First-Timer Home Theater Speaker Build

Ok I would have to say that confirms it. I can actually understand that a batch of drivers might be defective, but I am having I hard time understanding why I haven't heard about it anywhere else?

Guess it's time to contact PE again. Let's hope the reputation for good customer service is true. And I would ask them to test any replacements they send you before hand btw.
 
Contact PE?
Shall I just email that same guy you emailed before? I cannot find a customer service email on their website.

Also, I am curious. So, with the FR tests. Yesterday at the end of the day I took measurements for one of the tower. I had the mic at the height of the tweeter, 25in away. I also had the tower on this padded wooden box I found which I also stuffed with pillows so that it did not resonate much, if at all.
Without touching anything but the connections to the output, I took the measurements of all the speakers individually and then the pairs of WT(top), WT(bottom), MT(top), and MT(bottom).
So, the individual measurements for both the woofers just look weird, of course because it is off axis.

My big question is:
Do I still blend in the NF to the FF? In my head it seems odd to be blending an off-axis FR with an on-axis FR. But, without a doubt, I could be wrong. Prove me wrong! haha
 
PE's contact page: Contact Us

The riser underneath the speaker isn't so much about resonance as it is about using something that won't alter the baffle diffraction effect. Although generally, extending the bottom of a cabinet doesn't change the baffle diffraction too much, you are better off with something like a little stool so there are just a couple of legs and the rest is open space.

But your question about vertical off-axis angles is a good one and something I've been thinking about lately and funny enough, was just trying to work out some numbers and graphs involved with it today.

So I mentioned before that when we actually listen to real speakers, if the tweeter is at seated ear height, then all the other drivers are going to be listened to vertically off-axis. So it does makes sense to measure them this way too and therefore, yes one should be blending the FF off-axis response (for off-axis drivers) with the NF on-axis measurements. But Let's take a closer look at it.

So when the vertical distance between the tweeter axis and a mid/woofer isn't very large (with a 2-way for eg.), the mic can be fairly close because the vertical off-axis angle between the mid/woofer and the mic position at the close distance won't be very different than the angle between the mid/woofer and farther listening position (given an average home environment listening distance, say ~9ft). So the measured FR will be fairly accurate.

But when the distance between the tweeter axis and a woofer in a taller 3-way is fairly large, you are going to see greater differences in the vertical off-axis angles between the woofer and the different mic/listening distances, meaning the diffraction effects and therefore the measured FR are also going to be more and more different too. So that's not so good.

This is why in Jeff's measuring paper, he specifies that for the taller multi driver speakers, your mic distance should be larger and why I suggested about 36" away. Also note though that the main differences in the FR due to increased off-axis angles are all in the higher frequencies which are generally going to be out of the passband of the woofers in a 3-way. Still, I think accuracy suffers if the woofer roll-off isn't quite correct either.

I just popped open Sketchup and worked out the vertical off-axis angles for your towers. Here's what I found. Let's assume a listening distance of 9ft (108").

Drvr --- 30"--- 39"-- 60"--- 108" (mic/listening distances; "*"=my substitute for the degree symbol)

Mid ---- 7* -- 5* --- 3.5* -- 2*
Wfr1 -- 21* - 17* - 11.3* - 6.3*
Wfr2 -- 33* - 27* - 18* --- 10*

Open up Jeff's Baffle Diffraction simulator if you like and enter your tower dimensions and then play with the mic distance and the vertical off-axis variables yourself to see what effects they have on the diffraction responses. You should see that as the off-axis angles get larger and larger, the diffraction effects are also affected to a greater and greater extent. I think I still have more experimenting to do with this, but what it sort of looks like to me is that limiting a given driver's vertical off-axis angle in relation to the mic position to less than something like twice the vertical off-axis angle in relation to the listening position would result in fairly small differences in the different baffle diffraction effects and therefore more accurate FR measurements. So for eg., from the above little chart of numbers, it's only the angles with the mic out at 60" that are less than 2 x the angles at 108". The other ones don't qualify although for the mid, working with such small angles makes the differences also rather benign.

