+1 on the above.
I used to be militaristic about preventing any possibility of the level being boosted internally to over 0dB digital level. I've softened my tone on this lately.
First off, what Davey says is correct - music has a power distribution that is highest in the midrange and falls off above 2kHz and below 100Hz even if you think you have "a lot of bass" in your signal. Since boosting happens primarily in the bass region, you can allow for some leeway of up to 10dB or more at the lowest frequencies. This is because, apart from sine wave testing, you are not likely to encounter any input that really has much power in this audio region.
Secondly, I recently took part in a conversation where Nelson Pass and Siegfried Linkwitz were present. They were using a miniDSP along with some experimental amps from Nelson, to power some LXminis. They had to carefully plan the gain structure of the system, with the goal of amp and miniDSP clipping at about the same input level. As part of this they measured the miniDSP clipping behavior on a scope (they related this to me, I did not see it first hand) and found that the miniDSP clipping was not terrible, and was similar to analog clipping.
In the past I avoided digital clipping because I thought it would result in some horrible distorted noises. Now I am not as terrified of encountering digital clipping, at least with miniDSP products (although I don't personally use them at this time). This is encouraging, because reducing the internal levels in the DSP processor to ensure that 0dB is never encountered can make the maximum system output rather anemic if boost like an LT is used. This is because if the LT boost is 12dB (just as an example) then all the levels must be reduced by 12dB (under the do-not-exceed-0dB mantra) making your maximum output level 12dB lower. That's A LOT QUIETER. I sometimes found myself with the input volume (using a digital SPDIF input) maxed out, but the system just wasn't very loud. It was kind of disappointing at the time, but now I feel that I can raise levels carefully if I keep in mind how and where clipping might occur and if this would really be audible, or not.
Last edited:
I agree that occasional clipping is nothing to worry about at both frequencies extremes.
Now, as all program dependent regards, it depends on what stuff is your daily musical program. Nobody i know listen to pure tones, but it's all about personnal tastes...
Btw, i noticed trying to apply RIAA by dsp, that clipping made the clicks of vinyl much more audible and upsetting...
Now, as all program dependent regards, it depends on what stuff is your daily musical program. Nobody i know listen to pure tones, but it's all about personnal tastes...
Btw, i noticed trying to apply RIAA by dsp, that clipping made the clicks of vinyl much more audible and upsetting...
HI,
I don't understand this inputtsorry ! DO you mean about the low-pass slope choice of the boosted bass ? In relation also to the high-pass of the mid ?
Where phase behavior is critical when boosting (linkwitz transform?) ?
A normal design process might be as Juhazi indicated:
EQ drivers to flat response above and below their designed range. (So the electrical crossover when applied will result in matching acoustic crossover.)
Apply gain correction to the "hotter" drivers to yield a better relative match.
Apply crossover filters to achieve the integration between drivers.
Either (or both) of the second two steps will probably alleviate EQ boosting concerns you might have implemented in the first step.
IMHO, there is no need to be obsessed with 100% clipping prevention within the ADC/DSP/DAC structure. If you are, as Charlie mentioned, it requires a reduction in the nominal levels within the DSP.....for really no good reason since the program material is quite band-limited relative to testing noise/tonals/etc.
Phase behavior of a Linkwitz Transform is not of concern. You're equalizing the amplitude response to a lower cutoff frequency. Phase response is inherently corrected as a result. This is all minimum-phase.
Look here: http://www.linkwitzlab.com/filters.htm#9
Hope that makes sense.
Cheers,
Dave.
Last edited:
^^I didn't want to say that, just to be careful. When you have basically everything ok you can carefully use some positive eq/gain (above 0dbfs) in eq, when crossover slope's effect wins that. Latest software allows +6dB setting.
My room ambient noise level is 30-40dB. Measurement sweep must be around 90dB before I start to see distortion peaks. No hum or buzzing when I put my ear on a driver.
My room ambient noise level is 30-40dB. Measurement sweep must be around 90dB before I start to see distortion peaks. No hum or buzzing when I put my ear on a driver.
Last edited:
Ok, just to check if I understand:
Say I am working on a dipole woofer, which falls 6 db per octave from dipole peak. I am using it over Three octaves, from 350 - 44 hz.
First I apply a -6 db per octave lowshelf at 44 hz, i.e. electronically reduce the output With 6 db per octave from 44 hz and upwards. This mean that at 350 hz, I will have reduced the output With 18 db. (Correct?)
Secondly; I use the volume/gain-control for that channel, and set the overall gain = +18 db, thus the overall volume for that channel will be 0db.
