Calculating Amp Input Sensitivity

[mikekars]

If the amp sensitivity for a given amp / load configuration is missing from the manufacturer's specs can the power and amp gain ratings be used to calculate the input sensitivity?

 

For example, the PLX 3402 amp specs are:

 

• 3400w in bridged mode into a 4 Ohm load. T
  
• amp gain 32dB (40x) in stereo. Bridge mode add 6dB = 38dB (79x).
  

 

From those numbers I could calculate the maximum output voltage as

 

Vmax = sqrt ( 3400w * 4 Ohm ) = 117v.

 

Then I can derive the input sensitivity in this configuration by dividing by amps linear gain

 

Vin = Vmax / gain = 117v / 79x = 1.48v

 

Is this a valid way to calculate the input sensitivity?

 

If so it seems the input sensitivities in bridge mode are generally a dB or so less than in stereo mode.

[agedhorse]

Yes, that's close enough, though what are you driving at 4 ohms bridged with that amp?

[RoadRanger]

Yes' date=' that's close enough, though what are you driving at 4 ohms bridged with that amp?[/quote']Fryolator? LOL

 

[mikekars]

I am putting some more subs in the back of my civic hatchback because I can't seem to hear the old ones any more.

 

Actually it is more educational for me as I am trying to deepen my understanding of manufacturer specs and how they correlate with what I measure with a VOM. The whole measurement process has been very confusing, especially when trying to measure amplifier output volts. The open circuit measurements seem to be significantly different from what I get if I hook up some par cans as dummy loads of approximately 4 ohms. I am trying to get good measurement data that ties well to published specs so I can set limiters accurately.

 

The original question though relates to taking JBL published SRX speaker tunings and properly applying the recommended crossover gains in a DR260. My assumption is that JBL's suggested crossover gains are valid only if the amps being driven for the different band passes have identical input sensitivity and amp gain.

 

In my case I have amps with different input sensitivities as well as different gains on each pass band. It seems to me that if I want to replicate the factory tuning I need to take both of those into account and make relative adjustments to the factory crossover gains. The exercise for me has been to understand this well enough to properly calculate the correct crossover gains.

 

It is mostly an academic exercise but I would like to do an A/B listen test what I am using now to what the "factory" tuning would be using my amps.

[agedhorse]

I never use the recommended tunings as this is something I develop as a day job sort of thing, but what I generally do is set the band gains such that with the amp sensivities all the way up the sub band will start at +3dB, adjust level to 85dB spl average, bring up the mid band to where the average level is 85dB spl, then do the same with the high band. Now adjust band delays slightly to see what happens at the crossover points (noting that patterns in response will change perodically with 360 degree phase rotations that come with the changes in delay. Next adjust the output band eq using an RTA AND either understanding and avoiding frequency domain corrections for time domain issues. A dual fft approach helps identify this too. At this point I may go back and slightly adjust the band gains back to 85dB average. Now, listen. If the high frequency noise is objectionable I may decrease the hf amp sensitivity and increase the hf xover output starting at 6dB. Now I tend to power around the RMS rating because I generally find this to be the point of diminishing returns with today's drivers and the clip limiters in most amps guarantee no tracking errors or clipping since the detectors actually monitor the amps feedback loop rather than the external limiter approximating this. Otherwise, if you want to run drivers close to their limits there is a lot of data that we generate internally (including destructive data) to do so that most folks just can't do.

[mikekars]

That is a lot to digest but I think I have done something similar in the "shop" (basement). The direct measurement approach is probably superior to the theoretical approach in most cases. Sadly the basement is a poor representation of most venues and since I am not that fast with those measurements I seldom seem to have time at an actual show. So I am trying to get things as close as possible ahead of time.

 

With respect to setting relative crossover gains, I believe I was mistaken thinking the amp sensitivity is a factor. Assuming the amps are run WFO, it is only the fixed voltage gain of each amp that needs to be considered. That is, if the "factory" tuning calls for the low frequency band to be 0dB and the mid frequency band to be -3dB, then an offset would only be needed if the amps for those two bands had different fixed voltage gain. Some of the old Peavey white papers spell this out pretty well.

