Perry, (or anyone else who knows the answer), I had a question about how these speakers are powered. I had noticed many times that you stated they needed to be bi-amped, but I always assumed that meant that the 4 big woofers would be powered by one amp and the other would be for the 2 midrange drivers and the 4 tweeters. According to your schematic, one amp is doing all the larger drivers, and the other is just powering the 4 tweeters. From a power consumption standpoint that obviously doesn't make sense, as the tweeters would require next to no power in relation to the big woofers. Not being a speaker designer myself, I know there is a reason why they are set up this way, but why? I have also seen you mention tri-amping them - which if the tweeters NEED to be powered by a different amp I think I might consider. That way I could use some sort of solid state or newer style class D amp for the woofers, my best tube amp for the midrange drivers, and maybe a super lower power 300b tube amp or something for the tweeter section. If I were to do that, would I be able to eliminate a significant portion of the crossover components? Would the zobel network change at all? Please excuse my ignorance - I'm great at building speakers - but a speaker designer I am not.
Great Question @Subsonic1050. Part of the answer is here
https://www.diyaudio.com/community/...ompetition-updated-design.405104/post-7499720
You are right, in both Bitches Brew and Live Edge Dipoles, one amp is powering the woofers and mids and the other amp is powering the tweeters.
This choice has nothing to do with amplifier power consumption, and everything to do with achieving linear phase in the crossovers, which is rarely attempted even now with DSP. I think linear phase pays off in better stereo imaging and higher resolution. The tweeters do need their own amp no matter what and I wouldn't recommend doing either design 100% passive.
As you point out, you can use a MUCH lower power amp for the tweeter section. Even a tiny flea watt SET tube amp on the tweeters will probably put out plenty of SPL for most people, because the sensitivity is 95-100dB. (You could even wire the tweeters a little differently and get even more SPL if you needed it.) And yes you could use a solid state or digital amp for the woofers.
You can eliminate a significant number of the crossover components by omitting the zobel networks. The reason the zobels are there is:
1) If you use tube amps, they minimize the variability that the high output impedance of tubes imposes on the response
2) I have a Resistor-Inductor "tweeter zobel" on the Bitches Brew because I found that my solid state tweeter amps made a funny noise when I switched them off without it. The amp didn't like the capacitive load.
The only way to eliminate the passive components entirely is to triamp and use 6 channels of DSP instead of 4. I agree that option is better than the current design; I just didn't want that additional complexity. I felt the series LC network between the woofers and mids was the most elegant solution.
https://www.diyaudio.com/community/...ompetition-updated-design.405104/post-7499720
You are right, in both Bitches Brew and Live Edge Dipoles, one amp is powering the woofers and mids and the other amp is powering the tweeters.
This choice has nothing to do with amplifier power consumption, and everything to do with achieving linear phase in the crossovers, which is rarely attempted even now with DSP. I think linear phase pays off in better stereo imaging and higher resolution. The tweeters do need their own amp no matter what and I wouldn't recommend doing either design 100% passive.
As you point out, you can use a MUCH lower power amp for the tweeter section. Even a tiny flea watt SET tube amp on the tweeters will probably put out plenty of SPL for most people, because the sensitivity is 95-100dB. (You could even wire the tweeters a little differently and get even more SPL if you needed it.) And yes you could use a solid state or digital amp for the woofers.
You can eliminate a significant number of the crossover components by omitting the zobel networks. The reason the zobels are there is:
1) If you use tube amps, they minimize the variability that the high output impedance of tubes imposes on the response
2) I have a Resistor-Inductor "tweeter zobel" on the Bitches Brew because I found that my solid state tweeter amps made a funny noise when I switched them off without it. The amp didn't like the capacitive load.
The only way to eliminate the passive components entirely is to triamp and use 6 channels of DSP instead of 4. I agree that option is better than the current design; I just didn't want that additional complexity. I felt the series LC network between the woofers and mids was the most elegant solution.
Last edited:
Thanks so much for the prompt response Perry. Your knowledge on this subject is amazing. I have no problem with spending time or money to get the most out of this design. My general philosophy with "things in the signal path" is to use the highest quality parts I can, and wherever possible to eliminate things in the signal path entirely. Since I'm already going to have DSP in the signal path, I might as well try to eliminate the passive components if I can? Why are you saying that adds additional complexity? Again, I know there must be a reason why you say this, but eliminating all these parts seems to simplify things rather than complicate them. Do you say that because now you are adding another amplifier into the mix or because programming the 2 additional channels of DSP is very difficult?
