Wishing a warm welcome to Rythmik Audio

[GirgleMirt]

Yep, due to servo

See here: http://www.rythmikaudio.com/technology.html

*1. Amplified signal

The plate amplifier receives a signal which is filtered on the pre-amp board to achieve correct integration. The power amplifier sends a signal to the driver.

*2. Cone excursion

As the amplified signal passes through the voice coil, the cone moves.

*3. Correction signal

A very thin sensing coil is attached to the voice coil. As the cone moves, a signal is generated. This signal is used to accurately track the movement of the cone. Effectively the sensing coil is a specialized microphone which measures the precise output of the driver.

*4. Feedback

The correction signal is fed back into the summing points of the amplifier. It is compared to the original signal and corrections are made without the use of any active circuits.

*5. Corrected signal is accurately reproduced

The driver now reproduces a signal that is corrected. The sources of distortion which plague other subwoofers are dramatically reduced.

[RythmikAudio]

I was looking at these Rythmik subs on their website, and the frequency response in the specs seem to be the identical curve for all 3. Is that possible?

I need to discuss with the web designer. I may not have communicated them well with him. I have published DS1200 and DS1500 curves before. They are almost idential below 60hz for the reason GirgleMirt mentioned. Above 60hz, the roll-off is different for 12" and 15" due to the magnitude of inductance are different in those two models. Anyway, I will correct that as soon as possible.

BTW, Paul Spencer has helped me designed these webs and he has done a great job. He is also an audiophile and owns a pair of directservo subs.

There is a historical reason for me try to clone the frequency response. When I was designing DS15 (our first 15" driver), I had problem deciding what roll-off contouring I should use for the low end. I want it sound as close to our 12" as possible. I constantly switch between the two in my listening room. And I struggled for a month and then the face of Bob Carver gave me a hint and then I decide I should clone the frequency response as much as possible. Then if there is any difference, it is due to driver and not due to frequency reponse. Ever since then, that is our design philosiphy and it has helped us to design good driver and subwoofer. Whenever a driver does not sound as dynamic, we go look at the spider, the surround, and all the components that can affect sound quality.

For those not familiar with what Bob Carver has done, he is the first one successfully implemented transfer function cloning for pre-amp and power amps and demostrated in front of a panel of audiophiles, not just once, but twice. The second time was from Stereophile. The first time was from TAS. Stereophile felt the first event was not carefully designed so they gave him a tube power amplifier wrapped in a box or case that Bob cannot see what is inside. The rest is history.

-Brian

[clarke68]

I want to clarify a bit though because if we just focus on pace, rhythm and timing, it may sound very mechanical. If we analyze music techniques such as portamento (techniques of gliding between notes in singing or strings), what does it take to faithfully reproduce that type of sound? I think it is contrast and articulation (and not so much of rhythm as defined as periodic beat). Yes, human voice and strings are two of my favorite instruments, so my view may be biased.

Good point...you've obviously thought about this quite a bit! You may be biased, but a bias towards the human voice is probably a good thing for any speaker designer.

I love to hear what some of your favorite recordings are, both for general listening enjoyment and for subwoofer testing. For the latter, I'd particularly like to know specific things you listen for, and what they "mean" to you when evaluating/improving a product.

I need to discuss with the web designer. I may not have communicated them well with him.

I think you may have similar issues in the "specifications" section...the F15 and the D15SE are listed as having the same dimensions.

[millerwill]

I need to discuss with the web designer. I may not have communicated them well with him. I have published DS1200 and DS1500 curves before. They are almost idential below 60hz for the reason GirgleMirt mentioned. Above 60hz, the roll-off is different for 12" and 15" due to the magnitude of inductance are different in those two models. Anyway, I will correct that as soon as possible.

Thanks much for the detailed reply; much appreciated.

BTW, I presume that the 12" sub would be quite adequate for my medium small room, 2200 ct ft; is that right? I've had a SVS PB10 ISD for ~ 3 yrs and liked it, though I'm always open to something better.

Also, is the only difference between the F12 and F12SE the finish? (I actually prefer the black oak of the F12, to make it 'disappear' more easily.)

[Blake1214]

There is a historical reason for me try to clone the frequency response. When I was designing DS15 (our first 15" driver), I had problem deciding what roll-off contouring I should use for the low end. I want it sound as close to our 12" as possible. I constantly switch between the two in my listening room.

