Kleos - Dipole Line Array

[Johnloudb]

Interesting looking unit, John... will have to see if I can get one and check it out over the holidays- it may be useful doing some checks on the NCORE 400 amps I'm building, and easier to setup and document stuff than my HP8903 analyzer. Looks like a lot of performance for $200.

I was impressed with what it offers for the price too, about the same as a midline sound card. Checked out the Hypex NCORE 400 module, Wow! Shouldn't be much distortion to see if their published specs are accurate. Would be very interested to see your amplifier build when you do that. That would be cool!

[Johnloudb]

Wow....that analyzer looks V interesting!

Your plot looks pretty flat, too!
Glad they sound good!

I am sure you will the rest of it figured out soon.

Thanks! I hope I do get this figured out soon. Ready to be done with it and listen to some good sounds. It's been a long but interesting road for sure!

[cjd]

Don't rule out the possibility the iPhone isn't up to the task. That rings a bell as something someone else ran into and I can notb remember where...

[Johnloudb]

Don't rule out the possibility the iPhone isn't up to the task. That rings a bell as something someone else ran into and I can notb remember where...

Okay, thanks! I'll be sure to check out the iPhone too.

[Johnloudb]

Don't rule out the possibility the iPhone isn't up to the task. That rings a bell as something someone else ran into and I can notb remember where...

Yep, it was distortion from the iphone that I was hearing when I turned it up.

Well, my Quant Asylum audio analyzer came Monday. It's a nice looking unit and they got it out to me real quick. Going to download the software on my computer and see if I can get it working.

[Hank]

Apple is NOT the answer to everything. You can quote me on that. John, that's a mighty flat plot you've got going! Congrats on nearing the end of your odyssey!:T

[Johnloudb]

Apple is NOT the answer to everything. You can quote me on that. John, that's a mighty flat plot you've got going! Congrats on nearing the end of your odyssey!:T

Thank you Hank! Wise words about Apple. Still got a ways to go and speaker placement and room acoustics are going to play a big role as always.

I LOVE how intuitive and easy it is to set up this Quant Asylum QA400. One page of instructions (12pt type) takes care of software installation, calibration, and measurement taking! It's all right there in front of you ... no hidden menus, technical jargon, confusing instructions.

It works as advertised ... here is the distortion in loop-back mode:

[Johnloudb]

Well, I measured the distortion 0.024% at 1V (0.707 VRMS), and you know, it could be better. At least it's mostly 2nd and 3rd harmonic. May try a slightly higher bias on the JFETs so that's it's closer to the Idss. I'll take another measurement at less voltage and see what happens with distortion.

[Johnloudb]

Here's another distortion measurement I took, this time at 0.174 volts (0.123 VRMS). As expected the distortion is less, 0.004 percent. But I think I'll try upping the bias on one or two of the JFET buffers by putting a 20-30 ohm resistor in parallel with the 10 ohm one already there, and see what that does.

[Johnloudb]

I increased the bias on the JFET buffer on the equalization stage before the op amp. Then I measured the distortion after the op amp and it has really low distortion now ... 0.00069 percent! Mostly 2nd and 3rd harmonic and close the distortion of the analyzer itself. The sweet spot for the Analyzer distortion seems to be around -8dB. A lot of the noise just has to do with changes in my measurement setup, as the lid won't completely close because of measurement cables. Still I should work more on the low frequency noise from the power supply.

[Johnloudb]

(note the graph I posted above) So, since the increased bias helped so much on the JFET buffers on equalization stage, I need to order more 22.1 ohm resistors for the crossover JFET buffers. Overall the distortion should be quit low now, at least that's what I'm hopping for.

EDIT: just remembered the op amp provides a gain of 5 (14dB) so the JFET is operating at a much low output voltage which will affect disortion. Will just have to see what happens when I get the new parts installed.

[Johnloudb]

Okay, the resistors came and I increased the bias current on all the JFET buffers. It made no difference in the distortion as you can see. I don't know, any opinions out there???!!! I think, JFETs make sense at the input, but I think I'm going to redo the crossover section with the LME49722 opamp. I have some more of my hybrid surface mount generic PCBs that have a pads for the surface mount op amps, so will use one of those.

