Trinnov AltitudeCI Review and 32 Channel Build Thread

[etc6849]

Looks safe to update lol. I might try it later today. The bad report I saw was from an Altitude 16/32 user (note the FW difference), but they got theirs working.

 

[darth]

Do you guys normally use some monitoring headphones to benchmark or more of I know this song very well and that's how it should sound, beyond the measurements?

 

[etc6849]

Sure, I listen after calibrating, but I rely on REW measurements for the most part. I definitely do not make changes by ear. Distortion wise, my system measures better than any headphone in the world btw. I think just the bass distortion I show above proves that, but the bed speakers all have acoustically measured THD even below the subs. I will post final measurements of my setup within a week or so when I have time to experiment with the new firmware features.

Do you guys normally use some monitoring headphones to benchmark or more of I know this song very well and that's how it should sound, beyond the measurements?

 

[PGAMiami]

Install went smoothly. Waveforming settings changed a bit as I optimized the Trinnov before and after measurements. I haven’t checked them with REW though. Negative effects of turning on Pressurization Mode seem less now. Amplitude response in the very low bass improves slightly now if I turn Pressurization Mode is on but at the expense of slightly worse decay control. Now best performance is with high settings on Decay Control. So does it sound better? Maybe the bass seems more present. But maybe it’s placebo.

That said, they surely made some changes to Waveforming. I still like it best with my front Perlisten R210s as emitters and rear Rythmik F12SE as full time absorbers. Given the size of my room the two R210s should have no problems reaching THX Dominus levels without help from the F12SE. And the F12SE have more than enough capacity to absorb the reflections from the back wall.

Bass with Dirac ART was more fun to listen to but there is definitely more definition and detail coming through with Waveforming. I still don’t understand how ART magically knows how to use some speakers as absorbers since it has no idea where they are located. ART MIMO makes sense but the absorber part seems like voodoo.

 

[PGAMiami]

Actually went back to the older settings, they measure slightly worse, sound better. And I have Pressurization Mode ON now, measures worse, sounds better. This is on music, not not LFE subsonic bass that I'm hearing a more musical tone even though none of the content is below 16hz which is where the Pressurization Mode frequency is set.

 

[etc6849]

Actually optimizing the Trinnov requires external AES snakes, an expensive measurement rig that can accurately measure acoustic THD (not a USB mic), ability to externally turn off amps and subs, etc... Good thing I have all that

The Default BM crossover was off by about 1.2ms through the crossover region and the L+R setting was off too I just did my own method that measured better. It is good enough, sure, but I wanted phase to be as perfect as possible, especially through the BM XO region. And yes I did improve phase and get rid of some XO cancellation fixing it. Bass sounds 100% real and natural again, like it used to without the Trinnov and the 6 Xilica XD4080 DSPs.

The optimizer could have handled phase better (see below), but it did cancel most of the phase rotation from the Linkwitz–Riley 4th order filters I settled on for tri-amping. However, why can't they offer FIR filters for tri/quad amping?!? Seems stupid and sloppy to introduce that much phase rotation, then later "optimize" to get rid of it. Can you go higher than 4th order and have the optimizer just fix all phase rotation; no, it leaves more in my testing. I would stick with 4th order...

I promised some remeasurements. These are with 1/48 curve smoothing, not the 1/3 or 1/6 the Trinnov GUI shows. Note acoustically measured THD for FL crossed to the subs is better and lower than any headphone available. It was like that pre-Trinnov, but it wasn't simple to set it up by hand before. Might be a little easier with the Trinnov, but you have to jump through hoops to adjust stuff sometimes and absolutely need an external measurement rig to do what I did.

Note, my Optimzer settings. I do have a small room, but have 5 racks in there and 8 subs. So yeah, there are reflections. I don't want to bump correction up too much. The values I am using are from the Nova manual, so make sense in my studio like home theater.

 

[Frunse]

8 Subs in an smal Room....???? What youre Naighbour's say to that .....LOOOL I gues you has an own House for that???

One question: why youre not using the RME Card with DSP Software (like CamillaDSP , FIR) for Crossover instead of an DSP Device that only has 1200 Taps per Channel ???

The Phase Change on the Tweeter would be easier to fix, or are that passive Speaker???

Here i measured my Prototyp Waveguide with AM25 AMT and 70db/Octave @96kHz with CamillaDSP and rePhase Linear Phase Filter Calculation (16384 Taps):

https://www.diyaudio.com/community/...uides-for-cnc-3d-printing.318190/post-8024431

No Phaseshift at all Also Distortions are realy Low at Listeninglevel (sitting 2,5m away)

It runs also on an Windows PC.

