Lifting the Veil from the 2405

[toddalin]

When I built the crossovers in my modified L200/300s, the system was using LE175s on the short horns and 075s as the tweeters. I modified the L200 Studio Master and we came up with "Zilch tested and approved" the "Keeper Crossovers."

These used the a variation of the N7000/N8000 circuit for the high frequencies substituting a 0.5mH choke in place of the 0.7mH choke in the N7000 and 0.2 mH choke in the N8000. Let's call it a second order ~7,300 Hz.

The problem was that the 075 was so "hot" that I could never turn the L-pad up past ~1/4-1/3 of the way without the sound becoming too bright. At Zilch recommendation, I added the two resistors as a voltage divider (don't recall the values), and this knocked the 075s down to where the L-pad became useful throughout its range.

The 075s were later replaced by 2405s. Of course these are much quieter than the 075s and with the crossover components now hot glued in place on the circuit boards, it would be difficult to remove the voltage divider. I would typically run with the 2405 "wide open" just to get a decent match.

But with lots of listening and A/Bing with other speakers, I knew that I was missing the high highs and even though the voltage dividers should attenuate all frequencies similarly. Particually what was missing was the "tink" of the ride cybols, etc. Yes, I know that you hear it with the 2405, but until you can instaneously compare them with other offerings, you just don't know what you are missing.

I wanted a circuit that I could apply externally to the speakers (I can access the driver wires from the back at the crossovers) without opening them up, at least to develop it.

This is what I came up with. It adds a very high pass, 1st order filter to the existing 2nd order filter to flatten out the higher range. Because the 1st order is connected directly to the speaker, it doesn't pass through the L-pad windings and thus attenuates at a much slower rate. The 2nd order attenuates because the L-pad selects less voltage to go to the speaker AND the speaker goes to ground. The 1st order only attenuates because the speaker goes to ground.

All caps are Solen Fast Caps with AuraCap Theta 0.10 By-pass caps riding piggyback. All resistors are Mills.

[ivica]

When I built the crossovers in my modified L200/300s, the system was using LE175s on the short horns and 075s as the tweeters. I modified the L200 Studio Master and we came up with "Zilch tested and approved" the "Keeper Crossovers."

These used the a variation of the N7000/N8000 circuit for the high frequencies substituting a 0.5mH choke in place of the 0.7mH choke in the N7000 and 0.2 mH choke in the N8000. Let's call it a second order ~7,300 Hz.

The problem was that the 075 was so "hot" that I could never turn the L-pad up past ~1/4-1/3 of the way without the sound becoming too bright. At Zilch recommendation, I added the two resistors as a voltage divider (don't recall the values), and this knocked the 075s down to where the L-pad became useful throughout its range.

The 075s were later replaced by 2405s. Of course these are much quieter than the 075s and with the crossover components now hot glued in place on the circuit boards, it would be difficult to remove the voltage divider. I would typically run with the 2405 "wide open" just to get a decent match.

But with lots of listening and A/Bing with other speakers, I knew that I was missing the high highs and even though the voltage dividers should attenuate all frequencies similarly. Particually what was missing was the "tink" of the ride cybols, etc. Yes, I know that you hear it with the 2405, but until you can instaneously compare them with other offerings, you just don't know what you are missing.

I wanted a circuit that I could apply externally to the speakers (I can access the driver wires from the back at the crossovers) without opening them up, at least to develop it.

This is what I came up with. It adds a very high pass, 1st order filter to the existing 2nd order filter to flatten out the higher range. Because the 1st order is connected directly to the speaker, it doesn't pass through the L-pad windings and thus attenuates at a much slower rate. The 2nd order attenuates because the L-pad selects less voltage to go to the speaker AND the speaker goes to ground. The 1st order only attenuates because the speaker goes to ground.

All caps are Solen Fast Caps with AuraCap Theta 0.10 By-pass caps riding piggyback. All resistors are Mills.

Hi todd,
I think R1 has to be on the other side of R2
L1 about 0.1mH, so maximum round 21kHz
no idea for R3, R4, say R3=R4=10.....
but using 2405 i think that it would be better to use one capacitor more, C3, between L2 and R3 saying C1=1uF, C3=2uF, L2=0.16mH....

regards
ivica

[toddalin]

Hi,

Thanks for looking. I've tried both ways and it seems to make no difference on the Berringer 61-band and the ear can't readily hear a difference.

You also need to consider that the 2nd order though the L-pad also goes to ground through R1, so R2 adds another 2 ohms of isolation between the two stages that wouldn't be there if R1 was on the other side (my original throught).

In fact, I was worried that when the L-pad is turned to "0" that this would short out the amp. But the 0.56 mfd cap takes care of this and it has not been a problem, even without a resistor. When I tried a 1.2 ohm value here, I kept coming back to the 2 ohm value as it sounded a bit better to my ear. If someone is concerned, the 2 ohm value for R2 can be increased considerably without consequence. You need at least 8 ohms in here before you see any real difference in dB on the meter.

R3 and R4 are already in place and will be determined after these other values are finalized and the box is opened.

Adding cap C3 is essentally the L300 network except that it would be 1.5 mfd and C1 would be 1 mfd and IIRC, the choke is 0.18mH. This creates a spike in the area of the depression that is inherent in the ring radiators. But it also further reduces the volume and would require that I go into the networks and remove the hot glue that basically "pots" the network. Then there are those (e.g., Nelson Pass) that feel that the tweeter sounds more natual without the double cap spike.

I am trying to add volume in the upper reaches while "thinking outside of the box" both figureatively and literally.

[ivica]

Hi,

Thanks for looking. I've tried both ways and it seems to make no difference on the Berringer 61-band and the ear can't readily hear a difference.

You also need to consider that the 2nd order though the L-pad also goes to ground through R1, so R2 adds another 2 ohms of isolation between the two stages that wouldn't be there if R1 was on the other side (my original throught).

In fact, I was worried that when the L-pad is turned to "0" that this would short out the amp. But the 0.56 mfd cap takes care of this and it has not been a problem, even without a resistor. When I tried a 1.2 ohm value here, I kept coming back to the 2 ohm value as it sounded a bit better to my ear. If someone is concerned, the 2 ohm value for R2 can be increased considerably without consequence. You need at least 8 ohms in here before you see any real difference in dB on the meter.

R3 and R4 are already in place and will be determined after these other values are finalized and the box is opened.

Adding cap C3 is essentally the L300 network except that it would be 1.5 mfd and C1 would be 1 mfd and IIRC, the choke is 0.18mH. This creates a spike in the area of the depression that is inherent in the ring radiators. But it also further reduces the volume and would require that I go into the networks and remove the hot glue that basically "pots" the network. Then there are those (e.g., Nelson Pass) that feel that the tweeter sounds more natual without the double cap spike.

I am trying to add volume in the upper reaches while "thinking outside of the box" both figureatively and literally.

Hi todd,

Yesterday I have done some network simulation, and it has shown that LF attenuation is not so good for 2405 (behaved almost as -26dB/dec as the first order network, so I have suggested C3 to be introduced).
Using JBL standard 2405 network 1uF, 1.8uF, 0.12mH, and adding about the same L=0.12mH 'in-series' to the rest (L_pad and attenuation) almost the same HF peaking round 20kHz can be get.....

regards
ivica

[toddalin]

Thanks,

It would be an interesting excercize for someone with an N7000/N8000/equivelent on the 2402/2404/2405 to add this circuit to see the outcome recognizing that there would be some tweaking involved with the resistor values and turning of the L-pad to get the best sound.

Again, lots of people have access to the amplifier input to the speaker box and the wire output to the drivers/tweeter, but nothing else. This saves getting into the box and messing with the existing circuit, which may be potted.

Certainly simple enough to try for some who probably have all of these parts lying around.

[Ian Mackenzie]

I played around with it in LEAP using 2405 modelling amplitude and impedance minimum phase transformation.

Its a touchy little sucker but you can get one nice bump instead of spikes

Seeing as you are using the 2405 with another driver it's best to see how they work together and you seem to have done that subjective. That's the only way.

Let me know if you want to see the curve

[toddalin]

Still toying with the values a bit. While the 2.0 ohm resistor seems to help with the vocals (brings them forward and adds intelligability) and adds a bit more high end than a greater resistance value in this position, it can also be a bit harsh on some tracks that already have a preponderance of sound in this range.

Replacing this with 3.2 ohms takes away the vocal coloration while still bringing in the high highs and lifting the veil. And as per the doctor's creedo, "First do no harm."

I also have a trio (L,C,R) of higher value chokes to try.

Ian, I would love to see how this models out.