So if you are still with me so far, the problem with this is now how do we properly gate the measurement with a mic so far away for the FF measurement. Of course if you just take a look at the impulse response, it will tell you what the maximum gate gate time will be or for that matter if it will work at all. Because the gate has to actually be long enough for the direct sound to come from the individual drivers, but also has to be short enough to exclude any reflected sounds.

So what I was actually doing this morning was looking into this and writing a simple little spreadsheet program to calculate reflected path lengths and time of flight for sound at those path lengths. So with your towers, if you raise them up so the bottom woofer is about 24" off the ground, and if I've done the math correctly*, the minimum gating time needed with the mic at 60" is ~4.7ms and the maximum time before reflections reach the mic is ~6.75ms. That's for floor reflections. If the center of the speaker and the mic are also both 3ft away from side walls, the time it will take for reflections to reach the mic is ~7ms. So it looks to me like The FF plus NF measuring technique will work fine under these conditions with the FF gate being set anywhere between 4.7ms and 6.75ms. I would choose closer to the smaller end and call 5ms perfectly acceptable.

Anyways, that's my working theory right now. I reserve the right to change it any time due to further insight or contradictory experimentation. lol

On the other hand, there have been many, many speakers successfully built without putting the mic that far away from a 3-way tower. So what do I know.

* I posted a formula for the reflected pathway length the other day which I discovered today wasn't quite the right one. That formula gave you the direct path length between 2 points instead of the reflected one . The correct formula for the reflected pathlength (RP) if you are interested is as follows:

RP = SQRT[(Hm + Hd)^2 + d^2]

A more explanatory diagram is attached below.
 

Attachments

  • Refleted Path Length Calculation.png
    Refleted Path Length Calculation.png
    18.8 KB · Views: 125
When I suggested contacting PE, it was under the assumption that that was where you got the tweeters from. I can't remember if maybe you were dealing with Dayton directly before for that tweeter exchange a while ago. Anyways, maybe contact whoever it was with whom you were dealing with before.

Note that some of the above is just my own musings at this point. The larger distance for the FF measurement actually somewhat contradicts what Jeff has to say about FF mic distances. Just trying to be clear for you about that.
 
Oh PE as in Parts Express haha I couldnt figure it out, but duh!
Yeah I bought them through PE, I will send that out today.

It kinda seems like I shouldn't test anymore until I have new tweeters. Or we realize that the spec sheet was off and we gotta work with what we got.

You discussion about the off-axis measurements is pretty interesting and I will be looking at the Baffle Diffraction Simulator with the different angles.
So, I dont know enough about all this to contradict it or add. Haha purely absorbing it all.

But something that I cannot grip yet about it all is what to essentially do. Like, should I do what I've been doing? Mic now at ~36" on tweeter height and take all FF measurements (individual and pairs). Then take the NF measurements for the woofers/ports and mids, then blend it.

So up for discussion. I understand how we want to get the FR FF of the woofers and mids where the mic is at the listening position. To get a more true response to what the listener would be hearing. But that feels like a constraint that the listener is going to be exactly infront of the speaker at all times and exactly at the tweeter height. The listener is technically going to be off-axis horizontally for the majority of the listening. So, do we account for that in the xo as well?
Like, I guess I honestly need to know what I am measuring for. For the xo of course. But for what? To my knowledge from my class, which wasnt too much tbh, the xo is obviously a filter that directs the certain frequency bands to the corresponding drivers. But thats all I learned. I know there are xo techniques to balance the SPL levels, impedance match, achieve a smooth FR, and more.
So, if we want to achieve a smooth FR, wouldnt we want to get accurate on-axis FF and NF measurements for all drivers? To clearly see what the FR looks like when they are in the box and on the baffle. To ensure that what we are looking at is a pure response from that particular driver and not having to figure out what is going on since it is out of phase.

