SS 13m/8640 and 13m/8636 motors

[dlr]

I'm sending two 13m/8640 drivers to Madisound to be re-coned into 13m/8636. The motors are the same and for a few bucks more on reconing I can get the kevlar version that I've wanted to try out.

Before boxing them, I dissected them to inspect the motors. I thought that you guys might like to see a couple of the pics. If either Mark or zaph is interested, when I get them back I could send in one each of a 8636 and 8640 to see some distortion measurements. I had liked the 8640, but the 12m was an improvement. I think, though, that it may have had as much to do with doing a better job of the highpass on the 12m (after living with the 8640 for several years), so I'm planning on making comparisons later.

I'm impressed with the amount of copper in these things. Once the FR is smoothed, I would expect good results from these. In raw form the 8640 will likely look pretty bad above about 4K due to the on-axis breakup, but an LR2 at 2K has a pretty clean FR. The 8636 is smooth to start, so I would think that it might fare better. In any case, I'm curious to compare them to the 12m in the distortion arena.

I can't recall anyone having tested them. If that's been done, please point me to the results.

[Feyz]

And up to this time I used to naively think that Usher's the copper sleeve shape on 8945P was original, which I had taken a look years back, the picture attached. Stupid me!



I always wondered why Scan speak moved from full sleve inside gap to the up and below gap motors. Looking at their impedance curves their some older 7" drivers also should have full sleeve on the pole piece like 8546 Mark K had tested with good results. 8554 8" woofer also is a good one with full sleeve and an old design.

Thanks for the pictures.

[Jed]

I have some old madisound catalogues with distortion measurements by Scanspeak on these 2 drivers you mention. The graph is very small and hard to read, but it appears that distortion rises starting at 600Hz-200HZ, especially 2nd order harmonics for both drivers- .5%-1%. From 600HZ on up its around .3%, with the 8636 slightly better (less 2nd order products across the FR). Looks like these drivers are best used from 600HZ to 3K.

Jed

[capslock]

Yes, this seems to be the original SD motor. Dave, does the pole piece extend straight down below the lower end of the gap (so that the pole piece forms one long cylindrical column) or is it undercut, forming a T-pole?

Is the top plane (the non-copper-clad steel around the pole piece bore) level with the top of the outer top plate or is it elevated by 1 - 2 mm?

Feyz, the newer woofers and midwoofers should use a design very similar to the Excel's, I forget whether it is called SD-1 or SD-2. I can find the patent for you if you're interested. The reasoning is that you obtain higher Bxl for the same size magnet and thicker shorting rings, resulting in a more effective suppression of flux modulation. The penalty, of course, is some midrange distortion.

[Feyz]

Feyz, the newer woofers and midwoofers should use a design very similar to the Excel's, I forget whether it is called SD-1 or SD-2. I can find the patent for you if you're interested. The reasoning is that you obtain higher Bxl for the same size magnet and thicker shorting rings, resulting in a more effective suppression of flux modulation. The penalty, of course, is some midrange distortion.

Thanks but, I am very well aware of these, and Scan's patent, and some others, etc.... Actually it has been years by now first I had started looking into these and known all that My point is, Usher 8945P can do better than Scan's new motor SD-2 overall for a midwoofer, and even using a bigger voice coil than the 7" Revelator. So it doesn't really look like to me an improvement going to SD-2, unless it would be for a woofer where midrange is not important. Even there a combination can be used, but cost may become an issue then.

[dlr]

Thanks

I have some old madisound catalogues with distortion measurements by Scanspeak on these 2 drivers you mention. The graph is very small and hard to read, but it appears that distortion rises starting at 600Hz-200HZ, especially 2nd order harmonics for both drivers- .5%-1%. From 600HZ on up its around .3%, with the 8636 slightly better (less 2nd order products across the FR). Looks like these drivers are best used from 600HZ to 3K.

Jed

I had always used it down to 300-400 2nd, though it was a stretch given the Fs of 65 turned out to measure about 105 in my pair. I never drove them too hard, so I suppose that they should be good at this point when used that way.

However, I'll may pair them with a smaller woofer that will let me cross at higher so I'll keep that in mind.

[dlr]

I'll have to check that out

Yes, this seems to be the original SD motor. Dave, does the pole piece extend straight down below the lower end of the gap (so that the pole piece forms one long cylindrical column) or is it undercut, forming a T-pole?