Have I understood correctly?
Best regards
Gisle
Say I am working on a dipole woofer, which falls 6 db per octave from dipole peak. I am using it over Three octaves, from 350 - 44 hz.
First I apply a -6 db per octave lowshelf at 44 hz, i.e. electronically reduce the output With 6 db per octave from 44 hz and upwards. This mean that at 350 hz, I will have reduced the output With 18 db. (Correct?)
Secondly; I use the volume/gain-control for that channel, and set the overall gain = +18 db, thus the overall volume for that channel will be 0db.
Have I understood correctly?
Best regards
Gisle
That would be one way to do it.
I usually use the opposite approach though. I program the dipole shelving equalization using boosts only....or maybe a combination of a +db low-shelving filter and a -db high-shelving filter. I would not apply any level correction unless it's needed.
At first glance, you might think this programming scheme would cause much clipping in the low frequency area, but most likely it won't. As noted previously, there isn't much low-frequency information in most program material.
Also of consideration is where your system volume control resides. That might change the scheme, but for most it's upstream of the miniDSP unit so possible clip-inducing levels are already attenuated.
Dave.
In your case you would need to cascade two miniDSP shelving filters (with differing center frequencies) since a shelf from 44Hz-350Hz requires 18db. (The miniDSP units only support 16db per filter.)
So, say 9db each with Fc of 74Hz and 210Hz.
Make sense?
Dave.
^cascaded shelves are needed in this case yes. With my dipoles I use also peak correction. The 12" nude driver has total of roughly 22dB of correction, only the lowest end above 0dbfs level. My strategy is to first equalize flat almost 2 octaves past the intended xo-point, this gives easily good phase match for shallow slopes like LR2.
A trick to be mentioned is that one can use either positive or negative value for low/high shelf. This helps cascading and low end tilt (LT) a lot!
We must remember that Linkwitz transform comes with increased excursion and distortion - be careful with loud measurement sweeps! Not a problem with music, but many movies have huge rumbles in subwoofer range!
A trick to be mentioned is that one can use either positive or negative value for low/high shelf. This helps cascading and low end tilt (LT) a lot!
We must remember that Linkwitz transform comes with increased excursion and distortion - be careful with loud measurement sweeps! Not a problem with music, but many movies have huge rumbles in subwoofer range!
Regarding the plugin response difference: Do any of your speaker configs include shelving EQs? The reason I ask is, there is apparently a difference in the way shelving EQ is implemented in various miniDSP plugins.
For example, those in my nanoAVR set the shelving frequency at the midpoint between the "knees" whereas some others (2x4 I think) set the freq. nearer to one knee or the other. Sorry, I don't own a 2x4 so don't know exactly how it works. But I know it's different because of an exchange I had with a 2x4 user in this thread on the miniDSP forum.
If the shelving filters in the HD plugin are now behaving in a manner similar to the nanoAVR for example, it may be that the configs are not directly interchangeable between the original 2x4 and the new HD model.
For example, those in my nanoAVR set the shelving frequency at the midpoint between the "knees" whereas some others (2x4 I think) set the freq. nearer to one knee or the other. Sorry, I don't own a 2x4 so don't know exactly how it works. But I know it's different because of an exchange I had with a 2x4 user in this thread on the miniDSP forum.
If the shelving filters in the HD plugin are now behaving in a manner similar to the nanoAVR for example, it may be that the configs are not directly interchangeable between the original 2x4 and the new HD model.
yes, all of them. pretty annoying thinking one could input the same settings from platform to platform, then end up with a slightly different speaker.
As I've noted numerous times now on the OPLUG forum, you can't directly translate settings from one miniDSP unit to another (and especially to different brand units) without comparison testing of the results. The new LXmini+2 configuration was a good example of this.
There are a couple of threads regarding this on the miniDSP forum also. Especially regarding low-frequency errors on the 2x8 platforms.
I will recheck the two versions for 2x4 and 2x4HD.
Cheers,
Dave.
ok, thats why you made one and its the official xml on the owner support page?
not surprised if the 4x10 file got the same issue.
2x4: 12882hz = 4db 15849hz = 18db 5012hz = 19db
2x4HD: = 3db = 17db =20db
Last edited:
Has anyone done a comparison between the miniDSP 2x4 and the new miniDSP 2x4 HD? I just did a comparison between the two and the miniDSP 2x4 HD sounded better than the old miniDSP 2x4 in subtle but noticeable ways in clarity, which is unexpected as I did not expect to hear a difference. SPL levels were matched and the FR measurements were exactly the same. miniDSP 2x4 HD did use a digital input.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.