CSS Criton 3TD-X Kit Speaker Review

[musicisair]

What baffle would they measure on? If they used 8" wide and you were using 9-1/2" wide, it wouldn't be good enough. Just like going from their infinite baffle to finite using diffraction simulations you'd still have to do the same. And while the on axis sim might be close, no telling about the off axis.

I guess I was just assuming they'd test and publish results using an infinite baffle, and normalize for other variables as well.

 

[Paweł L]

I remember when Erin measured kit from Parts Express with Klippel and I played in Vituix to make new X-O. To me it was new experience having all up and down, forward and back set of measurements, not just few points in off axis direction. I never build it, but it might be interesting how good or bad was the simulation compared to real loudspeaker. Anyway, forget the straight line on axis target, as the tweeters power response and directivity just past the crossover point will dominate the room response. To counter act this one has to design X-O with a few dB dip. And actually it is very tricky to get the voicing right, half dB here and there might ruin the tonal balance. This was mentioned in earlier post by some of members of the ASR.
I tried really hard to use manufacturers data, then run the baffle step simulation and use the new frd to design X-O. It was sort of OK, but the final tuned by ear, based on actual measured data was always different than the first few raw examples. On top of that you have to take into account 'driver sound', and most seasoned designers know that alu cone will sound way different than paper or polypropylene or kevlar, carbon. And there are sonic preferences, some like laid back and easy going, some like it bright and very detailed. So yeah, measurements and simulations might not right of the box give the effect one is hoping for. It is still a bit a art, despite the fantastic software like Vituix and measuring tools available today to hobbyists (my 25 old Clio is archaic piece of junk comparing to latest tools one can buy) - getting the meaningful, error free set of data is always the key. Even applying EQ to my headphones based on data published here and there, only very, very few EQ curves will do, as some measuring systems and reference curves might be slightly off from what I consider natural.

 

[Paweł L]

What baffle would they measure on? If they used 8" wide and you were using 9-1/2" wide, it wouldn't be good enough. Just like going from their infinite baffle to finite using diffraction simulations you'd still have to do the same. And while the on axis sim might be close, no telling about the off axis.

Most of the time these days, it's a specific IEC baffle, as per standard for given size of the cone. At least we would like to. However if one takes a look at Seas, they also publish frd of driver mounted in a box.
Freq comparison from French diy site.

 

[Rick Sykora]

What baffle would they measure on? If they used 8" wide and you were using 9-1/2" wide, it wouldn't be good enough. Just like going from their infinite baffle to finite using diffraction simulations you'd still have to do the same. And while the on axis sim might be close, no telling about the off axis.

Have only used the diffraction tool to consider baffle step but I concur. If the off-axis sim was really good, the directivity diagram would approximate the Klippel measurement. Also, the driver manufacturer measurement conditions need to be comparable. This would seem to explain messy results.

 

[musicisair]

I remember when Erin measured kit from Parts Express with Klippel and I played in Vituix to make new X-O. To me it was new experience having all up and down, forward and back set of measurements, not just few points in off axis direction. I never build it, but it might be interesting how good or bad was the simulation compared to real loudspeaker. Anyway, forget the straight line on axis target, as the tweeters power response and directivity just past the crossover point will dominate the room response. To counter act this one has to design X-O with a few dB dip. And actually it is very tricky to get the voicing right, half dB here and there might ruin the tonal balance. This was mentioned in earlier post by some of members of the ASR.
I tried really hard to use manufacturers data, then run the baffle step simulation and use the new frd to design X-O. It was sort of OK, but the final tuned by ear, based on actual measured data was always different than the first few raw examples. On top of that you have to take into account 'driver sound', and most seasoned designers know that alu cone will sound way different than paper or polypropylene or kevlar, carbon. And there are sonic preferences, some like laid back and easy going, some like it bright and very detailed. So yeah, measurements and simulations might not right of the box give the effect one is hoping for. It is still a bit a art, despite the fantastic software like Vituix and measuring tools available today to hobbyists (my 25 old Clio is archaic piece of junk comparing to latest tools one can buy) - getting the meaningful, error free set of data is always the key. Even applying EQ to my headphones based on data published here and there, only very, very few EQ curves will do, as some measuring systems and reference curves might be slightly off from what I consider natural.