I use for now UBUNTU Studio and had some Problems with installing REW!

Get only white Screen after starting REW, after fixing that i get REW freezes when i try to change Samplerate on REW Setup so no luck for now.

But anyway i can Measure it with an second Windows Machine with HDMI Out and an USB Interface for Analogue in like Motu UL mk5 or so.

On an RPI5 REW was working on Linux but a bit slow one year ago when i testet it......

BTW: My RTX2080Ti is gone, got this stupid RAM error, lots of strange Symbols on the Screen, well known for an specified RAM Chip used, after the Waranty was gone like always LOOOOL.....

 

[etc6849]

Lol, nah. They'll be putting their house on the market soon. Great selling point for them. Just don't tell the new buyers.

I don't normally play them that loud. I did get the system up around 130 dB while playing around with the Fury Road intro, but not for long lol.

As the measurements show, eight subs aren't really about maximum SPL. The goal is better modal and decay control, along with very low distortion because no individual sub is working hard at all.

The Trinnov AltitudeCI is doing all of the processing now, so the 1200-tap limitation of the Xilica XD4080 is no longer relevant.

I'm also not convinced it's worth chasing a mathematically perfect phase response when the amplitude response is already doing exactly what I want. The overall slope and shape are intentional. I could increase the optimization in the CI, but at some point you'll add pre-ringing.

I could certainly add external FIR processing through the RME AES card, but that would complicate the system and introduce another potential failure point. These drivers are essentially irreplaceable, so I prefer keeping protection and crossover processing inside dedicated hardware.

Also, the 4th-order acoustic crossover behavior is consistent with how the speakers were designed. Going to an extremely steep linear-phase crossover just because I can would change the overlap between drivers and potentially alter the speaker's polar response.

8 Subs in an smal Room....???? What youre Naighbour's say to that .....LOOOL I gues you has an own House for that???

 

[Frunse]

I learned for my self, the less the overlap the cleaner the Speaker sound's, also less efected if you move the Head up or down while listening.

But i design my Xover by my self for long Time now, and don't know youre Speaker.

And yes Ringing is an Problem that's also the one Point i don't like on my SU-2 DAC'S (no switchable Outputfilter, you can see pre and post ringing) but for that Pricepoint i need to life with

 

[etc6849]

Man, I wish you guys could hear how the system sounds... It honestly sounds truly incredible. I have never heard bass, imaging and midrange that is this good. It 100% sounds like it measures.

I'm still having my CI hiccup every now and then though, even on FW 2.2. Nothing a reboot can't fix, but support is working on resolving the issues thankfully. They did fix several bugs in the new FW, and the unit sounds fantastic, which is the most frustrating part to have it randomly hiccup.

 

[beatelund]

You dont need to do measurements or run optimizer to get audio!

 

[etc6849]

You're right. Thanks! I'll just ignore the red setup and never ever click calibration

You dont need to do measurements or run optimizer to get audio!

 

[PGAMiami]

Man, I wish you guys could hear how the system sounds... It honestly sounds truly incredible. I have never heard bass, imaging and midrange that is this good. It 100% sounds like it measures.

I'm still having my CI hiccup every now and then though, even on FW 2.2. Nothing a reboot can't fix, but support is working on resolving the issues thankfully. They did fix several bugs in the new FW, and the unit sounds fantastic, which is the most frustrating part to have it randomly hiccup.

Still figuring out here how to take multichannel REW measurements so I may send you my measurements with REW. My only sources are Apple TV, Shield and Roon. Perhaps it could be done if I can get my MacBook Air to output via hdmi as multichannel.

Some questions if you don’t mind. Are you still using 1/6 octave and 5 reflections as your settings? Any other important changes vs default? Did you take a few measurements near the MLP to average out any comb filtering artifacts? I settled back to the default 1/3 and 3 reflections. Not much difference audibly.

Overall I also feel my system sounds better than ever. And the GUI does freeze every now and then if I’m making changes. If left alone it works 99% of the time. Reboots fix everything fine.

Remaining problem I have is SBIR above the HPF crossover and below the room’s Schroeder frequency. I have enough RPG room treatment such that the decay in the room is fast, but I still have an uneven response 80-300hz. Below 80-90 hertz Waveforming essentially makes the room modes disappear. Ironically Waveforming is not improving much the decay time, even increased it at 70 hz, but the decay was fast to begin with. This is with 2 Perlisten R210s on the front and two Rythmik F12SE on the back as absorbers. All on the floor.

All this above is assuming the Trinnov graphs are accurate. The one measurement I was able to confirm with REW was 2 channel stereo and those generally coincide with Trinnov’s graphs.

 

[etc6849]

Some questions if you don’t mind. Are you still using 1/6 octave and 5 reflections as your settings? Any other important changes vs default? Did you take a few measurements near the MLP to average out any comb filtering artifacts? I settled back to the default 1/3 and 3 reflections. Not much difference audibly.

Of course. I changed tons of settings I also changed settings in JRiver too. It would take me a 1-2 days to properly document every single thing I did.

Trinnov makes tons of money, I'd start emailing their support or ask your dealer and see what additional hints they give. Trinnov could allow customers to attend their calibration training... I think they used to, but there's a lot of nonesense now like you must use a dealer to get the most out of your system.

Since you asked, here are a bunch of hints. Nothing I did was a secret, it's all exposed in the GUI. Some of these may not apply to you if you use WF, but here's some things for others to ponder:

• I put custom filters in place before ever doing optimization.
• I have custom tri and quad amp settings as polar response matters and I'd never let Trinnov just calibrate my active crossover, because when it did, it didn't sound right at all.
• I have modified delays slightly after "optimization" too as the sub alignment to the speakers was slightly off and didn't sound like it should.
• I have used different optimizer settings, the one's I posted earlier...
• I have custom target curves that also match the natural roll-off of the speakers.
• I limit excursion in some places.
• I changed all kinds of decoder settings to ensure maximum dynamic range.
• I changed a few settings under the Spatialization screen.

You need a way to externally measure to even approach several of these items, but sure many of the others can be done from the Trinnov graphs. I would use something similar to my rig: soundcard with AES in/out, Earthworks M30 mic, Benchmark ADC1 and Benchmark MPA1 mic preamp. I have serious doubts about using a USB mic for this type of precision work.

Remaining problem I have is SBIR above the HPF crossover and below the room’s Schroeder frequency. I have enough RPG room treatment such that the decay in the room is fast, but I still have an uneven response 80-300hz.

Sadly, I can't really help much with this remotely, and I don't have the time to give it the attention it would require since I have a four-year-old and a day job. Of course, I'm happy to help in person or even remotely via video conferencing, but we'd have to discuss a flat fee plus an hourly rate since I'd need to take time away from work.

If you look at my room pictures, you can see that the room acoustics were heavily optimized before applying any EQ. It is very likely you aren't going to get the results I have without extensive room treatments. Granted, I am aiming for the best possible performance within the constraints of the room I have.

Honestly, there's only so much the CI can do. If you ask it to do too much, remember that a large positive correction consumes system headroom and can ultimately drive the digital output, DAC, amplifier, or loudspeaker into overload. I suspect their algorithm is smart enough to limit or avoid trying to correct issues like deep cancellations in the first place, but that's somewhat speculative on my part.

You're more than welcome to stop by when you're in my area and decide is it worth the work to further refine your setup, but you may not want to put all the acoustic treatments necessary in your room, spend the money on an external measurement rig, etc...

 

[Fidji]

Did you take a few measurements near the MLP to average out any comb filtering artifacts? I settled back to the default 1/3 and 3 reflections. Not much difference audibly.

I am taking multi-point measurements. helps greatly, once you know how to do it, especially in mid-range. It is combination of placement and weighing to get results.

re resolution - i think 1/6 is a sweet-spot, also needs to be properly combined with some Advanced Settings [e.g. FIR length]. Number of reflections - this depends on your specific situation and preferences. All those things have their internal logic and work in combination. Couple of years ago I had settings done by Trinnov people and they changed a lot of default settings. I would say there are couple of real levers, rest is "if you don't know what it is, don't touch it"

in my experience - all those adjustments can still improve experience over vanilla default settings, especially for music and MCH music.

 

[etc6849]

We’re actually pretty close on those settings. I’m already using 1/6-octave energy-response resolution and 5 cycles for early reflections. My FIR length is still 100 ms with a 5 ms reference though. According to the NOVA manual (since we don't get a manual with the CI ), 100ms should provide plenty of resolution:

I did experiment with allowing more correction at higher frequencies, and I heard what I considered artifacts even without increasing the FIR length. That is exactly why I ultimately reduced the high-frequency excursion limits to the minimum. At higher frequencies in particular, I’d rather constrain the correction than have it aggressively chase narrow, position-dependent features.

That may also be part of why taking multiple measurement positions worked well for you, since in theory it gives the Optimizer a better indication of which response features vary significantly with position. Per Trinnov’s documentation, that spatial variation should help it avoid unnecessarily correcting localized features.

https://kb.trinnov.com/en/knowledge-base/single-or-multiple-measurements

https://www.trinnov.com/en/blog/pos...-remove-a-loudspeakers-sound-characteristics/

So yeah, increasing the resolution and allowing the Optimizer to react to finer response features likely made the additional measurement positions really really important, since they gave it more spatial information to distinguish consistent problems from localized ones.

I’d argue that in a room where the first-order reflections are already well controlled, you’re using $20k–$40k speakers with good native directivity and response, and you’re really the only audiophile in the house, the better approach is to optimize speaker positioning and toe-in for the money seat. Then address the remaining reflection problems acoustically and minimize processing wherever possible. At that point, I’d rather preserve what the speakers are already doing well than ask the Optimizer to make increasingly fine corrections to relatively small residual errors, only to then add more measurement positions so it backs off from making some of those corrections in the first place.

You are certainly right that these settings interrelate. At some point, though, it can become a bit like a dog chasing its tail...

I am taking multi-point measurements. helps greatly, once you know how to do it, especially in mid-range. It is combination of placement and weighing to get results.

re resolution - i think 1/6 is a sweet-spot, also needs to be properly combined with some Advanced Settings [e.g. FIR length]. Number of reflections - this depends on your specific situation and preferences. All those things have their internal logic and work in combination. Couple of years ago I had settings done by Trinnov people and they changed a lot of default settings. I would say there are couple of real levers, rest is "if you don't know what it is, don't touch it"

in my experience - all those adjustments can still improve experience over vanilla default settings, especially for music and MCH music.

 

[PGAMiami]

One reason for getting the Trinnov was so I could do the calibrations myself. Thank you though for the offer. My room was professionally treated by RPG years ago, including Modex Plate bass traps, Abfusors, Diffusers … in all the right places. All done with an eye on acoustics as well as practicality and aesthetics. Some of my panels ended up in the attic, like the ones that were to go in front of the TV and the ones to control reflections on a side wall that has a large window. You could measure the difference when these were up or down. But in practice for me they were not worth the trouble.

The benchmark for me is how does this sound versus the convolution filter Mitch Barnett made for me. From the mid bass up I think Mitch wins. But Waveforming is really petty magical marrying subs to my R1s. Multichannel set ups with Mitch require more boxes than I’m willing to have these days. But he may have a hardware product out next year that makes it easier to implement.

Mitch’s filter runs on Roon. One alternative I may try in the meantime is having a setting in my Ci that only engages Waveforming and then running the convolution on Roon.

 

[PGAMiami]

I am taking multi-point measurements. helps greatly, once you know how to do it, especially in mid-range. It is combination of placement and weighing to get results.

re resolution - i think 1/6 is a sweet-spot, also needs to be properly combined with some Advanced Settings [e.g. FIR length]. Number of reflections - this depends on your specific situation and preferences. All those things have their internal logic and work in combination. Couple of years ago I had settings done by Trinnov people and they changed a lot of default settings. I would say there are couple of real levers, rest is "if you don't know what it is, don't touch it"

in my experience - all those adjustments can still improve experience over vanilla default settings, especially for music and MCH music.

Could you expand on your experience taking multiple measurements? How many, how far from MLP? I’m thinking 12” at most, give a big weight still like 40%,

 

[Fidji]

The benchmark for me is how does this sound versus the convolution filter Mitch Barnett made for me. From the mid bass up I think Mitch wins. But Waveforming is really petty magical marrying subs to my R1s. Multichannel set ups with Mitch require more boxes than I’m willing to have these days. But he may have a hardware product out next year that makes it easier to implement.

Mitch’s filter runs on Roon. One alternative I may try in the meantime is having a setting in my Ci that only engages Waveforming and then running the convolution on Roon.

I have started from the similar point [Roon convolution done by pro] and meanwhile I get better results from Trinnov only. [required some tinkering]. All the tools and CPU power are available inside Trinnov. I have dropped the Roon meanwhile [use Linn streamer and app instead]

One reason for getting the Trinnov was so I could do the calibrations myself. .

Go for it - you have fabulous speakers [TADs are one of two "dream" speakers for me, but probably going to end up with second hand Magicos ;-)] and you have done acoustic treatments + you have the best processor. All the conditions to get the best out of your system are there. And learning and getting improvements by yourself is bonus in itself. But you will need to invest very scarce currencies - time and patience

What sets Trinnov apart [IMO} is not Optimizer vs Dirac [or WF vs ART], but capability to do all the individualizations and adjustments. Dirac is black-box, Optimizer is basically open. You'll get very similar results out of the box with all the Wizards between Dirac and Trinnov, but once you start to fine-tune there is no contest. This is why Ttrinnov is still used in approx. 85% of high-end installs out there, despite Storm etc being available for some time.

You can get 20-30% improvement between auto and fully tweaked system, at least in my experience. I am on the 5th iteration [4 done by myself, one by pro] and I need to say every was significatn improvement. My 5th iteration is building on lot of work done by calibrator [speaker correction files], but I think I have better sound and measurements compared to pro calibration. Especially in WF and overall "immersion" quality.

I think if you want to move ahead more efficiently you should spend some time on getting "Trinnov advanced tutorial' - helped me tremendously. I was trying to figure out what to do on forums, contacting Trinnov directly etc. - but it was more of trial and error. Then I managed to get one guy taking pity on me and we spent some time discussing his and mine setting via Zoom and it all started to make sense.

If you feel like it could make sense, I could spend some time with you on the Zoom/Teams and show you my workflow and settings I use and why I use them.

Could you expand on your experience taking multiple measurements? How many, how far from MLP? I’m thinking 12” at most, give a big weight still like 40%,

Firstly - high quality measurements are absolutely crucial in order to get proper results and it will make great deal of difference

See above - I could explain you, why I am using custom grid with 14 measurements for WF and why it yields better results compared to standard measurements, I have also one specialty - my reference point is slightly different to MLP [it has very specific reason] , how and why I do 12 measurements for non-subs on my 2 seater and why I use it differently for different setups. But this is based on my speakers/room/preferences and it might not be useful 1:1 to you. -> this is why I think it is much easier to talk, than to spend lot of time typing and missing some important aspects.

 

[Fidji]

We’re actually pretty close on those settings. I’m already using 1/6-octave energy-response resolution and 5 cycles for early reflections. My FIR length is still 100 ms with a 5 ms reference though. According to the NOVA manual (since we don't get a manual with the CI ), 100ms should provide plenty of resolution:
View attachment 559986

I did experiment with allowing more correction at higher frequencies, and I heard what I considered artifacts even without increasing the FIR length. That is exactly why I ultimately reduced the high-frequency excursion limits to the minimum. At higher frequencies in particular, I’d rather constrain the correction than have it aggressively chase narrow, position-dependent features.

That may also be part of why taking multiple measurement positions worked well for you, since in theory it gives the Optimizer a better indication of which response features vary significantly with position. Per Trinnov’s documentation, that spatial variation should help it avoid unnecessarily correcting localized features.

https://kb.trinnov.com/en/knowledge-base/single-or-multiple-measurements
View attachment 559977

https://www.trinnov.com/en/blog/pos...-remove-a-loudspeakers-sound-characteristics/
View attachment 559981

So yeah, increasing the resolution and allowing the Optimizer to react to finer response features likely made the additional measurement positions really really important, since they gave it more spatial information to distinguish consistent problems from localized ones.

I’d argue that in a room where the first-order reflections are already well controlled, you’re using $20k–$40k speakers with good native directivity and response, and you’re really the only audiophile in the house, the better approach is to optimize speaker positioning and toe-in for the money seat. Then address the remaining reflection problems acoustically and minimize processing wherever possible. At that point, I’d rather preserve what the speakers are already doing well than ask the Optimizer to make increasingly fine corrections to relatively small residual errors, only to then add more measurement positions so it backs off from making some of those corrections in the first place.

You are certainly right that these settings interrelate. At some point, though, it can become a bit like a dog chasing its tail...

This is also my approach - get best speakers you can, place them properly, do acoustic treatements you can live with and only then deploy DRC for Bass and final touches. You will not EQ out shi...y speakers or glass aquarium. And of course - get as many subs as you can squeeze into room, only overkill cuts it for me;-).

BTW - out of my experience - I have speakers [electrostats] that are quite demanding re proper placement [dipoles with quite specific off-axis] and I think I am quite good at placing them properly. But optimal placement for Trinnov DRC is slightly different to "analogue" position - as I do not need to focus on bass area and I can prioritize mids/treble.

I have focused lately on "no distortion/compression" aspect and linearity of acoustics, and this was a step forward. I do not need to employ filters above 2kHz for L-R and overall I use filters much less. WF does most of the heavy lifting.

This is my T20/T30 after latest acoustics re-shuffle and it was bigger change than swapping amps some time ago.

Also overkill on woofage - 8x18inx2500W each yields very low distortion. You can see exactly where Pressurization kicks in and it goes from 0.5% to 1% in infrasonic. FWIW - I am big infrabass fan so I also ask myself "how loud can I get this 8Hz earthquake playing ;-).

I am not on board with "10% distortion is good enough for bass" type of installs and I also put big focus on Group Delay, as this is one thing you can defintily hear if is rigjht or wrong. Here WF is doing great job [and I help it by not having any Bass Reflex speakers/subs]