C173-6-096E and 6640 project

[bvbellomo]

I know I already had a thread about this project, but it was called "Ardent Questions", this design is no longer that similar to the Ardents, and other people were asking legitimate Ardent questions on that thread.

The first implementation will be a pair of small 2-way front-ported floorstanders built of the Accuton C173-6-096E and SS 6640. Depending on how I like that after extended listening, and how much my wife likes me, I plan to eventually use the drivers and crossover parts to build a larger 3-way floorstander. If the sound quality of the ported 2-way is accepted, I will retain a ported 2-way for my center channel.

Is anyone familiar with the Accuton C173-6-096E? I am in the process of breaking it in and measuring it. Working from memory, I had the FS as 56Hz, so that was the test tone I was breaking it in with. This is also the impedance peak I measured. So I went back to compare my measurements with the official specs, and I see 36Hz listed on Accuton's site, Madisound (which lists somewhat different specs) also has 36Hz. No vendor I can find lists this as 56Hz. Is it possible I got a C173-6-090? They look the same, and it has an FS of 57Hz. This is almost the same price, so I doubt someone pulled a fast one, and the drivers came with stickers that say "C173-6-096E" as does the packing slip and the invoice. I have yet to try the 2nd woofer, and I really hope it at least matches.

Here is my impedance data so far:

36|40.1265823
55|147.894737
57|147.894737
58|137.317073
60|108.461538
62|95.0847458
65|73.1578947
70|52.6923077
80|33.5897436
160|11.0828025
320|7.35632184

[sdl2112]

From the picture you posted in the other thread I believe it is the correct version. The -096 uses a rubber half roll surround and the -090 uses a fabric surround. It appears to be rubber in your picture. With break-in the Fs should become lower. It is also somewhat dependent on drive level...higher drive level yields lower Fs. Run it for a couple hours at ~2/3 xmax and measure again. If you can adjust drive voltage, try 1 or 2 volts.

[bvbellomo]

Thanks, surrounds are definitely rubber. I ran it overnight at just over 1 volt at 36Hz, and measured the 40 ohms this morning, the other measurements were last night after 24 hours running a 57 Hz sine wave also just over 1 volt.

How do I know how loud I can go without damaging it? I can check voltage easily with a multimeter. Xmax I can only guess at.

[JonMarsh]

If you put the generator at resonance, and put a 10-50 ohm resistor in series, it will only drive it to high excursion at the Fs resonance. (this is assuming you're using an amplifier and not a signal generator with a 50 ohm source impedance. The power dissipation is quite low, when driven at the peak of Fs (because mechanical and electrical impedance is so high).

Driving it at 1/4" P-P excursion is quite reasonable, and plan to do that for 48-72 hours. Then measure T/S parameters. I'd recommend getting the Smith and Larsen Woofer Tester 2 for checking impedance and resonance, and developing your ZMA files.

http://www.woofertester.com/wt2product.htm

[bvbellomo]

Impedence at 36Hz went from 40.1265823 this morning to 40.3174603 tonight. This is a small change, but measurements are repeatable, so it is still breaking in.

Going by VituixCAD, 10 volts at 36Hz is exactly 5mm excursion, which is my xmax. Does this mean it is safe and advisable to break this in at 2.5 volts? I still don't understand any other way of judging how close I am to xmax.

[JonMarsh]

That would be fine- it will not hurt the driver, with it sitting out of a box. The thermals are quite low.

OTOH, the Fs being off this far isn't really right at all. The original C290-T6 Fs was 72Hz; the revised design dropped that to 56-58. If you are actually seeing peak impedance at the 56 Hz range, there is NO way (IMO) the suspension is going to "break in" enough to drop it below 40 Hz. It's quite possible, I'm theorizing, that someone messed up and your parts were assembled with the wrong spider, though the correct surround. That would explain what you're seeing- that you have a 090 spider.

Have you measure both drivers, and are they both the same? This could be a screw up at Accuton, where they knowingly put these into the DIY channel, possibly at a reduced wholesale price. I note that this dealer you worked through carries none of the newer product. Mike visited Accuton in Hamburg while I was in Munich this year, and their comment was that the cell products are their focus moving forward; the others will only be available a little while longer to fulfill needs of existing OEMs until they make model transitions.

You got a good price, but you might seriously want to plan on a reduced bandwidth and using a sub. (80 or 100Hz). Or making a three way. The sub path would be easier as long as you can find an active crossover you're happy with.

