Except for the fact that most systems have the horn and bass speaker mounted in the same enclosure and are usually flat with the front I don't see any reason why with a separate horn like I am seeing here you are not at least moving the horn forward so that the voicecoils are in a vertical line? This would take care of the time misalignment in at least one plane, obviously only a coaxial can get that right in the vertical plane but would ease the design of the cross-over network.
On the subject of the inductors in the network for a high frequency device I don't see any real issue with using the Erse cored inductors as they are usually under a lower loading and are typically in the shunt position. For a bass speaker where you can easily saturate the core at high power and they are in the signal path I would go with an air-cored inductor for that position. You can incorporate the dcr of those inductors if you are designing a conjugate network to flatten the impedance rise of the devices.
Just my 2 cents while watching the thread.
On the subject of the inductors in the network for a high frequency device I don't see any real issue with using the Erse cored inductors as they are usually under a lower loading and are typically in the shunt position. For a bass speaker where you can easily saturate the core at high power and they are in the signal path I would go with an air-cored inductor for that position. You can incorporate the dcr of those inductors if you are designing a conjugate network to flatten the impedance rise of the devices.
Just my 2 cents while watching the thread.
When the system is lined up (drivers in-phase, and coherent impulse response), the woofer voice coil is seven to twelve inches in front of the compression driver diaphragm. The reason is twofold: the woofer has its own inherent lowpass function thanks to voice coil inductance and the HF rolloff of the cone, which is then summed with the lowpass function of the crossover (the crossover is the dominant delay). The combined effect of both lowpass functions is to push the woofer back in space.
If the voice coils are visually lined up, the woofer will be out-of-sync with the compression driver by nearly a millisecond, which is a big deal in the context of a 700 Hz crossover. What matters is not optical coincidence, but acoustical coincidence, and this is determined by measurement.
I was surprised when I was working with the first rough prototype that small front-to-back errors, on the order of 1/2" (12mm) were clearly audible, despite having only a very minor effect on the vertical polar pattern. Fortunately, the subjective impression of "coherence" matched up pretty well with the most compact and best-looking impulse response.
Considering that quite small errors were audible, I didn't even audition much grosser errors, like several inches. There are commercial systems that actually have two discrete peaks in the impulse response, but that's not what I wanted to design.
Last edited:
I've been using Bessel/Gaussian/LR-family crossovers since the late Seventies. My first vented system, the Audionics T-52, used a Bessel vented alignment, instead of the more common QB3 or B4. These low-Q alignments are less sensitive to amplifier source impedance issues, as well as dynamic Qts and FR variations in the drivers. There's substantially less overshoot in the time domain, as well.
They don't have the steep attenuation of the classical Butterworth alignments, but they still have much more than 1st-order crossovers, which are not that good for the tweeter (too much LF energy). It's a compromise approach that works for me; other folks go for 1st-order (which I avoid), and others go for classical filters (which I occasionally use if I really need the out-of-band attenuation). I try to balance inter-driver phase matching (within 5 degrees), out-of-band attenuation, and time-to-decay (less than 0.5 to 1 mSec preferred), in about that order.
The modern prosound approach uses very steep digital filters and FIR correction, but this system is not intended for multi-amp, digital-crossover applications. (Although it certainly could be used that way.)
They don't have the steep attenuation of the classical Butterworth alignments, but they still have much more than 1st-order crossovers, which are not that good for the tweeter (too much LF energy). It's a compromise approach that works for me; other folks go for 1st-order (which I avoid), and others go for classical filters (which I occasionally use if I really need the out-of-band attenuation). I try to balance inter-driver phase matching (within 5 degrees), out-of-band attenuation, and time-to-decay (less than 0.5 to 1 mSec preferred), in about that order.
The modern prosound approach uses very steep digital filters and FIR correction, but this system is not intended for multi-amp, digital-crossover applications. (Although it certainly could be used that way.)
Last edited:
You need a DC blocing capacitor before the autoformer. This will lose some of the subtlety in the sound. It is needed to avoid saturation of the core in a mumetal HF autoformer. If the amp has no DC on the output you can then clearly leave the cap out. If you use an ordinary silicon steel cored autoformer it will roll the treble a bit early. I prefer to avoid auto transformers.
Thanks again Lynn. I am a fan of the L/R Sallen Key network designs and usually will go with 4th order filters. I am not a fan of 1st order for anything that I have developed and I have issues with second order and phase alignment. I am moving away from passive networks besides as a conjugate network to correct impedance rise in devices and am on to active networks before multiple amplification. No network besides the conjugate between the amp and device.
ps. I understand your desire to remain n the vacuum tube realm and using high efficiency devices to allow that to happen. I will never be a fan of SE tube devices though, I just think of these as tone controlled devices unless you can get away with extremely low output power. Fine for a compression driver on a horn with 110+ db efficiency but not on any direct radiator device.
ps. I understand your desire to remain n the vacuum tube realm and using high efficiency devices to allow that to happen. I will never be a fan of SE tube devices though, I just think of these as tone controlled devices unless you can get away with extremely low output power. Fine for a compression driver on a horn with 110+ db efficiency but not on any direct radiator device.
Last edited:
Not discussing power amplifiers, but using autoformers or transformers as an attenuator for very efficient compression drivers. Note the swamping resistor prevents the LC of the input cap and the inductance of the autoformer from creating a LF peak. The swamping resistor is an essential part of the circuit, since it creates the effective load for the HP filter, and the very small source impedance as seen by the compression driver. It's also where the majority of the HF current is dissipated.
I've briefly wrestled with parallel-feed (parafeed) transformers in SET amplifiers, and I avoid them. The LC circuit is prone to LF peaking, so the selection of cap values is restricted to a fairly narrow range that matches the inductance of the transformer. Since I don't use or design SET amplifiers, it's not something I worry about.
