MiniDSP

MartinC

Wammer
Wammer
Jul 29, 2005
9,828
6,148
158
HiFi Trade?
  1. No
Just had a look at the file you uploaded. Do you have some acoustic absorbers around your room? The impulse response shows an unusual lack of early reflections, although possibly it's the more nearfield listening position reducing their significance.
 

zym

Newbie
Wammer
Dec 5, 2022
42
22
28
AKA
Not at Home
HiFi Trade?
  1. No
I don't have any purpose built acoustic absorbers in the room but there is a large bookcase immediately behind the listening area and the nearfield positioning 'should' reduce room effects.
 

MartinC

Wammer
Wammer
Jul 29, 2005
9,828
6,148
158
HiFi Trade?
  1. No
I don't have any purpose built acoustic absorbers in the room but there is a large bookcase immediately behind the listening area and the nearfield positioning 'should' reduce room effects.
What listening nearfield does is make the amplitude of the direct signal from the speakers higher relative to the reflections. This must be what I noticed but I'll be honest it's a stronger effect than I'd have guessed you'd have seen.

A different thought... Do Harbeth make a recommendation of where ear height should be at relative to the speakers, and how does your ear/microphone height compare to this? Just asking as the dip around the 2-3 kHz region is the sort of thing you get in the crossover region where this is too high or low. I am familiar with the 'BBC dip' concept so it may be this but this looks a bit pronounced to me... (I know this isn't the part of the spectrum you asked about but it's a feature that stands out to me.)
 

MartinC

Wammer
Wammer
Jul 29, 2005
9,828
6,148
158
HiFi Trade?
  1. No
Thinking about it the optimal height will vary with listening distance. You could just try measuring a bit higher and a bit lower to get an idea of the effect, although obviously if you decided you wanted to do anything about this you'd change the speaker height, or tilt them up/down.
 

zym

Newbie
Wammer
Dec 5, 2022
42
22
28
AKA
Not at Home
HiFi Trade?
  1. No
Harbeth do state that listening must be at the same height as the tweeters +/- a few degrees. This is pretty close for my listening, the mic may be a little lower (maybe 2-3cm) so it would be worth adjusting the mic for a few measurements.

I'll have to check this later tomorrow and respond.
 

MartinC

Wammer
Wammer
Jul 29, 2005
9,828
6,148
158
HiFi Trade?
  1. No
Harbeth do state that listening must be at the same height as the tweeters +/- a few degrees. This is pretty close for my listening, the mic may be a little lower (maybe 2-3cm) so it would be worth adjusting the mic for a few measurements.

I'll have to check this later tomorrow and respond.
Do any tests like this just with one speaker playing. Maybe try 5 and 10 cm both higher and lower just to get an idea. The ultimate decision on what is 'best' needs to come from actually listening but I'm pretty confident you'll see a difference re. the dip I mentioned.

Am I right that there isn't any sort of delay or phase adjustment on your subs? Just checking to make sure there isn't any point making specific measurements to try to optimise this.
 

zym

Newbie
Wammer
Dec 5, 2022
42
22
28
AKA
Not at Home
HiFi Trade?
  1. No
The RELs have a phase switch but only 0 or 90deg. No delay adjustment.
 

MartinC

Wammer
Wammer
Jul 29, 2005
9,828
6,148
158
HiFi Trade?
  1. No
The RELs have a phase switch but only 0 or 90deg. No delay adjustment.
I'm guessing you've tried making measurements to see which setting works best? changing it will affect the combined mains and sub response in the crossover region between the two. I'd check what works best for left and right separately, rather than just looking at everything playing at once.
 

zym

Newbie
Wammer
Dec 5, 2022
42
22
28
AKA
Not at Home
HiFi Trade?
  1. No
I repeated the main speaker measurements, this time taking special note of accurately placing the mic at optimum height for the Harbeths and the near field triangle. This has reduced or smoothed out the 2KHZz dip but has not improved the 80Hz position. The combined mains are shown below followed by the L main for comparison with previous.
Main Measurements 19th Dec.png
L Main Measurement 19th Dec.png

I did vary the speaker positions, widening & reducing the width by 10cm, both came out with worse measurements.
Regarding the subs, I have checked the phase alignment by both listening for the loudest response and measuring the alternative settings to ensure sub and main are aligned.

Work on minidsp will have to wait a while until I can get a suitable product.
 
