4430 help!

[Robh3606]

This is a great post Ian made a while back. Should help clarify whats going on with the pots R201, R202. R201 can be set to a low initial value as this is where the "0" range lies.

Rob



I will need a great German Beer to quench my thirst when I finsh the explanation okay!

I recall explaining this on the earlier forums and Rob in certainly on the money.

Starting with the 4430/4435 sensitivity, the latter is 96 db and the former is 93 db.

Thus the horn on axis response must match these levels and within the design of the overall network offer a degree of mid band level shift boost (in the case of the 4430) and cut.

Looking at the schematics, R5 effectively sets the overall sensitivity of the horn driver.

As you can see R5 is shunted across the driver and forms a voltage divider with the series resister R4 and the bandpass Eq contour network defined by R201, C9/10 and L1.

The bandpass filter provides an inverse of the mass rolloff response of the 2425/6 driver and associated 2344 bi radial horn.

The effect of the variation in values for the C9/10 in either network reflects the low pass time constant of the filter and the resulting response shape. The larger value of c9/10 in the 3135 provides more voltage peaking of the filter and resulting output in the 1-5 kertz range.

The effect of R201 is to raise lower the Hf extension by varying the Q of the filter.

The effect of R 202 is to provide boost/cut in th 2 khertz area which is attenuated to a preset level by R4.

The crossover values are different in either network to allow for summation of the drivers with the all pass 2 nd order filter, the crossover points are identical.

Because the drivers are in phase, the networks function is to cancel the resulting peak/null at the crossover point. The effect of horizontal displacement is critical for a smooth response.
The 2234/2235 may have different acoustic centres at this frequency hence the chosen values.

I hope this explanation helps.....time for a beer.

Ian
Attached Images

[Zilch]

So, R201 is not merely attenuation, it concurrently varies the time constant (and thus the slope, or Q) of the RCL filter R201-C9-L1.

I read that before, but didn't get that part until just now. Cool. Giskard's beautiful Sim is here:

http://audioheritage.org/vbulletin/s...page=7&p=43075

[We're talkin' the yellow part.]

I think I'd rather have R201 be variable so's I could play all that stuff.

Note: We've got another lead on a switch through member Tom Loizeaux, looks like....

[Earl K]

Rob,

- Have you done the math that would support Ians' contentions and your assumptions ?

- In particular, the characterisation about which of the two legs ( in the HiPass Section ) is the "bandpass" circuit seems flipped from my perspective .

FWIW : & just for this one example , assume the ac impedance of the 2425/6 at higher frequencies is 7.5 ohm ,,,, parallel a 7.5 resistor with that,,, the new working ac impedance becomes 3.75 ohms ( not accurate , but will suffice in this example ).
- 1.8uf times 3.75 ohms gives 6.75 ( I've excluded the DCR of L1 ). ( 1.8 is an approximate product derived from 20uf in series with 2uf ).
- Divide 6.75 into 159000 ,, the answer is 23555 (hz ). Divide 8 into 159000 ( likely a more accurate approximation when dcr values are included along with accurate ac driver/horn impedances ) and the answer is @ 20,000 hz.
- These answers/products are the Fc points for C9/C10 in series with C7 ( in biamp mode ).

- Say the working impedance around the driver / R5 area is 4 ohms. Then 159 times 4 = 636 . 636 divided by .04 (mh) = 15900 hz. ( this would be the so called 16K shelving circuit built into the N3134 )

- This circuit is best built step by step & then measured along the way as a hands-on educational tool .

[Zilch]

Rob,

L1 and L2 are apparently NLA at JBL. Can you get us DCR's on those two, please?

I hate to tear into my factory crossovers to get them, but I will if I have to....

Maybe Giskard knows, too....

[Guido]

Rob,

L1 and L2 are apparently NLA at JBL. Can you get us DCR's on those two, please?

I hate to tear into my factory crossovers to get them, but I will if I have to....

Maybe Giskard knows, too....

L102 = 0.45 ohms
L1 = 0.15 ohms
L2 = 0.38 ohms

Posted earlier from subwoof

[Robh3606]

Rob,

- Have you done the math that would support Ians' contentions and your assumptions ?

No, this supposed to be fun . Earl you kind of lost me in your example. Where did the 159000 come from??? I run that band pass leg without the 20uf cap and it seems to work as described. That R201 changes the contour on the top end of the range. If I am missing something please poke me in the eye

Rob

[Zilch]

Finding the damn resistors is the hardest part. There's only three of them (2 values, actually,) but if we want truly non-inductive, that's hard to come by.

Wirewounds are inherently inductive, but there's a non-inductive winding method (Aryton-Perry) that would be suitable. That's used in Dale type NH, but I can't find them. The Vishay Dale website is impossible.

Also, there's power thick films, inherently non-inductive, in TO-220 packages, but proper heat-sinking's not part of our typical "screw on a piece of plywood" assembly technology here.

Maybe just use the aluminum chassis-mount type RH. The two values we need are available in 25-Watt for under 3 bucks each.

Using those, the total's below for each, including the expensive switch:



Any crossover builders out there have resistor options or other better choices? The spreadsheet's an image I can update....

[Earl K]

- I'd suggest that one simply & fully disconnect R202 to discover what frequency information is being carried within that "leg" of the circuit.

Where did 159000 come from ?

- I realize that electronics' math isn't "fun" - so here's where that formula originates.