Essentially my thoughts are why is the vertical off-axis response being taken into account so much? Because there are so many other factors haha.
I understand that we should get the off-axis response to find all the SPL levels at the listener height in order to achieve a balanced SPL from all speakers.

My direction is just kinda lost, which is fine! The good thing is I got the time to figure this out and learn much more before I dive in.
 
Trust me - the FR you measured for those tweeters is way off and you really don't want to work with them.

Confused about measuring? Don't worry about it too much. It's a complicated topic and it's going to take some time, reading and experience to get a good grasp on it. For now just follow Jeff's paper. Take note of the particular criteria he's describing for the measurements and then figure out how it applies to each driver in each speaker. But when you get the new tweeters, I'll go over it again with you before re-measuring.

So what we've been looking at so far is the on-axis stuff but off-axis definitely matters too both for people sitting off-axis as well as for people sitting in the sweet spot on-axis. Although the sound that we hear that is coming straight at us from a speaker plays the most important role in our perception of the sound, what is happening in the range of 0-30* or maybe 0-45* off-axis plays a quite important role too. As does the room as well. Sometimes you may want to change the on-axis response so that the sum of both the on- and off-axis responses sound better. Sometimes you want to be particularly careful with the xo so that the off-axis response stays as similar as possible to the on-axis response. Sometimes you want to select drivers right from the start that have better off-axis responses too or a more specific type of off-axis response, for eg. with waveguides. And sometimes you want to make driver positioning choices that produce better off-axis responses, like with a CC where the tweeter is above the mids and not between them.

One other thing I can say is that for xo sims to work out properly, you need to have the correct phase info for the drivers and by this I mean the correct timing info of the sound wave which has to do with the x, y and z coordinates of the drivers on the baffle which will define the different path lengths of the sound from each driver to the mic. In XSim this is expressed by a single variable, delay, which is essential to get right in order for the different drivers' FR's to sum properly. And the only way to get this right is to place the mic at a stationary position and take measurements of the individual drivers and then afterwards, of the different drivers paired together with the tweeter. So moving the mic around for different drivers just want work out properly.

In terms of direction, yea, there's not much point to more measuring without the tweeters unless it's just for practice. So far you've focused on cab design and construction including box modeling and baffle diffraction sims as well as an introduction to measuring.

To round out your knowledge, you could do some more reading and/or I might suggest trying out some of the other aspects of the simulation procedure that we skipped over.

These would be:

- FR and impedance curve tracing
- frd and zma file combining and manipulation, including box effect splicing, combining baffle diffraction and FR curves and impedance modification or splicing
- xo sims
- the relative acoustic center determination procedure from measurements (vs from the spec sheets)
- practice some NF and FF blending with the measurements you have so far.

Choices are up to you.
 
One of the benefits to at least simming up the frd and zma files for each speaker based on the spec sheet info, so including at least the baffle diffraction effects, is that you actually have something to compare your real in-cabinet measurements to when you get around to it giving you a very quick idea of whether the measurements look right or not.

Something you can do to help your self understand how the horizontal off-axis responses of the different speakers are going to affect different people in a HT situation is to draw out your brother's room with the speaker and sitting locations included and then figure out the geometry and what all the angles are. Pencil, paper and a protractor will work but again, I find a drawing program like Sketchup to be quicker and more accurate. That will give you an idea of what the off-axis angle is for each person in relation to each speaker especially the front 3, and therefore also the angle up to which each speaker should have an off-axis response that is as similar as possible to its on-axis response. Notice also any differences that the toe-in angle of the front L and R speakers might make.

Next thing to note is that horizontal off-axis responses differ mostly in the HF's. Since the mid and woofer HF's are going to be rolled off in all the speakers, this means that the FR's in the off-axis positions are pretty much going to be dependent on the tweeter's off-axis response. Now go look at Dayton's spec sheet for the RST28F-4 and see for yourself exactly how much the HF response will change in your brother's HT for the different sitting locations. And the answer should show you 1 of the reasons I was happy with the selection of that tweeter for your design. Too bad it's turned out to be something of a pain in the......
 
