Negative amplifier output impedance to lower Qts?

[chris661]

Hi chaps,

Possibly the wrong place to post this - I was torn between here and Solid State.

Anyway, I have a couple of Eminence Kilomax 15 drivers, with one in a 100L 6th order bandpass enclosure. The drivers have a slightly high Qts which makes them "want" a bigger box. Unfortunately, that won't happen - 100L is already on the large side for a living room in the UK. FWIW, this is a PA subwoofer, and will occasionally be used as such. Isobaric isn't much of an option with the heatsinks in front of each driver.

The 6th order bandpass is split roughly 85L and 15L, with the bigger tuned to 36Hz, and the smaller tuned circa 120Hz.
The result off the too-small enclosure is a peaky 60Hz response.

I'd like to fix this. I can see a couple of options. The first is a full-blown ACEbass set-up: change all the T/S parameters to something more appropriate.
However, when I played with Hornresp, I found changing Re (the driver's DC resistance) to 3.3ohm (from 4.8ohm) was sufficient to get a lovely smooth response from 100Hz to 40Hz.

I gather that Qes (and thus Qts) is derived from Re, but is there anything else?

More to the point, would an amplifier with an output impedance of -1.5ohm apply the change I'm after?

The circuits for variable output impedance are well-known and simple to implement, so I'd much rather try that than ACEbass. There is an issue, though: all of the schematics I've seen use single-ended amplifiers. Is it possible to get negative output impedance using bridged amplifiers?

I know its an odd one, but if I can get this subwoofer sounding nice without having to spend £200 on drivers, that'd be lovely.

Cheers
Chris

 

[GM]

Variable Amplifier Impedance

GM

 

[CharlieLaub]

You might want to review this thread on the topic:
http://www.diyaudio.com/forums/mult...er-qts-you-cant-tuna-fish.html?highlight=tuna

It turns out that negative amplifier impedance (as well as positive impedance) are essentially "tone controls" or "EQ". They both change the output voltage delivered to the driver depending on the driver impedance. This is why a negative output impedance will also raise the output at high frequency (or at whatever frequency) where the driver impedance is rising. But doing EQ by using feedback (especially positive feedback like that used for negative output impedance) is not a good idea when you can just do it at line level before the amp.

Loudspeakers are a tradeoff. There are other approaches for using a driver in a "small" box but you will have to live with less LF extension, or apply more power to extend the LF. Even with more power, you will just get more distortion compared to the same driver in a larger box. You can tame the response hump using EQ or a HP filter. Negative impedance. ACE-BASS. Mount it in a wall in an free-air application to have a low Q response with more extension.

The simplest solution is get another driver that is better suited to home woofer/subwoofer applications instead of a pro drier like the one you mentioned.

 

[Inductor]

15" Eminence Kilomax Pro-15A

You can try small changes to test a better speaker response according to your final target.
Let's start with your initial 6th order bandpass 85L@36Hz/15L@120Hz with a +2dB@55Hz.
-3dB: 43Hz - 219Hz
-6dB: 38Hz - 243Hz
example of portA and portB (diam/length)
9.4 cm/11.80 cm
15.0 cm/13.24 cm

We are going to (/you can) lower a little the frequency (ports), so for a start we take about 1L from each enclosure for alignments purposes only. (The volume of a longer port will take some space inside the speaker.)
Target 6th order bandpass 84L@25Hz/14L@110Hz with a +1dB@50Hz.
-3dB: 38Hz - 222Hz
-6dB: 33Hz - 247Hz
example of portA and portB (diam/length)
9.4 cm/32.51 cm
15.0 cm/20.0 cm
You can try to smooth the high frequencies inherent with this speaker @170Hz with some kind of (test: aperiodic/port treatment/wool?) or an inductor/filter for the unwanted >100Hz frequencies.
Note also the better response at 30Hz of +4dB's. 😀

Note: best for this driver 131L@20Hz/71L@61Hz
-3dB: 37Hz - 89Hz
-6dB: 31Hz - 107Hz

 

[head_unit]

From a different thread:
http://www.diyaudio.com/forums/subwoofers/185482-ace-bass-guide.html

...Keep in mind that ACE BASS is essentially a fancy boost circuit! As Svante will tell you, the passband efficiency of the driver is unchanged. This does not mean that you can model the driver with the "modified" TC parameters in a box modeling program and assume that the entire response is at the raw driver's passband SPL. It is not. This is because the low frequency SPL of the driver is essentially "equalized" up to the passband level by ACE BASS by applying additional power to the driver. That's it! No magic.

