[johnk...]

I'm not getting this circuit. Your pic shows the amp + connected to the CT of the secondary of the transformer. One end of the secondary is connected to the speaker +. A cap is connected between the amp + and the other end of the secondary. The cap end connected to the transformer is then connected to an inductor in series with a resistor and connected to the amp -. The amp - is also connected to the speaker. This is redrawn in the picture below to the left. ( I edited the figure.)

[scottw]

What (if anything) do I have wrong?

Probably nothing, but is there an effect with the inductors/windings sharing the same laminations (as is the case on the left with the transformer but not on the right with the separate inductors)?

Scott

[Graham Maynard]

Hi John.

Thank you for your enquiry. 
Scott is right.
We cannot draw two windings which are on the same core as if they are two separate windings existing in isolation.  When both are on a common transformer core they are very tightly coupled.
This is why crossover chokes must be kept apart to minimise unwanted coupling, and why AB1 push-pull tube amps still work when only one tube is left conducting through part of an audio cycle.

Any voltage over the first transformer half winding is equally developed (minus winding and core losses) over the second half winding.
At low frequencies the capacitor has highest impedance so the amplifier develops voltage across half the winding via the choke and resistor.  This voltage is stepped up by the second half winding feeding the loudspeaker in series additive connection with respect to amplifier output.
At high frequencies the capacitor shorts one part of the transformer winding and the choke has higher impedance.  The windings are closely coupled so the entire winding is effectively short circuited and amplifier output passes directly to the driver.

The strange bit happens where the capacitor and inductor form a damped series tuned resonant circuit with the resistor, the centre resonant connection then becoming live in time with the driver resonance when the capacitor, inductor and resistor values are adjusted to match the driver's natural resonant frequency and 'Q'.

Via the transformer the driver load across the capacitor damps the series tuned circuit in return, and when this is balance tuned it reduces the driver's natural resonance at the same time as boosting driver output below resonance where the SPL would otherwise go into free-fall.

This circuit is intended for use with the likes of 8 to 15 inch augmentation drivers normally used on a baffle where driver resonance is within the audible bass range (circa 30-50Hz), and it provides some useful extra boost down to 20Hz which counters the normal roll-off when a SS amplifier is being used.

David says his driver resonance already occurs at 23Hz, which means that a larger series choke value might be necessary to match tune the circuit (possibly 10 to 12mH) so that the boost could come in around 15Hz, though a very large transformer core would then be necessary.  It all depends upon where the acoustic roll-off of his system arises, because this varies with baffle dimensions and room position as well as driver TS specifications.  The overall response from low resonance drivers on a baffle does not necessarily need correction only at their resonant frequency, whereupon additional circuit boost might still be better coming in higher than Fs.

With a 30-50Hz driver Fs however, and where driver 'Q' figures are not low, the real difference arises due to tonal change about the driver's resonant frequency.  This tonal change can be 'locked in' by a SS amplifier, yet 'opened up' by the transformer circuit in the same way as happens with some tube amplifiers, because the component values may be adjusted to provide an optimum loudspeaker driving impedance from the SS output.

The optimum driving impedance for a LF driver normally cannot be achieved without introducing power loss (choke or plain resistance) if that LF driver is driven by an already EQed SS NFB amplifier.  A SS NFB amplifier directly connected to a LF driver 'locks' a tonal change into the resulting reproduction due to driver stored energies (moving mass and suspension) being re-released by the driver in 'music-time'.  The lower bass frequencies can be there, but the notes simply do not sound right - as if slowed, stumpy and trapped - not outwardly free flowing with an open musically into the room !

This is why I have been imploring folks to try the circuit.

( I am going to add an original R back into the circuit diagram in series with the capacitor.  This to be between zero and 2.2 ohms to assist amplifiers which might trip due to phase change.  As stated I initially had resistance here, but my amplifier will drive any load angle, and I did not hear sufficient advantage to warrant retention so I removed it. )

Cheers ........ Graham.

[johnk...]

What (if anything) do I have wrong?

Probably nothing, but is there an effect with the inductors/windings sharing the same laminations

Scott

Yes, that dawned on me after I made the post. I edited the figure. But I'm still a little confused by Grahm's discription of the AC impedance. At low frequency isn't it just the upper section of the T winding (Zwinding)  in series with the driver, all that in parallel with the R + L + (C||Zwinding) of the shunt? Neglecting the impedance of the T windings isn't that just the driver Z in parallel with (R + L)? (It's been a very long time since I did anything with transformers in the loop.)