BTW, have you already worked out an alignment you want to use with the driver using the original T/S parameters? It's Q is under 0.2, like the 090, and while this provides very good damping, working out an extended range alignment is made more difficult.

[bvbellomo]

There's something happening here
What it is ain't exactly clear

I left the woofer on just under 4 volts last night, it isn't making the buzzing sound Jon suggested was amplifier clipping, and I am not worried about thermals. After 12 hours, you could feel warmth, but only if you were trying to. I don't know how far past xmax this will go before something bad happens, or what that will be, but I don't want to find out. Computer people are comfortable with heat, but suspicious of moving parts.

FS dropped to 48Hz, so it is still breaking in.

I measured the other woofer fresh out of its box, and right away noticed that buzzing sound, and measured it buzzing just under 1 volt. Fs on woofer 2 is 56Hz - the same as my first measurement on the first woofer, but that was after 12 hours break-in (although probably less than 1 volt).

Both woofers match, or at least I haven't seen anything suggesting they won't, and that is what I was most worried about.

If I had my way, I'd have 6 foot tall floor standers with a subwoofer. If my wife had her way, we'd use the TV speakers. I don't know where we will meet, but I am not giving up the sub I have now, unless it comes with a promise that I can build any 3-ways I want (or getting pregnant, we don't agree on anything, but this isn't the forum for that). For music, it would be nice to have full range mains, or cross them low. For movies, or a compromise, I just want to fill in the best I can between 60Hz (where the sub sounds great) to wherever the C173's works well without port reinforcement. I am hoping a port can fill that gap acceptably, even if I have to cross the sub higher. I always need a future upgrade anyway.

[Face]

Have you modeled it with your specs vs. Accuton's specs?

[bvbellomo]

I have modeled it with Accuton's specs. My specs keep changing, and I haven't even tried to measure Qts yet.

[bvbellomo]

I left this on overnight and most of yesterday just over 6 volts at 46Hz. Today I measured FS at 49Hz, so either it is going back up, behaves differently at higher power levels, or this is the margin of error for my measurements. Either way, it isn't changing much, and I don't think it will. I can work with this, but a lot still depends what Qts comes out at.

[JonMarsh]

It has never been my experience to see a starting Fs that was more than 10% off the published value, or to change more than 5-10% with breaking. So if you started in the mid 50's, I wouldn't expect it to go below 49 or so.

I stand by my first speculation. Wrong spider. Special deal to a secondary market vendor.

I HAVE seen differences in net Qts parameters which are a function of drive level, with ScanSpeak drivers, and different results depending on whether you're using 100mV drive or 4V drive. Scanspeak themselves says you have to use a few volts. for T/S measurements (this is for their older Revelator woofers) Modern stuff, I don't know. Haven't had issues with them.

As to Accuton, I've never used an Accuton driver as a woofer... note that the Dai Katana thread measurements of the AS190 seem to come in right on the money.

[bvbellomo]

I know everyone is going to tell me to buy a woofer tester to find Qts, but I want to be able to do this by hand. I know I made a mistake somewhere, but I don't know where. The best resource I found was http://www.mh-audio.nl/Measuring_Loudspeakers.asp

I am also have trouble measuring Re, but this could be my multimeter, I don't get stable readings either on the resistor or woofer. So I am using the Re = 6.61 ohms from Accuton's specs, which given a minimum impedance around 6.9 ohms I measured seems reasonable.

My highest impedance is 124.92228 at 48.7Hz.

RC = 124.92228 / 6.61 = 18.89898335854766
impedance of frequencies I am looking for = Re + sqrt(RC) = 10.95729609740902
I measured this at 12 Hz and 164 Hz
Qms = Fs * sqrt(Rc) / (f2 - f1) = 1.392850789104073, right here I bet I am wrong, as the spec says 4.07
Continuing on, I got Qes = 0.0778173129279388 and Qts = 0.0736997732988696 (there is an error in the Qts formula on that page).

So what did I do wrong?


Apologies to any followers of old-school Science for not rounding, with digital technology, it is faster not to round.

[JonMarsh]

A short suggestion: Do the math in an Excel sheet, in the same format he's using, and set the cell format to do the rounding for you.

[bvbellomo]

Re measured 6.9 ohms on my father in law's multimeter, which gives 0.8 ohms shorting the test leds. So Re is closer to 6.1 ohms. I was measuring with a 10 ohm resistor that is a less, so I am swapping it for one measuring closer to 10 ohms. I don't think either of these changes enough to explain Qms of 1.39 vs 4.07.

