SH-50 limiters.

[Helge A. Bentsen]

I run a pair of Danley SH-50s with a QSC PL 9.0(voltage gain 32dB), processed by a Lake LM-26.

Sounds good, no issues with broken drivers etc. so far except one crossover capacitor that needed replacement.

I've set my MaxRMS limiters for 1000w@4ohm(6.23dBu, 63,18V) and MaxPeak at 2000w(9.23dBu, 89,24V), and all is good. Up on browsing through my collection of old datasheets I came across the old spec sheet for a SH-50 stating that it's a 6 ohm box, not 4 as the new datasheet says.

I remember Ivan said that the change in ohm had more to do with end users having difficulties setting their limiters for 6 ohm operation, than any actually design changes. So I tried redoing my limiters for 6 ohm operation and I calculated a new MaxRMS at 8dBu( 77.45v) and a new MaxPeak at 11dBu (109.42v). Looks plausible to me, but I thought I'd run these numbers by you guys to check my calculations.

Is my numbers way off?
Do you guys treat the SH-50 as a 4 ohm box, 6 ohm box, or doesn't it matter?
What’s your recommended limiter setting for it?

 

[Art Welter]

Re: SH-50 limiters.

I run a pair of Danley SH-50s with a QSC PL 9.0(voltage gain 32dB), processed by a Lake LM-26.

Sounds good, no issues with broken drivers etc. so far except one crossover capacitor that needed replacement.

I've set my MaxRMS limiters for 1000w@4ohm(6.23dBu, 63,18V) and MaxPeak at 2000w(9.23dBu, 89,24V), and all is good. Up on browsing through my collection of old datasheets I came across the old spec sheet for a SH-50 stating that it's a 6 ohm box, not 4 as the new datasheet says.

I remember Ivan said that the change in ohm had more to do with end users having difficulties setting their limiters for 6 ohm operation, than any actually design changes. So I tried redoing my limiters for 6 ohm operation and I calculated a new MaxRMS at 8dBu( 77.45v) and a new MaxPeak at 11dBu (109.42v). Looks plausible to me, but I thought I'd run these numbers by you guys to check my calculations.

Is my numbers way off?
Do you guys treat the SH-50 as a 4 ohm box, 6 ohm box, or doesn't it matter?
What’s your recommended limiter setting for it?

The IEC standard (IEC60268-3) allows any impedance above the rated value, but limits the impedance below. It does not allow the rated impedance to fall below the 80 % of the nominal value at any frequency.
.
I don’t know if the SH-50 has been redesigned, the old “6 Ohm” spec sheet claimed a minimum impedance of 3.5 ohms at 16 K, obviously well below the IEC60268-3 standard.

According to the new spec sheet, though the SH-50 averages around 6 ohms, it dips below 4 ohms at 90 Hz, but the 16K impedance looks to be around 5 ohms now.

Setting an RMS or peak limiter to a specific voltage only limits the power relative to the specific impedance at a particular frequency, obviously the amp will produce a lot more power into the SH-50 at 90 Hz than 500-600 HZ, and the woofers will take far more power than the tweeter (HF driver).

Put 66 to 77 volts of sine wave at 16K into the SH-50 and you won’t have a tweeter after a few seconds, but that much average voltage in with “normal” music won’t be a problem.

Just watch out for Hi-Hat solos, your limiter won’t protect the tweeter from them ;^).

As far as your questions, your numbers are not off, I’d call the SH-50 a 4 ohm box, although nominal impedance and dispersion is just a name- the SH-50 is also a 50 x50 degree nominal dispersion box in the upper range, but not in the woofer range.

As far as a recommended limiter setting, that would depend on who (and what) you are trying to protect the box from.

 

[Bennett Prescott]

Re: SH-50 limiters.