[Graham Maynard]

Hi John,

There is series choke/transformer boost wrt amplifier output below Fs at low impedance.
The capacitor increasingly series passes all amplifier output above Fs at low impedance.
The C+L combine to increase series drive impedance wrt the amplifier at and around driver Fs; this reduces loudspeaker system mass/air storage capabilities.

http://server6.theimagehosting.com/image.php?img=T-bass.f83.png
Click on the image for clarity or saving.


Do you or a friend have a circuit simulator you can try it on ?
Even then what that will show cannot express what is heard, especially with the losses and transduction not being the same as those arising in real life.

Actually I am trying to pull everything together to construct a narrow OB project with LF drivers in the cabinet base too, and although there is cost and difficulty related to choosing a suitable transformer, the test reproduction has already been so transformed that it will be included.  I have tried LF EQ, but the bass becomes too 'heavy' like that.

I have read all of your own reference pages with interest, esp with regard to mixed cardioid/dipole bass, but I still want a simple single SS quality amplifier based system which is capable of driving this circuit and all other drivers without concern.

Cheers ........ Graham.

[D OB G]

Hi Graham,

I used a tapped inductor to test your circuit, and I did indeed need 12 mH to go sub-20Hz. (As you say, I might be better off seeking a boost higher up, especially as I too use a narrow baffle).
How would you specify a transformer that would be suitable for a 12mH choke?

David

[Graham Maynard]

Hi David,

Transformer ?

Needs to cover the amplifier output voltage over one winding at LF.
80W @ 8ohm equates to about 25Vrms, so two windings of 25 to 30VAC would cover this.
80W @ 4ohm equates to about 4.5Arms, so a 4 to 5A rating will cover this.

Thus I would have no qualms about running standard 250 to 330VA transformer ratings with an 80W amplifier, which means in general the transformer VA rating should be 3 to 4 times the rated amplifier output after its maximum rms. voltage output has been determined to choose the winding voltage.
A lower VA rating (2 to 3 times) will be more lossy but will still work !
A higher VA rating (4 to 5 times) will move most air, though only as long as the amplifier has genuinely good current driving capabilities and LF response.
(I note your NAD is -3dB at 2Hz, which is excellent for LF ambience.)

Also I would be confident at using 'old fashioned' and large 'E+I' or 'T+U' core constructions, but suggest that care is taken with modern toroidal types because their minimal centre cores can saturate much more easily, and may well do so at infra-bass frequencies. 
Some toroids intended for PSUs will saturate, some won't.  So if you do try a toroidal type then do check to see if there some threshold at which bass reproduction suddenly starts becoming distorted when it is not amplifier related.

This is why I mentioned size and expense, even though this circuit can still be cheaper and sound better than a separately EQed amplifier or additional LF driver.

( Beginning to sound like a salesman ! )

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

[scorpion]

Graham,

I have tested your circuit with good and bad results. Here are two measurements with the bass unit of my 'Volks-OB'. This speaker has a Qts = 0.73 and fs = 44 Hz so it should be a good candidate for an experiment. The T-circuit is your last published. Measurements are nearfield and with the T inserted after a passive crossover calculated at about 200 Hz with Linkwitz-Riley 12 dB/oct.

First figure is a before T measurement:



The second is the after insertion of T-circuit:



Positive, you does gain some 10 Hz in Bass ! This is an OB speaker and you will go level lower than fs.
Negative, it seems to affect frequency response above the resonance area, rising it. This is unwanted !

Comments ?

Cheers /Erling

[Graham Maynard]

Hi Erling,

Thank you for the technical feedback which supplements my listening observations.

The rise you find around 400Hz is obviously due to some circuit or baffle interection.  I did not hear anything like that here, and the simulator does not show anything either, so I wonder what might be causing this.

Given that this circuit could most suit the likes of the 'Volks-OB' once the R+L+C values have been sorted I should like to look further into this problem.

First though I trust that any other drivers are still being independently driven directly from amplifier output, for this circuit is intended to drive the bass crossover section only.

Possibly a slightly higher choke value would help your application by reducing the frequency at which maximum lift is arising, and maybe an increase in the resistor value in series with the capacitor to reduce the dip around 150Hz.

If the choke you are using happens to be air-core, then it is very easy to increase its value for tuning purposes by doing no more than slip a short length of ferrite rod from a scrap AM radio into the centre hole.  Clip the rod in a clothes peg to hold just the right distance into coil.  Then when you have finished making adjustments, you can measure the inductance and fit an appropriate choke.

Now it might be possible that trying both of these changes will also take out the lift around 400Hz.  If not a possible solution might be a Zobel between input and output of the circuit;  ie.  between the two '+' terminals.  A pure guess first try suggestion here would be 8 ohm + 47uF in series between the '+' terminals.

Looking at your before and after measurements, I am sure you agree there is something of real value to be gained here, and maybe just a little more tweaking to achieve optimum flatness will provide truly rewarding results.


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

PS.  Erling I just noticed you wrote this
>>  with the T inserted after a passive crossover  <<

The T circuit must be directly connected to the amplifier, between it and the and bass crossover section !
So I wonder if you could check for this before trying anything I mention above.
Maybe this is where the 400Hz lift is coming from, and why the LF boost seems more resonant and slightly higher in frequency than expected.

[scorpion]

Hi Graham,

You were right. Next measurements, before and after, is taken with T-circuit before crossover. Levels are directly comparable.
1st figure: nearfield before T, 2nd figure: nearfield after T.





Now it is up to the room to support the new low frequencies.

/Erling

[Graham Maynard]

Hi Erling,

From those plots there should be a little more LF gain.

With a large driver on a baffle it might be possible to extend the LF a little more by removing the resistor in series with the choke.

Worth a try !

To me the most notable improvement is in reproduction, the bass becoming deeper and more open at the same time. 

How do you find it ?


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

[scorpion]

Graham,

I am at loss hear. I only had components for one channel. Then I tried to compare mono signals but component signal loss made that very difficult.
I do think I have to refrain to make any judgement. However 1m distance measurements didn't show any improvment in bass response frequencywise, which may be room dependent.

/Erling

[Graham Maynard]

Hi Erling,

Your before and after measurements certainly showed little difference at low frequencies.

I use loudspeaker sited low impedance SS monoblocks with short thick cables, and I did not just hear the increased LF but could see it too, as a much increased cone displacement on kick drum and bass guitar.

Cheers ....... Graham.

[D OB G]

Hi Graham,

I'm now using 22mH, 7A 0,12,24V toroidal, 2.2ohms in series with 800uF.

I'm getting 5-6 dB boost at 26-30 Hz as determined by the room measurement function of the Deqx HDP3 (having tried umpteen combinations).

I'm only testing one channel at the moment, so I'm off to get the necessaries for the other channel.

I'll then get Praxis out and do some comprehensive measurements.

Even so, I have a CD with very difficult bass- Cassandra Wilson's thunderbird, with a lot of electronica bass, and on the one channel, not only is there more bass, but it is better bass with more tone and especially tunefulness (there actually are melody lines going on!).

David 

[Graham Maynard]

Hi David,

That is so good for me to hear because there really IS more to be gained from dipole bass compared to all others.

The reason for the improved reproduction is the 'C' + 'L' increase in drive impedance.  Fundamentally, resonance is a driver characteristic, but one which becomes modified by amplifier, loudspeaker mounting and room positioning.  The increase in drive impedance prevents a driver from storing more electrical energy than it should in the resonance mechanism(s) and then releasing this with loudspeaker system determined characterisations in its *own time*, instead of the cone being free to reproduce the recorded music waveform.

We are presented with so many options for different types of monopole enclosures (sealed, reflex, ABR, horn, TL etc) or dipole, bipole, cardioid etc, and then the last one I've heard of - multiple bass monopoles per channel, but none are really worth having if those loudspeaker systems employ or embody resonant effects which store energy at one moment in music time and then reproduce it at another.  So many arrangements can provide impressive bass, but using 'energy storage' effects to flatten a frequency response can completely ruin an impulse response, and this can ruin music reproduction as well. 

The transformer circuit has become my own reason for examining a fresh start - with it being part of a dipole design at outset.  My approach is not to use driver 'Q' characteristics to 'flatten' the SPL response, but to use this circuit for both LF boost and that limitation of driver energy storage in real world 'music time', something which separate and more costly EQed amplifiers simply cannot do either, unless they too power loudspeaker drivers through some kind of series electrical network !

I believe that you are now able to hear 'melody' and 'tunefulness' because the muddying effects of energy storage followed by the harmonically disturbing monotonic release due to voltage drive has been overcome via a circuit which additionally allows for individual optimisation relating to individual amplifiers, drivers, their mounting and the listening room.

I wonder if I might ask you another question. 
I think this transformer circuit makes a dipole driver sound faster again (as would any equivalent impedance optimised drive) due to the 'heaviness' of direct voltage reproduction being relieved;  is this your experience ?

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

[D OB G]

Yes Graham, faster, clearer, but with more body!

I came home to my daughter playing NIN (Nine Inch Nails), which has some weird bass.
She asked me if it was all those wires and things that made the bass so much better (and only one channel with T-bass).

I have had better bass compared to what are currently interim speakers before, due to doubled and quadrupled drivers, larger sand filled baffles, and other factors, but not as good as these interim speakers with the T-bass.

When I have the other channel going, I will be able to recalibrate the DEQX response, including the increased drive impedance.  It'll be interesting to see.

All of us open-bafflers have, over time (I've been building open-baffle speakers exclusively since 1978), and between us, tried countless empirical permutations and combinations.
Others, and we all know how they are, have developed the sophisticated theoretical bases for, in the first instance, simulation, and then the realization of SOTA real world speakers.

I want to thank you Graham for the Maynard T-bass circuit, which I fully believe, when the word gets out, will come into wide circulation.

David 

[Graham Maynard]

Thanks for that David, feedback from your daughter too!

There is no fooling those who haven't a clue about the technicalities.

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

LATE  POSTSCRIPT. 
Theory can distract a mind's eye and fixate thought processess, such that established concepts blinker what otherwise might be free thinking towards other possible solutions. 
So often, theoretically based reasoning and education implant notions of impossibility for alternative outcome before novel investigations can get going.  Some who rely upon theory could end up denying the possibility of other real-world compromises which might be equally rewarding.
I think overall though that some very useful investigations are currently being published, though that theory can then also be used like advertising blurb for solutions which are neither inexpensive nor simple to implement, and which could then lock a purchaser into other problems which were not discussed.

[D OB G]

This is the graph of two 15" drivers, L and R (the two parallelish lines) taken from the listening position, with DEQX's room measurement function. The other line is the equalization that DEQX will apply for as flat an in-room response as possible.

Fs 23 Hz, Qe 0.23.

This is with the inclusion of GRAHAM MAYNARD'S T-BASS CIRCUIT (which I implore people to try).

Now 32 mH 1.5ohms, 747 uF 0.47ohms.

Prior to the T-bass the DEQX showed driver roll-off from about 35-40 Hz.  So I presume the low extension is correct (notwithstanding the inability for dipoles to sustain bass below the room fundamental).  Of course Xmax becomes the limiting factor.

With the room equalization, the bass is very good!  (Next version uses two drivers per side).

Thanks Graham,

David

[Graham Maynard]

Hi David,

I note your component values are now more for extending the LF response with your low Fs driver, and that you have a decent response down to 15Hz ! 
The original values I show are more for lifting/flattening the LF where there is no EQ and only a simple crossover to a typical Fs=35~40Hz driver.  Actually, I'm not sure what amplitude EQ you are using there.

I am sure you agree though, that adjusting the component values allows useful adjustment to help optimise for both driver choice and room.  Given that you have 1.5 ohms in series with the choke you are not using the maximum LF boost available (which could be used for parties) but which would become noticeable and tiring during quality music listening.

I wonder if that 50>80Hz dip-peak-dip is due to reflection from behind the baffles, and if this might be reduced by placing a vertical board covered in carpet at 45 degrees between each LF driver and wall to deflect the rear output along a wall instead of into a corner/wall and back to the listening position.

Which 15" drivers are you using ? 
(We can be limited for choice here in UK unless we import.)
I have my eyes on these.
http://www.eminence.com/proaudio_speaker_detail.asp?web_detail_link=KAPPALITE3015LF&speaker_size=15&SUB_CAT_ID=3

These have greater X.max than the other Eminence drivers, decent sensitivity, and a Qes low enough to maintain control without becoming too inefficient at LF.
One 3015LF is likely to be 2dB down in baffle mounted LF output compared to one Beta-15A, but two in parallel would end up being 4dB more sensitive (more output for the same power if the amp can drive them).  More importantly however, two will cleanly provide four times the air displacement of most Eminence 15"ers at the lowest frequencies before becoming X.max limited.

These drivers are intended for use in vented cabinets where they must be powerfully capable, but when unloaded on an open baffle  I would not expect them to remain linear beyond 100W of input to each at 15Hz and thus the copper voice coils should be adequate.

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

[ttan98]

Graham,

15" driver is too big for, WAF is low, I am more akeen towards 12" woofer, say Eminence Beta 12" LF with TS parameters

BETA 12LT

Thiele-Small Parameters
Resonant Frequency (fs)                                  45Hz
Qes                  0.55
Qms                  6.44
Qts                  0.51
Vas                  136.3 liters


What do you think? I think it is OK, if I intend to use 2 in parrallel per side. Also it is quite cheap, $60 each.