Driver cooling.

[Welder]

My new speaker cabinets are almost finished.
The crossover is now in its own vented enclosure on the back of the cabinets and I’m looking at other ways to reduce the impedance changes that make passive crossover design somewhat challenging.
The main contributing factor to impedance change in a driver is heat which gives a constantly changing crossover frequency.
My current high frequency unit is ferro liquid cooled and in general hi frequency units don’t dissipate a lot of heat compared to the more energetic mid and bass drivers.
In sealed enclosures the transfer of heat across the cabinet walls is governed by the temp inside the cab, the temp outside and the heat transfer properties of the construction material; not a lot to play with there.
I was wondering if by increasing the area of the driver magnet surface on mid and bass units with an alloy finned arrangement better control of the rate at which heat is removed from the driver itself might give a more predictable change in driver impedance.
Has anyone had any experience of such an arrangement?

[Techno Commander]

Theoretically, a heatsink fixed using a thermal epoxy would work in dissipating some heat.
However, most hi fi speakers generally dont get caned enough to need this. Even in PA applications, the size of the magnet is sufficient. The only notable exceptions being the Volt radial drivers, which use the basket as a heat sink.

[DSJR]

ATC did the same and abandoned the sinks when the SL series was launched with no ill effects. The rapid movement of the coils in air, coupled with the dynamic nature of music signals shouldn't cause any problems at all - if it did, there'd be far more melted coil assemblies out there and it's most often blown amps that damage bass units - and clipping amps which damage tweeters..

[Welder]

Thanks for the reply Andy.
I was aware of the Volt Radial config (well, I would be wouldn’t I given my Volt fan boy status) so yes, I had something vaguely like that in mind.
Of course, impedance correction can be achieved with the use of a Zobel network post crossover, but I’m not keen to add more electronics to an already fairly complex crossover and of course, any Zobel filter added becomes part of the driver and one needs to recalculate the crossover all over again.
What I’m aiming at is a reduction in the maximum temperature of the mid and bass drivers in the hope of reducing the impedance variation range.

Hi Dave.
Good Grief :) I’m not looking at this to prevent driver damage due to excessive heat. It’s the heat effect on the driver impedance and the effect that has on the crossover that interests me ;)

[Reid Malenfant]

Hi John, i don't think it'll have much effect even on your 3 way speakers ;) I think you'll find that the inductance of the voicecoil will swamp any impedance changes cause by the voicecoil heating up :)

As the inductance won't change with voicecoil temperature i don't think you have too much to worry about :eyebrows:

[Welder]

Hey Mark.
Yes, you’re probably right, just been doing the calculations.
I really need to put the drivers in the cab and measure what sort of temperatures I’m getting in the new cab before I start thinking about this any further.
That’s the bitch about 3 way crossovers in sealed cabs, the filter slopes wobble about sumat dreadful as the impedances change.
I’m considering active amping the volts and passive crossovers the the scans……………….hmm, maybe stick the quantums in...........

[DSJR]

Simples - GO ACTIVE!!!!!!!!! All your worries eliminated at a stroke - and a better sound as well :)

[Reid Malenfant]

Hey Mark.
Yes, you’re probably right, just been doing the calculations.
I really need to put the drivers in the cab and measure what sort of temperatures I’m getting in the new cab before I start thinking about this any further.
That’s the bitch about 3 way crossovers in sealed cabs, the filter slopes wobble about sumat dreadful as the impedances change.
I thought your enclosures had ABRs? I guess you could still kind of call them "sealed" in a heat sense at any rate ;)

The worst time for cooling in your case will be the ABR resonance frequency. At that frequency the main driver will be moving the least so it won't be pumping air in & out of the voicecoil assembly like it would at say double this frequency where it should be moving the most (if fed the same AC voltage at both frequencies).

Ok, it's a guess, but a pretty accurate one given your bass drivers parameters. They have a low driver Qts & fairly low VAS so maximum damping will occur at the enclosure (read ABR) resonant frequency & minimum at about twice this. It just so happens that the driver will have an impedance peak at the point where it'll also move the most which will lower apparent power even more.

