Keith's Acapella system

[zydeco]

Cross-over design / Amplifiers

Keith

It must be a joy to listen to music on the system. It's interesting that you've got different amplifiers on the bass/mid-bass and mid/high. No doubt the solid state amplifier provides much better bass control than when the valves are run full-range. Are you, though, able to detect the sonic signature of the two different amplifiers?

Also, out of interest, how did you go about deciding on the cross-over point / slope / design between the bass and mid/high? Is this an active replication of the original passive cross-over? Or was it through trial / error measurements?

Cheers
Zydeco

 

[Keith_W]

Hi zydeco, thanks for the questions.

It's interesting that you've got different amplifiers on the bass/mid-bass and mid/high. No doubt the solid state amplifier provides much better bass control than when the valves are run full-range. Are you, though, able to detect the sonic signature of the two different amplifiers?

Of course I can tell the difference in sonic signature between the two amplifiers. Before I converted my speakers to a fully active configuration, I had two amplifiers which I could choose to run in a bi-amp configuration, or in a mono-amp configuration. As was expected, the valve amps gave a much nicer midrange and top end, while the SS amp gave much better bass control with the expense of the mid and top end sounding a little sterile and mechanical. In any case, the speakers are quite unique with different speaker technologies being used for bass, midrange, and top end. My listening suggests that the different driver types contribute much more to the overall sound than the choice of amplifier.

Also, out of interest, how did you go about deciding on the cross-over point / slope / design between the bass and mid/high? Is this an active replication of the original passive cross-over? Or was it through trial / error measurements?

I measured the response of each individual driver in my speaker and selected the crossover points accordingly. The midrange horn is relatively linear until 600Hz, then drops off with a 18dB/oct slope (probably a combination of the natural rolloff and the electric crossover). I chose a crossover point of 600Hz, simulated my intended settings with the DEQX, and had a good listen. I then ordered my Marchand hard-wired in this configuration. I think I have a little more to do before it sounds right, but it won't be far off. I hope!!!

 

[zydeco]

Keith

Thanks. I wasn't meaning to question whether you could hear the difference in the two amplifiers but rather whether this difference was audible when playing back in the set-up in the active configuration. You see I'm struggling with similar issues with for a horn / bass speaker and have landed on solid-state bass / valve mid-range for more or less the same reasons. I've had a lot of advice on the need to have identical amplifiers across all channels but, to my ear, differences cross the channels aren't audible and I was wondering whether I'm missing something.

Zydeco

 

[Keith_W]

Been a few years since my last update. Thanks Microstrip for reminding me that it exists, and that I should update it!

I am in the middle of switching over from an analog active crossover to a digital crossover. I took Blizzard's suggestion in another thread and I am implementing a digital crossover based on Acourate generated filters. It is a little difficult to explain, so I will use some diagrams.



This was what the system looked like as of a couple of months ago. You can see that the system has been converted to active some time ago (i.e. all the passive crossovers in the speaker have been removed), with crossover functions taken over by a pair of active crossovers. The Marchand is an analog active crossover that splits signal to the main speakers, and the DEQX is a digital active crossover that sends signal to the subwoofers and also performs room correction and DSP. Each speaker driver has its own channel of amplification.

The main disadvantage of this system is the processing overhead introduced by the DEQX of about 30ms. This delays the subwoofers with respect to the main system by 30ms, meaning that bass transients arrive much later than the rest off the sound - causing bass to sound slow and disconnected.

I could of course run the whole system off the Marchand, but then I would lose DSP. If I ran the whole system through the DEQX, the awful analog to digital conversion on the DEQX would kill the midrange and top end. I could feed a digital signal into the DEQX, but that would only be 16 bit audio and the DAC's in the DEQX are only of so-so quality.

All this while I have believed in the potential of DSP as an avenue to solve my problems. This is why, when I read Blizzard's thread, I was glued to the screen. I could scarcely believe what I read.