Yeah I am in contact with PE and they have the product manager on the case haha. They have asked me about my full setup and everything. I sent that out Friday and will hopefully hear back by today.

My plans right now are to get the measurements done by Feb. 26th, hopefully.
Since I will now be going to visit my brother in Texas for a little over a week and then quarantine in the basement when I get back. But, my thoughts are to get all the measurements before then so I can start xo design and stuff while he is working during the day.
So, I guess we'll see how quickly PE can get back to me and resolve this issue.


To get correct phase info, is there anything different I need to do? Doesn't that all already get measured with REW?
I will for sure though be doing some more reading in the mean time. Since Friday, I have been actually shifting my gears to looking for jobs again as well.

I have gotten the NF and FF blending down actually which is good. I fixed the import problems for both Response Blender and PCD, so I was messing around on those last week.

One thing I have done before, yet would like to get much better at is the room simulation. Yet, my brother is looking into moving to a house around May. So, I dont know exactly what the room they will be in looks like. Since, its sooo cold right now, there is no way I can paint the speakers in the garage. So, I'll have to wait till it gets warmer anyways and then deliver the speakers when he gets a house.

If I got some time today I will also look into the frd and zma file combining and manipulation.
 
Hmm, just had a thought. Do you think that the 4ohm average for the tweeter has to do with something about the FR? Like maybe the Denon receiver I am using for an amp, with all internal EQ bypassed, cannot handle it well or something.

Ill look into it.

Update:
That does not seem to be the problem. The manual says if the output impedance is under 4ohms for an extended period of time then the protection circuit would over heat. But the sweeps are pretty dang short, only 2.1 seconds. So, I can rule out the receiver then.
 
Last edited:
And current consumption up in the HF's tends not to be so strong anyways.

Re phase: REW can include the phase of the FR measurements but they just do it mathematically from the FR (ie. phase is just a function of the shape and amplitude of the FR) so there shouldn't be much difference getting it from REW or extracting it in another program.

Note however that when you blend the NF and FF measurements, you change the FR. So therefore the phase from an REW FF measurement is no longer valid. That's why Frequency Response Blender includes the Minimum Phase Extractor. But here's a little idiosyncrasy of mine - I prefer to use the same program for all my phase extraction for consistency's sake, just in case the math is a tiny bit different in the different programs. It probably doesn't matter but there you go anyways. Since XSim has that function, that's the one I use for all the FR files including tweeters which of course don't get NF and FF blended.

Also notice that there is a bit of a difference between a driver's phase and having 2 sound sources in phase. The 1st one is what I'm talking about above, but the 2nd one is about locating the different drivers' relative acoustic centers to make sure that once they go into a xo program, they will sum correctly. Have a go at finding the XSim delay and/or the relative acoustic offset in PCD if you happened to have taken any individual driver measurements as well as the measurements of the same drivers together. It should be instructive.
 
So since the speaker project has been on hold, I have been pulled into projects to do around the house while waiting for PE to respond.

They finally have this morning after I sent multiple followup emails.
So, they sent me screenshot of what they got, attached below.
Everything seems to be in order for them. So, they think it is my baffle design, room reflections, or test procedure that it making me have inaccurate results. Which is odd, everything seemed to be in order.

One thing I did forget to do is break them in, so I will do that today and see if things change.
 

Attachments

  • Image.jpeg
    Image.jpeg
    155.5 KB · Views: 157
  • Image (1).jpeg
    Image (1).jpeg
    461.5 KB · Views: 149
Yes, I was afraid that it still might be a measurement problem and not something wrong with the tweeters. Not saying that the latter might not still be the case but let's see if anything changes with some break in time. My guess is that it won't.