[Saurav]

I don't know, any opinions out there???!!!

Here's an opinion - you're orders of magnitude below the distortion of your drivers, and above 5kHz, in the range of your measuring instrument's noise floor. So you can stop worrying about distortion in your crossover, and worry about other things

[Johnloudb]

Here's an opinion - you're orders of magnitude below the distortion of your drivers, and above 5kHz, in the range of your measuring instrument's noise floor. So you can stop worrying about distortion in your crossover, and worry about other things

Yeah, good point. I was just expecting distortion at around 0.01% or lower at 1 volt, for some reason. I could cascode the current source which would lower distortion. I know 2nd and 3rd harmonic distortion is not easily audible though and might make the sound more pleasant? Anyhow, will think about it.

[Johnloudb]

Light Bulb!!! if I put the EQ/gain stage after the crossover stage, instead of the input stage, the JFETs will operate in their low distortion region even at the maximum amp input voltage of 2.5 volts peak. I'm so happy!!!

[JonMarsh]

Progress! I haven't had enough free time to follow this the way I would like, but that makes complete sense, as did Saurav's input!

[Johnloudb]

Progress! I haven't had enough free time to follow this the way I would like, but that makes complete sense, as did Saurav's input!

Hi Jon, I hope your free time starts expanding soon! I rewired the crossover already and here are the results ... slightly more noise partly because of the rewiring (longer wires) and partly because the op amp is amplifying more of the circuit noise. But, distortion is much lower now.



EDIT: I think the higher order harmonics are just the test setup, and they show up at about -6 dB where the higher order harmonics are supposed to rise on the distortion analyzer. Those harmonics disappear at -8dB.

[Johnloudb]

Snubbers on rectifying diodes work! This reduced the noise quite a bit, and I also added some 0.47uF bi-pass caps across the large 10,000uF caps which may have helped with some high frequency noise. The diodes are soft recovery diodes too, so they also benefit. I'll post a schematic of the unregulated power supply when I get a chance.

[Johnloudb]

I'm working on the PCB, making the changes to lower distortion and improve layout. Getting there, will post when I'm done.

[Hank]

:T 2014 debut!

[JonMarsh]

Good progress... evolution and refinement. :T

[Johnloudb]

Thanks for the support you guys!!!

Okay, got the PCB done but still checking it over tooth and nail, and have found a couple errors I corrected.

[Johnloudb]

Here is the top layer ground plain, with some needed power supply traces.

[Hank]

Lookin' good, John. That hole below and to the left of C102 on the ground plane - is that a through-hole or for a component lead?

[cjd]

I'm curious about some of your routing choices, where you're sometimes branching to connect multiple components to a trace, other times not bothering.

IN>R25>C7 being the most curious to me, but R13/R15>P4 also - and others.

R130/R131/C20 lineup seems odd too - sometimes routing through a through-hole, sometimes branching off a main.

Why? I have only internet training (i.e. none at all of any particular value, probably) so this I'm not trying to say anything is wrong here, I just don't understand.

I admit the use of random angles bugs me too, though I have no idea why.

It's tough for me to follow due to the lack of screen for the surface mount components.

If it sounds good, that's what matters I think.

[JonMarsh]

John, does your software use net list tracking from schematics to PCB? That's the best way to assure proper connectivity of a completed board. Stuff I have to do at work is far to complex to do otherwise, but even a "simple" electronic crossover circuit like this is not so simple.

[Johnloudb]

Lookin' good, John. That hole below and to the left of C102 on the ground plane - is that a through-hole or for a component lead?

Oh, no component there, that is just where the power ground is connected to the ground plane. I put it close to the regulator ground for lower impedance ground, and less noise.

I'll label it as such on the silkscreen layer.

[Johnloudb]

I'm curious about some of your routing choices, where you're sometimes branching to connect multiple components to a trace, other times not bothering.

IN>R25>C7 being the most curious to me, but R13/R15>P4 also - and others.

