Can anyone help with Alignment?

[Keith_W]

If you want us to check your workings, please attach an MDAT (or 3 MDAT's) with the following graphs:

Woofer L, R native measurements
Woofer L, R with amplitude linearisation
Woofer XO
Woofer L, R with XO

Midrange L, R native measurements
Midrange L, R with amplitude linearisation
Midrange XO
Midrange L, R with XO

Tweeter L, R native measurements
Tweeter L, R with amplitude linearisation
Tweeter XO
Tweeter L, R with XO

 

[Robbie010]

If you want us to check your workings, please attach an MDAT (or 3 MDAT's) with the following graphs:

Woofer L, R native measurements
Woofer L, R with amplitude linearisation
Woofer XO
Woofer L, R with XO

Midrange L, R native measurements
Midrange L, R with amplitude linearisation
Midrange XO
Midrange L, R with XO

Tweeter L, R native measurements
Tweeter L, R with amplitude linearisation
Tweeter XO
Tweeter L, R with XO

These will be coming your way shortly….

Couple of questions, at what distance should these measurements be taken?

What does “amplitude linearisation” mean?

 

[Keith_W]

Couple of questions, at what distance should these measurements be taken?

Here are a few rules of thumb:

- For speakers mounted on a baffle, the minimum measuring distance should be 2 baffle widths away.
- For horn speakers, about 3-4 horn diameters away.

The actual ideal measurement distance can be found by experimentation. Take measurements at various distances - 0cm, 25cm, 50cm, 75cm, 100cm, 125cm, 150cm. Apply a FDW of 1 cycle and overlay all the measurements. You will see (1) the measurements changing shape and (2) the SPL dropping. The point at which the measurement stops changing shape is where you should take the driver measurement. For example, the 125cm and 150cm measurements should ideally be the same, the only difference being lower SPL and maybe some reflections.

When taking measurements of drivers, it is VERY IMPORTANT that reflections do not contaminate your measurement. In this post I showed you how you can use the ETC to determine the lowest frequency you can measure which is uncontaminated by reflections. I examine the ETC of every measurement I take to determine the quality of the measurement. You want to correct the speaker, not a measurement artefact. This is why manual DSP is superior to automatic DSP. You can tell the difference. You can think. Software can't

What does “amplitude linearisation” mean?

I am a bit particular about terminology. I refer to the frequency response graph as "amplitude response" or "magnitude response", i.e. it's a graph of SPL vs. frequency. "Amplitude linearisation" means that you flatten the FR curve across the driver's bandpass.

 

[dualazmak]

I refer to the frequency response graph as "amplitude response" or "magnitude response", i.e. it's a graph of SPL vs. frequency.

In English, I often call it as "Fq-SPL curve" or "Fq-SPL response curve"; it looks easier to be understood especially for my domestic (in Japan) audio-enthu friends/colleagues.

 

[Robbie010]

Here are a few rules of thumb:

- For speakers mounted on a baffle, the minimum measuring distance should be 2 baffle widths away.
- For horn speakers, about 3-4 horn diameters away.

The actual ideal measurement distance can be found by experimentation. Take measurements at various distances - 0cm, 25cm, 50cm, 75cm, 100cm, 125cm, 150cm. Apply a FDW of 1 cycle and overlay all the measurements. You will see (1) the measurements changing shape and (2) the SPL dropping. The point at which the measurement stops changing shape is where you should take the driver measurement. For example, the 125cm and 150cm measurements should ideally be the same, the only difference being lower SPL and maybe some reflections.

When taking measurements of drivers, it is VERY IMPORTANT that reflections do not contaminate your measurement. In this post I showed you how you can use the ETC to determine the lowest frequency you can measure which is uncontaminated by reflections. I examine the ETC of every measurement I take to determine the quality of the measurement. You want to correct the speaker, not a measurement artefact. This is why manual DSP is superior to automatic DSP. You can tell the difference. You can think. Software can't



I am a bit particular about terminology. I refer to the frequency response graph as "amplitude response" or "magnitude response", i.e. it's a graph of SPL vs. frequency. "Amplitude linearisation" means that you flatten the FR curve across the driver's bandpass.

Excellent, thank you. I really appreciate your patience.

So, for clarity, the woofer baffle width is 320mm, so, ideally, I need to start taking measurements at 640mm and work back from there?

While the midrange horn isn’t round, the average diameter is 323mm, so I should start measuring at around 975mm and work back?