If it is a measurement problem, my apologies that I can't seem to help you with it. It's seems to be beyond me especially because the other measurements seem to be ok. If the problem persists, it might be worth starting a new thread just on this topic and seeing if better minds than my own can find the source of the inaccuracies.
 
My guess is that it wont either.
But I have arranged so that I can exchange the tweeters I have now for the ones they tested which is great!

And it's no worries, I dont think anyone can figure this one out. I mean even the PE people have asked me twice about my setup and testing procedures with nothing to add or change about it. The fact that other measurements look good and these are just way off tells me is a quality control thing on this batch or something of that nature.

When I get the new ones, I really hope that the tests come out well. I have also asked PE if they can send me the .frd and .zma files from their tests. Just in case.
 
It has been awhile Chris!

So after a week out of town, I am back to it and I have the new tweeters.

Yet, unfortunately I am getting similar results. The screen capture is attached below.

The green trace is what I just measured. The blue trace is the measurements from before. The red trace is the .frd file I got from PE.

The comparison seems to be good until around 9kHz. I do not know what is going on, but maybe my mic isnt good around that range, which would honestly be weird since it is a measurement mic and I have a calibration file for it. I am really at a loss here, do you have any suggestions?
Do we take these measurements as true and design around that? Thats what I am thinking personally.
 

Attachments

  • Screen Shot 2021-03-09 at 3.11.58 PM.png
    Screen Shot 2021-03-09 at 3.11.58 PM.png
    433.8 KB · Views: 124
Hey Keil, good to see you back at it.

Looking at your results, they are in fact much better this time but there is still some discrepancy as you note. So I took a deeper look at all of your measurements to date. I had just eye balled them before but this time I traced them and then subtracted one from the other to get a closer look at the differences. Here's what I looked at:

1. the RS150 FF measured vs spec sheet with baffle diffraction added in
2. the first RS28 vs the measured FR by Dayton
3. your previous tweeter measured vs the spec sheet
4. the new RS28 vs the measured FR by Dayton

You can see the results below. These are the differences between what you measured and what the spec sheet data suggest. Note that if the measurements were a perfect match, the differences would be zero, so just a flat curve at 0dB. The 1st pic shows cases 1 and 2 above. So the green curve is the difference between the original problematic tweeters and the purple is the RS150 and you can see that both of them actually were actually measuring hot above about 2000Hz although the tweeter definitely got worse above about 8kHz. Way worse. That latter difference may have indeed been the tweeters fault. Not really sure. Sorry I didn't notice all of these discrepancies before btw, but I didn't look at them in quite enough detail.

Now in the 2nd pic, we've got a look at the differences bewteen that other tweeter of yours and the new RS28 plus I included the 1st RS28 difference as well. This time there is very good correlation between the 2 sets of comparisons. Both are pretty similar up to about 8kHz again and then they take off just a little bit. So the problem between about 2-8kHz seems to have been solved but the rise above 8kHz remains.

It's a little hard for me to form a solid conclusion because I was hoping that all 4 measurements might show a similar pattern which would imply something wrong in the measurement set up but they don't quite do that. Close though if we discard the RS150 measurement and assume the 1st batch of tweeters was faulty. Then it would seem to me that you've got a measuring error somewhere above ~8kHz or so.

Your mic would seem to be the most likely culprit but there is also an outside possibility that it could be some kind of hidden audio setting in your computer, an eq setting in the amp you are using or something to do with soundcard interface. That covers the whole measuring chain. Swapping each of those out one at a time and seeing if there is any difference would be the way to hunt the problem down I guess but somehow I doubt you have replacements for each of those sitting nearby. So you could double (or triple) check your computer and/or amp, maybe assume the Focusrite is ok and then conclude it's a mic problem. Maybe try with and without the calibration file. And/or maybe try a different mic. Any chance you have a measurement mic that comes with most AVR's these days sitting around? I guess you'd need a 1/8" to 1/4" connection adapter but that would give you another good piece of data for the sake of comparison.