When setting long term power limiters, my thought is always: What's 3dB? To you and me, it is a very small difference. To the drivers, it is all the difference in the world. Therefore, I would tend to take the worse case impedance (i.e. lowest impedance at the frequency you will be hitting the box hardest, that HF impedance minimum doesn't bother me but that one at 90Hz will surely be sucking up some power), calculate from there, and then back off 3dB. Keep the box out of power compression (which can eat up that 3dB you set the limiters higher... quick), rest a little easier during shows.

 

[Helge A. Bentsen]

Re: SH-50 limiters.

Thanks.

I'm trying to protect it from driver and component failure, but not direct abuse and 36 hour long EDM-partys.

When setting long term power limiters, my thought is always: What's 3dB? To you and me, it is a very small difference. To the drivers, it is all the difference in the world. Therefore, I would tend to take the worse case impedance (i.e. lowest impedance at the frequency you will be hitting the box hardest, that HF impedance minimum doesn't bother me but that one at 90Hz will surely be sucking up some power), calculate from there, and then back off 3dB. Keep the box out of power compression (which can eat up that 3dB you set the limiters higher... quick), rest a little easier during shows.

According to your method, I should substract 3 dB for safety reasons?

If we say that the SH-50 is 3.8 ohms@90hz, that would be 61.64v, 6.01 dBu calculated, but I should enter 3.01/6.01 dBu in my rms/peak-limiters?.

 

[Bennett Prescott]

Re: SH-50 limiters.

Peak limiting is a whole different deal. I probably just wouldn't worry about peak, but I would expect much more than 3dB of headroom between conservative RMS limiting and what the loudspeaker can do with peaks.

 

[Langston Holland]

Re: SH-50 limiters.

Hi Helge:

That is a wonderful rig you have - and the Lake limiters are the best I've ever worked with. Your calculations are perfect, but I do have a few thoughts:

1. I am confident that Danley intended you to calculate the voltage thresholds using the nominal 4Ω impedance, just as you have.

2. Danley will soon clear up this confusion by specifying max. long-term and peak voltages in addition to watts, right Ivan?

3. The Lake processors use look-ahead peak limiters and nothing gets past the peak threshold. I've fed DLP inputs at 15dBu above the threshold with 6.5ms sine bursts up to 45kHz to prove it to myself.

4. Because the peak limiter is so effective, you are free to use the remainder of the limiter functions to optimize sonics at higher SPL's. In that pursuit it is best to use the automatic time constant function for the attack and release values and a -6dB "MaxRMS Corner".

Therefore, given your 32dB amp gain, the 6dBu RMS threshold and 9dBu peak threshold follows the mfg recommendations perfectly and I'm confident will protect the loudspeaker from damage, though you may find those are not the best thresholds to use sonically. Also keep in mind that even the most compressed music has a long-term "RMS" level 15dB or more below the peaks.

In your case, this is what I'd do:

Drop your MaxRMS threshold to 3dBu and MaxPeak threshold to 6dBu. Play some well recorded very dynamic music with good vocals that you really like at the point where either your console's output or processor's input is just about to clip. Does it still sound good? It should with that processor if the loudspeakers are up to the task, which I think they will be. If you felt the loudspeaker sound was getting ugly and you weren't able to go to the console's full output, back both thresholds down 1dBu, rinse, repeat. If you felt everything sounded fine and want more output, increase both thresholds 1dBu, rinse, repeat, but don't go above the mfg recommendation unless you're a nut case like me.

===

My pet project lately has been limiting. I've been measuring loudspeaker passbands with short-term sine bursts while observing distortion plots, scope displays, and listening to find the mechanical and listenable peak limits of the device (two very different levels in many cases). After a couple of dozen hours at this I've made the following observations:

1. Limiter settings on processors are even more deceptive and less transferable between brands/models than filters. This is yet another thing that must be measured to be verified.

