View Single Post
      09-16-2012, 09:28 PM   #45
mob17
Brigadier General
 
mob17's Avatar
 
Drives: E92 335D
Join Date: Mar 2012
Location: East Middle Land, UK

Posts: 3,730
iTrader: (4)

Bet you're wishing you never got the ms-8 now Awbimmer!

Anyways this is a post by Andy Wehmeyer, the product manager of the MS-8, answering a question similar to this thread. Apologies for the length but its worth reading. He explains the whole input sensitivity/calibration nicely with some background info too:

Original Question:
Maybe I've been reading too much, but I've confused myself. I've set my gains to about 25%, HU volume at 35/40. MS-8 volume at -40. Now I've run the acoustic calibration, and the listening volume is too low. I want maximum clean gain for each set of speakers, and a high sound level from the system.

How do I increase my gains now to get the output voltage I'm looking for? I'm planning on using a multimeter per JL's instructions on how to verify gains, but I need to know I'm hitting the inputs on the amps with the high end of the limits in order to get an accurate gain output reading. What should the MS-8 level hitting my amp input be? -40? -20?

Thanks for the help, I've read through most of this thread and haven't seen any detail about verifying amp gain with accuracy, just some quotes about "turn it up from there". I'm sure there is a fairly simple answer.


Andy's answer:

Too much has been made over the years about gain and level setting. Some years ago, systems were made up of a head unit, an EQ, separate electronic crossovers and then amps. Setting levels in these kinds of systems was important, but we no longer use so many analog components in a lengthy signal chain. In a system like that, each component should be driven with the highest level possible before the outputs OR the inputs are clipped. and the signal sent over the wire should be maximized to maximize the signal to noise ratio. The reason for all of this was noise. Too much INPUT sensitivity and too little output voltage would result in system hiss and would boost the level of engine noise, which would be amplified by every component in the chain.

We no longer have to be so concerned with all of that because systems now are often a head unit, a signal processor of some kind and some amplifiers. Additionally, components now often have differential inputs, which are designed to eliminate the possibility of engine noise.

So, with MS-8, the analog signal from the source is converted into a digital signal in the unit before anything else happens and we've designed the unit to make setting this level as easy as it can be. The RCAs are fixed input sensitivity and are designed to work with all aftermarket head units without any user intervention. Just plug them in and go. The unit, without EQ, is designed to provide unity gain up to 2.8V, which is plenty. That means if you put 1V in, you get 1V out. If you put 2V in, you get 2V out. If you put 9V in, you'll get 2.8V out but it'll be seriously distorted. When you turn the output of your radio down so that you send 2.8V, you'll get 2.8V that isn't clipped.

Now, one of the reasons that all of this input sensitivity setting procedure was developed and that many of us suggest things like 10dB of gain "overlap" is because a system that can't clip doesn't sound very loud, especially with recordings that have a high crest factor and little dynamic range compression added to the final mixdown. You want your system to clip, but you want to balance that with the amount of noise (hiss) you allow the system to produce. You don't want much of this clipping to be digital distortion, because that sounds bad.

If you're using a 4V head unit with MS-8, that provides 3dB of gain overlap. You won't hear much distortion unless you listen to sine waves recorded at 0dB with the volume control turned all the way up. With normal music and the head unit's volume control all the way up, only the transients that are recorded at 0dB will be distorted, but only by 3dB. You won't hear that and if you do, simply backing the volume control of the head unit off by a couple of notches will take care of it.

Since the MS-8 is designed to provide unity gain, it's sufficient to set the input sensitivity of the amplifiers to the same setting as the output of your head unit for 0dB of gain overlap between MS-8 and your amps). This will ensure no input clipping of your amplifiers and will ensure the least noise possible. Double the input sensitivity will result in an additional 6dB of gain. So, if your head unit is a 4V unit, you can set the input sensitivity of the amps to about 1V, which will give you a total of about 9dB of "overlap". Precision isn't required.

Because MS-8 also has a volume control, you'll have to manage how you use it. You don't have to use the MS-8 control if you'd prefer to use the one on your radio. If you choose to do this, then you'll need to set the MS-8 volume control at some level that allows the right amount of "input sensitivity" but also allows enough digital headroom for the EQ inside of the MS-8 to operate without running out of bits (that causes digital distortion). I suggest setting the MS-8 control at -6dB to -9dB during listening, unless you boost the bass in MS-8. If you boost, then you should set it lower by about the same amount as you boost and then use your head unit's volume control.

If this in't enough "gain" for you, then turn the amplifiers up after calibration by the same amount to maintain MS-8's "tune".


You can determine the point at which your radio outputs about 2.8V by putting the setup disc in your aftermarket radio and running input setup. The point at which you get OK OK OK is the point at which your radio outputs about 2.8V. You can continue to run input setup and MS-8 will Un-EQ and remove any channel delay. If you're using an aftermarlet radio, none of that should be necessary so all of the EQ filters will be set to unity. That means what comes in goes out. If you choose "skip input setup", all of the filters will be set to unity.