  • Like
Reactions: MartinC

MartinC

Wammer
Wammer
Jul 29, 2005
9,828
6,148
158
HiFi Trade?
  1. No
I repeated the main speaker measurements, this time taking special note of accurately placing the mic at optimum height for the Harbeths and the near field triangle. This has reduced or smoothed out the 2KHZz dip but has not improved the 80Hz position.
For what it's worth I wasn't expecting it to make any significant difference in the bass region, just the higher frequency dip. Which as you say, it has :) . Just something to bear in mind in terms of your speaker height re. your ears.

As you keep posting graphs with both speakers playing together I'll just mention I'd never actually measure this. It's only really of any use at all in the very low bass, and if you do want to know about the sum there you can simply add together results measured separately for the left and right speakers (via the REW Controls cog when looking at the All SPL graph). I would though tick the 'acoustic timing reference' option on measurements to best do this, which will mean that when you take measurements you get a high frequency sound from one speaker at the start and end of each measurement.
 
  • Like
Reactions: newlash09

zym

Newbie
Wammer
Dec 5, 2022
42
22
28
AKA
Not at Home
HiFi Trade?
  1. No
Some developments over the last month. After consulting with MartinC and talking to some dealers, I decided the right way to go for my varied I/O requirements would be a Minisdsp SHD, even though it may duplicate the Graham Slee Majestic pre/dac function. While waiting for a suitable unit to come up for sale (used) I continued room / speaker measurements. A final placing with the main Harbeth P3s at 1/3 positions down the length of the room and dual REL T5x at 1/4 positions on the front wall now gives very respectable response, removing the worst of the room effects.
I managed to snag an SHD recently posted on WAM. Transfering the software licence only took a couple of days before I could get started setting up. There is a definite learning curve to get the software and device configuration right before starting any measurements.
I decided to initially feed a fixed analogue output from the Majestic to the SHD to see how it worked as effectively a 'post processor' taking balanced out from the SHD to GS Proprius monoblocks and phono out to the RELs. Using the basic Minidsp instruction to set up the routing, levels and cross over then applying Dirac gave the first pass at a smoothed response up to 200Hz. It worked but wasn't particulary impressive, hard to describe but sounded to me like a badly compressed cd.
For the second try I removed the Majestic and fead the SHD direct from Lindemann Bridge and used the white paper published by Deer Creek Audio to configure the set up. The result was much much better, integration of the subs worked seemlessly with Dirac smoothing the response up to 200Hz leading to +/-5dB including a room dip at about 160Hz. Very impressive! I've kept this as a baseline to look for potential improvements over time, to try small adjustments to PEQ and XO filters over time.
I found a way to split the output from my Accession phono amp to feed both SHD and Majestic and enable the option to drive the main speakers (no subs) from either preamps. In this case I prefer the sound through the Majestic, to my ears the SHD analogue is, if anything, too analytical and I prefer the smoother Graham Slee sound. So the Majestic stays put.
As for the remaining SHD functions -
-I have not been able to get Volumio up and working but this not an issue as I like the Lindemann interface and it works extremely well with Qobus.
-I have configured one preset for headphones configured so I can take analogue output from the SHD to my ULDE head amp and simply turn the volume of the main speakers. Adding PEQ setting published for the headphones proved very simple and effective.
-The only aspect, so far, that doesn't work very satisfactorily is trying to use the USB output from the SHD. It seems difficult get the output level satisfactory for any 'ripping'. However, the Accession phono amp gives a variable output which I previously feed through a Focusrite ADC to store vinyl rips and will stick to this.

Overall a very satisfactory result, with plenty of options to try 'tweeking' the response over the months to come and I can concentrate of the music rather than fiddling with sub integration.
 
  • Like
Reactions: MartinC

Mr.Ian

Wammer
Wammer
Mar 4, 2009
510
85
58
North Nottingham
AKA
Ian
HiFi Trade?
  1. No
Interesting observations, you beat me to the SHD in the classifieds :-( still looking for one.

But I am seeing lots of threads that suggest for music the sound is better with eq switched off other tham maybe DEQX or ARC. How do people find the SHD as a pre amp whith Dirac on?
 