Hi Brian,

If you guys tried to make the 15" sound as close to the 12" as possible, does this mean the 12" driver sounds better than the 15"? The 15" has more output but gives up slightly on SQ? If so, will it be correct to say we can just add more 12"s to our setup to increase output but won't be able to improve SQ with the 15". Which one of these drivers do you think will be a better setup with the Sierra 1's? (In a dual setup)

[scape]

Hi Brian,

If you guys tried to make the 15" sound as close to the 12" as possible, does this mean the 12" driver sounds better than the 15"? The 15" has more output but gives up slightly on SQ? If so, will it be correct to say we can just add more 12"s to our setup to increase output but won't be able to improve SQ with the 15". Which one of these drivers do you think will be a better setup with the Sierra 1's? (In a dual setup)

+1 on this question, I started to get that impression from this thread. over time of research I heard forward firing was also better for music, but I think that is too subjective to the layout of the sub and quality of components.
I did stumble upon this thread, ironically while reading this one, which may help. but I'm sure Brian will clear this up anyway

[RythmikAudio]

+1 on this question, I started to get that impression from this thread. over time of research I heard forward firing was also better for music, but I think that is too subjective to the layout of the sub and quality of components.
I did stumble upon this thread, ironically while reading this one, which may help. but I'm sure Brian will clear this up anyway

The difference in sound between 15" and 12" drivers are now less than DS15. You really have to play them in the same room and back to back to hear the difference and even then, the difference is very subtle. If you crossover at 60hz or below (the most likely scenario with Sierra-1), the difference is almost zero. I hope this will clear things up. If not, I need to explain how servo works.

Dave had both of my 15" and 12" complete subs in test boxes last year for his evaluation. He was surprised that the two frequency response are almost right on each other. That is a feature, but a mistake. Keep in mind, what is the difference between servo and nonservo? Servo has feedback connected from the sensor output and nonservo has feedback connected from the amplifier outputs. With feedback connected from amplifier output in nonservo, it can only control the amplifier output to follow a linear relation with respect to input signal. The actual speaker output still needs to go through the transformation from electrical energy to acoustic energy. With servo, we are controlling so that sensor signal is linearly proportional to the input signal, completely bypass the interface between amplifier and speaker. As a matter of fact, we can even insert a 6ohms resistor in series with the driver coil and the frequency response would still be the same. We can also change the enclosure volume to 2x as big in sealed box (the case for vented box is a bit different) and still frequency response is still very close +/-1db or less. All of these demonstrate the consistency and coherence of servo subwoofer.

The new driver DS1500 in both F15/D15 is different from DS15 (our first generation 15" driver) in that the moving mass has been reduced by 20%. I contribute that to a more dynamic sound in the new model.

I seldom disclose or explain how servo reduces distortion. And that have left a lot to imagination. I've even seen one comment that even in servo you need to have the same linearity as nonservo. My reply wa very simple: "have you written down the close form equation for distortion analysis?" As a remedy of the problem, I now want to design a PowerPoint presentation to describe how the servo and nonservo are different. As a matter of fact, the two can be compared side by side to get an apple to apple comparison through the method of equivalent circuit analysis.

For those who are versed in T/S parameters, let us now pretend we don't even know T/S parameters and think what the impedance of a bass driver is made of. It is an ideal motor (called back EMF) in series with a voice coil resistor. Let us ignore the voice coil inductance for now. The current goes through the driver and causes an IR drop (current times resistor) which is a voltage number. To keep the discussion complete, let us say the voice coil resistance is Rvc, and the current is Ivc. Back EMF is also a voltage number and it is related to BL*v. BL is also referred to as BL profile which may not be linear. The amplifier output, Vop = BL*v+Ivc*Rvc. Vop is distortion free. Let us further assume BL is perfectly linear and v is perfectly linear. Does that mean the equation is balanced now? No. If Rvc is 0 (which is type of assuming we use a superconductor), then the equation is balanced. If Rvc is not 0, there will be counter-distortion current to counteract with the nonlinearity of spider/surround, flux modulation, all sorts of distortion mechanism that has no bearing with BL linearity. That counter-distortion is still a distortion components. That components creates a nonlinear IR drop with Idist*Rvc. Now the equation is not balanced at all. Right hand side have disortion component and left hand side does not. So BL*v needs to have distortion component that is opposite sign of Idist*Rvc and that is how the velocity of cone becomes distorted even when you have perfectly linear BL profile. Yes, we have heard some manufacturers want you to believe as long as they have linear BL, then it will be the ultimate drivers. Here you hear the other side of that story. Also Rvc is voice coil resistor with a temperature coefficient of 0.3%/Celsius. As the Ivc*Rvc modulates over time, there will be additional distortion mechanism. So here is the summary for the distortion in nonservo subwoofer:
1) The distortion and hysteresis (memory effect) of spider/surround creates a Idist*Rvc component as an internal source of distortion, this source goes to affect BL*v and cause it the have distortion components and cause a new Idist value and this process iterate until the equation is balanced.
2) BL profile can cause a distortion component.
3) Voice coil resistance can modulate over time and cause a modulation distortion best described as thermal memory distortion.

So the total distortion is contribution from all 3 above.

Now how does servo's equivalent circuit compares to the above? Our servo has a unique feature of taking both velocity sensing and current sensing feedback. The result is it creates an equivalent circuit that is very similar to nonservo except a couple of "subtle differences".