When you run acoustic calibration, MS-8 will set all of the output levels according to the acoustic response in the car. This takes into account the sensitivity of the drivers and their frequency responses. Some outputs will be increased in level and some may be decreased. So long as there's no hiss, the output level and the input sensitivity control of your amps don't matter. Let MS-8 do what it does.

When you run acoustic calibration, the output of the system can't clip the mic, or things will be ugly. If the mic is clipped during the first set of sweeps, the unit won't be able to locate the initial peak in the response and it won't set time alignment correctly. You'll know this is the case, because there won't be a center image. This is all the volume control setting for acoustic calibration is doing--making sure the system doesn't clip the mics. To ensure this, turn MS-8's volume control DOWN. If you use MS-8's internal amps, -20dB is the highest setting you'll need. If you're using additional amps, you'll have to turn it down MORE. -40, maybe. This setting doesn't matter too much in terms of the final outcome, but the level does change the way MS-8 will boost and cut, because everything has to fit in a "window". The window is big, though.

This is no different than tuning with a regular EQ. If you boost all the bands to fix holes, you'll have a super loud car and you'll probably have lots of clipping and noise. You'll also probably boost in the interest of filling holes cause by acoustic cancellation, which may overdrive your speakers and cause additional distortion. MS-8 tries NOT to do this, but it can't know for sure. If your response has big holes because you're using an 8" midbass and a 1" tweeter and the wrong crossover point, you may hear distortion because MS-8 is boosting as much as it can to fill a hole that can't be filled. Fix the speaker system. MS-8 can't make gold out of crap.

If, when you're tuning with a manual EQ, you cut all the bands to remove peaks, then you'll reduce the output voltage of the EQ at frequencies where you've cut and you'll probably want to readjust the input sensitivity of all of the amps to increase the level of the whole system to make up for it. If you think about it, this is how we all tune systems anyway. So long as you don't introduce a bunch of system hiss, this is fine. Cutting a lot and boosting a little is the best method. This is how MS-8's algorithm is designed to work. It works well, but GIGO still applies.

Once acoustic calibration is complete, if you want to use your head unit's volume control, set MS-8's volume control to -6 and go crazy. If you boost a bunch with MS-8's EQ or sub level control, you may have to turn MS-8's volume control down to leave more digital headroom for the boost. This isn't a defect. This is how digital EQ works. Some digital EQs normalize the response. That would result in everything else being reduced in level as you boost the bass (or the EQ). We didn't do this because it adds complexity to the system and the result of boosting with these systems when you reach the level at which there are no more bits available is counterintuitive.

If you want to use MS-8's control, do this: Put in a music disc, turn MS-8's volume control DOWN to something like -40 or -50 (so you know you're not clipping the outputs of MS-8 or your amps) and turn up the head unit's volume control. When you hear distortion, you'll know you're either clipping the output of the radio or the inputs of MS-8. Turn the head unit's volume control down until you don't hear any more distortion. This is the maximum usable output of the head unit for MUSIC and for that particular disc. If you listen to another disc with more or less dynamic range compression in the recording, you may find a different volume control setting produces audible distortion. This is related to our ability to hear distortion on transients, the length of the transients and how often those transients are repeated. The idea here is that you have control over the amount of clipping you allow. It is what it is and balancing the two volume controls will help you get the most from your system. I think the vast majority of systems will be fine.

The bass can't clip the mics during sweeps 2-4 either, because the EQ doesn't know what to do. Be sure that the level of the bass is LOW during the sweeps. If you have twelve 15" woofers and you want to wake the neighbors, turn the amp gain WAY down during calibration. The bass should be heard during the sweeps, but you shouldn't feel it. Then, after calibration, turn the gain of the amp up until you're happy. You can't wake the neighbors and have 40dB more bass than mids and highs and maintain the illusion that the bass is in the front of the car, which is what MS-8 is designed to do. You'll have to adjust the gain of the amp to get what you want, because MS-8 tries to eliminate what you want. Let MS-8 do what it does, and then make your adjustment afterward.

Finally, if the system doesn't play loudly enough, adjust the amplifier input sensitivity control to increase the overall system level. There's no need to use a meter or a scope. Just adjust them all up or down by the same amount to maintain MS-8's EQ and relative level settings. The VMM method won't work because there's a bunch of EQ applied to the signal, which increases and decreases the output voltage at different frequencies. BTW, the VMM method is only slightly more accurate than just setting by ear.

Don't be confused by all of the "this has to be precisely set to get the most from your system" garbage. If there's no noise (or the noise is low enough that you're not annoyed by it) and your system plays loudly enough, then things are set appropriately.
__________________
2008 E92 335D SE
2010 E90 320D SE
1994 E36 318I SE
1996 E36 316I SE
mob17 is offline   United Kingdom
0
Reply With Quote