 

There are, of course, other reasons to make adjustments to the crossover gains. Such as the one you mentioned when the high frequency driver is hissing when the high frequency amp is run wide open. Then you have to turn down the amp input attenuator knob and increase the high frequency crossover gain by the same amount to compensate.

 

Another reason, at least I think it is a reason, is if the output of the system processor (DR260) is to high of a voltage for the amp inputs. That is, if you have your mixer gains set the way you want but the DR260 puts out a voltage that is frequently over driving the amps. If you don't want to turn down the input attenuator on the amp, then you can do it in the DR260 by reducing all the crossover gains by the same number of dB.

 

Is that correct thinking?

[agedhorse]

All tunings are done in an anechoic environment. Outside without a hard floor (reflection plane, grass works well) is a good approximation. Laying a speaker on it's back also works well, dig a pit so the face of the speaker is even with the ground is even better (but not often practical).

 

If the output of the system processor is too high, it will be very loud. You want ~10dB between the point that the amp limits and the fdrive electronics clip. Good practice which allows the drive electronics to work undistorted into the limiter. The amount of limiter gain reduction gets subtracted from available drive headroom.

[dboomer]

My approach to creating DSP settings for speaker systems is all about crossover summation. I mean there are some generalities you start out with in picking crossover points depending on the drivers you are using and the outcome you expect. But once you are there getting the minimum amount of cancellation at the summation is my prime focus. I would play with filter types (butterworth, bessel, LR,etc), slopes, symmetrical vs non symmetrical, underlap and overlap and small delays to see if i can lessen the hole caused by the crossover summation. I might end up with something reasonably odd by the face of it but if the summation was optimized then you can go back and get more precise levels and EQ.

 

Just my 2 cents

[agedhorse]

I almost always use 4th order LR alignments for filters because they are by design electrically constant power summing. The smoothness of summing can sometimes be improved by under or overlapping the electrical frequency, but with well chosen drivers, horns and cabinet design it's often the best choice. Certainly it's in the ballpark and if you don't have the heavy duty analysis tools, is the best choice.

[dboomer]

While that's all true, it really the acoustic summing that is important. The electrical can look all kind of crazy if it produces the desired acoustical response.

[agedhorse]

If the electrical side has to be crazy to get smooth response, that's telling me that perhaps there are issues with the transducer/system design that might produce better results if improved.

[dboomer]

Sometimes yes ... but remember there are sonic consequences out in the air after the sound has passed through the speakers. And that's the sound that you are worried about getting as good as possible.

[Craig Vecchione]

Don't make me stop this car and come back there!!!!

[agedhorse]

Are we there yet, are we there yet?

[mikekars]

What kills me is that when measuring the system processor output with a dual fft tool like Systune it is relatively easy to get an electrical trace that looks like the smooth flat response in the textbooks. But then when you hook up a reference mic and measure the acoustic response it is anything but flat and very difficult to interpret.

 

Using delay to match the phase curves at the crossover point as well as possible is not too hard and generally seems worthwhile. But the magnitude response often has some wild fluctuations that have to be taken with a grain of salt. Outdoors is usually not too bad but indoors there are so many peaks and dips that all change depending on the mic position that it is next to impossible to smooth out the response curve.

 

If I made a bunch of adjustments and ended up with a crazy looking EQ preset like Don says, I doubt I would have the confidence to stick with it. It would have to sound noticeably better or I would just accept the limitations of my skill and flatten everything back to square one and move on until next time.

[dboomer]

Just remember that what your are sending to the speaker isn't the goal. What you are trying to achieve is what comes out of the speaker.

[agedhorse]

Correct Don, and it's generally better to design so that you avoid as much correction as possible. Correction carries with it the possibility (or probability) of adding other artifacts that in some cases can be worse than what you are trying to correct.

[dboomer]

it's generally better to design so that you avoid as much correction as possible.

Well you always start from there but since you are dealing with mechanical devices you can only go so far until you spend more money for the drivers. So at a price point you hit a limit. Again depending on the price point DSP offers a way to more forward that generally yields better sonic results at a much lower price. Y