A short suggestion: Do the math in an Excel sheet, in the same format he's using, and set the cell format to do the rounding for you.

I am not sure who "he" is. If you mean to round in general, I can, and that makes it easier to read (but doesn't change the math). If you mean the author of the article, I see an online calculator, but no spreadsheet. The calculator uses an Rx (-9db) instead of Rc, so I don't think it uses the exact same method. But I get a much more reasonable Qms = 4.32, Qes = 0.222, Qts = 0.211.

[JonMarsh]

Well, you're in the ball park. But it is a very low Qts value, which will likely limit the LF level due to being a bit over damped. That's part of what contributes to the high efficiency, of course... and EQ might be a possiblity. I'd probably think about an alignment that wold give a classic 2nd order LR high pass at 75-80 Hz; this is easily mated to THX sub setups, or to conventional electronic crossovers. Just one thought to consider...

[bvbellomo]

A few more questions:

1) I should have denim insulation tomorrow. I won't be able to line the entire cabinet, as it is 2 inches thick. Is it better to line the back wall? Or would a side do just as well? I do plan to do the top and bottom. I realize I can play around, but it isn't easy to seal the box and make changes multiple times.

2) Jon, I asked you once how one works on a crossover without sealing it in the box, and you said the easiest way was to run the wires out the port. That would work well here, however when I have the finished crossover, how do I get it inside the box? I am tempted to leave the back panel completely unsealed and just clamp it on. Will that let too much air out to measure? I could silicone it, and that would be air tight and easy to remove, but I question how well wood glue would hold over silicone later when I went to make it permanent. I could remove the woofer, but that means the crossover has to go closer to the woofer than the nearest brace. Or I could put it through the hole for the terminal cup, but that restricts me to a crossover that fits through a 2 7/8" hole.

3) You still recommend the ClarityCap MR? If I only have 2 caps (2nd order) is it worth paying for both? Or will a Dayton work just as well across the woofer?

4) Some people are saying it is better to use 2 caps in place of one, for example a 10uF and a 2uF instead of a 12uF, is there any truth to this? Obviously that is what you do if you can't buy a 12uF or already have a 10uF and 2uF laying around, but if you can buy a 12uF, is it worth spending more money on a 10uF and a 2uF or should I just buy a 12uF?

[JonMarsh]

Let's tackle these in sequence...

1. 1. Ideally, you would have each wall covered; at the least, have one of each parallel wall covered (back, one side, top or bottom.
2. 2. Try building your crossovers to a form factor that will fit through the woofer hole. Have done that with many projects. Or build a false base with a hollow space in the bottom where you can add a final cover to it, to install the crossover. If you decide to go the removable back route, use weather seal for the initial install for measurements, then glue in place with wood glue or epoxy for the final install. But then your crossovers aren't readily available if they need service. I would recommend against that approach if possible.
3. 3. I do recommend the Clarity MR cap, BUT they have a new series, CMR, you might want to check out. I'm trying out the Duelund RS series in my next few builds, have some on hand; they combine the best aspects of the CAST and VSF series in a much more affordable and compact package. They're getting very good early reviews. A good distributor is HiFi Collective in the UK. I would not go to the same lengths for the LV shunt cap as for the tweeter - something like Jantzen Superior Z or the Obligatto Gold series should work well. If you're not sure about your crossover values or configuration, then prototype with much less expensive caps until things measure the way you like, and replace with the good stuff when you finalize things.
4. 4. I have never noted improved performance paralleling more caps unless one needed a specific ESR in a power supply application or were running into dissipation limits for the series resonant capacitor in an LLC converter. (I've seen engineers pull some bonehead things in that regards, with caps running at 100C because of the wrong component choice). The most important thing you might want to chose for is a mechanical construction that minimizes and damps mechanical resonance, and high quality film and foils or foil metallization. A hall mark of the right approach is a heavy tube that the cap is potted in- such as the Jantzen Superior Z series, the Clarity MR and CMR, the Obligatto Gold, and the Duelund RS and Cast. I've never found paralleling a small expensive cap with a larger cheap cap to work out very well... You'll see multiple caps in some of my design because of packaging issues or that that is simply necessary to get the right value.

