Page 2 of 2 FirstFirst 12
Results 16 to 30 of 30

Thread: Klark Teknik DN-800

  1. #16
    Junior Member
    Join Date
    Apr 2016
    Location
    US
    Posts
    7

    My Journey

    Hey Guys,

    I doubt there's much interest anymore in active analog crossovers. But, I still fancy analog gear.

    This also ended up being a much more involved project than I had originally anticipated. Looking back, if I had known how much work this was going to take, I'm not sure I would have embarked on this.

    Nonetheless, I've completed this project and I'm very pleased with the results. I thought I'd share my story. Even if you don't learn anything from this, hopefully you'll be entertained

    My goal was to have two KT DN800, in 4-way mode. I had purchased three units, set in 2-way mode. I ended up cannibalizing one of the units.

    The units were originally set to cross at 1.2kHz. I had decided that a simple decades-based crossover would work well in my setup: 80Hz / 800Hz / 8kHz

    I was unable to find the software for calculating these crossover points. But I did find an electronic copy of the manual. This helped me calculate the correct component values. One of the main reasons why I chose the decades-based crossover points was being able to use the same resistor values for everything.

    My immediate challenge with this was converting from 2-way mode to 4-way. The guide wasn't super obvious about this. I also had several versions of the filter boards. Only one of which seemed to match the component layout in the manual. I made some educated guesses and thankfully came-up with the correct layout on my first try.



    I basically stripped all of the components from 8 filter boards. This is where the project starts to get tedious:





    I decided to solder sockets for the opamps. I figured since I'm gutting these crossovers, I might as well upgrade the opamps. I also decided to drastically lower the high-pass filter for the low-end signal. Here are the boards, populated with my calculated values:





    You will notice that I opted for single-channel SOIC-8 opamps, rather than the composite chips. Two reasons for this... First, single channel opamps tend to have better performance, compared to their composite brothers. Second, I believe this also helps with channel separation.

    What's even more sad, I couldn't find any adapter boards for the 4-channel opamps. So, I ended up making those PCBs. Just when you thought this project couldn't get anymore tedious

    I did a quick check of the frequency responses of my crossover points. Much to my amazement, not only did this even work at all, everything was responding in the expected ranges. I reached the image limit for this post. I uploaded these measurements in another post...

    I figured, why stop here? Why not refresh the limiter boards? I decided to upgrade the opamps and replace all of the capacitors in the signal path:



    And since I've updated the limiter boards, why not update the main boards? Originally, I tried swapping out the +/- 18V regulators, in favor of +/- 15V. It opens the door to a whole lot more opamps (i.e. AD797). And I've done this successfully in other analog active crossovers. But, the KT DN800 was not liking this idea.

    So, I instead dropped in 1.5A +/- 18V regulators. I know what you're thinking. What's the point? The fuses are rated for 1A. But, my thought was a higher-rated voltage regulator would have less voltage droop.

    The new regulators worked just fine in my first crossover. However, in my second crossover, the power kept shutting off (after about 30-60 seconds). This led to about a half a day of swearing and troubleshooting. I couldn't understand why the first crossover was working, but not the second.

    My initial thought was "I must have a short somewhere". But, I simply could not find one. And I wasn't blowing fuses. Then I noticed that each crossover had a different looking toroidal power transformer. I decided to measure the voltage coming off of each one. Sure enough, the first crossover read +/- 24V. And the second one was reading +/- 36V. This input voltage was causing my regulators to overheat! This was resolved by cannibalizing the power transformer from my third crossover. Problem solved.

    Here is the main board completed:



    Since I made so many changes, I figured it would be a good idea to tune the limiter boards. I used a function generator and an oscilloscope:



    Thus concludes my story to active analog crossover perfection...

    -Brian

  2. #17
    Junior Member
    Join Date
    Apr 2016
    Location
    US
    Posts
    7

    Crossover Frequency Response Measurements

    10Hz - 80Hz:



    80Hz - 800Hz:



    800Hz - 8kHz:



    8kHz+:


  3. #18
    Senior Seņor boputnam's Avatar
    Join Date
    Apr 2003
    Location
    northern california
    Posts
    7,007
    Impressive, and interesting...
    What's the slopes of the filters - they seem less than -18 dB but more than say, -12 dB (but I'm having a difficult time interpolating off the bar graphs...).
    bo

    "Indeed, not!!"

