SevenSixTwo - InDIYana 2018 Coax Design

[augerpro]

That's 2 dB scales, not 5dB.

Very true. Still worth playing with though, and gives him another perspective about what metric to design around.

Yeah I hate inductor humps. The parallel resistor mentioned above can work well. Sometimes you can play with box tuning to get the response you want too.

[rpb]

Can you post the transfer function of the tweeter?

R2 concerns me, you'll be putting ALOT power through it. And I'm not sure why you would want to pad a woofer anyway. There must be a way to get rid of it.

Looking at those off axis measurements you can see why it was recommended to model off axis too. That energy from 7-10khz almost has to be audible, it is only down directly on axis, and room reflections will send all that energy to the listener. More generally, looking at the 45 degree measurement, the peak at 2.5khz, dip 2.5khz-7khz, and peak 7-10khz will *probably* be audible. The only way to know is make a crossover smoothing that area out and listen to both. Time for a MiniDSP

That's 2 dB scales, not 5dB.

[isaeagle4031]

Even at 25W, that resistor is going to cook. Might try an inductor with a higher dcr or maybe a 50 ohm in parallel

[KEtheredge87]

Can you post the transfer function of the tweeter?

R2 concerns me, you'll be putting ALOT power through it. And I'm not sure why you would want to pad a woofer anyway. There must be a way to get rid of it.

Looking at those off axis measurements you can see why it was recommended to model off axis too. That energy from 7-10khz almost has to be audible, it is only down directly on axis, and room reflections will send all that energy to the listener. More generally, looking at the 45 degree measurement, the peak at 2.5khz, dip 2.5khz-7khz, and peak 7-10khz will *probably* be audible. The only way to know is make a crossover smoothing that area out and listen to both. Time for a MiniDSP

HI Brandon, thanks for the feedback! You are right about the power on R2... pretty high dissipation is predicted from XSim, which is why I spec'd out a 25W "Zister" for that spot, as well as a few others in the mid/tweet circuits. The reason I stuck that resistor in place was to deal with inductor bump I was getting from the 4.5mH coil. WIthout it, PCD has me up another 1.3 dB at 100 Hz. If I decide to get rid of it, I'll probably be adjusting the capacitor on that woofer section too... which in turn will cause me to retool the mid and tweeter sections too

With your comments regarding the 7-10kHz energy... maybe I need to listen to the speaker again while sitting off at 45 degrees. I could use a mini-DSP for listening tests, but I'll would need to convert it back to a passive version for the competition. Rules are rules

Here's the tweeter transfer function you requested. I had to get it from XSim since I used all my PCD spaces on the tweeter and couldn't add that final capacitor.

[augerpro]

Can you post the transfer function of the tweeter?

R2 concerns me, you'll be putting ALOT power through it. And I'm not sure why you would want to pad a woofer anyway. There must be a way to get rid of it.

Looking at those off axis measurements you can see why it was recommended to model off axis too. That energy from 7-10khz almost has to be audible, it is only down directly on axis, and room reflections will send all that energy to the listener. More generally, looking at the 45 degree measurement, the peak at 2.5khz, dip 2.5khz-7khz, and peak 7-10khz will *probably* be audible. The only way to know is make a crossover smoothing that area out and listen to both. Time for a MiniDSP

[KEtheredge87]

Hi Everyone, Quick update for you this afternoon. I think I have the prototype XO finished and ready to advance to the next stage (making it smaller!) I added one 3 uF capacitor in parallel to the tweeter section to roll the top end down just a bit. I did try a home-brew coil a well with some of the wire left over from custom sizing the inductors, but it wasn't enough to make a useful difference, and would add even more complexity, so that idea got scrapped. Below are some measurements from back inside the basement. I have the full system and individual drivers, the reverse nulls of mid/tweeter/both being reversed, as well as some 1/24th octave smoothed measurements at 0, 15, 30, 45, and 60 deg.

The reverse null between the mid and woofer wasn't in a great measurement range from the indoor work, so I included the outdoor measurement for that one. Much better definition there. Overall I think I'm sounding pretty good, and I hope I can get all these inductors close enough to fit nicely without causing any inductive coupling. We'll see!

[KEtheredge87]

Hi Guys,

Saturday night update for you. Today I moved the whole XO setup outdoors for some low frequency measurements with higher gating times. I wanted to see if the dip at 350 Hz was real, or room interactions. Based on the outdoor measurements, the 350 Hz dip I saw indoors doesn't seem to be a major issue, so it must be room interactions at work.