2. The Lake processor means what is says when you set the peak threshold. Example: set it to +10dBu and feed it with a sine burst at +15dBu and put an RMS reading meter (almost all of them) on the output and you'll read +7dBu. This is because a sine wave has a 3dB crest factor and the Lake limiter really is limiting at the +10dBu peaks. I'm aware of no other processor at this point that does this, i.e., tells the truth.

3. Most mfg's of high quality professional loudspeakers correctly specify the long-term power capability of their passbands. They are all over the map on specifying the correct peak capability. Most just add 6dB to the long-term rating and call it a day, but this is a lie (sonically) except in the case of the better designed low and sub passbands. Mids and Highs? With good sound? Forget it.

4. The guys that design the limiters in loudspeaker processors apparently have never seen a loudspeaker (excluding the Lake). You put a pretty number in your limiter's dBu field and you get that at the output plus another 3dB due to the crest factor - with continuous sine waves - but much, much higher peaks with real world short-term signals even with the ratio set infinity:1 in many cases. You think you're protecting your loudspeakers from over excursion when those limiter LED's come on? Not if the BE keeps pushing the faders up. Are you getting uncomfortable yet?

 

[Bennett Prescott]

Re: SH-50 limiters.

Limiter settings on processors are even more deceptive and less transferable between brands/models than filters. This is yet another thing that must be measured to be verified.

And is a real pain in the ass to measure, I don't have a good way to do it. I usually set threshold for long term power by math, attack and release roughly by ear, and peak entirely by ear. This is something that might come up at Rich Frembes' AES panel this fall, but I haven't put anything like the time in on it that you have apparently!

The guys that design the limiters in loudspeaker processors apparently have never seen a loudspeaker (excluding the Lake). You put a pretty number in your limiter's dBu field and you get that at the output plus another 3dB due to the crest factor - with continuous sine waves - but much, much higher peaks with real world short-term signals even with the ratio set infinity:1 in many cases. You think you're protecting your loudspeakers from over excursion when those limiter LED's come on? Not if the BE keeps pushing the faders up. Are you getting uncomfortable yet?

Actually, you're not protecting your loudspeakers from over excursion properly with any peak limiter I am currently aware of. I thought as you do, and then was summarily corrected in many thorough conversations with Phil Graham and then by Nick Hickman when I still had it wrong! I'm not such a math person, it turns out.

http://www.soundforums.net/live/threads/1086-Help-me-understand-system-limiting-setting-it/page2

 

[Langston Holland]

Re: SH-50 limiters.

And is a real pain in the ass to measure,...

I was going to mention that and what I went through to come up with something that made sense, but it would have been quite the rabbit trail. Maybe on another post, another day. My preference is to use a sine burst length </= the shortest event that can happen in use. Sine bursts because they are so easy to see deformation in with a scope or to plot harmonics (distortion) of.

Actually, you're not protecting your loudspeakers from over excursion properly with any peak limiter I am currently aware of. I thought as you do, and then was summarily corrected in many thorough conversations with Phil Graham and then by Nick Hickman when I still had it wrong! I'm not such a math person, it turns out.

http://www.soundforums.net/live/threads/1086-Help-me-understand-system-limiting-setting-it/page2

This is interesting, I'd love to see the math on why a driver doesn't behave like the display on a scope - i.e. changing frequency has no affect on amplitude. We certainly need increasing output with decreasing frequency to perceive equal volume due to how our hearing is designed, thus a useful loudspeaker driver would have to be designed to have progressively greater excursion at progressively lower frequencies within its passband because of its constant radiating area. To the extent the driver/horn/box/room system isn't like this is the extent that we try to force it into being with EQ. How they pulled this off I don't know, whether by design or one of the laws of physics that happened to land in our favor - hope the math'll help...

It's also interesting in that it would mean the effective limit of a driver at the upper end of its passband would be an RMS only issue - convenient with compression drivers considering HF absorption with distance and lower humidity.