[bvbellomo]

Denim insulation arrived today, and works well on 2 sides with a staple gun. I don't know how it sounds yet, but it looks like it will work well. If it does, I'd recommend this to other people.

My crossover probably has to fit through my terminal plate hole. I have a nice big woofer hole, but I put a brace underneath it, and there isn't enough space above it.

I'd love to know how the Duelund RS sounds, but a 10uF is almost the same price as a SS beryllium 6640! And I need a 12uF (at least in VirtuixCAD). The CMR is priced reasonably enough.

Cabinets are looking good, mostly done except for painting - I am just going to match the walls with some paint I have left over so we aren't looking at bare mdf. Can't wait to hear how this sounds.

Just broke my Jasper Jig! Be careful with these, the router doesn't have to fall far with the jig attached to have plastic shatter everywhere. I am okay on this speaker except the terminal plate, but I will need to buy another before I can start on the other speaker.

[Face]

If the value capacitor I'm looking for is unavailable, I try to split it by 50% or 33% when possible. For woofer caps or larger values, I'm a big fan of the Claritycap ESA/CSA.

[bvbellomo]

One temporary cabinet is done.

Image not available

If this were going to stay around a long time, I'd do more sanding and a few more coats, and probably would have done better prep before starting. Its is the same color paint as the wall, and blends with the room ok.

Now, I am trying to decide what port length to use. I bought a Goldwood 1-3/4" ID x 4". I thought a good way to start would be to measure it uncut and see where I am at. With the multimeter, I get an impedance of 10 ohms at 52Hz, with higher impedance on either side (zMin is around 500Hz). I also see a 180 degree phase change at 75 Hz using RoomEQ wizard. So is my cabinet currently tuned to 52Hz or 75Hz? Estimating cabinet volume, I am inclined to believe the latter, but don't understand why I measure an almost 40 ohm impediance at 75Hz?

Does running the wires out the port effectively reduce the port diameter? My crossover will also add some volume, so I am wondering if the tuning changes enough to worry about.

If I meant to tune to 80Hz and end up tuned to 52Hz, is there anything wrong with that? As long as I don't get a dip between 50Hz and 100Hz, I appreciate the lower extension, even though I didn't think I'd get it. Unfortunately, I can't get a good measurement - my computer is right next to a 300 gallon fish tank, so not only are room acoustics really bad, I have about a 45 decibel noise floor. I need to borrow my wife's laptop and go outside, but the weather has other plans.

I am really anxious to hook up a crossover and see how this sounds, but I want to get a port length set first.

[bvbellomo]

A quick update, hooked up a crossover and did some measurement and listening. I am definitely tuned to 75Hz, I might raise that to 80Hz or so, as I have a small dip. Some tall peaks and dips, probably 10db from trough to valley without smoothing. Of course I have lots of room issues, and I am still learning ARTA. I think these are still a little bright, I might have to up the L-Pad a little. Lots to tweek.

Overall, I am amazed. I just got done listening to "Best of The Doors" I grabbed the FLAC illegally because I can't find my CD, so some of the differences could be in the recordings (or it could be the same recording). I did listen to this with my RS100 surrounds, so I know most of the differences are in the speakers. This is not the most 'audiophile' material - lots of weird electric sound effects, so hearing something you didn't before could be awesome speakers revealing something new or broken speakers adding something new. I am able to distinguish live instruments (tamporine, rainsticks, cymbals, strings with minimal effects, obviously electrical guitar, and outright sound effects). Human voice is really impressive. The midrange is so much better than what I am used to, I can't tell if the SS6400's are really better than the 9700s or not. I need to get familiar with another "reference" track so I have something more audiophile to compare in the future.

I also grabbed some low brass and full orchestra - I played for 10 years so I know what that sounds like.

So where do I go from here? I ordered another Jensen Jig so I can listen in stereo. I think I just need lots of listening time so I understand if any changes I make (such as better caps) makes a difference or not.

[bvbellomo]

I am having trouble measuring my 6640's. This is a cheap amp, sound card and capacitor, so there could be some inductance rolloff, but I am surprised to see this much, and I have never been able to measure above 10k very well. I am also surprised to see how many ups and downs there are - the green lines are 10 decibels. This is very consistent if I use any kind of MLS or Periodic Noise and Single Gated Frequency Response. Distortion measurements are what I'd expect to see, and the tweeter sounds good listening to music. So I don't know if I have a real tweeter problem, or just a measurement problem. I always measure a 10k+ fall off with other speakers, haven't figured out why, but it never seems this bad. Room condition is horrible, but I don't expect that to affect the tweeter.

