I’ve been playing with the House Curve app on my iPad as a precursor to getting REW. I have an annoying mode at low frequencies in my room. The convolution filter that I’ve loaded into Roon is addressing that reasonably well, but to avoid the clipping metre coming on I’ve had to add 10dB of headroom. If I want to listen relatively loud (say 72 dB) I have to have the volume on my 222 at 80 - is that OK for the amp over a long period? Sorry if that’s a daft question, but I recall posts where people seem to have there volume control much lower.
Yes, most folks have their volume control between 8.00am and 11.00am. This is a Naim signature “feature”! Having the volume at 80% (4.00pm?) shouldn’t do any harm provided that the volume is not deafening. I suspect you are getting a very low input to the 222 (or the 222 is sending a very low input to the 300) if you aren’t going deaf at 80%. I’d check all the settings throughout the signal path to make sure something is set way too low.
As somebody who also has a room mode (mine is at 50Hz) I too use convolution as a fix. It works really well and adjusts the bass individually for each speaker.
I don’t have to use much headroom though - 3dB covers the Convo nicely. I also use some parametric (my hearing) so I have to nudge the headroom to -5dB for the most clippiest test tracks I could find.
I did use REW and a calibrated mic though.
I don’t really know how the headroom thing works, only that I have to turn up the volume to compensate. It’s just seeing the volume counter at 80 that gets me thinking is this a good thing to be doing. I’ll try to address the room mode via some room treatment in the future - but in the meantime……
Even with -10dB headroom I wouldn’t expect that amount of gain being needed on the amp. I’m only in the high 30s of the 272 range.
You’ve not enabled volume levelling or anything else?
With no DSP on at all I typically have the volume level in the 60s same with my Atom
Seems very high. Do you have any input trim enabled to match your phono input?
I’ve just discovered that your speakers are quite low sensitivity 82dB/W. 7ish less than typical.
I haven’t played with input trim so I guess still on factory setting. The 606s are 84dB sensitivity so not the most sensitive. I guess it would be interesting if anyone else is running 222 with 606s what volume levels they are using?
First of all, the physical volume control setting on the amp is of no consequence: provided nothing os being driven to the point that distortion occurs it will not do any harm to the amp, at worst if the amp gets to hot it’s thermal safety cutout will trip, leaving you in silence but resetting on cooling (at least, my understanding is that Naim amps have such a device).
If your DSP is simply cutting the level at places where there are peaks, then you don’t need to adjust anything for “headroom”. But if you boost anything, the maximum amount of boost at whatever the frequency means you loose that much headroom compare to playing flat, and unless you have a very high powered amp you could rapidly run out of headroom on peaks, causing them to overload, potentially causing speaker damage. I assume the headroom adjustment quoted is a reduction of the overall signal level so that boost at any given point is not too much - but it then would mean a lower average signal level, with the result that you are likely to simply turn up the volume control to compensate. Boosting with DSP to even out dips in response must be done with caution, and generally limited to only a few dB - remember 3dB boost is a doubling of power, and 10dB is 10x the power being demanded from the amp. Attempting to recover a frequency lost due to cancellation by boosting is doomed to failure and a very quick way to destroy speakers through putting the amp into clipping. Cutting peak frequencies on the other hand is risk free.
As for sound level, what you will get at your listening position for a given volume control setting is a result of many variable factors, from source output level, to the amp’s input sensitivity (and the new models may be different from previous), to speaker sensitivity, to distance from the speakers. For that reason whilst some people find it is loud at "eight o’clock’ on the dial, that could be very quiet on someone else’s setup.
Logic would suggest that this is true but it isn’t the case with Roon DSP (of which I am familiar) for both Convolution and PEQ. And Roon is what the OP is using.
Convolution tends to need less headroom than manually doing some PEQ and sounds better too IME.
The OPs situation where they need -10dB suggest that they either have a cuboid room or the measurements are not being done correctly. A calibrated microphone is a must.
Thanks for the detailed reply - the app measured a fair dip at 125hz, so I think that’s the problematic correction, but regardless, after listening for a while with the DSP on, I turned it off and felt the music was more lively, more sparkle, so DSP, or at least this filter, is not the one for me.
Yes, I’ll get a mic and see what REW suggests - I think ultimately some room treatment will be required.
You can’t really fix dips so I expect this is where your large headroom issue is coming from.
I wouldn’t really trust a 3 pence microphone in an iPad either.
It is not logic, but science! For what reason on Earth would you need to reduce level for headroom if you’re not boosting something, only cutting? The OP is boosting, so indeed that is his reason for it.
But cutting overall level for headroom is a false concept if you then wind up the volume control to compensate and give you the sound level you want. You have still lost the headroom, and your peak levels will suffer (and potentially damage tge speakers if you drive to clipping), unless tge amp has a lot more reserve power than you need for the music you play, or unless the processing also introduces dynamic compression to limit the peak levels (which damages the dynamic range of the music).
You’d have to ask on the Roon forum how their algorithm works but work it does. Any DSP is done in 64 bit floating point - a 10dB cut is two bits.
It’s only a small amount of headroom needed even with a 10dB cut using Convolution, 2 or 3db IME.
From that it sounds as if Roon is trimming 2 or 3 bits, perhaps to give room for the processing, and that is what they mean by headroom, as opposed to headroom as normally referenced in hi-fi matters, in the sense of an amplifier’s capacity to handle maximum peaks in music without going into clipping. Whether that means they have reduced dynamic range by 10dB or whatever, or whether they have simply reduced the signal level by that amount, is another question. If the former it may be detrimental to music compared to the original. If the latter (with nothing at a low enough level to be affected), then you would have to raise the amplifier volume control setting to compensate, which would not be a problem provided there was no or only trivial boosting involved.
As I don’t use Roon, and am unlikely to, their explanation is not something I’d be exploring their website or forum to discover.
Interesting thread; I use Housecurve in a similar way, but without the headroom adjustment.
Would you mind posting a photo of the adjusted curve please?
A couple asides, I wonder whether Home Audio Fidelity (HAF) would produce a different result than Housecurve? Certainly it’s something I want to tryout, but I had not considered REW. And, since adding a REL sub bass, occasionally the bass rumbles. I’ve tweaked the crossover and turned down the volume on the REL, but the rumble still occurs even when the volume on the preamp is turned down to around 8 o’clock.
Roon’s DSP headroom (as opposed to any other type) is mathematical process that reduces the volume by moving bits downwards in 64bit FP precision.
You only need to use it if you are getting digital clipping (as opposed to any other type) and the level needed can be checked by playing tracks known to be mastered close to the limit. There is no dynamic range.
This obviously makes the DSP’d music stream no longer ‘bit perfect’ but it is mathematically reversible and no data has been discarded.
This is what I had - the grey is the original and the green the corrected. As I mentioned though, I don’t like the overall effect so I’ve turned it off.
I can’t emphasise enough the use of REW and and a calibrated microphone to do this properly.
The room response can then measured using “Pink Noise” averaged over 1 to 2 minutes. Convolution can then be created from there.