Nevertheless, within the passband, limiting voltage will obviously limit the driver's excursion. Going outside the bounds of the recommended passbands in frequency or slope certainly require new measurements and will result in lower or higher peak voltage thresholds. That is one of the cool things about the high slope linear phase filters in the Lake - the potential of additional good sounding SPL even at the same crossover frequency, though higher slopes usually allow one to move the crossover frequency upward a bit as well without damaging polar response.

===

Edit: the first of many (probably) rudely copied items, this from Linkwitz discussing the negatives of a 2nd order crossover to a midrange driver: "Driver cone excursion increases at 12 dB/oct with decreasing frequency for constant SPL, yet the crossover rolls off at only 12 dB/oct which leaves the excursion constant."

This confirm's Nick's contention that one octave lower results in 4 times the driver excursion.

 

[Helge A. Bentsen]

Re: SH-50 limiters.

Drop your MaxRMS threshold to 3dBu and MaxPeak threshold to 6dBu. Play some well recorded very dynamic music with good vocals that you really like at the point where either your console's output or processor's input is just about to clip. Does it still sound good? It should with that processor if the loudspeakers are up to the task, which I think they will be. If you felt the loudspeaker sound was getting ugly and you weren't able to go to the console's full output, back both thresholds down 1dBu, rinse, repeat. If you felt everything sounded fine and want more output, increase both thresholds 1dBu, rinse, repeat, but don't go above the mfg recommendation unless you're a nut case like me.

Thanks, I'll give it a try next time I have some time to play in the warehouse. I agree in your praise of the LM-26 limiters, I find them very transparent and smooth when they kick in, but I've never measured them to verify their performance.

The sonic differences between limiters is something that drives me nuts. I think the worst limiters I've heard are those found in EAW mx8750 and Klark Teknik DN9848, but units like XTA dp226 and EAW UX8800 have a more pleasing performance. I often use my LM-26 for PA and subwoofer aligment and feeds delays, fills etc. from a UX8800 by AES/EBU, controlling everyting wirelessly on a tablet. This gives me a lot of flexibility and sounds pretty good.

Setting limiters actually became a lot easier as soon as I started thinking in terms of voltage instead in terms of power. +1 to max voltage specifications

 

[Bennett Prescott]

Re: SH-50 limiters.

Drop your MaxRMS threshold to 3dBu and MaxPeak threshold to 6dBu. Play some well recorded very dynamic music with good vocals that you really like at the point where either your console's output or processor's input is just about to clip. Does it still sound good? It should with that processor if the loudspeakers are up to the task, which I think they will be. If you felt the loudspeaker sound was getting ugly and you weren't able to go to the console's full output, back both thresholds down 1dBu, rinse, repeat. If you felt everything sounded fine and want more output, increase both thresholds 1dBu, rinse, repeat, but don't go above the mfg recommendation unless you're a nut case like me.

Langston,

If the Lake's peak limiter is peak limiting, won't setting the peak limiter 3dB above the RMS limiter simple cause them both to hit at the same time with sine waves, and with anything less dense than that (i.e. everything) the RMS limiter will never kick in because the peak limiter will catch everything?

 

[Phil Graham]

Re: Some math for you on voltage vs. excursion

This is interesting, I'd love to see the math on why a driver doesn't behave like the display on a scope

Langston,

The following is a gross oversimplification, as it considers only a sinusoidally driven mass. A real system also has to consider the suspension compliance and resistive losses, AND the equivalent behaviors of the box air (i.e. mass, compliance, and loss), the port, etc.