Thanks, that's a very helpful insight. I'm surprised aerospace-grade acoustic simulation (with 3D geometry, air/material density/synthesis, etc) hasn't trickled down into tools like VituixCAD yet; I think I was expecting something like ray-tracing, but for audio. Maybe in a few years, I know there game engines working on a version of this now ("Wave Tracing").

If I were going to build this, or something like this (maybe even both the 3td-X and the 3td-XCenter channel) with improvements/customizations for my room and preferences the process would need to look something like:

1. Build enclosure to to spec (maybe round over/chamfer corners).
2. Build crossover to spec, using the most inexpensive components that meet spec (might need to add in some resistors if replacing a high-DCR air-core inductor with an iron-core?).
3. Listen. If I love it, skip to step 6.
4. Measure in-room response and actual impedance (more tools to buy and new things to learn from scratch! yay!).
5. Use this real-room/world data in to VituixCAD.
6. Tweak the crossover design in VituixCAD.
7. Go back to step 2.
8. Optional: rebuild crossover with some less crappy components

Sounds fun (and expensive)! haha

 

[Ktacos]

4. Measure in-room response

You'll wanna measure the speaker away from boundaries if possible, outside is what most people do. You'll probably have to close mic the woofers and merge them with farfield measurements to get something accurate to design a filter around.

with improvements/customizations for my room and preferences

You're better off just sticking with what CSS came up with and using digital EQ to tune the speaker to your preference. CSS did a good job with the crossover on this speaker, I'm not sure there's really much room for modification that would lead to any improvements. I would change two things, the top end boost which is easier dealt with using EQ, and the mid range has a bit too much energy at 6k, it's only 20db down. You could also just do active filters and that would make your life a lot easier and give you a much better sounding speaker.

 

[musicisair]

You'll wanna measure the speaker away from boundaries if possible, outside is what most people do. You'll probably have to close mic the woofers and merge them with farfield measurements to get something accurate to design a filter around.



You're better off just sticking with what CSS came up with and using digital EQ to tune the speaker to your preference. CSS did a good job with the crossover on this speaker, I'm not sure there's really much room for modification that would lead to any improvements. I would change two things, the top end boost which is easier dealt with using EQ, and the mid range has a bit too much energy at 6k, it's only 20db down. You could also just do active filters and that would make your life a lot easier and give you a much better sounding speaker.

> You'll wanna measure the speaker away from boundaries if possible, outside is what most people do.

Ah, interesting. I only just recently learned that a flat in-room response makes for a boring and dull speaker, so that makes a lot of sense.

> You're better off just sticking with what CSS came up with

Most of my motivation is to learn and have fun with enclosure design and (passive crossover) electronics. But my end goal is to create some bespoke speakers, utilizing the same drivers, for use as rear/side surrounds and height channels. So was thinking I could play and experiment with a design that is already pretty great, like riding a bike with training wheels, before delving into more difficult project (though with far less important speakers). Maybe my approach isn't ideal?

 

[Ktacos]

Ah, interesting. I only just recently learned that a flat in-room response makes for a boring and dull speaker, so that makes a lot of sense.

That's not really why you measure as close to anechoic as possible, you do it because the reflections literally just make filter design near impossible. You'll certainly figure this out very quickly if you ever start measuring speakers.

For a 3 way speakers, you need to be quite far from boundaries and this includes the ground to get accuracy low enough to do the woofer to mid right. I'd really suggest starting with a two way first.

 

[Rick Sykora]

> You'll wanna measure the speaker away from boundaries if possible, outside is what most people do.

Ah, interesting. I only just recently learned that a flat in-room response makes for a boring and dull speaker, so that makes a lot of sense.

Suggest you ignore whoever wrote “boring and dull” btw. In a less damped room, would likely sound bright. In a highly damped room, it might be fine. In any case, would not be boring or dull. Some might call it fatiguing but is another subjective term that is overused imo.

 

[musicisair]

Suggest you ignore whoever wrote “boring and dull” btw. In a less damped room, would likely sound bright. In a highly damped room, it might be fine. In any case, would not be boring or dull. Some might call it fatiguing but is another subjective term that is overused imo.