I already have a MiniDSP SHD, which unfortunately only does 4 channel. But for the money I'd save on crossover components I could buy a Flex Eight which I assume would do the job. If I were to try the tri-amp option, I assume I still need the crossover components in the tweeter network because there are still 2 tweeters in the circuit? If I were to use a solid state amp for the woofers, a fairly powerful tube amp for the midrange, and another tube amp for the tweeters would I still need the zobel network?
Sorry for all the questions, but I'm pretty intrigued by the idea of this tri-amp setup. I'd like to take this design as far as it can go.
I already have a MiniDSP SHD, which unfortunately only does 4 channel. But for the money I'd save on crossover components I could buy a Flex Eight which I assume would do the job. If I were to try the tri-amp option, I assume I still need the crossover components in the tweeter network because there are still 2 tweeters in the circuit? If I were to use a solid state amp for the woofers, a fairly powerful tube amp for the midrange, and another tube amp for the tweeters would I still need the zobel network?
Sorry for all the questions, but I'm pretty intrigued by the idea of this tri-amp setup. I'd like to take this design as far as it can go.
I say the passive option is simpler because for two channels, you can achieve a complete crossover with two inductors and two capacitors. That’s in contrast to two extra DSP channels and two extra amp channels with cables and connectors. (Passive is not necessarily cheaper, however!)
If you’re comfortable with DSP, then programming a third channel is no big deal, and you just mimic the slopes that are already in the passive crossover, which you can see in the threads that I have linked to here. Basically that means a 6 dB low pass on the subwoofers and a steep low-cut shelf filter to reduce bass signals to the midrange.
If you Triamp, you still need the DC protection capacitors for the tweeters, and you still need the notch circuit for the rear tweeter.
There’s certainly no law that you need to use a Zobel, but the equalization differences between a tube and a transistor amp will be smaller if you do use it.
The vast majority of the amp power is consumed by the 14 dB bass boost at 25 Hz for the 15OB350 subwoofers.
PS I’m driving my Bitches Brew tweeters with a 200 watt amp. I bet the levels never get above 1 watt in real life. But they sound great 🌞
If you’re comfortable with DSP, then programming a third channel is no big deal, and you just mimic the slopes that are already in the passive crossover, which you can see in the threads that I have linked to here. Basically that means a 6 dB low pass on the subwoofers and a steep low-cut shelf filter to reduce bass signals to the midrange.
If you Triamp, you still need the DC protection capacitors for the tweeters, and you still need the notch circuit for the rear tweeter.
There’s certainly no law that you need to use a Zobel, but the equalization differences between a tube and a transistor amp will be smaller if you do use it.
The vast majority of the amp power is consumed by the 14 dB bass boost at 25 Hz for the 15OB350 subwoofers.
PS I’m driving my Bitches Brew tweeters with a 200 watt amp. I bet the levels never get above 1 watt in real life. But they sound great 🌞
Last edited:
Also, the design has a six ohm resistor in series with the front tweeter. That reduces the level of the front tweeter compared to the back and makes the sound more spacious.
If you used FOUR channels of DSP on each side, and 4 amplifiers, you could use a separate amp for the rear tweeter, and you could get rid of the six ohm resistor and the notch circuit on the rear tweeter. And you could control the level of the rear firing tweeters through the DSP.
Quad-amp.
Maybe that sounds like overkill, but you could do some very interesting measurements of reflected sound by turning off the front tweeters and EQ the rear firing sound any way you want.
My inner geek thinks that’s all very exciting; it’s also very complex.
If you used FOUR channels of DSP on each side, and 4 amplifiers, you could use a separate amp for the rear tweeter, and you could get rid of the six ohm resistor and the notch circuit on the rear tweeter. And you could control the level of the rear firing tweeters through the DSP.
Quad-amp.
Maybe that sounds like overkill, but you could do some very interesting measurements of reflected sound by turning off the front tweeters and EQ the rear firing sound any way you want.
My inner geek thinks that’s all very exciting; it’s also very complex.
Last edited:
Well that is the DEFINITION of overkill. I love it! I started to go down a whole crazy rabbit hole there, maybe I better just stick with the passive stuff. I think the thing that's the most challenging for me regarding this design is the need to have the MiniDSP also do the DA conversion. I have a Holo Audio Spring 3 KTE and the MiniDSP is not close in terms of the DA conversion. I wish there was a way to do the crossover portion in the MiniDSP, but still do the DA conversion with a better stand-alone DAC.
Perry, I've actually been looking into just that. There are 2 options that I know of, but both cost over 25k. I'd be willing to spend 7 to 10k on a good alternative to the MiniDSP, but that is more than I'm willing to spend at the moment.