  4. #19
    Senior Member grumpy's Avatar
    Join Date
    Oct 2003
    Location
    SoCal
    Posts
    5,338
    I think for < $100 TruRTA can do a swept measurement (not RTA, but higher resolution).
    Great story, and nice to see a project come to fruition.

  5. #20
    Junior Member
    Join Date
    Apr 2016
    Location
    US
    Posts
    7
    Hey Guys,

    Thanks for the feedback.

    I used the "free" version of TrueRTA. I just wanted to see if the crossover was functioning. You can license the app for higher resolutions.

    I just followed the values from the manual. I wasn't able to calculate my own. But, the manual indicates slopes should be -18 dB. I think the low resolution of my RTA readings is throwing this off.

    When I EQ my speakers for the room, I'll probably measure each individual band, rather than measuring a sweep. I have some 27-band cut EQs. I was thinking about using my function generator for each band and measure the levels of every speaker.

    Before, when I had a "simple" bi-amped setup, I just measured by ear. But now that I have 4-ranges per channel, I'm going to have to use a field mic.

    What tools, testing processes do you guys use? I'm not a pro...

    -Brian

  6. #21
    Senior Member grumpy's Avatar
    Join Date
    Oct 2003
    Location
    SoCal
    Posts
    5,338
    if you persist with the meter/RTA method, I suggest you use warble tones for the source
    and an analog meter representation (meter/bargraph/whatever), not numeric.
    Narrow resonances are more readily identified (often not worth trying to eq, but they can affect
    the RTA 1/3 octave filter which averages the in-band spectrum).

    Eventually, you might consider something like REW (can be a bit complicated, but it's also free).
    Something that uses a swept tone, or MLSSA-like signal can be very helpful. There are many
    products and end-to-end systems available (from near zero cost, to many thousands).

  7. #22
    Junior Member
    Join Date
    Apr 2016
    Location
    US
    Posts
    7
    These are good recommendations!

    Good points on using warble tones, over standing waves. I will also give REW a try.

    Kind regards,

    Brian

  8. #23
    Senior Seņor boputnam's Avatar
    Join Date
    Apr 2003
    Location
    northern california
    Posts
    7,007
    Quote Originally Posted by bbuchholtz View Post
    I just followed the values from the manual. I wasn't able to calculate my own. But, the manual indicates slopes should be -18 dB. I think the low resolution of my RTA readings is throwing this off.
    Cool.

    Quote Originally Posted by bbuchholtz View Post
    When I EQ my speakers for the room, I'll probably measure each individual band (pass)
    I like that plan, a lot.

    Quote Originally Posted by bbuchholtz View Post
    I was thinking about using my function generator for each band (pass) and measure the levels of every speaker.
    There's two subjects here - EQ and relative gain of each passband. Yes, do the EQ first, and then adjust the relative gains (and phase, if needed... ) to achieve the most appealing close-to-flat overall response.

    Quote Originally Posted by bbuchholtz View Post
    Before, when I had a "simple" bi-amped setup...
    That sir, is being too modest! Bi-amping is plenty tricky enough - you've just made your situation multiples more complicated!

    Quote Originally Posted by bbuchholtz View Post
    I just measured by ear. But now that I have 4-ranges (passbands) per channel, I'm going to have to use a field mic.

    What tools, testing processes do you guys use? I'm not a pro...
    A very simple set-up is to run a reliable source of Pink Noise and your TrueRTA. And, I suggest using a steady output of Pink for every passband, btw (not a sweep).

    The Pink source is flat. Running Pink through your system, ONE CHANNEL AT A TIME, ONE PASSBAND AT A TIME, using your TrueRTA as a guide (try and use at least 1/3 octave bands) filter the output for each passband to achieve a flat response on the TrueRTA, using the least amount of EQ (i.e., fewest filters) you can get away with. Each filter imparts a phase (delay) anomaly into the signal - the fewer, the better. And, it's always preferable to cut rather than boost. Boosting puts more energy into the listening environment and could cause more troubles...

