AV123 / GR Research X-Voce Speaker Review

Rate this speaker:

• 1. Poor (headless panther)

Votes: 281 93.4%
• 2. Not terrible (postman panther)

Votes: 10 3.3%
• 3. Fine (happy panther)

Votes: 2 0.7%
• 4. Great (golfing panther)

Votes: 8 2.7%

• Total voters
301

ctrl

Major Contributor
Forum Donor
My question is: Is it possible for a crossover circuit to create a peak that is higher than the raw driver response?
As mentioned by others, this is normal behavior of a woofer with low frequency low-pass filter. The impedance resonance peak of the woofer in the sealed cabinet and the the crossover interact in kind of a "resonant circuit".

As an example. If someone buys an 8 Ohm woofer and thinks that this woofer has constant 8 Ohm, that someone might use a online tool to calculate a LR 4th order low-pass filter for this woofer in a sealed box***.
It might look like this (it's a half room, 2pi simulation, but it doesn't matter in this case). In pink is the LR 4th order @150Hz low-pass filter target function. In red is the FR for the calculated XO filter (I did only a quick approximation) and in blue is the FR of the driver without XO:

It's perfect, but in reality (as everyone knows) the impedance isn't constant.

So when using the real woofer with its impedance peak at the resonance frequency of the sealed box and the filter interact with each other.
In green is the impedance of the low-pass filter interacting with the constant 8 Ohm. In black is the impedance of the low-pass filter interacting with the "real" impedance of the sealed woofer:

Around 82Hz we get an impedance minimum, which will influence the frequency response of the speaker.

In reality we end up with a FR with a peak around 82Hz (which surpasses the woofer SPL without XO) shown in red:

In the same way (low-pass filter and sealed woofer resonance interact with each other), the frequency response peak arises in your measurement of the AV123 woofer with XO compared to the FR measurement without XO.

*** Used a simulation of two GR Research M165-16 woofer with 15L sealed volume for each woofer.

Last edited:

Rick Sykora

Major Contributor
Forum Donor
As mentioned by others, this is normal behavior of a woofer with low frequency low-pass filter. The impedance resonance peak of the woofer in the sealed cabinet and the the crossover interact in kind of a "resonant circuit".

As an example. If someone buys an 8 Ohm woofer and thinks that this woofer has constant 8 Ohm, that someone might use a online tool to calculate a LR 4th order low-pass filter for this woofer in a sealed box***.
It might look like this (it's a half room, 2pi simulation, but it doesn't matter in this case). In pink is the LR 4th order @150Hz low-pass filter target function. In red is the FR for the calculated XO filter (I did only a quick approximation) and in blue is the FR of the driver without XO:
View attachment 330475
It' perfect, but in reality (as everyone knows) the impedance isn't constant.

So when using the real woofer with its impedance peak at the resonance frequency of the sealed box, the filter interacts with the impedance.
In green is the impedance of the low-pass filter interacting with the constant 8 Ohm. In black is the impedance of the low-pass filter interacting with the "real" impedance of the sealed woofer:
View attachment 330476
Around 82Hz we get an impedance minimum, which will influence the frequency response of the speaker.

In reality we end up with a FR with a peak around 82Hz (which surpasses the woofer SPL without XO) shown in red:

View attachment 330478

In the same way (low-pass filter and sealed woofer resonance interact with each other), the frequency response peak arises in your measurement of the AV123 woofer with XO compared to the FR measurement without XO.

*** Used a simulation of two GR Research M165-16 woofer with 15L sealed volume for each woofer.

Lol, had a simpler example but you beat me to it! Seems unlikely he used a 4th order crossover, but I get why you did.

For the sake of completing the example, please adjust the crossover and show how the response can be extended to better transition to the "midrange" section.

Last edited:

ctrl

Major Contributor
Forum Donor
For the sake of completing the example, please adjust the crossover and show how the response can be extended to better transition to the "midrange" section.
You can flatten the filter and/or add an additional notch filter to reduce the peak.

But the real solution to the problem would probably be a different design.
All four 16 Ohm woofers in a closed cabinet with 10-15L volume each (heavy dampened) and then tuned the speaker as a 2.5 way concept.

