Supravox 400-2000 - what to do with them?

[vinylnvalves]

I throught i would pose a question to this forum, as i expect to get a different view point (especially if Scott's around)
In a rash purchase i aquired a S/H pair of Supravox 4000-200's, and am now trying to decide what type of cabinet to put them in, to suit my application.
Background
I purchased these, as i am after low level detail, and somewhere in Mark's speaker thread (saga) i had read that these drivers had some of the best parameters for low level retreval. I also know that too much stuffing also kills low level detail also, so
liking the concept the of a french speaker which has an irregular lattice inside to negate the need for damping material, for example.
I hear FLH or OB being suggested i dont have the room for the first or the space behind the speaker for the later.
I have been listening in mono using my 160 ltr test cabinet, in a sealed configuration, sounds quite nice with a bit of tapering of the midrange to equalise the LF with my DSP. I seem to get enough LF for me, and i cannot really see the cone move from only a few watts. I am tempted to stick with sealed cabinets, but throught i would pose the question first to see if another style other than BR is worth trying. I have 3 months until Owston to build something - size limited to around 200 ltrs and 900 mm tall ( need to sit my midrange horn on it). Will be an active system also.
I have attached the TSP's for a driver.

aaa1.txt

(779 Bytes) Downloaded 220 times

[IslandPink]

Relatively low Q isn't it ?
"This low register driver unit of high efficiency (96 dB) offers a low register quick, deep and very pleasant to listening. Its exponential cone ensures you a broad band-width with a natural cut at 4000 Hz without accidents in run-out. Its engine made up of a ring of 3.5 kg of Alnico makes it possible to obtain the field necessary to reach a such efficiency.
These qualities enable him to be directly coupled with an 1 inch horn. Iin a bassreflex box or jensen, crossed with 500Hz, the response curve will be right with ± 2 dB from 40 Hz until the final frequency of engine 1 inch. "

I might just be able to tweak James into action for some advice on this one ( via email ) . I'll keep you posted.

[Scottmoose]

Based on the supplied data, with an Fh (-3dB mass corner frequency) of 183Hz before the output impedance of the amplifier is counted in, some variation of baffle, probably one of James's U-frame variations 'sounds' favourite to me also. Otherwise, it'd either be a critically damped sealed box, or possibly a nice 200 litre vented. Ideally the latter would be in a 'true' MLTL format, damped for optimal impulse response, but a 900mm height limits that unless you can fold it, which might eat into the floorspace. If you wanted a 'stock' reflex, 200 litres tuned to 30Hz should be a fairly practical starting alignment.

[vinylnvalves]

Thanks guys.

I have been playing with Jeff bagbys spreadsheet and don't really seem to get much of an advantage going from a 200 Ltr sealed to a BR, in the BR I had two impedance peaks of around the same magnitude as the single in the IB. I know as xmax is limited to around 6mm I need to be careful with EQ, the BR was giving me around 2db more output at 35 hz, but I have all the phase and port stuff to content with. The interesting stuff happens when I started introducing leakage, I suppose it's shades of grey between IB and OB.

The MLTL appeals, yes I can fold, I had thought of a TTL to minimise the back wave, but worry about how much stuffing would be needed in the transmission line, and if that will kill the low level detail.

Scott, maybe you can answer why is it important with a seal cabinet to go for a critical alignment, to my uneducated view it seems that a low Q (bigger box) has a shallower roll off and a better impulse response. The response can be tweaked with EQ using the DSP, gaining similar results to adding series resistance to raise Qts without the downsides. What am I missing?