    STAY WITH ONE SIDE ONLY. Do each passband separately - keep the others muted. Keep Pink running. Then, after all four passbands have been completed, ONE SIDE ONLY, open up all passbands and adjust their relative gains so that the TrueRTA has a close-to-flat overall response across the entire spectrum (you might see mesa-like features, up or down, representing each passband - try and get each level with the other by using only gain). You will note anomalies at crossover points - either dips (cancellation) or humps (summing). Before you try any more filters, try changing phase of any one passband and see how the response changes. Give the True RTA a minute to adapt to the changed signal.

    Ideally, you should have achieved a pretty flat overall response. Copy your settings to the other channel. RUN PINK but do not attempt to measure anything (the TrueRTA will be confused with too many sources). Walk across, and duck through, the sound stage - you will hear comb filtering at areas of passband overlap. Do not let this concern you - it is very audible with Pink but less so with Program material.

    Get busy...
    bo

    "Indeed, not!!"

  9. #24
    Senior Member grumpy's Avatar
    Join Date
    Oct 2003
    Location
    SoCal
    Posts
    5,338
    Excellent I totally blew past driver EQ and xover band balance . Nice save Bo!

  10. #25
    Senior Seņor boputnam's Avatar
    Join Date
    Apr 2003
    Location
    northern california
    Posts
    7,007
    Quote Originally Posted by grumpy View Post
    Excellent I totally blew past driver EQ and xover band balance . Nice save Bo!
    Not at all, man. This is right in my (limited...) wheelhouse - do it nightly...
    bo

    "Indeed, not!!"

  11. #26
    Junior Member
    Join Date
    Apr 2016
    Location
    US
    Posts
    7
    Hey Bo,

    Thank you for your input (and validation).

    I'll cut passbands first. I think I have a fairly good idea on how to work through that.

    Once done with that, I might need a little help setting phase offset. Aside from minimizing anomalies at the crossover points, could this also be done with an oscilloscope? My thought was to adjust phase, until amplitude is zeroed-out at the center.

    Sorry, my EE background is not with audio. I might very well be missing an obvious (and easier) way of doing things.

    Kind regards,

    Brian

  12. #27
    Senior Seņor boputnam's Avatar
    Join Date
    Apr 2003
    Location
    northern california
    Posts
    7,007
    Quote Originally Posted by bbuchholtz View Post
    Hey Bo,

    Thank you for your input (and validation).

    I'll cut passbands first. I think I have a fairly good idea on how to work through that.

    Once done with that, I might need a little help setting phase offset. Aside from minimizing anomalies at the crossover points, could this also be done with an oscilloscope? My thought was to adjust phase, until amplitude is zeroed-out at the center.

    Sorry, my EE background is not with audio. I might very well be missing an obvious (and easier) way of doing things.

    Kind regards,

    Brian
    I mirrored your entire post - for hopefully, good reason.

    I'm not an EE. I'm a self-trained FOH / systems engineer with 41-yrs of dealing with real world realities of typically poorly done installations with crappy gear (the two are not necessarily cojoined). For 30-yrs of that, going into clubs without the benefit of proper measurement gear, we "got by". About ten-years ago, Smaart entered the scene and I knew I had to learn it - even without knowing the circuitry behind everything I was running (because I had no chance/time to fix that), I could at least measure what I was hearing/subject to, and attempt a remedy it with the house's outboard gear. I slowly got good at it.

    Regardless of your background, what I shared is what every single FOH / systems guy does at every load-in, even if it's their PA. Why? Time is precious. As Jaime Anderson (Rational Acoustics) reminds: "How much time does it take to tune and time align a PA?". Answer: "As much time as you have". Period. You can spend, literally, man-days, but typically have maybe (maybe...) 30-minutes running Pink, running with your measurement mic around the room (unless you carry a multiple-input device) no matter the size of the venue or importance of the show. Sometimes I have no time. In any event, if you follow the steps I recommended (if they make sense), you will be very close to a very pleasing outcome.