It took me a year of reading and planning before I decided to make my move:



This is what the system will look like once I am finished. At the moment, I am only missing two pieces of the puzzle - the 8 channel Merging NADAC, and my new subwoofers. In its place, I am using an 8 channel RME Fireface DAC and I still have the JL sub.

As you can see, I have replaced the preamp, the Marchand, and the DEQX with two boxes - a PC (which I built myself) and the 8 channel DAC. All crossover functions and volume control have been moved to the digital domain, where I can do as much processing as I like and not lose anything in terms of quality. Once the signal has been converted to analog, what lies in the analog signal chain is far simpler than most audio systems - there is no preamp. There is no passive crossover network. There is therefore much less potential for signal degradation than most systems.

Furthermore, I will be able to download high resolution DSD files and play them back at their native rate without any loss whatsoever.

For those who are interested, this is the recipe for the PC:

- CPU: Intel Skylake i7-6700K
- Mobo: MSI B150i Gaming Pro AC Mini-ITX
- RAM: 16GB DDR3
- Storage: 1TB Samsung 850 Evo SSD
- PSU: Uptone Audio JS-2
- Cards: RME HDSPe AES 32 channel I/O sound card
- Still to come: JCAT SATA cables, SOtM CAT6 cables (yes, I might believe in measurements and might seem objectivist in my approach, but I can hear differences between cables so I will continue to buy them!)
- Also: RME Fireface UC microphone preamp and ADC/DAC, Dayton EMM-6 calibrated microphone.

But really, the heart of the system is in the software. I use:

- Acourate. All Acourate does is take measurements and generate filters. These filters can be installed in third party playback programs (such as HQPlayer, JRiver, Foobar, and so on), which then perform a variety of functions. But ... these filters are very powerful. It is possible to linearize the frequency response each individual speaker driver, correct phase, generate crossovers, perform room correction, time align the drivers, apply ISO226 loudness curves, and so on.

- HQPlayer. HQPlayer uses the filters generated by Acourate to play back audio files (these can be FLAC's located on the SSD or CD's). But the difference is - it is able to upsample the files up to DSD512, apply the filters, and output at this rate. The NADAC maxes out at DSD256, so I will "only" be able to use DSD256.

As mentioned, I have already implemented the active crossover. The preamp, Marchand, and DEQX are packed up in boxes waiting to be put on the secondhand market. What does it sound like?

Well, there are pros and cons. The main con is that the RME is nowhere near as good a DAC as the Playback Designs. Hopefully, the NADAC will be much better. But the sound has been improved in many ways - it is so much more coherent, imaging has improved, instrument separation and clarity has improved, and transients hit hard and with a bang.

From here onwards it will be waiting for my NADAC and my subs to arrive, and learning how to use some of the more advanced functions of Acourate. But the move does appear to be very promising.

 

[Ron Resnick]

Thank you for the update, Keith!

I did not see your thread originally but I, too, am incredibly impressed with what you have done. (I certainly could not have done it. Before even getting to the electrical aspects I would have been afraid even to remove that special German sand!)

Steve and I together heard an Acapella speaker at T.H.E. Show in Irvine this year. We both thought it sounded great! (Kedar predicted that I would like vocals on the Acapella and he was correct!)

1) Have you ever found the plasma tweeter to be at all forward or "hot"-sounding? Are you able to adjust independently the level of the plasma tweeter?

2) What prompted you to switch from the JL subwoofers to the Rythmik subwoofers?

3) What other SET amps did you audition before selecting the Carys?

Thank you!

 

[microstrip]

Been a few years since my last update. (...)

Great work Keith - it is really impressive. As far as I could understand your filters are modeled after measurements taken from the passive loudspeaker - please correct me if I am wrong! This fitting is carried by the Acourate program that generates the filters. Can you modify them in an interactive real time way when listening?

 

[Keith_W]

Hi gents, thank you for your interest.

1) Have you ever found the plasma tweeter to be at all forward or "hot"-sounding? Are you able to adjust independently the level of the plasma tweeter?

2) What prompted you to switch from the JL subwoofers to the Rythmik subwoofers?

3) What other SET amps did you audition before selecting the Carys?

1. The plasma tweeter has a volume control at the back - so yes, it is possible to independently adjust the volume of the plasma tweeter. And yes, it can be quite hot sounding if you turn it up too much.

2. The JL subs I bought (F110) were too small. That, and they had a tendency to blow up despite not being driven very hard. I heard a Rythmik sub at a friend's place and was quite impressed with the servo technology, so that prompted me to make a change.

3. Prior to choosing the Carys (I assume you are referring to the CAD-805AE which I owned before my current amp, the CAD-211AE), the only other SET amps I have heard was a lower powered Cary (CAD-300SE), a ridiculously priced Audio Note which I could never afford, and a Chinese amp whose name I have forgotten.

Great work Keith - it is really impressive. As far as I could understand your filters are modeled after measurements taken from the passive loudspeaker - please correct me if I am wrong!

That is partially correct, microstrip. I took a few sets of measurements when the passive crossovers were still in place. The Acapella has a modular construction where the cabling between the bass unit, the horn, and the plasma tweeter is exposed and can be removed. It is therefore possible to disconnect them and measure them separately to see exactly what they are doing. Using REW, I ran full range sweeps through each speaker module - that told me roughly where the crossover points and slopes were. Of course, that tells you the cumulative effect of the crossover (i.e. electric and mechanical) so if you want to know what the electric crossover is doing, you have to study it. Since I do not have the ability to do so (those capacitor and choke values mean nothing to me), I asked a speaker designer to look at it +/- send a waveform through it to ascertain its behaviour.

Once I knew what the electrical slopes were, I simply used that as a starting point and put those values into my DEQX. I played around with the crossover settings and experimented with the effect of steeper slopes and so on. Once I was happy, I ordered a Marchand analog crossover and specified those slopes which I designed on the DEQX.

Of course, the Marchand has significant disadvantages compared to the DEQX. But it did have one advantage which mattered a lot to me - it was by far more transparent.

This fitting is carried by the Acourate program that generates the filters. Can you modify them in an interactive real time way when listening?

No you can't. Once you generate the filters, the only way to modify them is to design a new target curve and re-generate the filters.

I should point out that I am using two pieces of software from AudioVero:

1. Acourate. Use this to take measurements, apply corrections, and generate filters.
2. AcourateConvolver. Uses the above filters to create the crossover. This program includes an equalizer function, which you are able to use to change the frequency response in real time if you like. But it is better to bake them in to the filters.

I should also admit that none of these ideas were mine. I am simply following the path trodden by others. The only thing you should be impressed with is my googling skills.

 

[Ron Resnick]

Thank you, Keith!

 

[Elberoth]

Wow, impressive !

 

[BobShermanEsq]

Been a few years since my last update. Thanks Microstrip for reminding me that it exists, and that I should update it!

I am in the middle of switching over from an analog active crossover to a digital crossover. I took Blizzard's suggestion in another thread and I am implementing a digital crossover based on Acourate generated filters. It is a little difficult to explain, so I will use some diagrams.

View attachment 28722

This was what the system looked like as of a couple of months ago. You can see that the system has been converted to active some time ago (i.e. all the passive crossovers in the speaker have been removed), with crossover functions taken over by a pair of active crossovers. The Marchand is an analog active crossover that splits signal to the main speakers, and the DEQX is a digital active crossover that sends signal to the subwoofers and also performs room correction and DSP. Each speaker driver has its own channel of amplification.

The main disadvantage of this system is the processing overhead introduced by the DEQX of about 30ms. This delays the subwoofers with respect to the main system by 30ms, meaning that bass transients arrive much later than the rest off the sound - causing bass to sound slow and disconnected.

I could of course run the whole system off the Marchand, but then I would lose DSP. If I ran the whole system through the DEQX, the awful analog to digital conversion on the DEQX would kill the midrange and top end. I could feed a digital signal into the DEQX, but that would only be 16 bit audio and the DAC's in the DEQX are only of so-so quality.

All this while I have believed in the potential of DSP as an avenue to solve my problems. This is why, when I read Blizzard's thread, I was glued to the screen. I could scarcely believe what I read.

It took me a year of reading and planning before I decided to make my move:

View attachment 28723

This is what the system will look like once I am finished. At the moment, I am only missing two pieces of the puzzle - the 8 channel Merging NADAC, and my new subwoofers. In its place, I am using an 8 channel RME Fireface DAC and I still have the JL sub.

As you can see, I have replaced the preamp, the Marchand, and the DEQX with two boxes - a PC (which I built myself) and the 8 channel DAC. All crossover functions and volume control have been moved to the digital domain, where I can do as much processing as I like and not lose anything in terms of quality. Once the signal has been converted to analog, what lies in the analog signal chain is far simpler than most audio systems - there is no preamp. There is no passive crossover network. There is therefore much less potential for signal degradation than most systems.

Furthermore, I will be able to download high resolution DSD files and play them back at their native rate without any loss whatsoever.

For those who are interested, this is the recipe for the PC:

- CPU: Intel Skylake i7-6700K
- Mobo: MSI B150i Gaming Pro AC Mini-ITX
- RAM: 16GB DDR3
- Storage: 1TB Samsung 850 Evo SSD
- PSU: Uptone Audio JS-2
- Cards: RME HDSPe AES 32 channel I/O sound card
- Still to come: JCAT SATA cables, SOtM CAT6 cables (yes, I might believe in measurements and might seem objectivist in my approach, but I can hear differences between cables so I will continue to buy them!)
- Also: RME Fireface UC microphone preamp and ADC/DAC, Dayton EMM-6 calibrated microphone.

But really, the heart of the system is in the software. I use:

- Acourate. All Acourate does is take measurements and generate filters. These filters can be installed in third party playback programs (such as HQPlayer, JRiver, Foobar, and so on), which then perform a variety of functions. But ... these filters are very powerful. It is possible to linearize the frequency response each individual speaker driver, correct phase, generate crossovers, perform room correction, time align the drivers, apply ISO226 loudness curves, and so on.

- HQPlayer. HQPlayer uses the filters generated by Acourate to play back audio files (these can be FLAC's located on the SSD or CD's). But the difference is - it is able to upsample the files up to DSD512, apply the filters, and output at this rate. The NADAC maxes out at DSD256, so I will "only" be able to use DSD256.

As mentioned, I have already implemented the active crossover. The preamp, Marchand, and DEQX are packed up in boxes waiting to be put on the secondhand market. What does it sound like?

Well, there are pros and cons. The main con is that the RME is nowhere near as good a DAC as the Playback Designs. Hopefully, the NADAC will be much better. But the sound has been improved in many ways - it is so much more coherent, imaging has improved, instrument separation and clarity has improved, and transients hit hard and with a bang.

From here onwards it will be waiting for my NADAC and my subs to arrive, and learning how to use some of the more advanced functions of Acourate. But the move does appear to be very promising.

Very impressive system. A bit off topic but I noticed you had asked someone about the current state of systems that can do 4K video and handle your HQPlayer needs. Hope this helps.

https://airtop-pc.com/airtop/airtop-product-tour/graphics-and-display/

 

[Keith_W]

Very impressive system. A bit off topic but I noticed you had asked someone about the current state of systems that can do 4K video and handle your HQPlayer needs. Hope this helps.

https://airtop-pc.com/airtop/airtop-product-tour/graphics-and-display/

Thank you! I prefer to roll my own system, actually. So far I have managed to build a PC more powerful than that Airtop (minus the graphics card) for less. And on top of it, my PC has a linear power supply.