ACE BASS "changes" the driver parameters by providing a power lift with some phase shift as well - these are the only things that can change in the electrical circuit after all! This is all implemented using a feedback network, which allows the parameter modification of the ACE BASS to be possible, but also brings some problems if/when the real driver Rdc (voice coil resistance) changes, e.g. when you deliver a lot of power to the driver and the voice coil heats up. Because the entire feedback scheme is based on near full cancellation of the driver's Rdc, when Rdc increases with V.C. heating all of the modified parameters change as well. This causes drift in the performance of the system away from the target, and is one of the main "problems" with the ACE BASS system, in my opinion. There have been many reports of people successfully building and using the circuit, however many of these just use it for a closed box system, for which the Linkwitz Transform would be simpler. The LT does not suffer from the performance drift of the ACE BASS but can only be used for CB systems. The ACE BASS is the only circuit that I am aware of that can be used with vented or passive radiator systems, so it definitely has a niche there. You can just use an amplifier with negative output resistance to "reduce" driver Qes, but this does not extend the response like ACE BASS or the LT.

The other main "problem" with ACE BASS is that in theory it requires that Rdc is completely canceled by the amplifier's negative output resistance. This can often result in an unstable system that goes in to run away feedback, so often only some partial cancellation can be achieved. I believe that this is due to the inductance of the voice coil creating a high-Q circuit within the feedback loop of ACE BASS, so using a low inductance driver (or multiple drivers in parallel to reduce inductance) may make this a non-issue. But when only partial cancellation of Rdc is used to increase stability, ACE BASS may not work well, although exactly how much cancellation and stability are achieved, and the resulting performance of the ACE BASS circuit, are not crystal clear to me. Maybe Svante can comment on this here?

If you use the modeling program Basta!, which has built in modeling of the ACE BASS circuit, you can see some of these issues that I have described above...
-Charlie

I would venture that the same criticisms apply to ACE BASS, Linkwitz Transform, negative output impedance, etc etc etc.

All those schemes take a polynomial representation of driver response and mathematically modify it. The problem is, the driver response is not really polynomial. The Thiele-Small parameters the polynomials come out of are just approximations to what the driver does under actual conditions: varying atmosphere, heating voice coil, aging, and most of all the fact that the impedance and response CHANGE according to the driving input! So you can *kinda* change the driver response with these approximate fixes, but there will be errors in doing so.

How serious are those errors is the big question.

 

[sreten]

Hi,

Such modifications require you to be clever. Noting TS parameters
are essentially small signal, attempting extensive correction of any
sort can all go horribly wrong at high levels in practical reallity.

E.g. you can design subwoofers with high power ports. However
in reallity using smaller open foam lined ports can gracefully "block"
at higher levels without excessive chuffing to compensate for
the overall Q of a driver going up as you driver it harder.

Analysis of ACE Bass versus simple EQ might resolve to them
being exactly equivalent at small signal levels assuming all
sorts of very linear parameters, but large signal they will
be very different, that you can very much guarantee.

Analysing 6th order bandpass boxes shows them to highly
sensitive to driver parameters, exact box tuning etc, you
name it and it must be right to get a flat response.
IMO even if you get it all right at low level, it all goes to
pot at high level, unless say designing for a PA, you
design it to roll off gradually both ends at low level.

rgds, sreten.

 

[bjorno]

...Such modifications require you to be clever. Noting TS parameters
are essentially small signal, attempting extensive correction of any
sort can all go horribly wrong at high levels in practical reallity.
...very linear parameters, but large signal they will
be very different, that you can very much guarantee.

Hi Sreten,

Agree.There are many attempts to overcome erratic Amp-Speaker interactions.Here is an example of an ACE -Bass looka alike Design:

http://pdfs.oppedahl.com/US/6104817.pdf

b🙂

 

[consort_ee_um]

There is no need to resort to such techniques, in this case. The Eminence is an excellent driver having a low Qts and good Xmax so you can use a Keele alignment defined here:
http://www.xlrtechs.com/dbkeele.com/PDF/Keele (1975-07 AES Published) - New Set of VB Alignments
The equations say a box and electronic filter tuned to 31.5Hz and with a volume of 96 litres will give you good response. Try modelling in Winisd which has the 6db boost filter option.
If you have already built the box I guess it will be difficult to modify to a standard bass reflex needed.

 

[doctordata]

There is an issue, though: all of the schematics I've seen use single-ended amplifiers. Is it possible to get negative output impedance using bridged amplifiers?