With sufficient volume and damping, one could further reduce the boost around 130Hz - free field/ 4pi simulation with baffle diffraction included, LR4@2kHz ***:

*** The passive crossover is only valid for the simulation, in reality it will likely have significantly different values.

Last edited:

Rick Sykora

Major Contributor
Forum Donor
You can flatten the filter and/or add an additional notch filter to reduce the peak.

But the real solution to the problem would probably be a different design.
All four 16 Ohm woofers in a closed cabinet with 10-15L volume each (heavy dampened) and then tuned the speaker as a 2.5 way concept.

With sufficient volume and damping, one could further reduce the boost around 130Hz - free field/ 4pi simulation with baffle diffraction included, LR4@2kHz:

View attachment 330487

Thanks, may be will share simpler version or wait until we get the crossover schematic. Would be helpful to have a driver to get an impedance profile. For someone who supposedly supports DIY, Danny’s driver info is decidedly lacking.

Based on @bconline’s info, agree the design change would have to be much more than a single cap change. For that matter, decreasing the iron core inductor is more likely to full some of the gap. To get it right would need to know more about the midrange filter too.

Last edited:

bconline

Member
As mentioned by others, this is normal behavior of a woofer with low frequency low-pass filter. The impedance resonance peak of the woofer in the sealed cabinet and the the crossover interact in kind of a "resonant circuit".

As an example. If someone buys an 8 Ohm woofer and thinks that this woofer has constant 8 Ohm, that someone might use a online tool to calculate a LR 4th order low-pass filter for this woofer in a sealed box***.
It might look like this (it's a half room, 2pi simulation, but it doesn't matter in this case). In pink is the LR 4th order @150Hz low-pass filter target function. In red is the FR for the calculated XO filter (I did only a quick approximation) and in blue is the FR of the driver without XO:
View attachment 330475
It' perfect, but in reality (as everyone knows) the impedance isn't constant.

So when using the real woofer with its impedance peak at the resonance frequency of the sealed box, the filter interacts with the impedance.
In green is the impedance of the low-pass filter interacting with the constant 8 Ohm. In black is the impedance of the low-pass filter interacting with the "real" impedance of the sealed woofer:
View attachment 330476
Around 82Hz we get an impedance minimum, which will influence the frequency response of the speaker.

In reality we end up with a FR with a peak around 82Hz (which surpasses the woofer SPL without XO) shown in red:

View attachment 330478

In the same way (low-pass filter and sealed woofer resonance interact with each other), the frequency response peak arises in your measurement of the AV123 woofer with XO compared to the FR measurement without XO.

*** Used a simulation of two GR Research M165-16 woofer with 15L sealed volume for each woofer.
Thank you very much for the detailed explanation and example. It really helps to understand what's going on. And that peak you simulated looks very familiar!

bennybbbx

Addicted to Fun and Learning
Back in 2019 I asked a technical question on the other forum and never got an answer. Maybe one of you can educate me. Below are nearfield measurements of the woofers with no crossover (blue) and going through the crossover (red). The green traces are with a crossover mod the Danny suggested (remove a capacitor). You can ignore those.
View attachment 330459
My question is: Is it possible for a crossover circuit to create a peak that is higher than the raw driver response? I had assumed that because crossovers are passive they cannot add energy at any frequency. But @kdp mentions "ringing" and @Toni Mas mentions impedance trouble with low crossover points.

Could it be that Danny tried to create a generic 200hz crossover, but the physics ended up creating a 100hz crossover?

you can try rew room sim and try posittion speaker so it fit good with the dips. or depend on your hearing use a DSP or a new crossover for 3 way speakers. in this thread the measures show clear the bass LPF is too low and the mid hpf filter begin too high and the bass woofer is too loud. I guess this is a special hifi design the developer like this way and it fits on his hear position and small room. it depend on room . in some rooms and placements can sound better as others

there can be REW room calculator function available that can read the spinorama data and create the FR in this room dimensions allow place of hearing position and speaker position . room have a very big influence

Rick Sykora

Major Contributor
Forum Donor
Ok, since I made the effort and was able to confirm the crossover is 2nd order electrical, here is a look based on Amir's measures and @bconline's post. Since I do not have the woofer's impedance profile, used something with comparable specs and adjusted the crossover to roughly match the curves. Here is how that looks...