 

[BobShermanEsq]

Thank you! I prefer to roll my own system, actually. So far I have managed to build a PC more powerful than that Airtop (minus the graphics card) for less. And on top of it, my PC has a linear power supply.

I build my own systems also, and yes linear supplies do make a difference.

regards
Bob

 

[Folsom]

It looks like you made all the right moves.... but what happens when you want to play a vinyl record?

 

[Keith_W]

Folsom, after owning two turntables in the last 20 years, I realized that I don't play vinyl enough to justify having a TT in the system. Prior to this I had kept it all analog after the CDP as a matter of philosophy, but then 99.9% of my listening was digital. It occurred to me one day that sticking with analog was holding me back and I would get much better results if I were to embrace digital.

I STILL believe that vinyl sounds better. I have had enough experience with turntables to convince me of that. But - doing what I did - you gain some and you lose some. I lose the ability to play vinyl. But I have gained transparency (through loss of the preamp), coherence (through DSP), imaging and soundstaging (also DSP), incredible dynamics (through going active). All the pluses add up and most people who have heard the system through its evolution would agree.

In any case, since the last post, there have been a few changes:

1. The NADAC MC-8 has arrived.
This is a massive improvement over the RME that I was using as an interim solution. But the question is - how does it compare to the Playback Designs? Comparison is quite difficult in my system, because the NADAC directly drives the power amps. The Playback Designs had to go through a preamp (Cary SLP-05) and an analog crossover (Marchand XM-44), so in some ways it is disadvantaged in my system. In my system, both sound about equal - both are neutral, relaxed sounding, and full of detail. Perhaps the Playback had a better sense of space. So I would have to call the MPS-5 the winner in this shootout.

2. I have replaced the woofer with a new one.
I spoke to a friend (Stuart of SGR Audio and complained that when I measured the response of the woofer, it was impressively linear - from 1000Hz to 5000Hz. Unfortunately, I want it to be linear from 50-500Hz. He said he had just the thing, and brought it over.



This is a comparison of the old and the new driver. Red = old, Green = new. You can see that the new driver is nicely linear down to the crossover point of 70Hz. The new driver has a double magnet and cast iron spider and is really heavy.



Once it is corrected with Acourate, it is even more impressive. This is a very zoomed in view of an actual frequency sweep. Note that it is +/- 0.5 dB over the range I am using it!

Nothing is ever free in audio, and in this case what I lose is sensitivity. The old driver was 90dB, the new driver is 85dB. This means the new driver needs almost 4 times the amplifier power of the old one, and boy - don't I know it. I was rather surprised to see the VU meter of my 150W solid state amp max out and hear clipping! You would think that 150W of SS power, directly connected to the woofer with no lossy crossover in between, would be enough. But no, it isn't.

I guess what this means is that I am now in the market for a 600W (into 8 Ohm) amplifier.


3. The crossover in the horns has been bypassed.
This wasn't an easy project. Why - because the cabinets at the rear of the horn are filled with sand. This makes them absurdly heavy to remove, and of course all the sand has to be drained out before you can access the crossover. It is easier to explain with pictures.



This is the horn removed. You will note a few details of the construction. Horn is made of fibreglass. The rear of the midrange horn driver is covered with a rubber boot. The driver itself is labelled "DYNAUDIO D-52 ACAPELLA", presumably a special version of the D-52 manufactured for Acapella by Dynaudio.



This is the box itself. As a friend said, it's very German. Only a German would construct a box out of thick ply, then line it with lead, then think to himself ... "let's fill it with sand". I usually invite people to perform the knuckle test, and I can tell you that NOTHING out there passes the knuckle test like this enclosure. It is not hollow, it is like rapping your knuckles against a block of granite.



Once you get all the sand out, the crossover is at the bottom of the box. It is a simple matter to snip the wires, solder the speaker cable directly to the binding post cables, and sleeve it all up. This is how you get the sand back in. There is an access hole at the bottom, and the sand goes in cup by cup into a tiny funnel. It takes about 20 minutes of tedious pouring to get the sand back in.