I know i can forget about subsonics as i recall you mentioned brick wall low frequency filtering. So unless you feed the thing high power deep bass it really won't warm up that much at all, besides that the voicecoil is 3" diameter if i remember correctly - you have nothing to worry about ;)

E2A:- The magnet assembly on those drivers is also huge! It'll suck up heat from the voicecoil very easily & it'd take a long time for the whole thing to be run on high power with some seriously low bass for the voicecoil to get hot as it'd need to heat up the whole magnetic assembly ;)

[Welder]

Yes of course Dave that is the ideal solution and which as you can see from my sig I’ve attempted.
However, “going active” isn’t so straightforward unless you nip out and buy a ready made system.

For a start, I would need 4 more amps, then I would have to redesign and built new speaker enclosures to house the amps. Then I would need an active crossover and because my speakers are custom built it would need to be a crossover I could adjust. Course, I could just bin my speakers and start again but I’ve had them for 30 odd years now, spent a couple of grand on upgrades and alterations and even if I sold them I wouldn’t get anything like what they are worth.
I do hope you’re not going to suggest I buy an AVI system :doh:

Anyway, I believe Marco is starting to panic at the thought that I might get Moko to measure something setting up his deck and I’m interested to read any advice you give the guy :lol:

[Welder]

Hey Mark.
Sure, in the strictest sense the cabs aren’t sealed because of the ABR’s.
Yep I’ve brick walled below 20Hz..
The magnets are rather large aren’t they and I take your point and thank you for bothering to have a look at what I have, and considering what I’m thinking about :)

It’s just I’ve got a couple of ideas in my head and want to explore them before I finish the speakers off.
I’m hoping that having the crossovers out of the main cab will make at least in theory some improvements, along with the shorter internal wiring and of course the new damping and bracing arrangement.
I had as I mentioned considered a zobel network primarily for the midrange to iron out a small difference in actual impedance; slightly pissed of given these were sold as matched drivers but still….

I may not hear any difference at all, but at least I will have attempted to get the best out of the current crossover/driver arrangement.
One cool thing is I can attenuate the drivers with regard to the listening environment more easily now and I may even be able to iron out a couple of small peaks in the frequency response now.
Never been quite sure that flatter response necessarily means better listening experience but I’ll see when everything is back together and run in.

[Reid Malenfant]

John, the fact that the crossover is outside of the enclosure is a good idea in itself :) There will be far less microphony, after all it'll be much harder for the speaker to vibrate components such as capacitors which definately suffer from this.

As they are passive they'll have losses. The law of conservation of energy means that the losses will ultimately be given up as heat, so you have removed another source of heat from the enclosurers as well - double benefit ;)

The crossovers (original Yamaha NS1000M) that feed the lower mid, mid & tweeter in my speakers are situated in the stands. It does make a difference getting them away from the interior of the enclosure :) Definate improvement imo, more detail & less smearing...

Kef have tried similar arrangements & given it the fancy moniker of "conjugate load matching"... In laymans terms it means they designed the crossover to work along with the drivers to give a much flatter impedance curve. This is definately a good idea if you happen to use a low feedback power amplifier as the gain will remain just about the same. With a class B amp with high feedback it probably won't make any discernable difference.

[Welder]

Hi Mark :)

”As the inductance won't change with voicecoil temperature i don't think you have too much to worry about”

Is this right? Doesn’t one get a rise in impedance due to semi inductance of the voice coil and that is also affected by temperature? I have always believed this was what Zobel networks attempted to rectify…………. a rise in inductive reactance gives a drop in capacitive reactance giving a resistance equal to the DC resistance of the voice coil.

You may well be right that the voice coil inductance will swamp any benefit of attempting to physically reduce the heating of the drive unit though.