- It's from a normalized scaling applied to the standard formula for finding capacitance :
" C = 1/(2<pi> * f * Xc ) "
( * is my short hand for the multipier sign )
f = desired crossover frequency
Xc= capacitance reactance at the crossover frequency

- This standard formula results in capacitance values in Farads, which isn't entirely useful.

Normalized and scaled to deliver an answer in microfarads ( uf ) the formula becomes ;

" C = 159000/( f*Rr ) "
f = desired crossover frequency
Rr= impedance of the tweeter ( AC ohms at the crossover point )

[4313B]

Dayton Film and Foils with Dayton Metallized Polypropylenes taste bad. Use the AudioCap Theta Film and Foils with the Dayton Metallized Polypropylenes. Also, the non-inductive resistors PE sells are perfectly fine. I'd skip the biamp switches myself.

Try biased passive, it's funner.

Finding the damn resistors is the hardest part. There's only three of them (2 values, actually,) but if we want truly non-inductive, that's hard to come by.


Wirewounds are inherently inductive, but there's a non-inductive winding method (Aryton-Perry) that would be suitable. That's used in Dale type NH, but I can't find them. The Vishay Dale website is impossible.

Also, there's power thick films, inherently non-inductive, in TO-220 packages, but proper heat-sinking's not part of our typical "screw on a piece of plywood" assembly technology here.

Maybe just use the aluminum chassis-mount type RH. The two values we need are available in 25-Watt for under 3 bucks each.

Using those, the total's below for each, including the expensive switch:



Any crossover builders out there have resistor options or other better choices? The spreadsheet's an image I can update....

[Zilch]

Thanks, Giskard. Thetas it will be.

Maybe I look a little more for the Dale NH resistors before doubling up using PE's.

Any suggestions on the inductors, or are they "close enough?"

L1 should be 20 or 22 Ga., looks like, but PE don't offer that.

Shall I hunt more for the 14 and 6 uF MPP capacitors?

A 10 anna 4 for 14 and a 2 anna 4 for the 6, maybe?

Biased passives gonna be another list.

[Robh3606]

"I'd suggest that one simply & fully disconnect R202 to discover what frequency information is being carried within that "leg" of the circuit."

I did I opened up one of the CC networks and cut the leg and the 1.2K-4k response just dropped out by several dB.

Rob???

[4313B]

What about http://www.partsconnexion.com/catalog/resistors.html? Anything there you like? Also maybe something here? http://www.percyaudio.com/

I'd get as close to the schematic values as possible. Buy inductors one size larger and unwind to spec.

Why are you doubling the resistor power handling over stock? The Mills 12 W aren't enough?

[Earl K]

Nice Work Zilch !

Good List I too would recommend "DC-Biasing" as a significant "step-up" that will give better sonic results.

- If money is an issue because of DC biasing, I'd still spend the money on the extra caps ( at least in the horn circuit ) while delaying ( or cutting back ) spending in the UHF contour circuit. ie; I'd build the UHF circuit without R201 ( and possibly L1 ) . If it turned out there really was too much UHF information present ( from a 2426 ? ), then at that time I'd spend the money on those parts.

[Earl K]

I did, I opened up one of the CC networks and cut the leg and the 1.2K-4k response just dropped out by several dB.

Thanks Rob for demonstrating that the "purpose" for that circuit leg ( in which R202 is located ) - is to "restore balance" to the horns' lower midrange frequencies. This particular frequency range has been severely depressed in the other ( UHF contour ) circuit leg.

Of course, if a person knew their own preference for "midrange balance' - then R202 could be omitted & replaced by either 2 or 3 resistors ( depending on if a person wanted an Lpad or Tpad ).

[Zilch]

Why are you doubling the resistor power handling over stock? The Mills 12 W aren't enough?

Because me STOOPID yesterday.

Factory board uses 10 Watt el-cheapos. Percy Audio DOES have the Dale chassis mount ones I was lookin' for, tho....

I've updated the parts list to reflect forum recommendations. Hit "ctrl + refresh" if your cache still says $88.34 total. Giskard spent our money on better stuff. Biased version wouldn't be much more, probably, since it wouldn't require them expensive bypass caps (though it looks like Rob used 'em anyway from his pic, above).

Right click the spreadsheet and hit "Print Picture" for a working hard copy.

I also independently confirmed DCR measurements of the inductors using 3-point resistance measurement at 400 mA.

L1 = 0.143 Ohms
L2 = 0.429 Ohms
L102 = 0.507 Ohms

Subwoof's numbers are good!

I'd say the parts list is good to purchase. Parts Express doesn't list the 7.5 Ohm Mills resistors, but they may have them. If not, go to Parts Connection for both them and the 20 Ohms.

Building for just biamp cuts the cost in half, or if you can do without the biamp option, knock out the switch, C7, and C8 to cut it by a third.

The only item in question is L1 with too low DCR. Maybe put a resistor in series with it, 0.06 Ohms? I do find them in surface-mount metal strip and wirewound versions, about $1.50. That's more like a fuse, actually.

We're lookin' for similar for the AM crossover in the "Q&D" thread, so if anybody has a lead to smaller-wire low vaue inductors, please post it.

I think I'll build these for the updated constant directivity L200's, maybe. I'll have to change the diaphragms in the LE 85's; they likely need it, anyway. I maybe use the aluminum C8R2421, though they're kinda "dear."

Here's what the board actually looks like, including the cheap switch. It's downright silly how these few components make such a HUGE difference. These are revision "E", 1989, from what it says on 'em. Missing components would convert them to N3135 for 4435 :