    However, I tune to flat. For many users, that is annoying and they prefer different amounts of EQ intervention . Fine, but TUNE YOUR SYSTEM FIRST. Ideally, you should have EQ control over your outputs that is SET (system EQ and delay); use a preceding L/R EQ (PEQ always preferred, like for instance if your pre-amp has EQ [mine at home do not]) to tune to taste.

    To your phase question - I'd ignore what the oscilloscope or your maths say. Those are interesting, but the response of your room (environment) to the energy you are putting into it is what matters. Do some simply 180-degree flips to phase of the different passbands. I expect, you will find that the MF and/or LowHF might blend-in more readily for your ears than if every passband is like-phase, but that's just a guess on seeing so many JBL networks where Lansing did just that.

    FWIW, (and off-topic for the 4345s) in systems where your subs are spatially separated from the mains, it's always best to invert the subs and start from there (pay attention all you homies with isolated, mono subs hidden behind the planter... ). Think about half-lambda and the arrival times to your ears. Electronic phase (approx. speed of light) is one thing - acoustic phase (speed of sound off the diaphragms) is an entirely different kettle of fish, and the only one that matters. Smeared arrival times pretty much suck.
    bo

    "Indeed, not!!"

  13. #28
    Junior Member
    Join Date
    Apr 2016
    Location
    US
    Posts
    7
    Hey Bo,

    I also prefer a "flat" sound. For me, the accurate reproduction of sound is most important. I don't like any passband overemphasized.

    Good points on phase. I hadn't considered running my subs 180 degrees out of phase.

    It might be a while before you hear from me a again. I'll be busy positioning speakers, cutting cables and taking measurements. But, I will be back to post results. Thank you for your help and guidance!

    I also captured some high-resolution RTA readings. There were some questions around slope. For completion, here are the new readings. The higher resolution exposes more ringing, but at least the slopes look steeper...

    10Hz - 80Hz:



    80Hz - 800Hz:



    800Hz - 8kHz:



    8kHz+:




    -Brian






  14. #29
    Senior Seņor boputnam's Avatar
    Join Date
    Apr 2003
    Location
    northern california
    Posts
    7,007
    Quote Originally Posted by bbuchholtz View Post
    I also captured some high-resolution RTA readings. There were some questions around slope. For completion, here are the new readings. The higher resolution exposes more ringing, but at least the slopes look steeper...
    Hi, Brian...

    10Hz - 80Hz: This one looks very well behaved. I'm a little skeptical of the information down to 10 Hz, though...

    80Hz - 800Hz: Those two humps, approx. 175 Hz and 400 Hz, if real, look easy to remedy.

    800Hz - 8kHz: This one is troubling - that is some nasty comb filtering. Are you measuring just one side, L or R? Those are octaves and their harmonics. Me no likie...
    I'm trying to understand where that could have originated... Could it be a ripple in your filter? Those are nearly 6 dB peaks.

    8kHz+: This one also appears to have comb filtering - not as pronounced, but why is it there?

    bo

    "Indeed, not!!"

  15. #30
    Senior Member bldozier's Avatar
    Join Date
    Jun 2014
    Location
    mount kisco
    Posts
    825
    why did i open this.

Thread Information

Users Browsing this Thread

There are currently 1 users browsing this thread. (0 members and 1 guests)

Similar Threads

  1. Need 800 series Altec driver info
    By CarlG in forum Lansing Product Technical Help
    Replies: 2
    Last Post: 04-14-2007, 01:23 PM
  2. 800 Array
    By Techbot in forum System Information
    Replies: 9
    Last Post: 08-26-2006, 08:16 PM
  3. Hurricane Katrina
    By Don McRitchie in forum General Audio Discussion
    Replies: 163
    Last Post: 09-18-2005, 03:51 PM
  4. Eq suggestions?
    By Figge in forum Miscellaneous Gear
    Replies: 41
    Last Post: 10-08-2004, 08:35 PM

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •