PDA

View Full Version : Ethyl Merman Crossover Schematic



toddalin
05-01-2019, 05:09 PM
After much listening, tweaking, changing, and even an unintentional mistake that led to the final product, this is the current crossover schematic for the 2241H/2251J/Heil AMT.

The unintentional mistake that led to an unconventional wiring arrangement lets the 2251H sing so clearly that the detail is unreal, probably about the best I've heard, certainly in my home at concert levels. (Perhaps the Harbeths that were here for a day or my LS3/5s sound better at lower volumes, but I've not done a head-to-head comparison.)

If you look at the schematic you see that the 6.8 mfd cap for the 2251 actually attaches to the woofer's positive lead rather than directly to ground. The connection was supposed to go to ground (the other side of the 47 mfd cap), and did in one crossover. But I accidently put the wire on the wrong side of the cap in the other crossover, and in head-to-head comparisons, one was head-over-heels better. (Had to figure out why.) This works with the wedgie in the AMT and the UHF control brings the "air" back in, not that there's not plenty without it. But when I would listen to the 2405 I could hear more "top end sizzle." This takes care of that top end without adding sibilence, and the 2405 looses its advantage here.

http://www.audioheritage.org/photopost/data//500/medium/ScreenHunter_205_May_01_16_51.jpg

toddalin
05-02-2019, 11:22 AM
Actually, can someone model what's going one here? Seems like it could be interesting.

Thanks

toddalin
05-02-2019, 01:03 PM
Truely an interesting situation.

The inductance/impedience of the 2241H woofer is creating a second order filter for the 2251J mid. When the 2241 woofer is disconnected, the slope on the mid appears as a 1st order (on the 61-band RTA) letting the speaker play to <100 Hz petering as it goes down, with some peaks, from ~1kHz. But there is a suck-out/drop-out at ~350 Hz.

When the 2241H is connected, the mid goes to a second (or more) order filter falling like a stone at ~350Hz. But the entire band above this is pushed up flattening it out up to the ~1kHz peak. If the 2241 is replaced by an 8-ohm non-inductive resistor, the mid does not fall like a stone at ~350Hz and largely resembles the curve without the 2241 connected, just falling a bit more at toward bottom end of <100 Hz. The size of the cap (i.e., 6.8 mfd) controls the top end of the 2251J up to its upper crossover point. A bigger cap rolls off the high end sooner. The 6.8 mfd looks about right to get the flattest spectrum and best detail.

The arrangement seems to have little measureable effects on the 2241H and its spectrum looks pretty much the same with or without the additional connection to the 6.8 mfd cap. Perhaps there is a bit more roll-off above the 350 Hz crossover frequency (notable at ~500Hz and up) which is a welcome trait.


Recognize that unlike the 2241H, the 2251J has no inductance (differential drive with a neodynium magnet) and any loss to the upper frequencies of the 2241 is from the combination of the 6.8 mfd through the 16 ohm load (of the L-pad/2251J) to ground which is very minor when compared to the 47 mfd dumping to ground directly.

Robh3606
05-04-2019, 01:57 PM
Here is what its doing

Rob:)

toddalin
05-04-2019, 02:40 PM
Here is what its doing

Rob:)

Thanks Rob!

I don't think that you are considering the effects of the inductance of the 2241H (R2) on the 2251J and that it's slope is steeper than shown. This inductance seems to change the slope from first to second order as shown on the RTA. The 2241H has an Le of 1.75.

Also, shouldn't the 0.9 mH choke cause the 2251 to roll off at the top end rather than tilt upwards? Seems to in practice.

Robh3606
05-05-2019, 07:22 AM
Hello Toddalin

You are right that inductor is not in the mix. Either I have a wrong value or an open node I will re-run and re-post. I can't model the 2241 without an impedance and phase curve that's why it's all with simple loads.

Rob:)

toddalin
05-05-2019, 12:00 PM
Thanks Rob.

I think you have the 0.9 mH as 900 mH (or such).

Note that the 12 mfd on the Heil actually pulls it quite low into the 2251 territory. This was to get the best transition based on actual conditions rather than modeling. But reminents of the signal were still present down to <600 Hz in the Heil, and while this is no strain on the Heil, it is unnecessary to burden it with this load.

I added 0.6 mH to gnd to make this a 2nd order crossover. Little changes up at the transistion, but there is was a bump ~2000-2,500 Hz that is smoothed where the drivers overlap, and the increased slope pulls the all traces of lower midrange out of the Heil. Perhaps the 12 mfd can now be reduced to roughly half this value.

Also, 0.68 mfd on the UHF boost was too bright (and will probably be brighter with the Claritycaps when they arrive). The addition of the 5 ohms is like a "loudness" contour for the highs. When the tweeter is turned down at the crossover it turns down the UHF at a lesser rate. But when the L-pad is all the way up, this only adds about 0.2 dB at the high end..., if that. (Hard to reliably measure such small changes due to the nature of pink noise.)

"We'll listen and see."

Revised schematic is included.
http://www.audioheritage.org/photopost/data//500/medium/ScreenHunter_207_May_05_11_43.jpg

Robh3606
05-05-2019, 12:26 PM
Hello Toddalin

It's 900 u so same as .9Mh if I had a 900Mh there would be nothing at all in that leg.

Rob:)

toddalin
05-05-2019, 12:36 PM
Hello Toddalin

It's 900 u so same as .9Mh if I had a 900Mh there would be nothing at all in that leg.

Rob:)

Would still come through the resistor at a reduced level.

Robh3606
05-05-2019, 12:46 PM
You will get something but way down in level.

Rob:)

toddalin
08-01-2019, 10:57 PM
Latest and Greatest!

http://www.audioheritage.org/photopost/data//500/medium/ScreenHunter_211_Jul_30_11_40.jpg

Joni Mitchell doing "Woodstock" on the Ethyl Mermans. This is why I call them the Ethyl Mermans..., they do females so well and can do it VERY LOUD if desired.

Joni is my favorite female vocalist and has one of the hardest voices to get right, and this is the best I've been able to do, or hear for that matter.

Again, this is just a CD directly through the Yamaha (room warts and all) with no eq, digital processing, or subs added.

https://youtu.be/-fMHV5G33PI

DogBox
08-05-2019, 09:40 PM
Latest and Greatest!.... (room warts and all) with no eq, digital processing, or subs added.

https://youtu.be/-fMHV5G33PI

Sounded really nice and 'did' have that "live sound" - even through YouTube and my 'nothing' computer speakers [Logitech 2.1]

Did you design this crossover yourself or 'borrow' the design and tweak to your own needs..? I don't quite understand how that 'Cap' combines

the LF and MF..?!! Interesting! :blink:

Well Done! :applaud:

Steve [DogBox]

toddalin
08-05-2019, 10:53 PM
Yes, designed them myself with lots of tweaking and some dumb-azz luck due to a miswiring that sounded "right" (the black art everyone talks about).

I don't quite understand how the 6.8 through the 0.065 through the woofer circuit combine of the 2251, but they certainly do! They were supposed to go to ground, but one was miswired and sounded better that way. I changed teh other and from there is was tuning with the cap and choke.

I know that lowering/increasing the value of the 6.8 cap increases/rolls off the ~4KHz to the 2251 and the 0.065 choke removes what's left over (from the RTA). Also, if the 2241 is disconnected, or changed to a resistor, it's inductance is lost from the mid changing the mid's slope.