R130/R131/C20 lineup seems odd too - sometimes routing through a through-hole, sometimes branching off a main.

Why? I have only internet training (i.e. none at all of any particular value, probably) so this I'm not trying to say anything is wrong here, I just don't understand.

I admit the use of random angles bugs me too, though I have no idea why.

It's tough for me to follow due to the lack of screen for the surface mount components.

If it sounds good, that's what matters I think.

Good catch, well what you're seeing there is my own ambivalence about whether or not it's a good idea to run a trace through a component pad. I've decided it doesn't matter and I think I'll clean it up a bit. Sometimes I have traces at weird angles so the trace is father away from adjacent traces or to keep the trace length shorter. Just don't want any possibility of interference or stray capacitance problems.

[Johnloudb]

John, does your software use net list tracking from schematics to PCB? That's the best way to assure proper connectivity of a completed board. Stuff I have to do at work is far to complex to do otherwise, but even a "simple" electronic crossover circuit like this is not so simple.

Very good point. It does have net list tracking of schematics to PCBs. It doesn't support automatic routing though. With a different circuit, I tried doing a schematic once and then producing a PCB from the schematic and it didn't work out as expected. I'll have to look into that more. I think I'll be alright with this though. Going over it very carefully.

[Johnloudb]

I've checked out the PCB, made some more adjustments with layout of parts, with short trace paths. Still working on drawing the power supply schematic and will post that soon. I also changed some parts. I'm using the LM7815A and LM7915A positive and negative regulators now as they have lower noise, and will keep the opamp cooler as it won't have dissipate as much power.

I changed the dual opamp to LME49880 with JFET inputs. It has very similar performance and I've heard very good reports about it's use in crossover circuits. Someone mentioned on DIY that the distortion LME49720 rises significantly with much resistance at it's inputs, which my EQ filter has below 1KHz.

Schematic -


Bottom Layer and Silkscreen -


Top Layer and Silkscreen -

[Johnloudb]

Okay, got the power supply schematic done. So I guess that does it!

EDIT: oh yeah, the unregulated supply on the left is not on the board! I'm not doing a board for that.

[Johnloudb]

If there are people interested in this PCB just send me a PM, would be happy to share it. I will do a group buy on this if there are enough people interested. The price for 4 boards is going to be $120 which is not too bad. To get the price down to $10 a board I need to buy 50 boards, that would be 23 other people buying two boards.

This project does need matched JFETs pairs (within a milliamp) for each buffer. I have not had difficulty with this, but I don't mind dealing with surface mount components either. I know a lot people wont touch 'em though.

I'll work out a BOM which shouldn't be too much, I think. There is a crossover calculator here which can help with design.

http://sim.okawa-denshi.jp/en/OPseikiHikeisan.htm

[Johnloudb]

I ordered all my parts yesterday, which came in around $115 minus the $12 transformer which I already have.

[Johnloudb]

I ordered boards yesterday for me and one other, after checking the board out another few times. Probably gone through about ten times total, so if there are still errors, I'm an idiot! But, I'm don't expect that.

Anxious to get this together and enjoy some music through these babies.

[wkhanna]

Progress:T

[Johnloudb]

Progress:T

Hi Bill, Can't argue with ya there. Moving along slowly but surely and expect things to pick up once I get my boards on Thursday.

Well, I just got home from swimming at the pool, and getting a caramel mocha coffee at Barne's and Noble (Starkbuck's inside). On the way home I just thoroughly enjoyed some music on KUER (Univ of Utah NPR) which plays Jazz at night. This music was so nice and relaxing ... I had to just sit in the driveway and listen to it till it was over, and got the artists name - Joshua Breakstone "With the Wind and the Rain." Will get that CD! I don't listen to Jazz much but very glad I did tonight. Got check out Jazz on KUER much more.

http://wemu.org/post/wind-and-rain-pure-pleasure-guitarist-joshua-breakstone

Anyway, this week I got a new (used) Toyota Camry, a very nice little '98 LE V6 that came with a nice little 45 watt Kenwood Stereo. It has a Jazz setting too, which sounds great. Just got this car fixed up for a tidy little sum of $1650, it had some brake issues and stuff, and got it tuned up and all the fluids changed too. Runs great! The whole aura of this car and radio is just relaxing and mellow, soothing ... I love it!