Yes there is , put a low R resistor (.1 ohm 5 watts) in series with your speaker , differential opamp over the resistor , that's your non inverting feedback signal after some math's , invert the signal with an inverting opamp that's your inverting feedback .I would go for the Acebass though . Cost you 20 bucks to build and very flexible . Enjoy it for years now ! Just search for K140 Acebass in the forum .

Cheers ,

Rens

 

[CharlieLaub]

There is no need to resort to such techniques, in this case. The Eminence is an excellent driver having a low Qts and good Xmax so you can use a Keele alignment defined here:
http://www.xlrtechs.com/dbkeele.com/PDF/Keele (1975-07 AES Published) - New Set of VB Alignments
The equations say a box and electronic filter tuned to 31.5Hz and with a volume of 96 litres will give you good response. Try modelling in Winisd which has the 6db boost filter option.
If you have already built the box I guess it will be difficult to modify to a standard bass reflex needed.

If you read the opening post in this thread, it says that 100L is "too big". Your suggestion of Keele's 6th order alignment is 96 liters, essentially the same as the O/P's 6th order bandpass volume.

 

[CharlieLaub]

Actually, there is one more way to use this driver that has not been mentioned. It's a good way to use any high Vas low Qts driver in a small box. It's the use of the driver completely below its resonance frequency, pioneered by Bag-End in their ELF/Infrasub line. Essentially you put the driver in a very small box compared to Vas so that Qb is around 1.0 or more, although the actual value isn't critical. Fb should be around where you want to cross over to the main speakers. You then apply a boost circuit that uses a double integrator below Fb to boost up the power and flatten the response. You can read an article about this on Rod Elliot's web site:
Sub-Woofer Controller a la Bag End ELF

This is similar to the Linkwitz Transform in many regards. You will be able to use the driver in a very small box indeed, but you will need A HUGE amount of power to raise up the response at the lowest frequencies. But if you pair a pro subwoofer driver like the Kilomax with a pro class-D amplifier with a built in limiter you can just crank up the gain in the circuit on the input to the amp and enjoy some serious bass.

 

[chris661]

There is no need to resort to such techniques, in this case. The Eminence is an excellent driver having a low Qts and good Xmax so you can use a Keele alignment defined here:
http://www.xlrtechs.com/dbkeele.com/PDF/Keele (1975-07 AES Published) - New Set of VB Alignments
The equations say a box and electronic filter tuned to 31.5Hz and with a volume of 96 litres will give you good response. Try modelling in Winisd which has the 6db boost filter option.
If you have already built the box I guess it will be difficult to modify to a standard bass reflex needed.

Hi,
Thanks for posting this. I've experimented with similar set-ups for HT (JBL GTO1214 in 75L tuned 18Hz), but it hadn't occurred to me to do something similar for these 15"s.
I've a QSC USA 850 that'll bridge nicely into a single 15", giving roughly a kilowatt of continuous power. Should give enough headroom for considerable 30Hz output.
FWIW, the link's broken.

I also simulated a 40Hz version of this cabinet. This got me ~4dB of SPL over the 30Hz version, so there's a fair chance I'll be building this, but have multiple ports so that one (or more) can be blocked to change the port tuning. A pair of 4"x6" square ports 17.5" long seems reasonable, though may start chuffing with only one in use.
It'd also be much easier to build a pair of them this way (rather than trying to replicate the Tannoy 6th order BP cab): one of my drivers is still sat there unused.

Charlie, I think you might've missed the bit about this still being a PA subwoofer. A tiny sealed box that'll hit 15Hz without a problem would be cool, but I'm afraid my JBLs are currently serving that purpose already.

Thanks everyone for replying, I'll make my way through the papers in-between exam revision.

Chris

 

[bjorno]

Hi,

FYI.. Testing Qm-Damping to 'Shoehorn' the Driver into the 85 +15 L Box:

b🙂

 

[chris661]

Hi chaps, reviving an old thread.

I'm trying to figure out how low I ought to go for this subwoofer, so I figure the best bet is to find the lowest frequency of the music it's likely to see.

Could anyone recommend any software that'll let me run tracks through to see a spectrum analysis?

TIA
Chris

 

[bentoronto]

Chris' post started with a question that's really about voice-coil sensed motional feedback and quickly turned into (the usual) discussion of mental-games with simulation software in order to trick the speaker closer to nice-curve behaviour.

True, if you are playing mental-games there are many ways to make a curve nice and they all seem OK. But a VAST difference in system performance between making a nice-curve using simple EQ tricks or equally simple T-S what-ifs and going to motional feedback (such as ACE Bass).

Ben