The actual measurements required are:

L&R Driver Native
L&R Driver with EQ flattened response
L&R Driver with EQ flattened response & xovers applied

Did you also need L&R driver with xover applied but no EQ flattening?

Is this correct?

 

[voodooless]

For the horn, use the widest width. And because it’s placed asymmetrically, I would add a bit to the baffle width as well.

1.5m should be a fairly safe distance.

 

[Keith_W]

Excellent, thank you. I really appreciate your patience.

So, for clarity, the woofer baffle width is 320mm, so, ideally, I need to start taking measurements at 640mm and work back from there?

While the midrange horn isn’t round, the average diameter is 323mm, so I should start measuring at around 975mm and work back?

I usually start at 50cm and work backwards. Sometimes a 50cm horn can be measured at 50cm. The reason horns need to be measured at a distance is because there are internal reflections in the horn if you measure too close - you get out-of-phase reflections arriving at the mic. These appear and disappear at various distances, so they are called "evanascent waves". Beyond a certain distance, these disappear completely, you are now away from the acoustic nearfield, and SPL drops according to the inverse square law. Somewhat confusingly, this is known as the "critical distance" (NOT to be confused with "critical distance" in room acoustics!).

Bear in mind that as wavelengths get longer, you eventually reach a point where it is impossible to take a measurement which is not contaminated by reflections. This is true unless you are able to raise your speaker to a great height and your mic the same amount (a 20Hz wave is 17m long!). This lower frequency limit can be determined two ways. The first way is by examining the ETC, as I showed you. The second way is to calculate it:

... assuming no other reflections intervene first, measuring from further away AND at a greater height lowers the low frequency limit where you can obtain a reflection-free measurement. See above diagram to calculate it. Or you can determine it from the ETC.

The actual measurements required are:

L&R Driver Native
L&R Driver with EQ flattened response
L&R Driver with EQ flattened response & xovers applied

Yes. Only the first (L/R driver native) needs to be measured. The rest are simulations, not actual measurements. You also need to include the raw XO before convolution with the driver response. Only after you are satisfied with the simulation, you verify it with an actual measurement. It saves you time. You will be surprised how close the simulation will look like the actual measurement.

Did you also need L&R driver with xover applied but no EQ flattening?

No.

See this post upthread. What you need to do are a whole bunch of measurements at different distances. Examine them for quality as I detailed in my previous post. Discard all the measurements which are contaminated by reflections that can not be windowed out, and discard all measurements with poor SNR. Only keep and show us the measurements which you want to use for driver correction.

And finally ... have you ordered Vance Dickason's book yet?

 

[Robbie010]

I usually start at 50cm and work backwards. Sometimes a 50cm horn can be measured at 50cm. The reason horns need to be measured at a distance is because there are internal reflections in the horn if you measure too close - you get out-of-phase reflections arriving at the mic. These appear and disappear at various distances, so they are called "evanascent waves". Beyond a certain distance, these disappear completely, you are now away from the acoustic nearfield, and SPL drops according to the inverse square law. Somewhat confusingly, this is known as the "critical distance" (NOT to be confused with "critical distance" in room acoustics!).

Bear in mind that as wavelengths get longer, you eventually reach a point where it is impossible to take a measurement which is not contaminated by reflections. This is true unless you are able to raise your speaker to a great height and your mic the same amount (a 20Hz wave is 17m long!). This lower frequency limit can be determined two ways. The first way is by examining the ETC, as I showed you. The second way is to calculate it:

View attachment 480471

... assuming no other reflections intervene first, measuring from further away AND at a greater height lowers the low frequency limit where you can obtain a reflection-free measurement. See above diagram to calculate it. Or you can determine it from the ETC.



Yes. Only the first (L/R driver native) needs to be measured. The rest are simulations, not actual measurements. You also need to include the raw XO before convolution with the driver response. Only after you are satisfied with the simulation, you verify it with an actual measurement. It saves you time. You will be surprised how close the simulation will look like the actual measurement.



No.

See this post upthread. What you need to do are a whole bunch of measurements at different distances. Examine them for quality as I detailed in my previous post. Discard all the measurements which are contaminated by reflections that can not be windowed out, and discard all measurements with poor SNR. Only keep and show us the measurements which you want to use for driver correction.

And finally ... have you ordered Vance Dickason's book yet?

Already read it…..