MartinC

Wammer
Wammer
Jul 29, 2005
9,828
6,148
158
HiFi Trade?
  1. No
But I am seeing lots of threads that suggest for music the sound is better with eq switched off other tham maybe DEQX or ARC. How do people find the SHD as a pre amp whith Dirac on?
I only use Dirac in the low bass region as I don't like what it does when run full-range. Others like the latter though.

The low bass (<100 Hz say) is where most domestic rooms will have both the most issues and also the most easily managed issues, so this is where the biggest benefit of the likes of Dirac Live is in my opinion. This is most notably the case if there are strong room resonances that give obviously louder and boomy bass at particular frequencies. In my own room any music that excites the 35 Hz room mode sounds pretty awful if I don't correct this for example. The more obviously someone has a bass issue to begin with then the clearer the benefit of applying corrections will be, although now I'm used to a more even bass response I do find I'm more sensitive to it not being so with either DL off or more so in other systems.
 

Camverton

Wammer
Wammer
Jul 20, 2009
4,621
2,155
158
Herefordshire
AKA
Malcolm
HiFi Trade?
  1. No
For what it's worth I wasn't expecting it to make any significant difference in the bass region, just the higher frequency dip. Which as you say, it has :) . Just something to bear in mind in terms of your speaker height re. your ears.

As you keep posting graphs with both speakers playing together I'll just mention I'd never actually measure this. It's only really of any use at all in the very low bass, and if you do want to know about the sum there you can simply add together results measured separately for the left and right speakers (via the REW Controls cog when looking at the All SPL graph). I would though tick the 'acoustic timing reference' option on measurements to best do this, which will mean that when you take measurements you get a high frequency sound from one speaker at the start and end of each measurement.
Measuring both left and right together is only really useful for lower frequencies going through to misleading at higher frequencies. Having said that I normally measure L+R and left and right separately. It gives a quick and dirty view of what is going on, especially at lower frequencies. It is true that instead of measuring both left and right together you can add them together but I iirc using a timing reference is necessary and that the alignment tool, rather than trace arithmetic should be used. Not all analysis can be done on traces that are “added” or aligned (but without powering up REW I can’t remember what). The alignment tool is also a fantastic way of predicting the effects of gain and delay when integrating subs without having to run umpteen trial and error sweeps.

As for dirac, it is worth experimenting with. In some situations I have had best results with full range and manipulating the target curve, in others I have found it best to just correct the lower frequencies. Either way it should be borne in mind that improvements to impulse response apply throughout the frequency range even if only the bass frequencies are selected for processing. This also affects reflected sounds by reducing them, which may or may not be a good thing sound quality/effect wise. Martin and I disagree over this but I can only go by my etc charts which show a 15dB reduction at about 3ms. This is with omni speakers which are a very different kettle of sound waves to conventional speakers of course!

The important thing with dirac imho is to keep comparing with dirac filters on and off. It is surprising how much the final result can vary depending on microphone positions and, correction range and the target curve.

Quick question for REW users, do you remove the listening chair, or maybe cover it with a duvet etc, before measuring? Now there’s a whole new pot of worms!
 

MartinC

Wammer
Wammer
Jul 29, 2005
9,828
6,148
158
HiFi Trade?
  1. No
Measuring both left and right together is only really useful for lower frequencies going through to misleading at higher frequencies. Having said that I normally measure L+R and left and right separately. It gives a quick and dirty view of what is going on, especially at lower frequencies. It is true that instead of measuring both left and right together you can add them together but I iirc using a timing reference is necessary and that the alignment tool, rather than trace arithmetic should be used. Not all analysis can be done on traces that are “added” or aligned (but without powering up REW I can’t remember what). The alignment tool is also a fantastic way of predicting the effects of gain and delay when integrating subs without having to run umpteen trial and error sweeps.

It is necessary to use the same timing reference for all measurements (e.g. left speaker) but there is no need to do anything with the alignment tool (and it would be actively wrong if any alignment adjustments were applied). Using an acoustic timing reference is as simple at ticking a box though so no big deal.

As for dirac, it is worth experimenting with. In some situations I have had best results with full range and manipulating the target curve, in others I have found it best to just correct the lower frequencies. Either way it should be borne in mind that improvements to impulse response apply throughout the frequency range even if only the bass frequencies are selected for processing.

The part in bold is not true I'm afraid. It's possible to test this and I've confirmed that Dirac Live phase adjustments are limited to the frequency range defined by the curtains just as the amplitude response corrections are. I can dig out some data to prove this at some point if you'd like me to?

