L300 with DSP

[alskinner]

My goal is to use DSP to get the L300s to sound better. I have always enjoyed the L300s but as most of us know there are phase and timing issues between the woofer and the HL92 due to its length. Also equalization has to be performed with an outside unit. With DSP a delay in timing is easier. Also crossover settings can be changed in a matter of seconds and equalization can be changed rapidly. driver polarity can be changed easily. I also want to swap the 136As and 2216NDs to see which sound better in this configuration.

My first concern is to protect the drivers which I may be able to do with capacitors. I noticed in the M2 what JBL calls a crossover appears more to protect the compression driver than anything else. Early on I almost learned this lesson the hard way when I was using a trial version of TrueRta. I failed to notice that the sweeps started at 10hz. When I executed the sweep I saw the 2235s violently move back in forth until the cone broke free of the surround. Luckily, I was able to put on new surrounds and after sweeping the driver all was well. Point being if you don't pay careful attention, dsp can damage or destroy a irreplaceable driver.

I researched the forum and the internet for information so far can't find much on this type of implementation with the L300; so I am going to start from scratch. Anytime you stray from engineering specifications it will not be a walk in the park. This will be time consuming as I will have to take careful measurements as I make changes to see if I am heading in the right direction. I was fortunate on the M2 project since most of the settings and adjustments were provided by forum members.

The 2216NDs won't be in till the middle of February. The MiniDsp flex eight will be in Saturday. In the mean time I am trying to find at all I can about the L300 and DSP.

I do have a question. The original L300 was crossed at 800 and 8500. It appears a 12db/octave slope on the low section and a 18db/octave is this correct, also was a Butterworth or a Linkwitz-riley used?

Any thoughts or comments would be appreciated.

Regards
Al

[jmpsmash]

You just have to be careful when you do the frequency sweep. Whatever frontend (most likely a USB DAC or some sort) need to have a volume control. Before that, use your software to generate a sweep or a constant signal (REW can do it) and calibrate it with the DAC/Pre volume control. In REW, you can calibrate it such that 2.83V in REW will actually generate 2.83V at the amplifier output. Then u can very precisely limit the voltage/power to the drivers.

Most ppl will use 2.83V as the reference voltage, but you don't really need to sweep at 2.83V, and for efficient drivers like compression drivers and 15", 2.83V is way too loud. For purposes of crossover tuning, I find that even as low as 1V is fine. 1V at a 8R driver is only 1/8W. Any driver can handle that.

For high frequency compression horn, I find that they can handle 100Hz+ no problem, even at 2.83V. 1V is definitely fine.

[alskinner]

Thank you for the information. I have been using REW for some time and like you said there is great control of output. That's why before any thing else I build the crossovers in dsp to help insure that that unwanted frequencies are not passed on to a driver not able to handle them. I like to start with the original slopes and work from there.

Regards
AL

[Mr. Widget]

I like to start with the original slopes and work from there.

That's what I would do.

With a few measurements and a bit of care and patience, I bet you will create a system that far surpasses the original. My vote is on the 2216Nd.


Widget

[jmpsmash]

That's what I would do.

With a few measurements and a bit of care and patience, I bet you will create a system that far surpasses the original. My vote is on the 2216Nd.


Widget

Agree. Even just something simple as the auto-EQ in REW will produce a much flatter FR than any passive network can. Add back the original xover points and slope you will have a great combo. I did it in a speaker with M2 horn and it sounded great.

[Ian Mackenzie]

https://www.audioheritage.org/vbulle...re-broken-HELP

The L300 crossover (electrical)
Woofer 12 db LP
Horn 12 db HP, 12 db LP
Slot 18 db HP

Have a look at the Nelson Pass L300 project (Google search). Unfortunately you can’t make a silk purse out of a sour’s ear. But using a systematic approach you can a reasonable way to minimising some of the L300’s problems. You will find some of the problems are less obvious than first thought. The dsp can’t think and determine these issues.

My suggestion is to become proficient with REW using a Laptop and a calibrated microphone so that you can do your own research. Relatively speaking the networks JBL put together back then were crude. Mainly built for power handling and reliability.