I think I'll have to dig out some text books as I learnt this stuff many years ago and I'm not sure i remember some of it that well :scratch:

[YNWaN]

The problem with adding heatsinking to the magnet is that it doesn't really require it. The component that actually heats up is the voicecoil and this is connected to the cone and not to the magnet. For this reason, metal cones are often claimed to reduce thermal compression as they act as a heatsink for the voicecoil (depending on how the voicecoil is attached).

[Reid Malenfant]

Hi Mark :)

”As the inductance won't change with voicecoil temperature i don't think you have too much to worry about”

Is this right? Doesn’t one get a rise in impedance due to semi inductance of the voice coil and that is also affected by temperature?
Hi John. yes it's right as far as i know ;) The only thing that will alter is the DC & AC resistance of the voicecoil itself. To change the inductance you'd need to increase or decrease the amount of turns in the gap, as this isn't going to happen only the resistance can change :)

As i say, the inductance will or should damn near swamp the increased AC resistance due to coil heating. You'll find a rising impedance from as little as 80Hz or there abouts that is entirely down to this inductance & i think you'll find it'll cover up any increase in coil resistance as seen by the crossover.

You may well find an inductive reactance that equates to about 20 - 30 ohms at about 500Hz & you'd need to heat the voicecoil well beyond the temperature where the copper windings would turn to slag to get an 8 ohm coil to the same resistance :doh:

[Welder]

Thanks Mark :)

I’m sure I used to have a better grip on this stuff years ago :scratch:
I think I’ll concentrate on getting the bloody things put together and get listening.
It’s rather exciting. I may even weld up some new stands if I’ve got any gas left :eyebrows:

[Reid Malenfant]

I’m sure I used to have a better grip on this stuff years ago :scratch:
I did to believe it or not :eyebrows: all this alcohol is killing the old grey matter :doh:

When i mentioned AC resistance earlier i should have said AC impedance at very low frequency. The zobel type network you mention is probably a decent attempt at keeping the AC impedance flatish up to the crossover point, thus making things easier.

We both know that capacitive reactance increases with frequency so they probably selected a low resistance resistor similar to the DC voice coil resistance & added a capacitor so that more of this resistor is loaded as frequency increases. Get it right & they could create a near perfectly resistive load above 80 - 100Hz. Similar to Kefs' conjugate load matching i mentioned earlier (i think :scratch:).

Obviously it'd also act to some extent as a crossover filter in itself & this would need to be part of the equation. Life & speakers are never simple :doh:

Aye, get it together my man :)

[Welder]

Right Mark, as forewarned ;)

My two mid range drivers are supposed to be a matched pair, but they’re not.

http://www.rumoh.nl/index.php?item=scan-speak-13m8621-00-classic&action=article&group_id=20000258&aid=11987&lang=EN

I’ve got matched sensitivity and impedance for mid and high range apart from this one driver.
The Volts give Idb more at 2.83V/1m at 91dB which is deliberate.
I’m using measured values and not based on nominal impedances or manufacturers spec.

I want to sort this one midrange out (yep, its prolly inaudible but ..)
I was going to add an L pad, for the low value driver placed between the mid range filter arrangement and the driver, that way the crossover doesn’t need recalculating; it just sees the L pad (1 series resistor and 1 parallel resistor) as part of the driver.

I believe the correct way of doing this is too calculate the resistive drop in the high band inductor (because I’m trying to match mid to high, 1 mid and two high correct)

Db=20log (( Inductor wire resistance/measured impedance) +1) = frequency loss (Fl)

I now need to add frequency loss to the measured sensitivity of the “best” mid driver and then calculate for a series resistor and parallel capacitor for the out of spec mid

For the series resistor, current = voltage/capacitance
Using 1 volt to keep the maths simple

Series resistor voltage drop given by Rs=Vs/I
Given constant current

For parallel resistor
Rp=Vr/Ip where Vr =1

So, Rs=Vs/1and Rp=Vr/Ip I think.
Thus the total attenuation is Rt = 1/ (1/Rp+1/Z) where Z is the capacitance.
And in db its db=20log (Rs/Rt)

Not quite sure if I’ve got all this right. Does any of this make any sense? :scratch:

[Reid Malenfant]

Hi John, unlike some i don't have a degree in this :eyebrows: I learnt by reading, experimenting & measuring because i wanted to be able to make my own kit. The same with the electronics side of it to, i have no qualifications yet i know more about electronics than my last boss who had an HND ;)

Are you trying to tell me you have one midrange driver that happens to be about 1Db more sensitive than the other & you'd like to reduce it's output by 1DB to match?

