver assembly can the driver still be faulty? A search online yields no used replacement drivers, but apparently a Scan-Speak 8545-01 is an "improved" replacement. Anyone have any experience with this mid driver replacement in the 3.8?

Hi kb3,

You can get drivers for Response 3.8 directly from ProAc, as I did several years ago. The two bottom drivers are identical. I paid around $800 each. In a review in Stereophile, I think they said these drivers were modified for ProAc, so I did not want to buy them on Internet for much less money.    

Once you make a transaction, you can get them in a couple of days. 

I love the speakers, and I hope this helps.

lectronjh50

You should check the capacitors in the crossover, they are the most likely thing to fail apart from the driver itself.  These have to be removed from the circuit to do an actual test.  If you have an ESR meter (equivalent series resistance) you can do an in circuit test that will give you a rough idea if the caps are good.  I know a lot of techs who think an ESR meter is more useful than a capacitance meter, and it is a lot cheaper too.

You can get a 9 volt battery and connect it to the tabs on that midbass driver to see if it moves in or out; if it does, you know the driver is not completely dead.  
 

I contacted the Sound Organization (Proac dealer) who will find out if the original drivers are still available. I'd like to stay original but $800 is a lot of money for one driver! I read the same thing about the Proac drivers being "modified" Scan Speaks but I have my doubts. Guess I'll pull the driver out tomorrow and unsolder it to perform the 9V test and go from there. If the caps on the crossover are electrolytics then they should be replaced anyways, so maybe a crossover overhaul is in my future. Fortunately the PC board is large and simple and the components are discrete and easily removable; the most uncrowded crossover I have ever seen. The components are pretty good quality for a 25 yr old design, with large gauge iron core inductors. Can't tell what caps were used, they appear to be Proac branded. I have requested a schematic form Proac to assist in identifying nominal values.