Measure the system as a whole on and horizontal off axis (important)
Measure each driver individually with the stock crossover
Measure each driver without the crossover.

Listen for enclosure resonances using a swept sine wave. You might find adding additional fibreglass beneficial in certain locations inside the enclosure.

Listen critically to each driver with different program material with the other drivers turned off using the DSP crossover.

Decide if you are going to use the original diaphragms and the original woofers?

While what you have said is in a general sense correct you will need to make some decisions about the crossover frequencies and the specific characteristics of each. The DSP won’t pull a gene out of a bottle for you.

If it were me….
Pull out the mass ring in the 2235H then measure and EQ in DSP (as above).
Measure the LE 85 on the horn (with a new D8R2425 diaphragm) on and off axis. Find the -6 db point off axis angle between 700-1200 hertz. Do the same measurement with the 2235H.
Crossover where the 2235 and the LE 85 off axis -6 frequencies intersect using a 24 db LR acoustic slope measured with REW.
Measure the acoustic time displacement with REW at the crossover point. Add dsp delay to the woofer
Match the woofer and LE 85 levels then measure the two drivers in phase and out of phase. Reverse phase should produce a uniform null at the crossover point.
Using a REW measure the LE85 break up point (around 8000 hz).
Repeat the above process for the LE85-077 crossover point.

The key is to rely on your own measurements and what you learn as you go along. Then listen critically. The above will get you in the ball park but personal preferences come into play. Oh the 2235H is not a great driver above 300 hertz. It’s dull and muted compared to how it sounds with mass ring out.

[alskinner]

Thanks Ian

Just what I needed. I have no illusions that this exercise will result in a world class speaker. my goal is to see how much improvement can be made to the original design using dsp and also to deepen my knowledge and skill set in the digital domain. You have laid out a plan that makes perfect logical sense. It is now up to me to execute.

You are correct in that the 136a does not like frequencies above 300 hz . the 2216nd can go to 800 without distress.

I will now go back into seclusion and work on it. Will re post when I have something useful to report or need further guidance.

Regards
Al

[Ian Mackenzie]

Cool.

You can obtain a calibrated usb mic from Minidsp (if you are using a Mac Laptop) or an audio interface and alternate calibrated mic in the REW help menu).

Once you get the hang of REW there is no going back….
It’s your sailors compass and sextant.

If you go to Charlie’s Audio Pages (audio.claim.net) there are some very good resources around getting sense of measurements. See White Paper - Accurate ln-Room Frequency Response Down to 10 hz. They important things to know about how to take near field and far field measurements (driver diameter dependant and baffle width dependent.

Enjoy and keep us posted.

[Ian Mackenzie]

Hi Al,

I hope the attached measurements inspire you.

Its an overlay of the 2420/L85 with the 2312 horn and 2308 lens. There is also the 2312 horn with the lens off and on and the 2235.
All the measurements where taken on a test baffle @ one metre, gated and 1/12 octave smoothed. It would otherwise be difficult to interpret.

I didn't do any vertical dispersion measurements and I recommend that by rotating the test baffle 90 degrees.

I don't have any 077/2405 measurements or any measurements with the L300 stock network.

Unfortunately measurements alone don't really paint a true subjective picture of how a driver or a system sounds. Any driver be it a direct radiator has a good starting point in its pass band and a good ending point. Balancing this up is as much an art as it is a science.

In this respect some respected loudspeaker designers focus on a smooth midrange though the 1000 - 3500 hertz region where they believe a loudspeaker’s errors are more troublesome than putting effort into managing peaks much further up on the frequency range.

The key is to understand what are the errors and which ones to focus on with a particular set of drivers and enclosure. For example if a set of drivers will never be enthusiastic with classical or opera then focus on making the system with some personality for popular music or rock so it’s enjoyable to listen to on those genres.

Because the upper bass is important in popular and rock genteel it might also mean improving the enclosure by stiffening some panels, adding stuffing in certain areas while adding an 25mm mdf panel to the base of the enclosure. The more a driver kicks the more strain it puts an enclosure under. If the box or panels flex or resonate then bass lines will be blurred. With REW you can identify benign and troubling (long ) resonances with measurements outside. REW will also analyse your room resonance (far field) and accurately model room modes. This will help you distinguish between what the loudspeaker and the room are doing.