This also affects reflected sounds by reducing them, which may or may not be a good thing sound quality/effect wise. Martin and I disagree over this but I can only go by my etc charts which show a 15dB reduction at about 3ms. This is with omni speakers which are a very different kettle of sound waves to conventional speakers of course!

We do indeed disagree :) . It is flat out impossible for Dirac Live to reduce reflections, and it does not claim to do so. It just affects how much the speaker drivers move, not alter their directivity. I can only think your unusual ETC change is the result of a coincidental combination of he DL filter effect and the filtering of the reflection itself but if the reflected signal at individual frequencies could be viewed I am certain there would be no change.

Edit: for example, lets say DL increased the amplitude at a frequency that is absorbed to a significant extent on reflection, then when combined with all the other frequencies in the ETC the relative level of a reflection peak may change. The relative levels of direct and reflected sound at this frequency would be unchanged by the application of the Dirac Live filter though.

Quick question for REW users, do you remove the listening chair, or maybe cover it with a duvet etc, before measuring? Now there’s a whole new pot of worms!

It depends what I'm wanting to measure. I removed the back of my sofa and cushions when assessing early reflections recently as it removed those coming off the sofa. If I want a measurement to include the bass response I leave these parts of my sofa in place though as they don't have an insignificant effect.
 
Last edited:

MartinC

Wammer
Wammer
Jul 29, 2005
9,828
6,148
158
HiFi Trade?
  1. No
@Camverton I've just spent a bit of time trying to see if I could work out what's going on better in terms of the ETC changes you've seen. I can't see anything like 15 dB reduction you've seen but I can see some peaks reduced by up to a couple of dB, and a smaller number actually increased in amplitude with DL on. I'm sure this is all tied up with the changes to the impulse response shape, how surfaces reflect different frequencies, and how the contributions of multiple reflections all sum up. It's complicated! Definitely something that's made me think but it still remains a physical impossibility for DL to reduce boundary reflections.

Just to check actually, your 15 dB change is for a single speaker measured on it's own I assume? Rather than two playing together?
 

Camverton

Wammer
Wammer
Jul 20, 2009
4,621
2,155
158
Herefordshire
AKA
Malcolm
HiFi Trade?
  1. No
@MartinC We’ll just have to carry on disagreeing! I write from my experience of using dirac and a read of the dirac website although I don’t suppose either of us are privy to the algorithms they use.

To be fair I’ve only used dirac in four different rooms, from up to 99Hz to full range correction. The speakers I’ve used it with include Kef LS50, Meridian dsp5500, BKS107, a couple of different Martin Logans, Quad 2812, MBL 126 and German Physiks, most of the above on their own and in conjunction with either a single ported SVS sub or a pair of sealed SVS subs.

Assuming we are both writing from our experience and perhaps suppositions of how the algorithm works it is quite possible that we might come to different conclusions. Of the speakers above, the effect of dirac, not surprisingly, varies and the effect varies according to the recording and genre of music. I certainly wouldn’t presume to say that you are wrong so much as to say that my experience has led me to other conclusions.

Interestingly, with dirac correction up to 99Hz the effect of cleaner impulse response is very clear when listening to harpsichord music, not the most bass rich music! As to whether it is right to do this with an omni speaker is a matter of taste and how I happen to feel on the night but with the memory slots it is great to have a choice from no subs and no EQ, through to with subs and dirac live either limited to low frequency correction or full range.
 

MartinC

Wammer
Wammer
Jul 29, 2005
9,828
6,148
158
HiFi Trade?
  1. No
@MartinC We’ll just have to carry on disagreeing! I write from my experience of using dirac and a read of the dirac website although I don’t suppose either of us are privy to the algorithms they use.

For clarity I'm not questioning the ETC changes happen but refute the conclusion that the changes demonstrate that DL reduces boundary reflections. This has nothing to do with the DL algorithms as it is physically impossible for them to alter loudspeaker directivity. The complication of looking at an ETC is it sums up what happens at every frequency all at once, which is why the factors I described can change what is seen but without there being any change in direct vs reflected signal amplitudes at each frequency.

Ultimately it makes no practical difference whether I'm right here of course. It's whether someone enjoys the sound of the result that does, no matter how it is arrived at.
 

Forum statistics

Threads
113,444
Messages
2,451,263
Members
70,783
Latest member
reg66

Latest Articles