Image not available

I'd appreciate any advice to get good measurements, so I can finish my crossover design.

[wkhanna]

this is an interesting project, i hope you get spur measurements figured out.

[TEK]

What kind of amp, soundcard and mic are you using?
What distance are you measureing at?
Are you directly on axis?
What level are you neasurering at?
What software are you using?
Habe you a calibrated mic and are the calibration data feeded into your measuring progtam?

Just a couple of questions to get a error search process started...

[JonMarsh]

I am having trouble measuring my 6640's. This is a cheap amp, sound card and capacitor, so there could be some inductance rolloff, but I am surprised to see this much, and I have never been able to measure above 10k very well. I am also surprised to see how many ups and downs there are - the green lines are 10 decibels. This is very consistent if I use any kind of MLS or Periodic Noise and Single Gated Frequency Response. Distortion measurements are what I'd expect to see, and the tweeter sounds good listening to music. So I don't know if I have a real tweeter problem, or just a measurement problem. I always measure a 10k+ fall off with other speakers, haven't figured out why, but it never seems this bad. Room condition is horrible, but I don't expect that to affect the tweeter.

Image not available

I'd appreciate any advice to get good measurements, so I can finish my crossover design.

To TEK's questions, I would add,

Why are you using a capacitor (for)? What size? Where is the zero in the response curve for your expected load impedance with that value of capacitor? (have you measured the tweeter impedance properly, not like you did with the Accuton driver?)

What do you mean by a cheap amp (specific?) I use a cheap amp for measurements when I go portable such as visiting Ron (a Behringer A500 stereo amp)- But I've benched it and I know it's limitations, frequency response wise and distortion wise. What are the limitations of the amplifier you are using? (published or personally verified?)

Where are you placing your mic relative to the DUT, and how close are the nearest bondaries, and are you using any acoustical damping on them? Keep in mind the travel time of sound- what is the measurement gate setting you're using, and is it short enough to keep out early reflections (or are you using a measurement designed to attenuate the longer reflections, such as Half Hamming or Half Hanning?) What measurement window shape are you using? What kind of mic? What kind of preamp?

What measurement stimulus are you using, with what program, and why?

What kind of audio interface are you using? USB, Firewire, ?

(I thought I recalled you mentioned in an earlier post having an engineering background- you're being remarkably UNspecific for an engineer... :W

I may be confusing you with someone else, though... but that doesn't change physics or the need to be specific.

[bvbellomo]

Will post more when I get home, but I am using Arta (double checked with room eq wizard) with various parameters and microphone placements with Periodic Noise and Single Gated Frequency Response. I haven't tried to measure off axis, but manually aligning the microphone probably puts it around 2 degrees off axis. I have tried measuring as close as half a meter to almost 3 meters away. I've tried a lot of different measurements, get very similar results, lose my top octave similar to the above in all of them.

I haven't tried to measure impedence yet (properly or improperly) but this is a good next step.

I am measuring mainly with my crossover in place (using Dayton caps) and have double checked with just a 12uF capacitor. I have not tried hooking the tweeter directly to the amp. I'd expect some impedence from the Dayton cap and Dayton resistor lpad, but I'd expect it to be barely measurable.

[Face]

It's been a long time since I've used Arta, but is there some sort of sine sweep that could be used instead Periodic Noise?

[bvbellomo]

There are various sweeps, and I have also used the to double check and still have a problem. What is everyone else using? I thought windowed periodic noise was supposed to be more accurate for high frequency.

[Face]

I moved to the OmniMic as soon as it became available.

[JonMarsh]

There are various sweeps, and I have also used the to double check and still have a problem. What is everyone else using? I thought windowed periodic noise was supposed to be more accurate for high frequency.

Why did you think that? And how are you windowing it? (sorry to be so grumpy, but I have personal reasons).

You also haven't answered any of the questions from TEK or myself.

