Problem with distortion with OB line array

Status
Not open for further replies.
Well now I have a stereo pair and have performed some (limited) mesurments using my pc microphone. Not certian of the test methordolagy at this moment in time but it apears that I have as predicted roll off >4.5Khz and bass rolloff @ 150Hz. So I purchased some BESTON RT100A ribbon tweeters. I have also ordered the microphone capsuals recomended by Linkwitz to do some better tests.

With alot of EQ putting the bass back in and cutting out most of the trebel (as anything >4.5Khz is garbage) they sound quite nice and my amplifier actualy blew up before they distorted :hot: (early chip amp had no thermal protection and bad heatsink bonding).

Anyway I have figured out how to make very good crossovers based on state variable filters. I plan on 4th order crossovers with the drivers been fed the following:
  • SUBS (20 - 150Hz)
  • MIDS (150Hz - 4Khz)
  • TWEETERS (4Khz +)
It may seem that I am lacking in power for trebel but I looked at the power spectral density of all the types of music I listen to (almost everything) and there was hardley any energy past 4Khz as no instrument has fundimentals greater than that frequancy.


I have ended up with quite complex electronics now with:
  • 6 Channel digital volume
  • 3 Amplifiers
  • Crossover Consiting of about 20 Opamps
  • Equaliser (yet to be designed)
  • Softstart Circuit
:smash: :smash: :smash:



*Having some troubles understanding the gui but is Speaker Workshop a good tool for mesurments? (I used RMAA for the first tests and it dosen't seem to be the right tool as it lacks gated mesurments and can't acutatly test <300Hz in demo version).
 

Attachments

  • obtwoofthem.jpg
    obtwoofthem.jpg
    95.4 KB · Views: 257
Got some nice Freq response mesurments. Purple is from aprox 2" infront of the lowest driver. Bass response isn't getting down to 150Hz unfortunatly and experimenting with test tones shows that for <180Hz the shiny dome rattles so I will have to crossover my woofers higher than I like.

However on the plus side 200 - 4K is very flat, far better than my existing monitors.

Response from 1.5m away shows room nodes all over the place (as expected this is only a small narrow room full of junk). More importantly however is shows comb filtering justifying the tweeters.

I used Room Eq wizard to get these graphs and a cheap trust pc microphone. Checking with a decent set of speakers (monitor audio referance class monsters from the late 80's) the mic apears pretty good. Mesurment mic is on order however for design of the overall system EQ.

For fun I put the inverse transfer funtion into foobars EQ and the result was quite nice sound with some rattle and weird/dead sounding treble. If only it was posible to get a very cheap very small driver capable of covering the entire frequancy spectrum, then I could have just made a huge grid of drivers and EQ'ed them into line. I do know grado headphone drivers would work (as they are capable of low distortion at high power and don't need head seal) but the expense would about £4000 per speaker.
 

Attachments

  • mid.png
    mid.png
    26.7 KB · Views: 203
Visaton FRS8 comes close to full-range for a small driver. Bit of eq and they sound good. I've used it in OB, and OB+BR type thingy, and it sounded good both times. You might want more in the bass section, but they'd be my recommendation.

Chris
 
actualy in hind sight one would need a driver center to driver center distance of 2cm which is slightly imposible. However using a driver with 5cm diameter it would be posible with alot of equalisation to fix the comb filtering I think and comb filtering would only onset at >6.8Khz which is beyond where the ear has fine resolution. When I learn how to construct FIR digital filters I will attempt it. As for the bass my experiance is that it can be EQ'ed in if one has massive amounts of amplifier power and enough drivers to overcome the ineficiancy (and there speakers don't have a dome that like to rattle like mine).
 
Status
Not open for further replies.