Last edited:
The game of getting the right balance between undesireable resonances or over damping is tricky but perhaps well worth the effort, if you experiment with some of the exellently engineered autoformers. I guess the larger ones sound better because they will take more DC offset or low frequency with less or no saturation. So it does make it a costly way but very good for those that set it up right.
My pleasure. The amplifier matter was solved through the course of a decade with my previous speakers. They took me from from Class B integrated amps (many brands), to Class A solid state amps following Nelson Pass designs: first the A75, then the First Watt F5 and presently the First Watt F4 (much better IMO). Now I feel no need to upgrade. Along the way I realized that the line and phono preamps are just as important, if not more, and upgraded these too. My brother has high end tube amplification, and it will be fun hearing how the speakers sound at his place. This will occur later this summer.
So I don't feel frustrated by my present audio hardware chain. However… Whereas well recorded and well engineered records sound very realistic (classical, rock and jazz alike), sometimes the limitations of the tape masters or poor engineering choices are very obvious. While playing some records, my brother and I believed at first that we had found a fault in the speaker design, only to realize shortly afterwards that they were revealing a limitation of the source material… As a result, I am now making room by disposing of records with such shortcomings that they are "not worth keeping". On the bright side, I understand better the fuss around some records that sounded good on my system before, and now sound spectacular through these speakers.
Indeed, at least for the two way with beryllium diaphragms… I have never heard speakers that could image better. Left to right extending beyond the speakers, depth behind the speakers, and… dare I say… subjective height. Listening to some organ records, one can visualize where the pipes are located. We found (like you) that vertical clearance below the horn contributes to this.
Also, although the speakers are physically large, they sound very detailed even when played at low volume. Every musical instrument, detail, ambiance remains there: only scaled down in volume. My previous speakers could never do this.
A most worthwhile read indeed. See also this nice compilation by Jimbee:
filtre "quasi-optimal" : modèles et variantes - Le blog de jimbee
Never thought that this vertical clearance would matter much, but it does!
I must admit that looks matter to me, and therefore I did not want the horn being recessed too much behind, or sticking out too much in front of the woofer baffle. I landed "flush" as I wanted with the throat-less Radians, but not so with the Yamahas: the horn sticks out by a couple of inches - the length of this driver's throat.
In the end, for the type of music listening I enjoy, results exceed expectations.
Pierre
Last edited:
A modern high speed op amp will do less damage to the music then a speaker level transformer , cap, resistor - multi-amping is the way to go
Maybe. But it will not let the driver interact on one amplifier. I think it sounds more correct, but more dull or dry with active XO's. Passive speaker level XO can come alive in a way, even if they are as poorly built as my first attempts have been for sure.
I refer to Behringer CX3400 (analog) and Mindsp 4x10HD (digital).
Last edited:
That's a excellent point about the ability to use a single amp verses being forced to use multiple amplifiers and the additional complexity of active crossovers, etc., to achieve somewhat similiar results.
Best Regards,
TerryO
The Behinger sounds really bad
, i never had the mindsp. I have like 200 lbs of inductors, autoformers, caps, resistors ect and sure it can sound good but never great like PLLXO or a good electronic crossover. In some cases where you try and stretch the bandwidth out of a compression driver (too low or too high) you have to resort to those pesky passive components - I try not to get caught in that mistake after tears of listening to those compromises
POOH,
When you start to get into having to add slope to a network and notch filters and such to a passive crossover things start to get rather difficult to say the least. Dealing with phase shifts and time alignment in the passive realm is not for the average or even above average diy builder. Then we need to consider the power loss in all those passive components as the network becomes more sophisticated. I see nothing wrong with an active electronic network working in the analog realm. I am not ready to go over to the dark side of digital filtering, I just don't have enough knowledge to know the pitfalls of that approach.
When you start to get into having to add slope to a network and notch filters and such to a passive crossover things start to get rather difficult to say the least. Dealing with phase shifts and time alignment in the passive realm is not for the average or even above average diy builder. Then we need to consider the power loss in all those passive components as the network becomes more sophisticated. I see nothing wrong with an active electronic network working in the analog realm. I am not ready to go over to the dark side of digital filtering, I just don't have enough knowledge to know the pitfalls of that approach.
Actually, after nearly 20 years on various Audio Forums and Internet Interest Groups, I'd say this thread has actually stuck closely to the original subject. They all drift a bit from time to time, but this one always seems to return to topic.
Best Regards,
TerryO
I think your right and when it does drift it's because there are valid angles to think about. Great stuff.
The Behinger sounds really bad
I know the CX3400 is not the best. I can control the filters pretty good, but I don't see exactly what frequency I set it on. Another problem except from the dull sounding components, is the only option of LR24dB filtes. Still sounds more accurate than if I try 2nd order or higher in passive speaker level XO.
The Minidsp is very accurate and fun to play around with but has a slightly toy-ish sound to it, but is still very good. A problem is that it produces a "tshhh"-noise together with a 2A3 tube amp and I need a RCA isolation transformer to remove most of the noise, and the trafo is made for car subwoofers, so the top end is awkwardly rolled off. Until I get a better crossover I don't mind.
From what I have read PLLXO needs to be customized to the amp and cannot be made above 1st or 2nd order, without making it impossibly complicated. I would need a steeper slope than that at 250Hz.
What is a good electronic crossover? Found a 1000 dollar Marchand on ebay yesterday. 1000 dollars is a bit too much, and since this is a DIY forum, I would prefer to build it myself.
Last edited:
Off the shelf used try the TDM, they are excellent. The Marchand modules are really inexpensive and you can just get the boards and use your own components - opamps, resistors, caps ect