If so then an L-pad made up of two fixed value resistors will be fine & you'll not need to recalculate any filter components for the crossovers as you'll be looking at the same impedance as before :) That's the beauty of the L-pad ;)

Effectively what you'll be doing is adding a certain resistance in series to give you the 1Db drop in output & adding a second resistor in parallel to give the same original impedance as the driver..

Let me know if this is what you are after & i'll help you out... If it isn't then try asking a little differently so i can absorb it :lol: Either that or i'll take a look tomorrow when 100% sober :cool:

[Welder]

Hey Mark :)

Yep, you got it. Its a couple of db out though which may just be audible.

I’m not an electronics engineer either.
I did learn this stuff a long time ago though.
I think I’ve got the right idea here, just checking I haven’t missed anything crucial in the method I’m going about it.
After all that I’ve found an L pad calculator but I may just do the sums anyway.
I would like to get it all welded up tomorrow coz I’m pissed off with listening to music through my headphones.

[Barry]

I did to believe it or not :eyebrows: all this alcohol is killing the old grey matter :doh:

When i mentioned AC resistance earlier i should have said AC impedance at very low frequency. The zobel type network you mention is probably a decent attempt at keeping the AC impedance flatish up to the crossover point, thus making things easier.

We both know that capacitive reactance increases with frequency so they probably selected a low resistance resistor similar to the DC voice coil resistance & added a capacitor so that more of this resistor is loaded as frequency increases. Get it right & they could create a near perfectly resistive load above 80 - 100Hz. Similar to Kefs' conjugate load matching i mentioned earlier (i think :scratch:).

Obviously it'd also act to some extent as a crossover filter in itself & this would need to be part of the equation. Life & speakers are never simple :doh:

Aye, get it together my man :)

No it doesn't: Xc = 1/( omega*C), where C is the capacitance in Farads and omega = 2*pi*frequency (in Hz). Thus as frequency goes up, Xc goes down.

Regards

[Reid Malenfant]

Uhhh :doh: Yeah barry, my misleading words.. What i meant to say was the impedance of the capacitor decreases with increasing frequency :doh:

I'd had a few bevvies by that point so got things arse about face, nothing unusual :eyebrows:

[Barry]

Uhhh :doh: Yeah barry, my misleading words.. What i meant to say was the impedance of the capacitor decreases with increasing frequency :doh:

I'd had a few bevvies by that point so got things arse about face, nothing unusual :eyebrows:

Thought as much - it seemed an unlikely mistake for you to make. Forgive the terse reply and 'lecture'. ;)

[Reid Malenfant]

Thought as much - it seemed an unlikely mistake for you to make. Forgive the terse reply and 'lecture'. ;)
No problem Barry, actually you were right in pointing it out at the end of the day :)

Must engage brain before typing in future :eyebrows:

[Welder]

Hi YNWaN

What you say is correct; the highest temperature will be at the voice coil. Despite the air gap and the cooling of the coil through air movement heat is transferred to the pole and thus to the magnet.

What I have are fairly high power handling drive units in a relatively small volume enclosure with high thermal insulation properties Cab construction is bitumen sheet sandwiched between ¾” HD chipboard and ½” beech ply on the inside. It gets hot in there :eyebrows:
I’m interested in seeing what effect if any improving overall heat removal from the enclosure has.
I have an idea that would link the driver magnets to the outer surface of the enclosure (and improve the driver’s ability to resist in cabinet vibrations).
In theory this provides a low temperature sink and what would be a steep thermal gradient that might just remove enough heat to make a difference.