[Scottmoose]

All rear-loaded reflex boxes have a double-saddle impedance peak of varying magnitude depending on tuning & damping; that's innate to all vented enclosures, other than a maximally flat impedance transmission line or a fully reactance-annulled horn. You'll get some variation between different software packages as they'll likely have slightly different air-temperature & pressure assumptions. For a 200 litre box tuned to 30Hz, I would anticipate 3dB more gain at that point over a sealed box of the same volume (approximate Qtc about 0.53). Driver excursion is much of a muchness down to about 50Hz, less for the reflex between roughly 24Hz - 50Hz, and increasing below that, as you'd expect as the box unloads 24dB/octave below Fb. Depending on what material you're planning on listening to, the vented cabinet should have an advantage in terms of power-handling until you hit it with large amounts of sub-bass. Step / impulse response is better for the sealed of course; if you could handle the size, a big EBS tuned very low would ~equal it, the price being size of course. That is typically less significant at lower tuning frequencies than it is higher up, given how rapidly our hearing acuity drops with frequency. Still, technically a critically damped sealed will by definition always have the edge on that front, with only the aperiodic TL matching it on that score.

Re Xmax, while it serves as a guide, I wouldn't place excessive value on it, since there is no single definition for what it actually is. There are a number of ways of generating a figure (I did a count a few years back and gave up when I reached a dozen) that could be called Xmax, all of them give different results for the same driver, and like THD, none of them actually tell you anything useful. For e.g., say you have a driver that has been rated at 5mm Xmax on the assumption that is the point THD hits 10%. a/ for all you know based purely on that number, HD may be at 9.9% from 0.1mm - 4.9mm deflection, which isn't going to sound a heck of a lot better. And b/ purely IMO, THD itself tells you nothing of value. The individual distortion levels might, & even then up to a point that has been disputed by Floyd Toole etc., and probably a lot of people here too (me included for what that's worth). With that all said & done, keeping the driver under the tightest control & the lowest practicable deflection within the design goals is always an advantage. High efficiency is where it's at.

Re the TL, it shouldn't. Done well, it should really have the opposite effect, since the object of the damping is to eliminate the effects of the back-wave, so you're only listening to the direct radiation from the cone. Placing too much damping in very close proximity to the cone however can mass-load the suspension & diaphragm, forcing it out of linear operation or from resonating as designed. A lot of designs where the line cross section is keyed purely off driver Sd without any consideration for what it might need also had that issue, since they were often excessively small, again mass-loading the driver diaphragm with unfortunate results.

For important to go for critical, it isn't, unless it's in line with your design goals. Then it is. In an ideal world, taking sealed alignments and assuming it is in line with the design goals in terms of size, output / power-handling, a Qtc of 0.5 (critically damped) is the optimum since it's transient perfect with the minimal overshoot. Granted, John Kreskovsky has pointed out that higher Qs may in a sense be more 'accurate' in terms of response linearity and even overhang, but that is providing the drivers can get low enough (and you can factor room-gain into that too), so for bass enclosures, this is more a matter of technical discussion. Of course, all of this is purely from the technical POV anyway, and isn't factoring in personal preference etc. Many people don't like critically damped bass alignments since while they're very 'tight' / 'fast', they can often be somewhat 'dry' / 'anaemic' given the shallow rolloff & inherently amped characteristics, so they're best used with drivers that can get very low, mitigating against that, especially when room-gain is factored in. A maximally flat Butterworth (Qtc = 0.707) is probably the most popular compromise.

[IslandPink]

Very cool info
I like what the small MLTL does for me in the 'bass' ( really upper bass on this one ) with this :
http://www.audio-talk.co.uk/phpBB3/view ... f=4&t=5806
and it seemed to work OK using Hornresp to design this - that's what I used, so you could use that for comparison to your own software.
I have light damping in the upper half of this enclosure , long-haired wool , plus 1/2" wool felt on some of the surfaces near the driver ; but not what you'd call 'stuffing' , and it does seem to preserve the dynamics.

[vinylnvalves]

Thanks Scott for that indepth response. Looking at my simulations in the cold light of day, i agree with you regarding sealed v BR.

The only thing i am not getting correlation with is the cabinet resonance in my test cabinet where i get 48hz not the 55hz suggested for a 160 ltr cab. Are the measured TSP out, can only mimic this by increasing qts.

The EBS - i think isnt appropiate as it will compromise the mid range too much. I like the idea of the terminated transmission line TTL to kill the backwave, but havent got my head around how big the cabinet will be so accomodate a 1/2 wave line. I assume it cannot taper much and it will be at least SD at the start.