Myself, I use

• An ACO Pacific instrumentation microphone and preamp/buffer
• With a TC Konnekt 8 Firewire audio interface; microphone output from BNC converrter into XLR into the balanced mic input
• TC Konnekt is connected to a Macbook Pro via Firewire to Thunderbolt adpater
• TC Konnekt sample rate is set to 88.2 kHz, for a minimum bandwidth of 40 kHz
• Software used is Fuzzmeasure, which like Liberty Praxis, uses a swept sine chirp which is mathematiclally convoluted into an MLS impulse response in the sofrtware, then windowed for the time gate to avoid nearby boundary settings, that determined in milliseconds by how far the nearest boundaries are
• Alternatively, I can run the outrput of the ACO preamp with BNC to Balanced XLR converter directly in to my Audio Precision APx555
• The AP also uses sine chirps with convolutionl and has its own function to automaticallyi detect reflections and display both the minimum time window as well as a longer window for analysis.

You can approximate the time gating by using a gate in milliseconds equal to 1 msec per foot for which the gating time is no more than the distance to the nearest boundaries + 50%.


Fuzzmeasure:




Windowing and sweep settings are quite important for getting good measurements.





AudioPrecision Acoustics

[bvbellomo]

Jon, your still helpful when you're grumpy, so feel free to be grumpy. I am 30 miles away from my setup and typing on a cell phone while solving an unrelated problem, I will answer your questions when I get home.

I haven't seen a definitive guide for how to measure drivers. I have information from this forum and tutorials for Arta including one that either came with Dayton's microphone or someone posted with it

[bvbellomo]

I am using ARTA on a Windows 10 PC built on a ASUS P8H77-V LE LGA, which has a Realtek ALC887 for output. A Dayton Audio UMM-6 is connected via USB for input. The UMM-6 is suspect, as I have typically measured a drop in the top octave with everything else being different (different speakers, amp, sound card, room, etc) but the 6640 on my current setup measures less 10kHz+ output than anything previously.

I am outputting via the headphone jack to a Lepai LP-2020TI Digital Hi-Fi Audio Mini Amplifier. The amp advertises 20 Hz - 22 kHz and 14 watts per channel at 1 kHz at 1% distortion. I have not personally verified it's frequency response or distortion.

more later

[TEK]

Have you downloaded the calibration file for the UMM-6 and ensured that is correctly loaded into Arta?

[bvbellomo]

Have you downloaded the calibration file for the UMM-6 and ensured that is correctly loaded into Arta?

I am loading a calibration file for the microphone. Am I loading it correctly? I don't know. Without having a reference setup, how I can verify the calibration is correct? I'd expect different results if it wasn't loading, or if it loaded the wrong way (treat a +6db as -6db).

[TEK]

If you do a measurement that has a consistent result you can do it with and without the calibration file loaded and verify that you get expected changes in the measurements. You can open the file in for example notepad and see the numbers to check how mutch it changes the different requences.
You wrote "a file", not "the calibration file for the mic" but that is of course what you ment, right? You have of course been to Dayton and used your mics serial number to get the right file(?).

[JonMarsh]

Two important words in a situation like yours, with so many equipment uncertainties.

Loopback Testing

Test without the power amp, and with the power amp. You might need to attenuate the input to the power amp so you don't overload the record input.

Second recommendation is to use a software package that supports sine chirps. Much better S/N than MLS and many other techniques, and easy on the drivers.

Third, pay attention to the length of the measurement window! It's almost certainly way too long for your measurement, resulting in pickup of lots of room reflections. You'll probably want to be down around 10-15 msec from the first part of the impulse arrival- you are windowing against an impulse convolution, right? picking the start and ending is quite important, which is why the sweep and window settings are circled in the image above. Note, in this case I'm measuring the filtered output of a midrange driver, and you can see from the impulse response that there are reflections at around 6 msec into the gated interval, but low enough (and suppressed by the half Hamming window shape) to keep things reasonably clean. total gate time was about 20msec. You may not need more than 10msec to have enough lower frequency bandwidth and resolution. Your software should be reading out the effective resolution as a function of window length; in this case it's about 50Hz, which means also that that is the lowest frequency data point of any use. I usually use 1/12 octave smoothing; that's enough to resolve differences equivalent to a half step on the musical scale, but keeps the fine hash to a minimum. I see many DIY's using 1/3 octave, and if they like living in fantasy land about how flat their driver or system is, well, it doesn't matter to me, but that's like designing in the dark.

Fourth, don't use the Realtek in it's default sampling mode, which is probably 48kHz/16 bit- set it up for 88.2 or 96 kHz at 24 bit. If it's not possible to do that, get a real audio interface- what you have may just be a digital noise maker... :W


My last grumpy comment for this morning is, consider this: You've just bought a fairly high end pair of drivers, as regards what is available on the DIY market- but it seems like you're trying to develop a system with them with low end tools. That would be like buying a really nice slab of coco bolo for a rosewood style front panel for the speaker, and then only working it with a cheap carpenter's saw, a hammer, and a basic chisel. Yes, they're technically wood working tools, but not really at all suited to the job.