One more question, when applying EQ to flatten the native response, do I apply the individual EQ that REW supply to each driver or do I apply the same EQ to each driver i.e. measure one driver, EQ it flat and then apply that same EQ to both drivers?

 

[Keith_W]

One more question, when applying EQ to flatten the native response, do I apply the individual EQ that REW supply to each driver or do I apply the same EQ to each driver i.e. measure one driver, EQ it flat and then apply that same EQ to both drivers?

I would EQ each driver individually. Be sure that the discrepancy you are measuring is not a measurement artefact, though.

 

[Flaesh]

https://dbmark.co.uk/products.php?prod_id=DBMXCA48PLUS - Robbie's processor.
I didn't see a nymber of PEqs, there are no FIR filters, there are all-pass filters.

I would EQ each driver individually

I would check the identity of the drivers and if everything is fine, I would connect the corresponding channels and work with one speaker.
One can experiment with the processor right away. Or study VituixCAD and simulate a loudspeaker with different crossovers.

 

[Robbie010]

Ok, gents here goes:

Tweeters MDAT attached.

 

[Robbie010]

Midrange MDAT attached:

 

[Robbie010]

Woofer MDAT attached:

 

[Robbie010]

So this pretty much the best I can do in my domestic setting. I followed your advice, took multiple measurements of the native drivers and settled on a specific distance for the final measurements, 90cm for the Midrange horn, 70cm for the woofer and 30cm for the tweeter. I applied the 1 cycle FWD and then flattened the response for each driver in accordance with the REW EQ settings, I then applied the crossover settings and re-measured again, at the same distance, without altering the gain/volume.

One small potential issue I noticed late in the day is that I left a delay of 0.708ms on the tweeter throughout the process, not sure if that makes a difference.

Do your worst.........

 

[Robbie010]

Just thinking through this process, when flattening the response with EQ, I only applied the REW EQ setting to cover the intended range of the driver i.e. I only EQ’d the midrange between 1,200hz and 6,000hz, even though the driver obviously plays outside of this range. Is that correct or ahould I have flattened the entire response of the driver before applying the crossover filters?

 

[Flaesh]

I only EQ’d the midrange between 1,200hz and 6,000hz, even though the driver obviously

I assume that (for example) Keith in this thread and Vance in his book have already written about this

Edit: HF144 driver has an atypical frequency response, not the most convenient for equalization. Was this driver a random choice, or why the HF144? In any case, equalize it to an acceptable frequency response.

 

[Robbie010]

I assume that (for example) Keith in this thread and Vance in his book have already written about this

Edit: HF144 driver has an atypical frequency response, not the most convenient for equalization. Was this driver a random choice, or why the HF144? In any case, equalize it to an acceptable frequency response.

Ok, so my thinking was to have 3 drivers that cover a wide frequency range in order to give me lots of flexibility in terms of chosen crossover points.

When looking at drivers, I assumed (possibly incorrectly) that 3 drivers that overlap in frequency response can’t be a bad thing.

However, as I’ve said previously, this is the start of my project, not the end. Ultimately, I will happily take on the advice of those more knowledgeable and if the right course of action is to the change drivers then thats what I’ll do.

 

[Robbie010]

I assume that (for example) Keith in this thread and Vance in his book have already written about this

Again, apologies for my ignorance, I can see I’m wearing patience thin……

I’m struggling to keep up with all the replies in this thread, there is lots of technical info, which is over my head and obviously, I can’t read and understand a 300 page book between replies.

At this point, I’m looking to take advantage of your years of experience and expertise in helping get this system setup and running in a very basic manner. This is my main hifi setup in my living room and while I am expecting this to be a long term project / learning experience, I obviously need to rely on your gracious help and patience in explaining these processes in a very basic (none technical) manner.

 

[Flaesh]

get this system setup and running in a very basic manner

I understand that it can be difficult to immediately understand something that's unclear)), even if it's quite simple.
Equalize the horn-driver combo.
Equalize the Fane.
Determine levels for equal SPL of both bands.
Apply some kind of crossover at 800 Hz. These will be your home Max and [Super]Wax:

Then you can start listening to music. And continue learning, observing, and understanding how things sound. And improve the initial rough XO version. With DSP it's much easier than a passive crossover.

Bass as such isn't noticeable yet, but that's a separate topic.

Don't rush to change the speakers; it's best to put that off until later. And put the super tweeters aside for now, too.

 

[Robbie010]

Theres are pretty sexy though.