[ivica]

Still toying with the values a bit. While the 2.0 ohm resistor seems to help with the vocals (brings them forward and adds intelligability) and adds a bit more high end than a greater resistance value in this position, it can also be a bit harsh on some tracks that already have a preponderance of sound in this range.

Replacing this with 3.2 ohms takes away the vocal coloration while still bringing in the high highs and lifting the veil. And as per the doctor's creedo, "First do no harm."

I also have a trio (L,C,R) of higher value chokes to try.

Ian, I would love to see how this models out.

Hi todd,

expecting that the voice has something over 8~9kHz , I think, it is very optimistic, so such spectrum (voice) I would put into VHF driver, and mid-bass section too, and so I would suggest to use 2405 over 9kHz (or even over 10kH) (as JBL done in 4344, 4345..) and may be some correction of its response can be done putting some UHF bump over 15kHz (even though most of us would not hear such high notes) , as You have done, or I have suggested to use small choke (round 0.1 ~ 0.15 mH, if 4344~4345 type of 3rd order network is used). Using not sharp filter (as You have done) for 2405 in the multi-way speaker where 24xx drivers with relatively long horn are used , comb filter effect would be very pronounced (due to the relative large position offset between the 2405 and VHF driver).
Using active digital networks with the programmable delays such com filer effect can be diminished, bit if such network is used, then other EQ can be arranged too.

regards
ivica

[toddalin]

Certainly we can, and do use eq, but I'm trying to get the best sound using the "direct" connection that turns off all digital processing and panel lights.

I put the other chokes on, (looks to be about 30 turns of #18 (maybe #16) gauge around a toroide ferrite core), and this took a bit of really high frequencies (~18-20K as seen on the Behringer), and harshness, out such that I can use the 2 ohm value.

[Ian Mackenzie]

Here is simulation

Red is the raw driver
Brown is the Toddalin crossover and blue is an adoption of the 3134 crossover from the 4430.

I found a lot more stability in adjustment of values and shaping of the curve and level shifting due to the impedance of the 3134 schematic aka the value of R20

The values chosen givevthe smoothest response in each circuit

R21/R22 is the level shift below 10000 hz
This can be replaced by an 8 ohm Load

R23 is an overall level shift

R19 is the amplitude peak of the series resonance L3 and R19
R2 can be half of an L pad

If you want to try other values sing out

I think this is a cool diy project

The curves are 5 Db per division

[toddalin]

The image is very close to what I see on the Behringer and you can see how this brings out detail and "air!"

But recognize that the two circuits "interact" differently though the L-pad as they blend. Is there any way to put that into the model?

[Ian Mackenzie]

Zactly

It's an be done

If you sketch the horn filter l can add that in if you like later
Are you using a short horn?


Green is the same fixed pad R16 and R17

I have increased R14 and R15 from -5 Db to -10 Db

Off for some Zzzs
4.45am from my iPhone

[toddalin]

Zactly

It's an be done

If you sketch the horn filter l can add that in if you like later
Are you using a short horn?


Green is the same fixed pad R16 and R17

I have increased R14 and R15 from -5 Db to -10 Db

Off for some Zzzs
4.45am from my iPhone

Thanks,


I use tangential 2420s on the longer (L300) horns.

[toddalin]

Zactly

It's an be done

If you sketch the horn filter l can add that in if you like later
Are you using a short horn?


Green is the same fixed pad R16 and R17

I have increased R14 and R15 from -5 Db to -10 Db

Off for some Zzzs
4.45am from my iPhone

Thanks,

I found my schematic on-line. Use 0.5 mH for L3 and L4 chokes and increase the two 8.2 mfd (16.4 mfd) to 18.5 mfd total. Also connect C7 directly to the hot input and I think move the MF L-pad to the "other side" of C6. And this doesn't include the voltage divider I added between L4 and the L-pad to reduce the volume of the the 2402 to where the L-pad is "usable." I don't recall those values and won't know until I open a cabinet, but the reduction was certainly over 6 dB.

[toddalin]

Ended up using a resistor value of 3.2 ohms for R2 of the new addition circuit. I think the integration is more seamless.

Been doing lots of listening, especially to SACDs. I can't say that these have ever sounded this good before! It truely is like taking a cover off of the tweeter to reveal new detail and clarity.


If there is anyone in the So Cal area who would like to hear the difference while these are still "out of the box" so I can instantaneously A/B them at the flip of a switch, just let me know. But don't wait too long because once they go into the cabinets, this will no longer be possible.