Based on @bconline measures, Danny's fix was a different cap. If you lower the above cap substantially, you can see the same sort of change he measured...

As this is clearly insufficient to fix the design error measured by Amir, the inductor needs to change too. So, let's redo the crossover to better bridge the gap.

As you can see, the main crossover change is to the value of the inductor. The resulting response looks like this...

Please note I am not suggesting this is the fix for this crossover, this is just illustrating the type of change required to address the ugly response dip. To actually fix, would require knowing more precise woofer specs and the crossover design for the midrange section of the speaker.

Attachments

• 1701347631114.png
6.4 KB · Views: 16

bconline

Member
Ok, since I made the effort and was able to confirm the crossover is 2nd order electrical, here is a look based on Amir's measures and @bconline's post. Since I do not have the woofer's impedance profile, used something with comparable specs and adjusted the crossover to roughly match the curves. Here is how that looks...

View attachment 330588View attachment 330589

Based on @bconline measures, Danny's fix was a different cap. If you lower the above cap substantially, you can see the same sort of change he measured...
View attachment 330594

As this is clearly insufficient to fix the design error measured by Amir, the inductor needs to change too. So, let's redo the crossover to better bridge the gap.

View attachment 330598
As you can see, the main crossover change is to the value of the inductor. The resulting response looks like this...

View attachment 330601

Please note I am not suggesting this is the fix for this crossover, this is just illustrating the type of change required to address the ugly response dip. To actually fix, would require knowing more precise woofer specs and the crossover design for the midrange section of the speaker.
This makes sense to me. The capacitor that Danny had me remove was 80uF, probably where you have the 250uF.

Rick Sykora

Major Contributor
Forum Donor
This makes sense to me. The capacitor that Danny had me remove was 80uF, probably where you have the 250uF. View attachment 330901

80 uF (capacitor C2) would be a typical value for a second order crossover. My largish value was tweaked to approximate Amir's curves since did not have impedance profile for the woofers. This is why I included the disclaimer at the end of my post.

If Danny does not issue a fix, will see if I can get enough details to supply a solid fix.

Rick Sykora

Major Contributor
Forum Donor
Should mention that you could eq a reasonable fix too. Amir has already supplied a good start. As the listening room is a big influence, would be best to include the room in the eq as well.

p.s.

Eq is potentially an even better fix for X-Statik or X-Voce. You may have noticed that the ”fix” I posted earlier would sacrifice some low end bass. Equalization could leave the existing bass rolloff intact but fill in the ugly dip.

Last edited:

RVCollura

Member
Hi. I'm the guy who did the extensive measurements of the X-Statik in 2019 that @amirm mentions in the review. I can tell you first hand that no measuring or golden ear was needed to know that something was very wrong with the bass. 30 seconds of casual listening made it obvious.

After all the testing, discussion, and experimentation, I attributed the problem to a bad batch of woofers that were out of spec. I did this becuase...
1. I thought I was the only person with this problem
2. I did not know enough (and still do not know enough) about crossover design to have any other explaination
3. It was a graceful way to say it's a bad speaker without saying it was a bad design
However, having read this thread...
1. I know that #1 is not true. It seems that many (all?) people who have this design (X-Voce and X-Statik) suffer from the same bass problem.
2. Other posters on this thread, who DO know enough about crossover design, have an explanation for the peak that is inherent in the design.
3. Maybe it IS a bad design.
Hi, thanks for joining and posting. This review and thread has given me much to ponder. I wish I had seen the original thread you posted in 2019. During the Covid lockdown I bought one of the last X-Statik kits with the original wired crossovers and built them as a project (I have been building speakers for decades but wanted something that I thought was well engineered). I have been quite happy with these speakers BUT I have only ever used them in a system with equalization and a pair of subs. After seeing this review, I did a bunch of testing and also listened to them using no subs or correction. I also decided to get a new Denon receiver so I could use Dirac to handle the corrections. This is the measured (before - raw) frequency response I had in my smallish room. Clearly shows the same pattern as the review...