I’m barking mad (?)……… possibly.
I’m wasting my time (?)…………..maybe, but considering this is a subjectivists forum where some of the most unlikely benefits are perceived from some of the most unscientific and expensive additions to peoples Hi Fi systems I shouldn’t get too many bricks hurled at me.
You never know, it may just work ;)

[Reid Malenfant]

Hey John, have the Volt bass drivers got a vented polepiece? Or if you like, is there a hole on the back of the magnet that goes through the centre pole to aid in cooling or is the rear magnet plate solid?

I'd look it up but i'm rather busy :eyebrows:

[Welder]

Hey Mark.
It’s solid.
They're back together now mate and I’m just warming them up with a bit of Flaming Red by Patty Griffin :)
Just cant say how nice it is to get away from the ghastly racket of my headphones (well, not really but….)
Got grunt again!! :ner:
Going to see how they sound for a bit and then (if I can lift the bloody things) drag them over to my mates and see what his Sugden kit can do with them.
I just can’t wind them up in my place coz its suspended floor, small area and I get more standing waves and vibration than an enthusiastic football crowd :lolsign:

[Reid Malenfant]

So what's this idea of yours then? This heatsink :scratch:

Only thing i can think of is to clamp a nice thick piece of alloy (12mm or thicker) to the rear of the magnet & pass it through the enclosure walls & onto a proper heatsink...

Guess i'm quite mad as well :eyebrows:

Or you could enclose the whole magnet assembly & oil cool the thing sort of like the way electrical substation transformers are cooled :eyebrows:

[Welder]

Yep, you’re definitely barking too :lol:
I did think briefly about liquid cooling but its way too complicated :doh:

I’ve got these 3” thick cast iron billets knocking about.
I was thinking of boring them out to fit the OD of the magnets and drilling and tapping two large holes at 180 degrees on their OD and connecting with metal rods, through the side of the cabinet to 8” diameter ¼” thick plates with threaded bosses and machined fronts basically.
I was considering doing this fro the Volts and the mid Scans.
Thermal paste should sort out the heat transfer at the joins and by tightening the plates on the rods everything should lock together quite nicely.

That’s it in essence :mental:

[Reid Malenfant]

Hmmm, i wouldn't stick any ferrous metals near the magnetic assembly or you could seriously alter the driver Q & make mincemeat of the proper enclosure response :doh:

Besides that iron is a pretty crap conductor of heat, what you need is aluminium or some form of alloy. You don't tend to find cast iron heatsinks probably because they wouldn't work very well ;)

[Welder]

Good thinking, cast iron wouldn’t do very well would it :doh:
Not sure why I mentioned the cast iron given I originally had mentioned ally; prolly coz I’ve been down to the shed and that’s what I looked at last :lol:
Suffering from a bit of substance abuse atm. Well one does when one just finishes a Hi Fi project doesn’t one? :eyebrows:

I’ll get back to this later.

[Welder]

Right, were was I? Oh arr talking a load of shite :lol:

I think I’m just going to do this with one speaker and see how it compares to the other with regard to the temp inside the cabinet and on the magnet itself.
Trying to work it all out in theory first is all very well, but the math’s gets quite complicated :scratch:
I can’t see it doing any harm. It’s pretty easy to do and easily reversible if necessary.
Securing the driver using the magnet seems a reasonable idea; whether it transmits more or less vibration to the cabinet remains to be seen. If it acts as an effective heat sink as well so much the better.
The basic idea partly comes from a friend of mine who has the magnets of his Radial Volts mounted in a vertical cabinet brace.

I also need to fit some handles on the sides. I think they must weigh 70kg + now! :eek:

The new cabinets, crossover out of the main enclosure, internal wiring changes and driver matching, plus oil and paper capacitors for the hi range have all combined to give a pretty stunning sound out of what are basically rather ugly unfinished boxes.

I’m thinking about loading them onto the back of the truck in March and bringing them to the show if someone there is prepared to stick them on the end of some suitable kit :eyebrows: