[D OB G]

Hi Graham,

Did you get my email?

I imported the EMI-1550s to Australia.  Cost of drivers $US360.  Cost of freight about $US200 ! ! ! ! ! ! !

I haven't found anything better, since I want a lowish Qt (unless you have a lazy million to buy some Supravox field coils- which would give a fantastic ability to control so many parameters).

The EQ I use is to correct for the room, so all the measurements were done with the pre-EQ function, fanatically changing component values and then measuring room response at the listening position.

By the way, I'm inclined to think that the room EQ fuction has limited usefullness with the DEQX, since the linear phase filter already takes the room into account (it can be set with all high-pass- no filter). 

Once EQ is applied there is no way in DEQX of measuring the total effect, so I get Praxis out.  Being dipoles, taking measurements at a few positions near the listening position and averaging reveals a minimum of irregularities. Your T-bass without EQ does seem to give the best sound.  My baffles are currently only 550mm X 850mm, so Xmax really is an issue.

And I don't think I'm getting any room loading. From below the -3dB point, where ever that might be chosen, the reponse seems to drop at the expected 18dB per octave.

David

[panomaniac]

+6dB would be the maximum gain possible with simulated 'lossless' components.  However a real choke and transformer will have winding resistance...

For sure, for sure.  I just wanted to get an idea of what is going on.  It would be possible to simulate the DCR of the transfo & the choke, but just wanted to keep it simple.

Increasing the R in series with the choke decreases the maximum LF boost and the effect of its value is increased due to the effective LS impedance transformation.

Oddly enough, I didn't see that in the sims.  At least staying not too far from 2.2R (+100 -50%).  But I'll check again.

The primary is not used.

 

Indeed.  I've been thinking about picking up a pair of autotransformers made for 70V commercial sound.  They have various taps that would allow different gains.

An equivalent LS load would be necessary having an impedance peak which matches driver characteristics to properly study the circuit.

Oh yeah, that's a tuff one!  I do have some circuits like that.  There is even a  way to emulate a certain driver's impedance and phase, if you've plotted it (I have). I think in some spice programs you can even play a wav file.  May try this sometime.  But it won't be soon.

  Remember that the voltage waveforms do not tell the whole story and there is a need to observe driver current as back-EMFs are reactively generated.

Right. And I have no idea how, or even if, that can be modeled.  That's the sticker.  Since so much of what you like about this circuit has to do with the reactions between amp and driver, it might not be possible to model it.  Well, at least I don't know how!

I see you have captured the lag introduced by the circuit.

That part come out well.  Always good to look at phase.

One fun thing is the circuit without the inductor.  I becomes a fairly easy way of getting a bass boost that falls off at the very lowest freqs were you don't really want it.  Tunable, too.   Very handy for OB bass.

 

[BrunoB]

Hi Bruno,

Good LF there, and you say a system resonance of 26~27Hz with drivers having a specified Fs of 39Hz.

David's driver has a Qes of 0.23;  yours 0.26;  both low and thus preventing unwanted driver resonance when the driver is not loaded by cabinet air.

The transformer circuit has an advantage other than simply boosting LF amplitude, which anyone can already do with EQ.  It improves the quality of the bass beyond what a directly coupled amplifier can achieve, whether separately EQed or not.

Hi Graham,

thanks four your comments.

I placed other measurements of my sub (impedance curve, waterfall plot) in this album:
http://www.audiocircle.com/index.php?action=gallery;area=browse;album=1451



Bruno

[Graham Maynard]

Hi David,

Yes thanks, I did receive your e-mail, and those Madisound EMI-1550s appear unique amongst the natural accordian edge drivers available.  The low Fs plus high VAS mean that the compliance induced roll-off self protects below wanted OB AF only.

Also the low Qes of the 1550's is ideal for tube amplifier drive and the T-bass circuit.  It is possible run a low Qes driver via a higher source impedance (ie. not directly connected to a SS NFB amplifier) such that its Qms characteristic can be allowed to raise LF output about Fs, though this must be done carefully (most say not allowing Qts to exceed 0.7) in order to achieve the cleanest sounding reproduction.  (This can also be done passively but then needs a higher voltage/power amplifier to make up for the introduced losses.)

You appear to be using the EQ more for taking out room peaks than boosting to achieve flatness, which must be the best way of using it.

I came across this the other day;-
http://www.landtime.com/perm/smallroomdipolebass.htm
where Eric Weitzman sets out his objective findings to challenge some of the oft repeated views about dipole bass reproduction which must put folk off trying dipole LS systems.


Hi Michael,

Yes those 'old fashioned' line transformers could increase this circuit's versatility.

Power transformers are now so specialised that these too have become less versatile.  Your comment makes me remember I still have a couple of old 0-24-30-40-48-60Vac multi-tapped components of at least 5A rating, the likes of which are just not available off the shelves any more, and yet which could be so useful with this circuit.

However, whether going beyond the 1:2 step-up ratio would be sensible or not must depend upon an individual situation.

I suppose that using a high step-up followed by a series choke would be like generating current drive which might extend the level response of a low Fs driver, though here too there would be a limit beyond which the reproduction will then become degraded.  An optimum driving impedance is necessary, and the 1:2 ratio does do a good job.

Simulation normally looks at sine derived SPL and amplitude response plots even though these cannot relate to the reproduced dynamics.

The choke and transformer provide phase linear boost for a leading LF waveform edge whilst the capacitor is still charging, then the capacitor plus inductor together limit the boost at the driver's naturally efficient resonant frequency.  I can hear these changes, but I am not sure whether simulation can adequately represent them.  The energies stored by a driver also relate to its electrical characteristics, which do not directly tally with the overall transduced amplitude/phase characteristics.  

This circuit improves reproduction due to the manner in which energy is transduced in time, and yet I am sure individual current and voltage waveforms must individually look more distorted if viewed in isolation.  Whilst running voltage EQ or a passive choke response into a driver's SPL plot might simulate a flatter response (as with MJK's Alpha-15A exercise), what is heard in real-life due to the eventual 'straight line derivation' can often sound either boomy or un-naturally 'trapped' compared to what is expected.

I have EQed before in many different ways, and yes the measured response can be flattened with an impressive looking LF output, but this T-bass circuit provides for impedance matching with optimum energy transfer, and thus bass energies become released and free to flow musically into the listening area.


Cheers ......... Graham.

Hi Bruno. 
You just posted.  I will follow your link now.  Thanks.

[D OB G]

Hi Graham,

Thanks for the reference to the Eric Weitzman post.

It may be OT, but since we are talking about low freq performance, maybe not.

He states that intuitively he disagrees with the perceived wisdom of room effects on dipole OB, and the maths and measurements of John Kreskovsky, Siegried Linkwitz, and Earl Geddes.

I intuitively disagree with Eric.

His objective "bathroom" test is incomplete.

We aren't given the size of the speaker, the size of the baffle, or the Fs.
The way I read the graph, we get a drop-off from about 100Hz (maybe where the calculated 75 Hz is actually applying) to about 50-60Hz at 18dB per octave (which might be what we would expect if we knew his method).
There is  then an essentially flat response below that, and with two of his measurements, a rising response at the lowest freqs.
How does he explain that?
It would be interesting to know his hypothesis for this behaviour, because with this exception, he is otherwise demonstrating the perceived wisdom.
We are not shown how this freq band is referenced to higher freqs.
For all we know, he might have propped a naked speaker against the wall, and we might be getting 18dB per octave below 1500Hz or something.

There are some interesting measurements in the DIYaudio thread called "Cardioid Bass" which I know you have contributed to.

I'm afraid I trust the rigour of these corroborating results over Eric's unclear methodology.

But then again, I DID measure 15Hz in room with my measurements.
And yet I couldn't hear (feel) this (AT ALL).
(I know what a 16Hz organ note feels like).

Very low freq measurements don't seem to necessarily correlate with what is heard.
Depending on the setup, FFT measurements may not be accurate at low freqs.
Swept sine waves need a long period to be accurate at low freqs.

Maybe you can illuminate the relationships of very low freq measurement behaviour with perceived sound, and the proposed dipole room dependent roll-off, especially as your T-bass effects what is heard?

David 

[Graham Maynard]

Hi David,

You raise many points and I can understand what you are saying. 

OT ? - not at all - entirely relevent;  merely what I term discursion.
Yes I have followed the diyAudio 'Cardioid Bass' thread, and I do not like the way that intensely technical discussion and investigations repeatedly focus on aspects which divert attention, when it is unavoidable fundamentals which should be being discussed.

From our point of view I guess that what it comes down to is that we like OB reproduction because there are no internal cabinet reflections coming back out through the driver cone, no air-spring reactions modifying transduced front of cone output, and no path delayed rear of cone output being internally or externally re-combined with front of cone output;  all of which introduce dips and peaks in the steady sine response and affect phase linearity, thus modify reproduction in 'music' time.

However the majority of us music listeners cannot afford to have IB rooms with which to provide 'ideal' driver mounting.  We can however utilise 'clean' dipole mounting, with any increase in baffle size, wings, U-shape etc. decreasing the frequency at which an axial pressure differential can develop, and then increase either cone surface area or cone displacement to counter the front/rear recombination.

So in your own room you could neither hear nor feel the 15Hz as a pressure sensation in spite of the microphone indicating wave presence at this frequency.
1/4 of a wavelength at 15Hz would likely need you to be outside of your room or even your entire home before you could properly sense pressure waves at this frequency due to baffle size/velocity related differentials around the LS and within the room.

And yet the same situation can apply with a monopole because it can be the room and where you are within the room which determines whether any pressure nodes will develop wrt a source, though of course the nature of the loudspeaker will also influence the room response. 

If there is any way you can get one of your baffles out into a garden or a wide open space, then do please try it. 
You will then hear reproduction to the lowest limit of your sensing capabilities and instantly know why you dislike 'boxy' reproducers.

Must go for now.  Back tomorrow.

Cheers ........... Graham.

[BrunoB]

But then again, I DID measure 15Hz in room with my measurements.
And yet I couldn't hear (feel) this (AT ALL).
(I know what a 16Hz organ note feels like).

Plausible explanation: our hearing/body at very low frequencies is not very sensitive, unlike the mic which has (or should) have the same sensitivity across all frequencies. In other words, to hear or feel very low frequencies, the sound pressure must be much higher than for other frequencies where the ear is more sensitive. I guess that organ that plays the 16 Hz  note is a very powerful sound emitter.

Bruno

[Graham Maynard]

Not thinking freely enough at the moment to cover thread raised points, so just a quick reply.

Hi Bruno,

Did you ever measure the phase characteristic with your microphone when making the amplitude plot of the ripole, between say 20Hz and 200Hz ?
Loudspeakers are normally not phase coherent through this range anyway.

In your link you show four plots.
No 1 if flat, No 4 is falling with falling frequency, yet both are from the listening position.
Does No 1 include EQ ?

In this German link the plot relating to the dipole response does not seem right;-
http://www.lautsprechershop.de/pdf/ripol/ridtahler.pdf

I wonder if their 1x1m^2 baffle response is simulated (or 'engineered') as if supported in free space and not on a floor in a real room ?

The ripole is obviously a very useful design through its intended working range, especially when size is considered, and I am not trying to pick holes, merely attempting understand the dynamics, which includes the phase response.

In the No 4 illustration you show the ripole as being 10dB down at 27Hz wrt 80Hz.  Yet a dipole mounted, low resonance and high VAS driver should be able to equal those figures, and maybe the likes of David's EMI-1550s can be less than 10dB down over the same range without notching becoming necessary around 200>300Hz, and with a single driver being more phase coherent (open clarity) whilst having similar output to two in a ripole (ripole less efficient)  ?

A fair test would be using *two* identical drivers on an easy to make open baffle and comparing with them placed in the same real listening room position.

Cheers ............ Graham.

[Graham Maynard]

Hi David,

In Reply#60 above, you say that the your baffles are 550x880mm.

It is my intention to go vertical with drivers mounted on a narrow baffle, even though this will further decrease LF SPL wrt driver excursion.

In Lynn Olsen's 'book' sized diyAudio thread - Beyond the Ariel - he expresses a wish to have larger drivers mounted beside each other at floor level for maximum SPL, but this is something I am wary of due to lack of considerations for possible floor or lower baffle mount reflected pressure waves affecting reproduction;  ie.  reflections back to individual driver cones, or from one cone to the other via the base or floor, which then modify on-going reproduction after the event which generated the reflecting wave.
I guess it is all down to how high the 'LF' driver is expected to crossover into the next driver's range, and even then there are the drivers above to consider.

I see two solutions for drivers beside each other;-
One being to 'curve' both the front and rear baffle-floor angles so they cannot border significant pressure reflection events:  The reflector at the rear could even be double curved and greatest at the centre line between the drivers in order to angle LF wavefronts towards the sides and not just rearwards. 
Or, to simply lift the main baffle up and gap it by about half an inch with spacer blocks from the floor or its base, so that front/rear air flow prevents pressure build up on both sides.

Erling notes the improved LF SPL of a 'U'-frame in his own 'MJK's OB Bass study' thread, but also reports that via the drivers they engender a recognisable 'sound' of their own which is different from plain open baffle reproduction.  Thus he suggests they should be crossovered low.

Hence I wonder if 'U'-frames might sound better, though also be less efficient, if their length/width/depth dimension tuned characteristics were relieved of pressure energisation by fractionally or resistively gapping the front and sides as well as  resistively damping the end of the 'U'-frame, though of course any motional loading behind a cone will still affect the 'open-ness' of forward reproduction.

Bruno also measures an improved LF response via his Ripole driver loading, where again the working upper limit must be sharply limited.

It is my concern that simply studying amplitude responses wrt frequency does not significantly reveal either dynamic capabilities nor reproduction qualities, such that a simple software generated solution might offer an optimum amplitude response, yet still not sound good due to its 'measured amplitude flatness' including waveform steady sine energised gain due to tuned 'Q' characteristics which subsequently become separately audible when music is being reproducing.

Cheers ............ Graham.

[Graham Maynard]

(Just lost an entire post again because of referencing via Windows Back/Forwards buttons.)

Re David's and Bruno's comments about what is measured compared to what we *hear*.

Minimalist 'audiophile' systems can only sound 'best' at one reproduction level, and that is why I designed this 'loudness' circuit to be inserted once a system has been set to sound correct at the maximum level it can cleanly sustain.  Unfortunately it would require a four gang potentiometer, or two twin ganged with gearing to adjust both channels simultaneously.

http://server6.theimagehosting.com/image.php?img=VC.74a.jpg

http://server6.theimagehosting.com/image.php?img=vcplots.jpg


Cheers ........... Graham.

[D OB G]

Hi Graham,

I didn't know you were intending to match the T-bass to the optimum loudness level.

This makes an interesting comparison to Peter Walker's (of Quad  fame) comment that for any given programme material, there is really only one "correct" volume level.

My current baffle is vertical, but with my next set-up it will be 850 mm wide, and 550 mm high, (the mid/tweeter baffle will be physically separate) and is already made up of two pieces of steel welded together, and when looked at from above at 135 degrees to each other.

This gives me the option to test splaying out the dual drivers, or pointing them inwards.
It will be interesting to hear and measure the differences, especially in light of what you and Lynn and Rudolf (a virtual absense of floor bounce delay, and consequent cancellation aberations) have said.

David

(I will damp the 10 mm thick steel with very thick layers of cork on both sides.  I have had a lot of success using steel).

[Graham Maynard]

Hi David,

I'm hoping to build a system including T-bass which will sound good when running loudly, and then 'loudness' compensate as the level is reduced.
(I have found that you cannot overcome this 'human' weakness without EQing or independently adjusting low/high frequency levels as the volume is turned down.)

It does not surprise me that Peter Walker stated that there is only one correct level for playback, all we do beyond that is please ourselves, and Quad stuck with tone/loudnesss and slope controls on their pre-amps in order that their customers could please themselves.

Sounds like you are going for a really 'dead' baffle with that steel and cork, and clearly you are leaving all your options open for bass driver positioning and alignment.

There have been measurements related to the Ripole, and I have thought about 'W' mounting LF drivers as well, but I must stick to principles relating to pistonic wave launch.  I don't want output from a cone at the inside back of any mounting structure being delayed with respect to any part of a driver cone which is closer to the cabinet front.  After all, kick drums always face the audience.

Cheers ......... Graham.

[BrunoB]

There have been measurements related to the Ripole, and I have thought about 'W' mounting LF drivers as well, but I must stick to principles relating to pistonic wave launch.  I don't want output from a cone at the inside back of any mounting structure being delayed with respect to any part of a driver cone which is closer to the cabinet front.  After all, kick drums always face the audience.

Cheers ......... Graham.

I had the same thought regarding pistonic wave launch: the mounting structure of  my W ripole is the front slot. The sound coming from the deepest part of the cone in the slot is delayed. On the other hand , if I would consider the opening of the slot itself as the sound source, with air moving fast in and out from the slot, then I would have an Air Motion Transformer (AMT) http://en.wikipedia.org/wiki/Air_Motion_Transformer.

Bruno

[Rudolf]

There have been measurements related to the Ripole, and I have thought about 'W' mounting LF drivers as well, but I must stick to principles relating to pistonic wave launch.  I don't want output from a cone at the inside back of any mounting structure being delayed with respect to any part of a driver cone which is closer to the cabinet front.  After all, kick drums always face the audience.

Cheers ......... Graham.

I had the same thought regarding pistonic wave launch: the mounting structure of  my W ripole is the front slot. The sound coming from the deepest part of the cone in the slot is delayed. On the other hand , if I would consider the opening of the slot itself as the sound source, with air moving fast in and out from the slot, then I would have an Air Motion Transformer (AMT) http://en.wikipedia.org/wiki/Air_Motion_Transformer.

Bruno

Come on guys, don´t talk yourself into a frenzy here!

For a 15" driver in a W-frame the distance between the "front" and "back" end of the cone will be no larger than 32 cm. At 100 Hz this is less than 10% of a wavelength.

For a 1" dome tweeter the distance (along the z axis) between the surround and the dome tip is typically 8 mm. This is less than 10 % of the wavelength of 4100 Hz.

Following your logic you would hesitate to use a 1" tweeter for anything above 4 kHz. Are you kidding?

[Graham Maynard]

Hi Bruno and Rudolph,

When there is pistonic wave launch from an OB, the cone displacement creates localised dimension/frequency dependent pressure differentials which act upon the cone, baffle, supports and wings, mountings etc.
Any local pressure alternation cannot energise an independent *cavity* response other than than those which are cone or wing dependent.

A ripole however, is going to be like a cavity resonator where the depth behind the radiating slot can resonate at a 1/4 wavelength depth.  With a 15" driver and say 16" depth this would be at about 850Hz/4 or about 210Hz.  This peak would then need to be notched, which means a sharp response turnover, which in turn means a sharp phase change which is likely to have a significant effect to beyond 1/2 of this frequency, and I suggest maybe even out to 1/3 frequency, at circa 70Hz.

Bruno shows in Reply#55 above that his ripole also has a sharp roll-off at 21Hz.  This really is quite a high 'Q' turnover so again this is going to lead to considerable phase change, and I suggest here out to a minimum of 3x turnover frequency, or about 65Hz.

These two phase characteristics are more or less running into each other and could become a continuously incoherent characteristic, which of course happens with bandpass cabinets anyway.

I don't like bandpass bass due to this lack of waveform coherence.  Bandpass (and Ripole) LS systems might be efficient, and can be rib shaking, but they do not sound 'open' and integrated with the rest of the music programme.  

Thus I must personally regard a Ripole as a 'Sub' and not for use as an integratable LF augmenter with an OB.

So I wonder Bruno if that cavity resonance might be further reduced in amplitude and raised in frequency by inserting a roll of carpet along the top/back/lower surfaces of the inner chambers - like toroids with the slot opening ends cut out.  The slot to back dimension, and, the top to bottom dimension, could then be rendered less constant, and this roll damping would have the greatest effect on the flat open centre line of each camber.  Such a roll might even reduce system 'Q' at LF which I see as essential for a Sub to integrate with room 'gain' characteristics anyway.
Maybe any difference in reproduction could best be observed by listening to music and then quickly pulling the roll out again.

I wonder if anyone ever tried rounding off the backs of Ripole chambers or moulding them in cement ?

Cheers .......... Graham.

[Rudolf]

Graham,
I don´t see how your last post connects in any way with my post before that. Time delays from a spatially extended source are a quite different effect than quarter-wave-resonances of a cavity.

Did it ever appear to you that all drums are "energized" by quarter-wave-resonances? Without the "drum" surround they would sound lean and bass-shy. I don´t see that anyone ever has tried to reduce that resonance "in amplitude and raised in frequency by inserting a roll of carpet along the top/back/lower surfaces of the inner chambers" of a drum. In fact I believe that compact W-, H- and U-frames with a somewhat controlled resonance peak will render most drums better than an equivalent plain flat baffle of large size.

[boudy]

Hi,

This is cool stuff.

I assume that since this circuit interacts with (or depends upon) the impedance characteristics of the driver, that any existing Zobel circuit should be removed?

[Graham Maynard]

Hi Rudolf,

Re your Reply#73.
I am still puzzled as to what you see as the 'Frenzy'.
You wrote >>
For a 15" driver in a W-frame the distance between the "front" and "back" end of the cone will be no larger than 32 cm. At 100 Hz this is less than 10% of a wavelength.<<

and then in Reply#75.
>>I don´t see how your last post connects in any way with my post before that. Time delays from a spatially extended source are a quite different effect than quarter-wave-resonances of a cavity.<<

Fair enough !  Let's discuss:

32cm represents a full 360 degrees of wavelength at approx 1kHz, and when two sources are in phase, plus in line with a receiver/listener there is waveform addition.
32cm between two sources radiating 500Hz are 180 degrees apart and there is total cancellation.
32cm between two sources radiating 250Hz are in quadrature, thus producing a phase shifted resultant.

I have said I consider the ripole to be like a bandpass sub, and, when compared to a standard OB mounted driver, the reproduced Ripole output must be transduced with a different (increasingly lagging with frequency) LF phase characteristic when both are driven by the same electrical source.

(Similarly, 64cm between the outer edges of two cones radiating 125Hz in line with a listener also leads to quadrature component generation such that two drivers beside each other at floor level will generate a radiation pattern which changes twice as much with angle away from the centre axis as will the same drivers when mounted one above the other;  thus the characteristic nature, and not just the amplitude, of upper bass sound reproduction will change more with angle away from axis.)

There is no need to mention tweeters, or to ask me if I have thought about the 'drum' - which is meant to sound the way it does.

You finish off by quite reasonably stating your opinion;-
>> In fact I believe that compact W-, H- and U-frames with a somewhat controlled resonance peak will render most drums better than an equivalent plain flat baffle of large size.<<

However, what about the occasions where the LF driver is also expected to still be contributing some output up to say 250-500Hz to compensate for baffle step changes, and/or a widerange which has greater efficiency above 500Hz than below.
Here all of the LF options you list will be running in the range where tuned peaks and phase changes arise compared to what a reasonable large driver could accomplish on an OB.

You mention a 'controlled resonant peak'. 
The very fact that there is one, as with the Ripole, means that the first half cycle response loses out on amplitude as 'resonant' energy becomes stored to build up the 'controlled resonant peak'. 
I stated similar in Reply#4 above. 
Maybe some do not like me stating this, but it is the case. 

For example look at MJK's own impulse responses on page 10 here:-
http://www.quarter-wave.com/OBs/U_and_H_Frames.pdf

where the OB has the best first cycle dynamic response capabilities (the example shown does however need a lower Qes driver to control subsequent cone motion)

also as illustrated by MJK here on page 11, where the lowest Qes driver provides the best impulse response;-
http://www.quarter-wave.com/OBs/OB_Design.pdf

OB with a lower Qes driver is going to give the most dynamic reproduction, no matter how extended the simulated or measured LF steady sine responses appear with higher Q chassis.

Cheers ......... Graham.

[Graham Maynard]

Hi boudy,

The capacitor and choke can impedance match and tune to the driver's resonance, or be used to tailor LF boost and then cut at or above resonance.

A Zobel is normally connected to counter a driver's rising impedance with frequency, and thus acts at much higher frequency, so should still be used as/if necessary.  Slightly different Zobel component values might sound better though.

Cheers ............ Graham.

[boudy]

Thanks for the clarification Graham.

When you refer to Qes I assume that you are referring to the electrical Q and not the overall Q (qts), n'est ce pas?

I'd appreciate suggestions for starting values of this circuit for the Visaton B200 running full range in a narrow (12" wide) desktop U baffle. I'm currently driving them with a Trends 10.1 but will be moving to a 100W amp when winter returns and I have time to build a kit.