This might all sound like a fuss but the more you can isolate and identify problem loudspeaker errors then dealing with them become a matter of prioritising what you can realistically do to get the best out of a system.

In the diy space you are not so constrained by manufacturing budgets. You also don’t necessarily have time constraints getting a system to market. This is why some systems have a mk2 version some 12 months after the release of a single system.

JBL originally used an empirical approach of listening, measurements, network modifications and listening until they were satisfied. I recommend that despite the DSP active crossover automation. These drivers have interesting peculiarities and some time spent investigating this will bring you subjective rewards.

https://www.theabsolutesound.com/art...er-roundtable/

[Ian Mackenzie]

If your keen search for interviews with Andrew Jones and Steve Gutenberg. He is by the the most generous designer who clearly communicates to the market clues and insights on his success which speaks for itself. As an influencer Steve Guttenberg bring value to his channel.

Steve Guttenberg was a reviewer and wrote the Audiophiliac blog for CNET.com for many years. He also wrote for the Absolute Sound, Stereophile, Listener, Audio, Home Theater, Sound & Vision, Fi, Home Entertainment, and many others. Steve also worked for Chesky Records as an Associate Producer on dozens of sessions.

[alskinner]

Hi Ian

Thanks for the resources and graphs. The article from Absolute Sound was very enlightening as to thought processes of speaker design and comparisons between speaker systems of yesteryear and the state of the art today.

I do have a minidsp UMIK-1 mic with the calibration file and a Windows computer. I've taken basic measurements in REW, but have to ask myself how accurate are those measurements. I purpose to learn all I can about it.

On another note a couple of things I like about the Minidsp Flex is being to mute any driver or the left or the right speaker, and the ability to invert the signal of any driver. This keeps from having to constantly disconnecting and reconnecting wires during measurements. I know there are better dsp units out there, but the minidsp works fine for my needs plus I have good understanding of how the device works.

As far as crossovers I have the Nelson Pass modification installed at the moment and also the original.

I plan to document the by saving the measurement readings and including how the measurement was performed as to height distance and angle and any changes to the cabinet, drivers or crossovers and the reasoning behind them.

Regards
Al

[Robh3606]

I do have a minidsp UMIK-1 mic with the calibration file and a Windows computer. I've taken basic measurements in REW, but have to ask myself how accurate are those measurements. I purpose to learn all I can about it.

I plan to document the by saving the measurement readings and including how the measurement was performed as to height distance and angle and any changes to the cabinet, drivers or crossovers and the reasoning behind them.

Regards
Al

Hello

Getting your methodology up to snuff is one of the most important aspects of getting repeatable measurements. Documenting the test conditions as you plan to do is a great first step and after going through the process a few times it will become intuitive.

Have Fun!

Rob

[alskinner]

Thanks Rob

You are absolutely correct. Without a step by step method to the problem and getting each step right before moving on, frustration and failure is likely to occur. I am confident that with some time and patience I'll be able to achieve my goal.

Just to get off topic for a moment, this was what the famous motivational speaker Zig Ziglar said about confidence:

"Confidence is going after Moby Dick in a rowboat and taking tartar sauce with you"

Zig Ziglar was born just a few miles from me.


All the best
Al

[Ian Mackenzie]

Those actual mics are individually calibrated with a serial number linked to each calibration file. They are as good as my Leap LMS mic. The REW is light years better than LMS in terms of its range of capabilities. REW has some well evolved features which make it very powerful. A whole range of professionals use REW these days. My acoustician who specified my own room acoustic treatments used my REW measurements when he discovered l used REW. Greg Timber’s uses REW and swears by it.

[Ian Mackenzie]

On the methodology this paper talks about various methodologies. More specifically it refers to some things that many are not aware like the upper limit of a near field measurement and the correct far field distance to capture enclosure baffle influences.

http://audio.claub.net/software/FRD_...0to%2010Hz.pdf

The key is consistency of your measurements and label what they are.

Best of all you Save a group of measurements as a file and email them to anyone with REW for comparison or discussion.