I was looking at the effects of a leaky sealed cabinet, changing QI which in turn affects qtc. Interestingly it doesn't seem to (in the simulation) change anything expect for the magnitude of the impedance peak reducing it significantly. I know this is variovent technology which is of questionable benefit when compared to a correctly designed BR.

What suprised me was i thought the very leaky cabinet QI=1 would be akin to a H frame so was expecting the cabinet resonance to drop to around Fs depending on the loading.

The leaky cabinet simulation looks to good to be true - not the aperodic cabinet i thought i was going to get. ??

[vinylnvalves]

thought i would share the freq response from the driver in my test cabinet. The curves represent increasing distance away from the cone, there is some evidence of a floor interaction as i move the mic further away from the driver.

Can the effects fo this floor interaction be minimised, by methods other than lowering the driver closer to the floor? I was thinking a vent close to the floor would maybe help, however i think the freq is too high for a BR vent to have any positive contribution.

[vinylnvalves]

I have started to knock up a pair of boxes - wishing a couple of days into it, that i hadn't gone for a non-rectanglar box, as the angular cuts are challenging my trig.

A question for Scott - i was wondering if the lower than predicted resonance in my test boxes, may not be the driver TSP but maybe that my "test" box - used to be MLTL cabinet which has a baffle up the middle to make a 1.6m long line. I had thought that blocking off the vent would convert them into a simple seal cabinet.

[Scottmoose]

Narry a problem.

The only thing i am not getting correlation with is the cabinet resonance in my test cabinet where i get 48hz not the 55hz suggested for a 160 ltr cab. Are the measured TSP out, can only mimic this by increasing qts.

Depends what you're driving them with. If it's something like a SET amp, then the output impedance will raise system Q; my US friends tend to assume this will fall in the region of about 2.5ohms - 4ohms. I probably wouldn't get too worried about it since the room is likely to swamp the effects.

The EBS - i think isnt appropiate as it will compromise the mid range too much.

Why would an extended bass-shelf box alignment compromise the midrange?

I like the idea of the terminated transmission line TTL to kill the backwave, but havent got my head around how big the cabinet will be so accomodate a 1/2 wave line. I assume it cannot taper much and it will be at least SD at the start.

You can taper a sealed TL as much as you like, providing you've sufficient internal volume. Driver Sd has no functional relationship to the cross section of a TL (or any other QW pipe), like any other cabinet, Q, Vas and F0 (Fs) dominate. There will be some differences in behaviour depending on the driver tap position along the pipe & degree of taper -e.g. the attached. This is a generic driver in a 30in line, unstuffed to make the differences clearer. Top = untapered, driver tapped into the line 6in from one end. Middle = line with same internal volume, but with a 1:8 expansion ratio, driver tapped in 6in from the smaller end. Lower = same pipe, but with an 8:1 contraction, driver tapped in 6in from the larger end.

I was looking at the effects of a leaky sealed cabinet, changing QI which in turn affects qtc. Interestingly it doesn't seem to (in the simulation) change anything expect for the magnitude of the impedance peak reducing it significantly. I know this is variovent technology which is of questionable benefit when compared to a correctly designed BR.

Small thought it was questionable, which is why it was in his doctoral thesis, but he left it out of his published papers. Personally, I disagree. Resistive vents can be very useful to lower the effective Q of a given alignment / sealed box volume. Done cautiously, you can also damp the Q of a reflex box, which can be a handy trick to have up your sleeve on occasion.

What suprised me was i thought the very leaky cabinet QI=1 would be akin to a H frame so was expecting the cabinet resonance to drop to around Fs depending on the loading.

The leaky cabinet simulation looks to good to be true - not the aperodic cabinet i thought i was going to get. ??

Nominally closer to a U-frame than an H frame per se, but varies with assumption & most software derived from Helmholtz unless otherwise stated assumes a coincident driver & vent location. Aperiodic technically means 'without period' so the closest you can get to that is a full-sized reactance-annuled horn, or a maximally flat impedance transmission line (hence the name). So when you get to some extremes, the assumptions a lot of software is based on tend to go a bit haywire.