That's the best I've got at this point anyways. Or you could just proceed and we can kind of ignore what's happening above about 8kHz. That might be ok though I'm not quite sure. I'll have to see how the rest of the measurements look to know for sure.
 

Attachments

  • Diff's 2.jpg
    Diff's 2.jpg
    69.8 KB · Views: 131
  • Diff's 1.jpg
    Diff's 1.jpg
    78.4 KB · Views: 121
Good to be back at it!

Thank you for the close inspection! It is clear that the old ones seemed to be a bad batch. Yet, these still arent quite there.

I have been looking everywhere and there are no hidden EQ settings on either my computer, amp, or Focusrite. I will try an AVR measurement mic I have though. See what I get.

My amp is on "Pure Direct" so it bypasses all internal EQ. My computer has no EQ on it that I can find and also a Mac is more optimized for audio, so I would doubt there would be any hidden audio setting that I cannot find.

The only thing I can thing of is the Focusrite. When I was donig the impedance testing, I plugged my computer into a power strip under my work bench and I heard popping or something like sparking. I instantly unplugged my computer from the charger.
Came to realize it was the very old power strip with weird "protection circuits" that had either gone bad or something. Because I also plugged my oscilloscope into the same plug a week prior and it was making some weird noise.
Thats the only thing that I could think of that would maybe mess up the focusrite, yet I looked it up and it has protection circuits built into it.

I do not know, but I will try the AVR mic and get back to you.
 
Yea, I use pure direct for measurements as well. Here's a thread from someone else who had a problem that ended up being in his computer - Unexplained low frequency measurement mystery. But I expect Macs are different.

Try the AVR mic. Way back when I was 1st starting out, that's what I used for my 1st build before getting a real mic. It might actually give you better results than what you're getting now. Or maybe not. 😀

I guess another alternative is to get a new mic that doesn't need the focusrite. I use a UMM-6 which I plug right into my laptop via a USB connection - Dayton Audio UMM-6 USB Measurement Microphone. And of course, wouldn't you know it, it's on back-order.

Starting a new thread just on this topic that might get some more participation might still be a good idea too if we can't seem to resolve anything. Just a thought anyways.
 
I currently do not have money for another mic at this time unfortunately. But remember when I tested another tweeter and the results came out just fine? Makes me believe that maybe the baffle is doing something to it the response.
Or even though it is very quiet down here, there might be some high frequency sound it is picking up. Though I doubt it.

The mic is also picking up the woofers and mid drivers response just fine.
It really is a frustrating mystery.

I will start a new thread to get some input!

[Update]
Here is the new thread:
https://www.diyaudio.com/forums/equ...-tweeter-testing-discrepancy.html#post6568204
 
Last edited:
Ok I guess I didn't explain things well enough.

The graphs I posted yesterday showed that both the woofer measurement and the measurement of that other tweeter also looked to be off in the HF's just like the old tweeters but not to quite as significant an extent. If they were actually ok, the difference between measurements would be closer to 0dB but in fact, that's not the case in both examples. Have a look at those graphs again.

Previously I thought they looked fine but that's because I didn't examine them in enough detail. They weren't as bad as the old tweeters but they also weren't quite right either.

If trying the AVR mic produces no changes, the next thing I was thinking of was to re-run the loopback function in REW and re-calibrate the soundcard and focusrite combination. Or let's see what other minds might come up with.
 
Oh dang you are right.

Well I tried the ACR mic yesterday and it either was only clipping like crazy or wouldnt get anything. I'm guessing it was built specifically for the Denon receiver, which doesnt make much sense either tbh. Like, a mic is just a mic.
Yet it did not work. So, I'm back to phase one. But I will try the recalibrate everything.

I should honestly download REW on my Windows side and try it out over there as well. Just to check.

But if this all doesnt work, I will have to invest into another mic.