Getting the whole setup going outdoors was pretty involved. Lots of fiddling with all the ladders, sawhorses, measurement tables, power, etc... Depending on how I interpret the impulse responses, I could make the gate time as large or as short as I wanted. There were only tiny fidgets on the impulse response line, but they didn't look like any reflections I had seen to date. I took them as electrical system noise rather than acoustic reflections, hence I chose to interpret the gate time as large for my intended purposes (low frequency measurement).

Additionally, while doing measurements indoors last night I experimented with the inductor values on the tweeter section. I measured the response with custom inductors (unwound to value) vs. stock inductors, and it turns out the custom versions measured better. Thanks Kevin K. for the idea to try both ways, as it was a great learning experience to play with the values and observe the response changes!

I'm zeroing in on my final bench XO before trying to compress them on to a board and get some real soldering done. The response seems fairly flat, with a slight upward slope above 3Khz. I will try to address this with some resistor adjustments tomorrow on the tweeter section.

Outside of all the frequency response measurement business, I think it's starting to sound very nice. To be honest, I haven't found anything objectionable about the current high frequency situation. Either the upward slope isn't within my own hearing band, or it's subtle enough that it doesn't stand out in any of my test material so far.

Below is a screen capture of the current XO measurements. This was from the outside measurement session, and shows the impulse response behavior that I chose to interpret as favorable to my needs. There's also some pictures of the testing setup.

Thanks for following along!

[KEtheredge87]

Oh Boy Oh Boy! First prototype XO impressions to share!

Tonight I finished cobbling together the prototype XO, and well... its BIG! I took a mixed approach, spot-soldering a few parts together like the notch filter before XO on the tweeter section and jumper clipping everything else. These few solder joints just cut down on the number of jumper wires that I needed to use, but there was still a spaghetti mess of jumpers to handle! When this gets refined I think I'll have to combine the woofer and mid XOs in the vented cabinet space and give all the Coax chamber space to the HF horn section. I'm a bit concerned about getting all those inductors closer together while still keeping them from coupling with each other, but I'll deal with that down the road. Oh, and to Kevin K. 's suggestion, I didn't mess with any of the inductor values. Everything is as-purchased from PE / others.

First listening impressions with one channel of stereo sound (mix of MP3 from iTunes and Amazon Prime Streaming):
Overall - Everything sounded quite pleasant. I don't recall hearing any peaks or noticeable dips. Nothing sounded harsh, and only a few tracks gave me anything to critque by ear. I did not hear anything that I would qualify as a "honk" from the horn section either, so that's a plus in my book! I also did not hear anything that presented as port noise from the slot. If anything some rap tracks might have a bit too much bass for this box tuning (or maybe my file quality isn't all that great?).

Individual track notes:
Tool - Jambi: The opening guitar riff and drums were giving the coax mid a challenge. I think I heard some kind of distortion or maybe cone breakup in that track. Might not have been the highest quality file either.
Tool - Forty Six & Two: This track didn't have the same issue that I noted from Jambi. The intro was strong and the electric guitar distortion sounded good!
A Perfect Circle - Lullaby: I picked this after hearing Scott use this at CKY-DIY back last year. The track's bass drum gave some smaller drivers fits then, but no issues on the SevenSixTwos tonight
Led Zeppelin - Going to California: Robert Plant's vocals came through cleanly and with good emotion. The high notes didn't pose any odd vocal issues for the coax.
Daft Punk's Tron Soundtrack - Overture: I like this track for the orchestral strings and brass sections. I got the strings more than the brass out of the single channel listen tonight, but I like what I heard.
Dire Straights - Money For Nothing, Sultans of Swing: The drums in Money were epic at the beginning! Knopfler's guitar solos in Sultans were very clean and sounded great!
The Matrix Revolution Soundtrack - Teahouse, Mona Lisa Overdrive: The japanese drums in the teahouse scene were nicely conveyed at moderate volume. The Mona Lisa Overdrive song from the freeway fight scene sounded good as well. Made me want to watch the movie!
T.I. - Top Back: The bass line in this song seemed like it was a bit overdone and muddy... I think... having issues remembering exactly what I thought was mildly objectionable, but something was a bit off. Probably worth trying again to confirm.
Lindsey Stirling - Crystalize, The Arena: Lindsey's violin sounded really good on this speaker. The drums and bass were well matched to the strings. Makes me wonder what this would really sound like in stereo!

I listened to a few other tracks here and there, but nothing else stuck out in my memory to write down. I'm pretty pleased so far and will continue to work on it. Maybe tomorrow I'll be able to take some measurements of the individual driver frequency responses and compare them against the XO simulation. I'd also like to take distortion measurements, but I'll have to figure out how to run those. That'll be another first for me. Thanks again for following along!