A question for Scott - i was wondering if the lower than predicted resonance in my test boxes, may not be the driver TSP but maybe that my "test" box - used to be MLTL cabinet which has a baffle up the middle to make a 1.6m long line. I had thought that blocking off the vent would convert them into a simple seal cabinet.

Essentially that makes for a 63in sealed TL, so depending on cross section & damping, you may have some effects creeping in from that.

[vinylnvalves]

Thanks Scott - a lot to consider.

You can taper a sealed TL as much as you like, providing you've sufficient internal volume. Driver Sd has no functional relationship to the cross section of a TL (or any other QW pipe), like any other cabinet, Q, Vas and F0 (Fs) dominate. There will be some differences in behaviour depending on the driver tap position along the pipe & degree of taper -e.g. the attached. This is a generic driver in a 30in line, unstuffed to make the differences clearer. Top = untapered, driver tapped into the line 6in from one end. Middle = line with same internal volume, but with a 1:8 expansion ratio, driver tapped in 6in from the smaller end. Lower = same pipe, but with an 8:1 contraction, driver tapped in 6in from the larger end.

I put the TSP's into the quick and dirty TL spreadsheet you recommend. Its suggesting a very big box - see

. I am assuming that the spreadsheet is about generating LF from the port. Dont see how i will fit a sealed TL into a 60x50x80 box.

Small thought it was questionable, which is why it was in his doctoral thesis, but he left it out of his published papers. Personally, I disagree. Resistive vents can be very useful to lower the effective Q of a given alignment / sealed box volume. Done cautiously, you can also damp the Q of a reflex box, which can be a handy trick to have up your sleeve on occasion.

If i understand what you saying and what i have read, the vents are used to make a small box seen bigger, to get the Q down. What i was looking at was dropping the box Q > 1.
It would suggest for a 200 ltr box that 4 variovents would be required, i am considering as part of the next "experiment" that i could use an oversize hole on the front panel which was destined for a reflex port to be a resistive port. So i can compare 3 enclosure types, with a blanking plug and wool felt.

Essentially that makes for a 63in sealed TL, so depending on cross section & damping, you may have some effects creeping in from that.

I might on my existing test cabinet add an additional baffle to extend the line, as its easy to see if i can drop the box resonance further. I am assuming that dropping the box resonance is benefical.

Will be incommunicardo - for then next week, as in the south of france.

[vinylnvalves]

Built a second test box, a 200ltr box. I am currently listening to the Supravox in stereo in the sealed cabinents. Don't sound to shabby - most noticeably improvement is that the midrange seems to be cleaner. Probably because all the group delays are reasonably close as all 3 ways are sealed. Bass extension not as much as I would like, nice and tight but maybe to dry for me. There is bass it's just subtle ( except when I put SAHB faith healer on - and it's definitely there).

Next experiment as I said was to try a low tuned BR, and then to stuff the port, so see what happens.

[IslandPink]

Nice !
That's a big box - I know what 160L was like with the Onkens.
I guess you've got to find the level of porting you can live with that doesn't screw the group delay.
Shame they don't do a slightly lower efficiency alnico 400 with a Q of 0.35 or so - that would be really useful.
I suppose that's the point of the field coil ones, but that's more faff than I could put up with.

[vinylnvalves]

Added a port to one of the test cabinets tuned to 26Hz. (EBS alignment i suppose) No experience of BR cabinets to some help required.

The measured response tied up well with the predictions, expect the higher frequency impedance peak is bigger than the lower one. Do i need to damp more

The frequency response didnt have as much LF extension as anticipated, and didnt do anything for what i thing is floor bounce. The attached trace showns the freq and phase responses of the supravox and the JBL2482 crossed at 350hz LR4. What i am confused about is how the phase changes at 200hz and around the xover. I show the phase of the LF changed in polarity. Neither way looks convincing to me..the blue trace looks marginally better.. Whats am i doing wrong?

[chris661]

There's a fair chance your measurement mic will roll off at each end.

The jumpy phase indicates something is going on. Might be a reflection or standing wave, or your crossover might need work. If you can get them outside a good distance away from reflective surfaces, that'll tell you a lot.

Chris