Thanks for checking me there. There are so many subjective and abstract words people use to describe audio I doubt anyone is really even describing the same thing.

 

[musicisair]

Thanks for posting this because it highlighted that we apparently have a misprint of the schematic. We actually use a 0.47 uF cap across the 0.75 mH inductor on the mid circuit.

You have two versions of the crossovers published for both the tower and the center.

Here is the other 3TD-X Tower schematic _with_ the 0.47uF cap:

https://519b00b6-c940-4d04-b319-f074686b5916.usrfiles.com/ugd/519b00_be5ad54f4eaa4fdfb1a3b83de5008908.pdf

And the two for the Center channel kit:

Without the 0.47 uF cap:

https://519b00b6-c940-4d04-b319-f074686b5916.usrfiles.com/ugd/519b00_0aacc20e0ad840fca3812b38938b6cb9.pdf

And with the 0.47 uF cap:

https://519b00b6-c940-4d04-b319-f074686b5916.usrfiles.com/ugd/519b00_ab299731432041a3a9462db9dc76f8e1.pdf

I'm not sure where these can be found on the website, as I couldn't find them again when clicking around just now. I had to search my browser history to find them.

 

[Paweł L]

The 0.47uF capacitor just increases the slope of filter a little and creates a deep narrow dip around a few kHz, it's a sort of a notch filter. Usually the capacitor is series with a few ohm resistor, this way one can control the Q factor, or in other words the depth of filter. Very useful, cheap little trick to attenuate cone resonance.

Suggest you ignore whoever wrote “boring and dull” btw. In a less damped room, would likely sound bright. In a highly damped room, it might be fine. In any case, would not be boring or dull. Some might call it fatiguing but is another subjective term that is overused imo.

Frankly, a learn to appreciate boring and dull. I also like Sennheiser HD 600 series, and just can't stand older Beyers. If anyone of the older diyers remembers Audio Concepts kits by Mike Dzurko, G3 kit was my first floorstanding 3way - smooth and boring, 10" woof close to floor in aperiodic box, 4,5" mid and 1" Vifa Alu dome D25AG35 if I remember correctly. Maybe a bit too laid back, but very easy to place in the room, never boomy and shrill.

 

[Paweł L]

That's not really why you measure as close to anechoic as possible, you do it because the reflections literally just make filter design near impossible. You'll certainly figure this out very quickly if you ever start measuring speakers.

For a 3 way speakers, you need to be quite far from boundaries and this includes the ground to get accuracy low enough to do the woofer to mid right. I'd really suggest starting with a two way first.

Possibly on very tall ladder, in the middle of the corn field in the spring Or Ground Plane in the driveway. The problem is the environmental noise. I will try to use my friends garage one day, when we are completely empty and free. Put the speaker on one leg of hoist and press the Up switch. As for now the best results where with 8' tall ladder, but it's pain in ... to do off axis measurements. And I hate when the neighbors have to cut their grass exactly in the middle of measurements

 

[Ktacos]

Possibly on very tall ladder, in the middle of the corn field in the spring Or Ground Plane in the driveway. The problem is the environmental noise. I will try to use my friends garage one day, when we are completely empty and free. Put the speaker on one leg of hoist and press the Up switch. As for now the best results where with 8' tall ladder, but it's pain in ... to do off axis measurements. And I hate when the neighbors have to cut their grass exactly in the middle of measurements

Yeah environmental noise makes things tough, I've had cicadas basically drown out whole measurement sets.

 

[Rick Sykora]

This speaker may be a better example crossover than any of the ones suggested earlier. All of the filters are simple second order.

The compact layout made it look more complex than it is. The only "unusual" parts are the .47 bypass cap and the resistor in the woofer circuit. Both have comparable impact on the FR just at different frequencies. Also note the reverse polarity on both the midrange and tweeter. Here is a simplified look that should help novices see more clearly...

If you play with shorting out inline components or opening others, you can see how they affect the overall response. By opening the individual drivers, you can isolate a particular driver and its crossover.

SPL Trace is useful but can also introduce some ugly artifacts if unchecked. Should always check to see how well the traced curve matches the source by simply adding the driver and connecting it straight to the source. Also check measurement footnotes as some drivers (notably tweeters) may be measured with a passive filter in place.