[JavadS]

Hey Javad, I bought these from PE earlier, and placed another order for 3 more sets on Saturday: https://www.parts-express.com/large-alligator-clip-20-awg-test-lead-set-10-pcs--360-154
They are crimped only, but that seems to be OK for basic test purposes. I only had one of them fail at the alligator spring clamp thus far... that rivet isn't the greatest.

Ok yea I saw those, 20 gauge but glad to hear they measured well, I have no issue with crimps as long as they're done well, thanks!

[KEtheredge87]

Keith interesting, have a link to the clips you got off of PE? I have a ton of the 22g ones but haven't measured them all, sometimes I double up on them for a higher power circuit, haven't noticed too many issues up till now, but now you have me thinking, thanks! BTW I love using the clips along with my test board, soldering stuff gets old real quick especially when you want to swap 3 components in 10 seconds while Omnimic is going, would be so much harder (and hotter) to do with soldering.

Javad

Hey Javad, I bought these from PE earlier, and placed another order for 3 more sets on Saturday: https://www.parts-express.com/large-alligator-clip-20-awg-test-lead-set-10-pcs--360-154
They are crimped only, but that seems to be OK for basic test purposes. I only had one of them fail at the alligator spring clamp thus far... that rivet isn't the greatest.

[JavadS]

Keith interesting, have a link to the clips you got off of PE? I have a ton of the 22g ones but haven't measured them all, sometimes I double up on them for a higher power circuit, haven't noticed too many issues up till now, but now you have me thinking, thanks! BTW I love using the clips along with my test board, soldering stuff gets old real quick especially when you want to swap 3 components in 10 seconds while Omnimic is going, would be so much harder (and hotter) to do with soldering.

Javad

[Paul K.]

Thought you might like to know this. I have a set of 4 alligator clip leads I bought from Radio Shack probably 15 years ago. When I first used them I was trying to measure some really small voltages and I couldn't get consistent measurements. I suspected the crimped connections in the leads being the problem, so I soldered them and the problem was eliminated. Today I took 2 of those leads and measured their end-to-end resistances with my True RMS Fluke multimeter; they measured 0.04 and 0.05 ohms.
Paul

Thanks Paul! I did a bit of googling to that end myself. I was surprised that the soldering didn't help much. I was measuring the resistance in two ways... one was my Extech EX330 multimeter that claims a resolution of 0.1 ohm with accuracy of +/- (1.2% of reading + 4 digits). The other was using DATS V2. The DATS was giving me 4 decimal places on the reading, and was generally more repeatable than watching the multimeter. I'm not sure what the accuracy specs are for the DATS in resistance measurement mode. Ultimately, all I can think is that the amazon wire wasn't really copper, or at the very least, wasn't good quality copper.

The engineer in me says absolutely solder the PE leads when they arrive, but then I'll have to solder 60 joints, and admittedly I'm a bit lazy there. That's 60 times to remove wires, strip off a bit more, open the crimp wings, put it all back together, then solder. The things I do for science...

[KEtheredge87]

With that many parts I'd just solder the tips of each component and skip the jumpers. I never had much success with jumper holding everything together.

Thanks John, I always appreciate a simple solution to counter my engineering neuroses. Maybe one day I'll conquer my irrational fear of screwing up components!

[jhollander]

With that many parts I'd just solder the tips of each component and skip the jumpers. I never had much success with jumper holding everything together.

[KEtheredge87]

According to the Wire Table in Introduction to Electrical Engineering by Mueller I've retained from using in college over 50 years ago, 22 AWG annealed copper wire shows a resistance of 16.14 ohms per 1000 feet, so a 2-foot long piece should have a resistance of ~0.03 ohms. I would think that after you soldered your El Cheapo Amazon crimps, you would have measured a much lower value than 0.5 ohms. Regardless, the PE leads are obviously better quality (and I would solder their crimps regardless).
Paul

Thanks Paul! I did a bit of googling to that end myself. I was surprised that the soldering didn't help much. I was measuring the resistance in two ways... one was my Extech EX330 multimeter that claims a resolution of 0.1 ohm with accuracy of +/- (1.2% of reading + 4 digits). The other was using DATS V2. The DATS was giving me 4 decimal places on the reading, and was generally more repeatable than watching the multimeter. I'm not sure what the accuracy specs are for the DATS in resistance measurement mode. Ultimately, all I can think is that the amazon wire wasn't really copper, or at the very least, wasn't good quality copper.

The engineer in me says absolutely solder the PE leads when they arrive, but then I'll have to solder 60 joints, and admittedly I'm a bit lazy there. That's 60 times to remove wires, strip off a bit more, open the crimp wings, put it all back together, then solder. The things I do for science...