That said, the following should be enough to get you thinking. I make no attempt at exposition, as I'm pretty sure you know how to do the math. Hopefully there are no mistakes, as I wrote it out quickly. This is for the 1D case:

1. Voltage to amplifier through the voice coil impedance Z (which is Rdc + vc inductance)
2. At low frequencies Rdc dominates, so V=IZ is (mostly) V=I(Rdc) and V and I are therefore mostly in phase
3. F(t)=Bl x I(t) (x ==cross product)
4. F=ma - where mass would be the inertial mass of the system (cone mass)
5. F(x,t)=m(d2x/dt2) - acceleration is the second derivative of displacement wrt to time
6. V(t)=A*sin(wt + p) where wt is angular frequency and p is phase
7. I(t) is then nearly (A*sin(wt + p))/Z
8. F(t)= Bl x [(A*sin(wt + p))/Z]
9. Ignoring all the constants, and the starting phase: sin(wt) α (d2x/dt2) - where α means proportional to
10. Integrate both sides twice with respect to t to find the displacement: x=∫(∫sin(wt) dt)dt
11. The first integral is: -1/w*cos(wt) + C
12. Second integral, ignoring the constant, is: (-1/w)∫cos(wt) = (-1/w)(1/w)sin(wt) + C
13. In the end, again ignoring the constant of integration: x(t) α -1/(w^2)sin(wt) - where x(t) is displacement as a function of time.

So, in the simplest case of an undamped mass being driven by a sinusoidally varying voltage source, the displacement of the mass has the opposite sign as the original driving voltage. Also the amplitude is inversely proportional to the square of the angular frequency. That means the lower the frequency, the higher the amplitude. For the spacing of an octave (i.e. freq. of 1 vs freq. of 2) it is straightforward to see where the 12dB/octave comes from:
20log(2^2/1^2) == 20log(4/1)=12.04dB

As an example of how this gets more complicated: For periodic driving of an under-damped simple harmonic system, which is closer to a loudspeaker in free air, the form of solution to the differential equation is:


Where the first term is the transient solution that depends on the initial conditions of system motion, and the second term is dependent on the driving force, and not the initial system motion. This solution is courtesy of the excellent GSU hyperphysics page.

The amplifier voltage is not a simple expression of the driver displacement.

 

[Phil Graham]

Re: SH-50 limiters.

Langston,

If the Lake's peak limiter is peak limiting, won't setting the peak limiter 3dB above the RMS limiter simple cause them both to hit at the same time with sine waves, and with anything less dense than that (i.e. everything) the RMS limiter will never kick in because the peak limiter will catch everything?

+1 to Bennett's speculation, as I read this the same way. Plus, I generally find a good starting point for the RMS limiting of a driver is 6dB below the AES rated power, using the Z_min impedance to calculate the voltage threshold equivalent.

 

[Langston Holland]

Re: SH-50 limiters.

If the Lake's peak limiter is peak limiting, won't setting the peak limiter 3dB above the RMS limiter simple cause them both to hit at the same time with sine waves, and with anything less dense than that (i.e. everything) the RMS limiter will never kick in because the peak limiter will catch everything?

Absolutely correct - I didn't notice that correlation until you mentioned it. I never intended for the Lake's RMS limiter to do anything - only the peak limiter has been important to me given the very forgiving crest factor in real music as you noted. In the case of the Lake, I've found that moving the RMS limiter more than 3dB below the peak limit threshold softens the music more than I like when used with the "-6dB MaxRMS Corner" setting - which I do like. The subjective result of my recommendation to Helge is the best sounding limiting I've ever heard - significantly better than I've been able to achieve thus far with the Xilica XD, UX8800 or Ashly processors.

I also haven't had the motivation to measure the Lake's RMS limiting function because it didn't make sense to me given the nature of music. Nick has me reconsidering this given his comments and the fact that he's obviously very sharp.

I love being ignorant, every day is full of surprises. Off to brush up on some calculus...

 

[Bennett Prescott]

Re: SH-50 limiters.

I never intended for the Lake's RMS limiter to do anything - only the peak limiter has been important to me given the very forgiving crest factor in real music as you noted.

Heh... Langston, I have the same view but use it for exactly the opposite justification. I care mostly about RMS limiters because the crest factor in music lets you still get excellent fidelity while protecting your drivers with only a few dB of power limiting.