[Hank]

Now you'll be cruising, listening to jazz instead of speaker-building. Beware the temptation! :W

[JonMarsh]

Everyone needs a little time off... even from speaker building! :B

[Johnloudb]

Now you'll be cruising, listening to jazz instead of speaker-building. Beware the temptation! :W

Just working, no cruising here ... wish I was though!

Everyone needs a little time off... even from speaker building! :B

Yep, I couldn't agree more, Jon! :T

Got boards today, and I made a mistake. Fortunately not the catastrophic one, just that I selected standard service and didn't notice that it didn't include silkscreen and solder mask. Anyway, it was still my best option cost wise, and I don't really need silk screen ... just disappointed that I didn't get to see my boards in all there glory.

So here's what I got and my work around for having no silkscreen layer:

The bottom layer with surface mount components -



The top layer ground plain -



Here a print out of the silkscreen and pads which the software prints out the same dimensions as the board. Very convenient!



And here I have taped the printout so the holes line up. Putting it up to the light helped me line everything up.



So, this worked out well. And you know I remember reading recently in Stereophile about an Audio Research product, a preamp I think, that used boards without solder mask because they said the resulting dielectric absorption hurt sound quality!!! I'm on right track, and you know that I've maintained golden ratio dimensions on this board!!! Going for gold here ....

John

[JonMarsh]

That whole Dielectric absorption thing depends on both board material and the type of solder mask. It's also a good idea if you go without solder mask to use gold plate, not standard tin/solder, as it oxidizes in the air, changing the surface conductivity.


For insights into optimum PCB material based on DI, and solder mask, study Ayre Acoustics products. A LOT of research went into that by those guys-

[Johnloudb]

That whole Dielectric absorption thing depends on both board material and the type of solder mask. It's also a good idea if you go without solder mask to use gold plate, not standard tin/solder, as it oxidizes in the air, changing the surface conductivity.


For insights into optimum PCB material based on DI, and solder mask, study Ayre Acoustics products. A LOT of research went into that by those guys-

I've got that covered too ... :W though do need to order their Pro Gold stuff for that last ounce of protection.




I know Ayre Acoustics leaves no stone unturned, high quality stuff! I have doubts I could hear a difference, but who knows and I do plan on getting REL polystyrene caps for the high frequency filter after I see how this all works out. The board was designed for those. Right now using Wima caps till I get everything worked out.

[Hank]

Methinks someone at AR forgot to order soldermask also, and then Marketing put a spin on it. Another thing you can do is after assembly and test, you could spray conformal coating on the board to keep out moisture. I bought a spray can from Mouser with my upgrade caps and resistors that I'll use on my Acoustat Model 3 interfaces. Now get to soldering!

[Johnloudb]

Methinks someone at AR forgot to order soldermask also, and then Marketing put a spin on it. Another thing you can do is after assembly and test, you could spray conformal coating on the board to keep out moisture. I bought a spray can from Mouser with my upgrade caps and resistors that I'll use on my Acoustat Model 3 interfaces. Now get to soldering!

Hank, I think you're right about Audio Research ... makes me feel better anyway.

Great, I'll try the conformal coating. Got one board mostly stuffed with parts ... Pictures at 11:00!

[Johnloudb]

Here's a picture, and I need to order some more parts tonight cause I forgot heat sinks, and some resistors weren't right. I ordered mostly Dale RN60 1/8 watt but they are the size of 1/2 - 1 watt resistors and a tight fit. I guess all RN60 resistors are the same size? Anyway, I wanted the RN55 which are considerably smaller but Mouser didn't have the values I needed in those.

[Johnloudb]

Here's the underside of the board with most the parts soldered in ... the yellow caps in the bottom are just extra bi-pass caps for the op amp. The LME49880 needs a lot of bi-pass capacitance acording to thd data sheet, and recommends paralleled ceramic caps for lower impedance. Need to get some flux remover too.