Bypassing the crossover has given me an extra 10dB in the horn. Which is great.

Unfortunately, I have also lost 5dB in the woofer. No matter, it is a simple matter for Acourate to correct this. But ideally I would like to gain another 15dB in the woofer. What this means is that I will need a new amp. I am already talking to another amp manufacturer about a custom amp designed for me that will produce all the power I need, but also have the gain structure to make the woofers match the horns. But that will be some time in the future.

 

[zog]

Folsom, after owning two turntables in the last 20 years, I realized that I don't play vinyl enough to justify having a TT in the system. Prior to this I had kept it all analog after the CDP as a matter of philosophy, but then 99.9% of my listening was digital. It occurred to me one day that sticking with analog was holding me back and I would get much better results if I were to embrace digital.

I STILL believe that vinyl sounds better. I have had enough experience with turntables to convince me of that. But - doing what I did - you gain some and you lose some. I lose the ability to play vinyl. But I have gained transparency (through loss of the preamp), coherence (through DSP), imaging and soundstaging (also DSP), incredible dynamics (through going active). All the pluses add up and most people who have heard the system through its evolution would agree.

I still keep a vinyl rig - it is what we use when we have friends around.
(Note to self: Must visit Melbourne).

 

[Folsom]

Thank you for posting the experience of your adventure, Keith.

Various manufacturers sell Hypex amps that will fit your requirement and probably give you just what you're looking for in the bass region.

The one thing I don't understand is where the second, non-internal, bass driver is located. Isobarik is a fine configuration, but I trust not having the one playing inside fullrange gave you the upgrade you where wanting.

 

[Keith_W]

Thank you for posting the experience of your adventure, Keith.

Various manufacturers sell Hypex amps that will fit your requirement and probably give you just what you're looking for in the bass region.

The one thing I don't understand is where the second, non-internal, bass driver is located. Isobarik is a fine configuration, but I trust not having the one playing inside fullrange gave you the upgrade you where wanting.

Thank you Folsom. Yesterday I borrowed a 300W solid state amp - i.e. it makes twice as much power as the amp I am currently using. I will try it to see what difference it makes.

As for where the drivers are:



... this is the external driver. It is mounted on a panel with lots of screws. The panel can be detached, which shows:



A nicely constructed cabinet with lead lining and walls covered in thick padding. Where is the second driver?



Right here. UNDER the padding. Obviously, this picture was taken after the driver was removed.

This is a strange arrangement. The internal driver is run full range from the crossover, relying on mechanical rolloff only. It fires through the thick padding into the back of the external-facing woofer. The external facing woofer is connected to the crossover network, so it has both electric and mechanical rolloff.

I really do not know what the designers were thinking when they implemented this. To me it seems really odd. But then, I am not a loudspeaker designer - so perhaps someone with more knowledge might like to comment.

 

[Folsom]

Ah, so it's hidden.

If you want another potential upgrade and own a bandsaw, or have access, I'd consider removing that padding and cover the inside with no-rez. Also I'd see what it sounds like without padding/stuff between the midbass drivers. Danny Richie, who sells the no-rez, could probably give some good comments on application inside of that box. I've found its vastly superior to using poly fill like stuff. My midrange quality takes a dive every time I try to use stuffing... it just works poorly. That's my finding anyhow, and for the price the only investment really is time between trying dampening material.

 

[Keith_W]

Thank you for your suggestion. I have never heard of it. I will ask a few of my friends (who are loudspeaker designers) what they think of it.

BTW, I know that my box resonances are fairly low. I have measured it. But a little bit more won't go astray.

 

[Keith_W]

Here is an update of the system. That diagram was made in 2016, since then there have been a few changes of equipment and refinements in the tuning of the DSP.

As you would guess with a DSP controlled system with an abundance of amplifier power and no passive crossover, and 4 subwoofers, it sounds pretty awesome