As ET might say if he notices this thread, you would benefit greatly from a judicious application of good tools to your development problem.

[bvbellomo]

Thanks, I thought this week I'd have a lot of time, but none so far. One quick question - how close is too close for the microphone? Can I put it 1 inch away and not worry about reflections (small window)?

[JonMarsh]

If you want to do nearfield measurements, to minimize the impact of reflections, 3"-6" works well. But be careful about your level, unless you're using an instrumentation microphone like I do, with 200V polarizing voltage and high voltage buffers. For an electret, you may overload the mic, if you test at standard 2.83VRMS level, because that's around 90 dB at 1 meter, but goes up or down 6 dB for each halving or doubling of distance. So 18-19" would take you to 96 dB at the mid, 9" would take you to 102 dB at the mic, and 4.5" would take you to 108 dB age the mic. You can see where this is going. My mic is rated for 140 dB- what's the spec on yours? (there is also the mic preamp to consider- adjust gain or pad the mic carefully. )

[bvbellomo]

If you want to do nearfield measurements, to minimize the impact of reflections, 3"-6" works well. But be careful about your level, unless you're using an instrumentation microphone like I do, with 200V polarizing voltage and high voltage buffers. For an electret, you may overload the mic, if you test at standard 2.83VRMS level, because that's around 90 dB at 1 meter, but goes up or down 6 dB for each halving or doubling of distance. So 18-19" would take you to 96 dB at the mid, 9" would take you to 102 dB at the mic, and 4.5" would take you to 108 dB age the mic. You can see where this is going. My mic is rated for 140 dB- what's the spec on yours? (there is also the mic preamp to consider- adjust gain or pad the mic carefully. )

So I started thinking, if I put my microphone at 3 inches and no objects are within 6 inches of the tweeter, I can start capturing in the time it takes for sound to travel 3 inches and stop in the time it takes for sound to travel 6 inches, and know I have no reflections.

In practice, this didn't work so well. If the microphone is 3 inches away, I could get reflections from whatever the microphone is placed on, as well as possibly the microphone itself. If sound travels 340 meters per second, then it takes 1/34th millisecond to travel a cm, which gives about .45 ms. But visually I see about 1.45ms before sound is recorded. So I don't know where the extra ms came from, unless it is delay in software or hardware before the signal is played.

[augerpro]

Can you post a pic of the impulse response and gating? Also a pic of your measurement environment. We need specifics and we're really not getting that yet.

[bvbellomo]

Image not available

Here is a 65535 length 48000Hz sampling rate MLS sweep which takes just over 1.3 seconds to record. Using a smaller sweep gives results that are consistent, except the lines get smoother (same drop), and I lose low frequency detail. I suspect the waviness is reflections. I am not concerned with the drop below 5kHz, 6 inches from the tweeter dead on axis puts me far off axis with the woofer and farther away.

I will try to play more with the sound settings, hardware specs and measure impedance when I have more time.

[augerpro]

That plot just tells us you have a problem with your measurements. We already know that!

We need responses to our specific requests.

BTW a properly gated response will "get smoother (same drop), and I lose low frequency detail", that's a function of the process and math. There are ways to fix that but first we need to get you to where you KNOW you are getting a clean far field measurement.

[bvbellomo]

I don't have a problem with my measurement! Or at least the main problem is output from my software, computer hardware, or amp.

At the same volume, a 10kHz sine wave outputs 0.863 volts from my amp. A 20kHz sine wave only measures 0.397 volts, I'd expect 0.4315 volts to be 6db lower, so my graph above is actually close to flat, at least closer than my output.

Now I can measure whether it is the amp or headphone jack with the issue, and replace whichever has the problem.

[bvbellomo]

I am willing to bet the Realtec is the weakest link here. What does everyone think of the DragonFly Black USB DAC? This would give me a 9010 Sabre DAC I can plug in to any computer, which I like a lot better than a higher end PC Express sound card for a computer I probably won't use with audio in the future.

[bvbellomo]

I know Jon isn't the biggest fan, but I like Emotiva's Ego products better. I'd take Burr Brown over ESS Sabre. This would have either a PCM5142 or PCM5141 DAC.