After setting this all up and adding the Dirac filter I have to say that I am still quite happy with the way these speakers sound but I’m more skeptical of anything GR-Research puts out and says about other speakers. I wonder what others who have these speakers have done with them and if they are satisfied with the corrected sound. IMHO, X-Statiks are just not worth using without a sub.

bconline

Member
Should mention that you could eq a reasonable fix too. Amir has already supplied a good start. As the listening room is a big influence, would be best to include the room in the eq as well.

p.s.

Eq is potentially an even better fix for X-Statik or X-Voce. You may have noticed that the ”fix” I posted earlier would sacrifice some low end bass. Equalization could leave the existing bass rolloff intact but fill in the ugly dip.
Yes, I would use eqaualization... IF I still had the set. At the end of 2019, I sold the the X-Statik and X-Voce I had, thinking they were unfixable.

Last edited:

Rick Sykora

Major Contributor
Forum Donor
Did manage to locate a X-Statik crossover schematic but will not post as it is GR intellectual property. In a simple sim (no cabinet comps) can see how Danny could miss but should have been measured to verify. Due to different cabinetry would expect it to have a different power response than the X-Voce too.

I know there is some interest in a posting a fix. Am convinced the X-Voce is more work (new cabinet, crossover redesign) than is worthwhile for a niche center channel. The X-Statik hopefully can be fixed with eq, but otherwise is Danny's issue. Does not appear he is going to publicly admit a design defect, but hoping he at least handles privately. If he does not, PM me and can share some potential remediations.

Last edited:

Steve1138

New Member
Guess you might not get invited to Danny's place for that BBQ after all...
he has been invited down as has Gene Della Salla WITH Audioholics... they just don't want to go. But Chris from Vinyl Attack and Jay of Jay's Iyagi have both been and presented a video response of their visits. I find it rather disappointing they can't get heads together and see what comes out of it for our benefit.

Rick Sykora

Major Contributor
Forum Donor
Danny has posted plans for new videos on X-Voce/X-Statik soon. He has found a sudden need to build new sets of these speakers! He apparently does not admit any design defect but will be “updating” his legacy stock and offering a new crossover with “better” components. As yet there is no mention of a retrofit for existing customers.

Rick Sykora

Major Contributor
Forum Donor
Yet another customer has posted the same poor bass response for his X-Voce on the GR forum. Danny's response is changing again. His latest claim is that most of the hump was due to a floor reflection. Will be interesting to whether he can show any measurements that correlate. Does not seem to understand how a Klippel operates though. Would actually better if he used Amir's measurements to fix but would mean admitting his error.

Danny did offer to work with the customer to help remediate. This seems further progress at least and interestingly fits another known model - Shock, Denial, Anger, Bargaining...

Last edited:
OP

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
He thinks a 70 Hz peak is "floor bounce???" The wavelength is 16 feet. How on earth do you get a directional bump that way? Does he not understand the modal vs statistical region of room response? Here is our measurements:

The power response completely tracks the on-axis indicating that it exists in all directions, floor or no floor. Same with the dip that follows it.

He is handed an anechoic response of the speaker showing that problem and he is still guessing about things: "We set up an older model and was able to replicate that response. Most of that is caused by the floor reflection. We have a newer crossover design that takes the floor reflection into account and has a flat response."

DanielT

Major Contributor
And here is the thread you are referring to. Incidentally where Danny says, and remember his forum doesn't have a music section, music tips like ASR has:

They hate all of us real audiophiles and listeners of music and we are just sellers of "snake oil".

Why didn't he just at least take his AV123 / GR Research X-Voce outside and make some measurements? He would have immediately seen the problems with them from the beginning. Very strange.

voodooless

Grand Contributor
Forum Donor
Why didn't he just at least take his AV123 / GR Research X-Voce outside and make some measurements? He would have immediately seen the problems with them from the beginning. Very strange.
There is zero need for this. Several people have shown the issue with simple simulations. You don’t even need to leave your chair to figure out that this is shit without ever building it…

Replies
98
Views
13K
Replies
32
Views
4K
Replies
149
Views
26K
Replies
225
Views
19K
Replies
327
Views
42K