 

[Kerry Armes]

You have two versions of the crossovers published for both the tower and the center.

Here is the other 3TD-X Tower schematic _with_ the 0.47uF cap:

https://519b00b6-c940-4d04-b319-f074686b5916.usrfiles.com/ugd/519b00_be5ad54f4eaa4fdfb1a3b83de5008908.pdf

View attachment 409793

And the two for the Center channel kit:

Without the 0.47 uF cap:

https://519b00b6-c940-4d04-b319-f074686b5916.usrfiles.com/ugd/519b00_0aacc20e0ad840fca3812b38938b6cb9.pdf

And with the 0.47 uF cap:

https://519b00b6-c940-4d04-b319-f074686b5916.usrfiles.com/ugd/519b00_ab299731432041a3a9462db9dc76f8e1.pdf

I'm not sure where these can be found on the website, as I couldn't find them again when clicking around just now. I had to search my browser history to find them.

The one with the 0.47 uF is the correct version

 

[Kerry Armes]

The only unusual parts are the .47 bypass cap and the resistor in the woofer circuit. Both have comparable impact on the FR just at different frequencies.

The resistor in the woofer circuit damps the "knee" that can happen sometimes on woofer crossovers and reduces some output bump we would get in the summing around the woofer to mid crossover point. This is a fairly common practice in woofer circuits since you get this benefit and it can also sometimes help align the phase a bit better if needed.

The bypass cap is a notch circuit for the mid. The mid has a little bit of cone breakup, which doesn't impact the distortion but was not as low in output as we wanted with the second order filter. This help notch out the peak and provide a better overall rolloff for the mid.

 

[Kerry Armes]

CSS did a good job with the crossover on this speaker, I'm not sure there's really much room for modification that would lead to any improvements. I would change two things, the top end boost which is easier dealt with using EQ, and the mid range has a bit too much energy at 6k, it's only 20db down. You could also just do active filters and that would make your life a lot easier and give you a much better sounding speaker.

Thanks you and appreciate the feedback. I want to reiterate though that the peaking in the woofer and mid that is being measured here is a result of the other drivers not being disconnected during near field measurement. You can see we verified this in a post on our 1TD-X as well. https://audiosciencereview.com/foru...n-1td-x-kit-speaker-review.41923/post-1488504

As for the tweeter, I personally think that bringing that area down results in a sound that is a bit too dull because the off axis of the tweeter falls off very quickly in that area. The peaking is part of the natural response of the tweeter and not "tilted up" to achieve that result. We just choose not to filter it out because we think it is an overall negative impact to the sound. But I can only hear up to 16 kHz anyway so younger ears might feel a bit different.

 

[Ktacos]

As for the tweeter, I personally think that bringing that area down results in a sound that is a bit too dull because the off axis of the tweeter falls off very quickly in that area. The peaking is part of the natural response of the tweeter and not "tilted up" to achieve that result. We just choose not to filter it out because we think it is an overall negative impact to the sound. But I can only hear up to 16 kHz anyway so younger ears might feel a bit different.

That's where my hearing tops out, but I still find elevations up there to be quite annoying. It's easily altered to ones preference with a single filter and I generally consider a speaker that just needs one filter in the tweeter range to meet preference to be a good speaker.

Thanks you and appreciate the feedback. I want to reiterate though that the peaking in the woofer and mid that is being measured here is a result of the other drivers not being disconnected during near field measurement. You can see we verified this in a post on our 1TD-X as well. https://audiosciencereview.com/foru...n-1td-x-kit-speaker-review.41923/post-1488504

Good to know, thanks for the clarification.

 

[Kerry Armes]

That's where my hearing tops out, but I still find elevations up there to be quite annoying. It's easily altered to ones preference with a single filter and I generally consider a speaker that just needs one filter in the tweeter range to meet preference to be a good speaker.


Good to know, thanks for the clarification.

I definitely find it annoying if the tweeter doesn't have a steep drop off axis. Tweeters like the XT25 or the SB29RDC I'm ok with a little lift there. Broader dispersion in that region with a lift can stand out a lot more to me.