Niels - I'm not sure why you sent me a link to the MiniDSP flex? That's already the product I was looking at. I think I better be clear about why the MiniDSP has to do the DA conversion, because most MiniDSP units have digital outputs so that you can do the DA conversion on a separate DAC. That's what I was doing with my MiniDSP SHD - doing the DSP with the MiniDSP, but then the DA conversion with my standalone DAC. The reason is that in this setup, I need to have digital outputs for each channel. If I was bi-amping I'd need a digital output for the woofers and midrange and then a separate digital output for the tweeters. If I were to tri-amp then I'd need 3 sets of digital outputs. The miniDSP can't do that as far as I know, but even if it did, I'd also need 2 or 3 separate standalone DAC's then to handle each of the outputs! It gets pretty out of hand.
The MiniDSP stuff is absolutely brilliant technology, and I think Perry is completely correct in his thinking that DSP is the way of the future. The problem I've been finding is with the hardware. Unless you're willing to spend 25-50k on a piece of gear you're limited to the MiniDSP as far as I can tell. From my own experience with their products I know that there is a very sizable difference in sound quality between the DAC in the SHD and in the Holo Spring. I have a few audiophile buddies and we have blind tested this as well as done blind testing on several other DAC's.
These speakers were never intended to replace my main listening speakers, but I must say I've been getting excited to see what they can do. I think I'm just going to build them as they are, try them with the MiniDSP and see how good that is. If they are close to my current setup that will be quite a feat, and would tell me that their potential is pretty amazing with the MiniDSP DA conversion holding them back. The technology will continue to progress, and hopefully someday soon someone will come out with a high quality piece of gear which accomplishes exactly what we're talking about for a somewhat reasonable price.
I'm going to link the 2 pieces of gear here that look like they can do what we need here:
Bachh-SP
Emperor DSP-X
Niels - I'm not sure why you sent me a link to the MiniDSP flex? That's already the product I was looking at. I think I better be clear about why the MiniDSP has to do the DA conversion, because most MiniDSP units have digital outputs so that you can do the DA conversion on a separate DAC. That's what I was doing with my MiniDSP SHD - doing the DSP with the MiniDSP, but then the DA conversion with my standalone DAC. The reason is that in this setup, I need to have digital outputs for each channel. If I was bi-amping I'd need a digital output for the woofers and midrange and then a separate digital output for the tweeters. If I were to tri-amp then I'd need 3 sets of digital outputs. The miniDSP can't do that as far as I know, but even if it did, I'd also need 2 or 3 separate standalone DAC's then to handle each of the outputs! It gets pretty out of hand.
The MiniDSP stuff is absolutely brilliant technology, and I think Perry is completely correct in his thinking that DSP is the way of the future. The problem I've been finding is with the hardware. Unless you're willing to spend 25-50k on a piece of gear you're limited to the MiniDSP as far as I can tell. From my own experience with their products I know that there is a very sizable difference in sound quality between the DAC in the SHD and in the Holo Spring. I have a few audiophile buddies and we have blind tested this as well as done blind testing on several other DAC's.
These speakers were never intended to replace my main listening speakers, but I must say I've been getting excited to see what they can do. I think I'm just going to build them as they are, try them with the MiniDSP and see how good that is. If they are close to my current setup that will be quite a feat, and would tell me that their potential is pretty amazing with the MiniDSP DA conversion holding them back. The technology will continue to progress, and hopefully someday soon someone will come out with a high quality piece of gear which accomplishes exactly what we're talking about for a somewhat reasonable price.
I'm going to link the 2 pieces of gear here that look like they can do what we need here:
Bachh-SP
Emperor DSP-X
There are different versions of the minidsp flex. The one that he linked to has two digital stereo outputs. So you will have to connect two dacs to it to get analog out!
Live Edge Birch Dipoles: Keep the 10 degree tilt, it works perfect.
Bitches Brew - 5 degree tilt instead of 10. That's cuz those tweeters are at a higher elevation.
You were reading my mind balthazarp - I actually messaged deer creek audio (one of the main dealers for minidsp here in the US) 10 minutes ago about if I could share the Dirac license from my SHD with the flex. Haven't heard back yet, but as you point out - I'm not sure I need to?
Well I got all the parts ordered now. The coaxial drivers I ordered last week and should be here tomorrow. I placed 2 other orders, one with Madisound and the other with parts express for all the other drivers, horns, and crossover components. I flattened some of my walnut slabs yesterday to start working on the baffle build. Thanks for all the help and input thus far.