1) The BL profile is replaced with the BL profile of sensing coil (not the driver coil). We have designed the sensing coil to be super overhang. It is way more linear and this does not reduce the efficiency of the speaker at all. I've seen several so-called "linear motors" widens the magnetic gap so much in order to get linear BL profile. It reduces efficiency as a side-effect and that leads to other type of distortion. Here we take a different approach. We can achieve linear BL profile while not widening the gap at all.

2) The equivalent Rvc value is almost resistive and it is only 1/3 of actual voice coil resistance. Whenever it is possible, we'd like to make equivalent Rvc as low as possible while maintain unconditional stable characteristic of the system. 1/3 is what we can achieve so far.

3) The temperature coefficient of Rvc is now 90ppm/Celsius because we use Mills MRA10 audiophile resistor as the current sensing resistor.

To complete the picture, the equivalent circuit of servo is again a back EMF (with profile from sensing coil) in series with a resistor only 1/3 of physical voice coil value and with a temperature coefficient of only 0.009%/C. Because the Rvc is now only 1/3, the distortion caused by spider/surround... is only 1/3 of nonservo and we are trying to put this number as small as possible and in the future it can be even lower. As I have mentioned before, surround/spider distortion is nonlinear memory effectdistortion. BL profile distortion is linear and coherent/consistent all the time. That 1/3x reduction addresses both types of distortions.

The physical aspect of the driver such as driver coil's BL profile and voice coil resistor are not in the equation at all. The distortion components of driver coil's BL nonlinearity is indirectly in the equation because the distortion caused by it is reduced to 1/3 because of 1/3 equivalent Rvc.

All these have contributed to improvement in sound. Another important feature of directservo is now we have 3 different Q characteristic in the system:

1) The physical Q value. Physical Q value determines the energy transfer ratio between our amps and speaker. For instance, even though 0 voice coil resistance can eliminate a lot of distortion sources, it causes an undesirable stall effect that the frequency response follows a -6db/oct starting from 100hz all the way to 5hz which is not very good at all. The best physical Q value is around 0.7 to 0.9 to ensure max energy transfer in the pass band.

2) The system Q value determined by the equivalent circuit should be as low as possible as it is proportional to "equivalent" Rvc. Currently it is 0.3 for 12" driver and 0.25 for 15" driver. Lower equivalent Rvc increases damping force. When you push the cone, you can feel a stronger resistive force pushing you back. This characteristic is best observed in free air because once you are in a box and you really need to push fast enough to emulate the frequencies above 10hz (otherwise, you will be just excercising the spring force of a sealed box). In a spring-mass-friction system, resistive force is the only one that dissipates energy (this is also how we are different accelerometer-based approach, which implements a very large spring force in the equivalent circuit, and spring force does not dissipate energy). It also means the cone would transfer less and dissipate more of the standing wave inside the enclosure. Moreover, if the system is accidentally push out of their linear operation range (such as clipping) or deviated from their linear path due to nonlearity of the system, it can recover 3x faster than nonservo as soon as it is back to linear range. The result is a sense of fast bass under almost all condition.

3) Final time domain/frequency domain Q value. Although system Q should be as low as possible as we push below 0.25, the final frequency response from the sub stay above 0.5. Anything below 0.5 can sound very lean. For this reason, we have provided three settings for everyone to try (0.5, 0.707, 1.0) and most listeners like 0.5.

Servo is the only approach that can completely decouple all three Q values without one stepping on another. Our directservo further improves on that as 1) our impedance loading component is more resistive than accelerometer-based approach, and 2) There is only two resistor and one capacitor between our sensing coil and the feedback summing point of power amplifier. It has the least coloration. So I also like to call our servo as no-catch servo technique. BTW, DirectServo means direct coupling, both electrically and mechanically.

[ccotenj]

man, that's a lot to chew on on a sunday morning...

thanks brian...

[buddhadas]

...Or any other morning for that matter. Thank God I was fortunate enough to stumble upon Ascend speakers and find speakers I like because they sound good. Knowing Dave, and how he always puts the best product out there, I can only assume that the Rythmik subs will sound equally as good. No offense Brian, but that description was just a schooch beyond me.
I do appreciate your willingness to discuss your product(s) with us here.

Peace,

Jim

[RythmikAudio]

...Or any other morning for that matter. Thank God I was fortunate enough to stumble upon Ascend speakers and find speakers I like because they sound good. Knowing Dave, and how he always puts the best product out there, I can only assume that the Rythmik subs will sound equally as good. No offense Brian, but that description was just a schooch beyond me.
I do appreciate your willingness to discuss your product(s) with us here.

Peace,

Jim

It is ok. For non-techies, I have another version to compare nonservo vs servo. It is like compare a dumb bomb and a smart bomb. A dumb bomb can home in a target when wheather condition is good. A smart bomb constantly adjust its direction between iis current position and target position and the result is a higher hit rate in practical use.