Hello Eelco
By transparent, I mean invisible. No veil. No loss in dynamics. It's like there's no crossover.
This transparency is no different in amplifiers and preamps. Some amps sound dull, lifeless, clinical, muffled etc. while others exhibit micro detail with crystal clarity coupled with wonderful tones. To some listeners, all this is rubbish. To others, it's everything. So, really depends on the listener.
Hi dave
Presently, the 5 spot freq are 2K to 4K5 in 500Hz intervals. Supplied are also components for half the range, ie from 1K to 2K25.
The crossover can easily be changed to cover lower frequencies like 200-400Hz. I can provide you with the network values if required.
Thanx. I'll have to look at these for my next FAST (it is just starting to gel in my head. Alpair6peN + 2 Foster 12" per side). Can the slopes be changed by just leaving parts out?
dave
Yes, by omitting the resistors, you can get 12dB/oct.
The Alpair6peN + 2 Foster 12" per side sounds like a nice project. I'm working on some boxes for the Tang Band muscle boy that I just bought (W5-1685). A bit of a fun project for the winter.
Michael, I'm not sure if you have seen it, but this link from Rod elliot I think should be of interest with respect the the BSC. Baffle Step Compensation
The line level passive network he has certainly meets the simplicity requirement, though you may need to add an extra buffer potentially.
One of the things that swayed me to going passive was I couldn't see how using an analog active crossover I could easily get the desired acoustic slope, something that I can do using measurements and simulation software with passive.
I guess using a combo of active for the actual crossover and passive for response shaping (eg notches or shelving) is an interesting approach
edit: it took me a while to realise but it seems we named our active crossovers the same, though mine is still yet to see the light of day. I did search at the time to make sure there was nothing already using the name
Tony.
Last edited:
I guess using a combo of active for the actual crossover and passive for response shaping (eg notches or shelving) is an interesting approach
On this note going passive, at line level, between known impedances, using spice and starting with a desired transfer function you may be able to implement all your corrections on a typical driver, whatever they may be in just one stage.
I probably had passive line level on the brain from an earlier post but worth mentioning. Had success with it some years back and I'm sure I'll do it again, probably soon.
I also use a spice assisted passive line level thing for the lower cross in my car, but that was a straightforward decision.
I also use a spice assisted passive line level thing for the lower cross in my car, but that was a straightforward decision.
I use 2nd order HP or LP filters with Q>1 or transformer or tapped inductor approaches to achieve these results. Results are worth it if you can get another half octave frequency extension than you could get otherwise, or fill in a response hole with essentially no penalty. Plus a 6db boost, for example, would make a 100 watt amplifier sound like a 400 watt amplifier in that frequency range - something beyond the realm of active xovers.
thoriated, I've noticed when simulating that I get at times undesirable peaks in the transfer function which I've assumed were due to CL interactions resulting in resonances. I had (perhaps naively) been thinking that this was a quirk of the simulation software that would not manifest in real life implementation.
My current design (yet to be implemented) has 0.3 ohms in series with a cap to damp this in the simulation, but I was thinking I could probably eliminate this in the actual, perhaps not....
Out of interest, I have a nasty dip at 2Khz that I've not been able to deal with (other than with circuits that resulted in drastically low impedance dips at that frequency) could you give an example of a circuit that will give a reasonably high Q boost at 2Khz without dropping everything else? perhaps 4 - 6 db...
I'm confused how you can get a boost with passives, other than with transformers, but even then how to only boost the desired range?
Tony.
My current design (yet to be implemented) has 0.3 ohms in series with a cap to damp this in the simulation, but I was thinking I could probably eliminate this in the actual, perhaps not....
Out of interest, I have a nasty dip at 2Khz that I've not been able to deal with (other than with circuits that resulted in drastically low impedance dips at that frequency) could you give an example of a circuit that will give a reasonably high Q boost at 2Khz without dropping everything else? perhaps 4 - 6 db...
I'm confused how you can get a boost with passives, other than with transformers, but even then how to only boost the desired range?
Tony.
From my first serious loudspeaker I ever designed, I was charmed by the active way of doing this
Especially in multi-way systems, the tuning and adapting all those different efficiencies, can be a pain the passive way. Every time replacing lots of resistors, and then to find out, the new load is more resistive or less, so the load on the filter has changed, and THUS the sound has too...
No, thank you. I just turn a pot, and ready.
Then, the sound of it. I like a powerful, dynamic sound, and that's just what you get with an active system. I design all my power amplifiers just the way they can handle the frequencies they should reproduce in an optimal way.
And, in the "loudspeaker cookbook" in just read that in a 2-way combination, a 60W amp on low and a 30W amp on high equals a "normal" system power of 175 Watts! Well, that's just what I always wondered about: HOW can my small system (built around a few TDA2030A integrated circuits) play SOOO loud without distortion? That was my first system, after someone from a recording studio came to visit me, he heard those speakers and that very moment, they were sold...
I had to build something new the day after
Especially in multi-way systems, the tuning and adapting all those different efficiencies, can be a pain the passive way. Every time replacing lots of resistors, and then to find out, the new load is more resistive or less, so the load on the filter has changed, and THUS the sound has too...
No, thank you. I just turn a pot, and ready.
Then, the sound of it. I like a powerful, dynamic sound, and that's just what you get with an active system. I design all my power amplifiers just the way they can handle the frequencies they should reproduce in an optimal way.
And, in the "loudspeaker cookbook" in just read that in a 2-way combination, a 60W amp on low and a 30W amp on high equals a "normal" system power of 175 Watts! Well, that's just what I always wondered about: HOW can my small system (built around a few TDA2030A integrated circuits) play SOOO loud without distortion? That was my first system, after someone from a recording studio came to visit me, he heard those speakers and that very moment, they were sold...
I had to build something new the day after
And, in the "loudspeaker cookbook" in just read that in a 2-way combination, a 60W amp on low and a 30W amp on high equals a "normal" system power of 175 Watts! Well, that's just what I always wondered about: HOW can my small system (built around a few TDA2030A integrated circuits) play SOOO loud without distortion? That was my first system, after someone from a recording studio came to visit me, he heard those speakers and that very moment, they were sold...
I had to build something new the day after
Passive xovers eat power.
According to Billy Woodman (founder of ATC) complex passives can use up to 9dB. My own passives, which were a very simple 12dB design ie an inductor and a cap for the lo pass, used up 3dB of power.
Depending on the Q of the slope , you can get a boost , then exagerated by a bad acoustic load = boom box
Just to qualify though, inductors and capacitors do not consume power. What goes in also comes out, notwithstanding relatively small losses.
Nevertheless my woofer went from 92dB/1w to 95dB/1w after removing the lo pass from between it and the amp. The lp consisted of a series inductor and a parallel cap, no resistors. Tannoy listed the woofers powerhandling with passive xover as 100w while the driver itself (ie no passive xover) is marked 50w.
This seems unusual to me. The only way a passive crossover will reduce power is through equivalent series resistance. Thus I would have to assume that the inductor you used had a high DC resistance.
On another note, Hybrid Design.
Passive networks can't boost the power, output power is always slightly lower than input power. (EG they're not over-unity...)
However they can boost the output voltage relative to the input voltage via impedance transformation, which supplies more power to the driver at the expense of a lower input impedance seen by the amplifier.
For example the passive 4Khz 18dB/oct butterworth high pass filter driving my ribbon tweeters actually has about 1dB more voltage output above about 6Khz than the input even with losses of the air cored inductors taken into account.
From the perspective of driver sensitivity matching you've achieved a 1dB "gain" in output from a passive network. However it does so at the expense of lowering the input impedance to about 7 ohms, compared to the nominal 8 ohms of the tweeter.
The reason for the (initially unintended) impedance transformation in this particular case is because of the gradual inductive rise in output impedance of the tweeter interacting with the 3rd order filter which is designed for a resistive 8 ohm load.
If I apply a zobel network to the tweeter to flatten the impedance rise, the voltage "gain" disappears, (reverting to a slight loss) and the input impedance rises from 7 ohms to 8 ohms.
However it turns out that in this case the rise in response is a fortuitous match with the overall response of the speaker leading to a flatter response with less components, so I leave it that way.
Any time you have a 2nd order or greater passive filter, you have the opportunity for (possibly unwanted) impedance transformation to occur - usually but not always at the corner frequency of the filter.
For example design a 2nd order butterworth low-pass filter with the right LC ratio for a 4 ohm load but apply an 8 ohm load and there will be peaking at the cutoff frequency where the output voltage is greater than the input voltage, but this will also cause a significant dip in the input impedance below the nominal driver impedance.
Leave the driver disconnected entirely and (assuming no L-Pad) you will get a dead short across the amplifier at the cutoff frequency, as you now basically have a series tuned shunt notch filter. In that sense a 2nd order low-pass/high-pass filter is just a tapped shunt notch filter... (This is why you never want to disconnect an individual driver from a 2nd order or higher crossover while testing other drivers - always disconnect the input of that crossover section, not the driver at the output, or substitute a load resistor in place of the driver)
There are quite a few different arrangements that will give an increase in output over a narrow or broad range.
The point though, is you can't get something for nothing - any increase in output voltage will always be at the expense of a dip in the input impedance. To get a 3dB increase in the output the input impedance would drop to half the actual load impedance of the driver at that frequency.
If the driver has a higher than nominal impedance in that frequency range (such as the top end of most drivers range) that drop in input impedance may not be below the nominal impedance though.
(For example if you were trying to increase the output at the top end of a midrange drivers response where it is falling off and the impedance has risen from 8 to 20 ohms, the reflected input impedance may still only drop to say 10 ohms, whilst still increasing voltage drive to the driver, and overall output)
So yes, although passive networks always have a net power loss, that doesn't mean there is always a net voltage loss between input and output, or that "cut only" filtering is the only option. If you are thinking about driver sensitivity in voltage terms and aren't too worried about significant dips in input impedance modest voltage gains are possible, certainly over narrow frequency ranges.
Last edited:
Thanks Simon! Your example of the tweeter output increasing at higher frequencies is something I also saw with some of my simulations, I discounted it as a bug and didn't proceed with that design, as my tweeter does roll off at higher frequencies perhaps I should have pursued...
The speakers in question already have a low impedance, and as they say, there's no such thing as a free lunch. Unfortunately in my case halving the impedance at 2K will result in a dip to approx 2 ohms at that freq so it would seem short of some active equalisation (or a nasty impedance dip) I have to live with my 2K freq dip.
Tony.
Your example of the tweeter output increasing at higher frequencies is something I also saw with some of my simulations, I discounted it as a bug and didn't proceed with that design, as my tweeter does roll off at higher frequencies perhaps I should have pursued... The speakers in question already have a low impedance, and as they say, there's no such thing as a free lunch. Unfortunately in my case halving the impedance at 2K will result in a dip to approx 2 ohms at that freq
so it would seem short of some active equalisation (or a nasty impedance dip) I have to live with my 2K freq dip. Tony.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.