I couldn't see down into it. I'll have to get something non-ferrous to poke around to find out. They're still in the box to go, so I'll check it out before I ship them.

Can that copper be removed/replaced without concern of it being loose afterwards?

Is the top plane (the non-copper-clad steel around the pole piece bore) level with the top of the outer top plate or is it elevated by 1 - 2 mm?

I think that it's level, but I'll check that, too.

One thing I wonder about is if it would benefit by building up some putty to shape the inner area on top of the pole piece to provide a smoother air flow in the pole piece vent. I did experiment with stuffing the vent from the back side with some lamb's wool that helped a small bit, but that would reduce the heat dissipation. Have any of you guys experimented this way (shape, not stuffing)?

[dlr]

Maybe sensitivity is the key

So it doesn't really look like to me an improvement going to SD-2, unless it would be for a woofer where midrange is not important. Even there a combination can be used, but cost may become an issue then.

If there is no increased benefit in distortion, I can't see it would seem to me that sensitivity is the benefit. It must cost more to undercut the pole piece.

Would there be a cost benefit to having a cheaper magnet (lower magnetization with subsequent lower B in the gap), yet still having good distortion characteristics and sensitivity?

Just throwing out ideas.

[jkrutke]

This thread is like speaker porn. Not suitable for viewing at work. :B

Here's my contribution, a naked Dayton RS90. It's got a .5mm coper sleeve through the gap. I can't tell yet if it's undercut.

I've got a couple Excels I could hack apart and measure, standard and hex neo magnet. It seems that those are fairly well known inside though.

I'm just beginning to play around with the free version of Maxwell myself, but I'm nowhere near what Feyz and Capslock are doing yet.

[dlr]

That picture raises another question

This thread is like speaker porn. Not suitable for viewing at work. :B

Here's my contribution, a naked Dayton RS90. It's got a .5mm coper sleeve through the gap. I can't tell yet if it's undercut.

Love those naked drivers! :E

Oh, the question, yeah. What difference is there in the impact of the copper sleeve with the flat top as in the RS90 and the version in the SS and Usher that is peaked well above the gap and rounded over? Is the rounded over section helping with Xmax or for some sort of smoother transition region of some sort?

[TacoD]

Ok speaker motor experts what about the Audiotechnology motors? (for the inexperienced some resemblence with older scanspeak motors is apparent)




edit: Skaaning was already involved in Dynaudio and Scanspeak

[JonMarsh]

Love those naked drivers! :E

Oh, the question, yeah. What difference is there in the impact of the copper sleeve with the flat top as in the RS90 and the version in the SS and Usher that is peaked well above the gap and rounded over? Is the rounded over section helping with Xmax or for some sort of smoother transition region of some sort?

Dave, the ratinoale I read years ago from Scanspeak was that the elevated copper shorting ring reduces inductivity modulation when the VC goes forward substantially from the rest position.

[dlr]

That makes sense

Dave, the ratinoale I read years ago from Scanspeak was that the elevated copper shorting ring reduces inductivity modulation when the VC goes forward substantially from the rest position.

My next question would then be, if the copper cladding is not symmetric about the gap, then would that not introduce some form of asymmetry in the inductance, unless it's some sort of counterbalance to the field below the gap.

On that latter thought, maybe it's a case of trying to get closer TO symmetry, since the copper may extend well down into the motor along the pole piece below the gap. That would make sense to me in that case. They can extend above the pole piece just so far, but it would be better than none.

[dlr]

That looks like the SD motor

only it's also got some felt underneath the copper rolled section.

[capslock]

Love those naked drivers! :E

Oh, the question, yeah. What difference is there in the impact of the copper sleeve with the flat top as in the RS90 and the version in the SS and Usher that is peaked well above the gap and rounded over? Is the rounded over section helping with Xmax or for some sort of smoother transition region of some sort?

No extension means the fringe fields will be different above and below the gap, hence Bxl asymmetry. With an undercut pole piece, things get better, but having an extension à la SD (rigdge) or Excel (cylindrical extension) will make Bxl more symmetrical. The extension also helps with inductance symmetry.

[capslock]

Ok speaker motor experts what about the Audiotechnology motors? (for the inexperienced some resemblence with older scanspeak motors is apparent)


edit: Skaaning was already involved in Dynaudio and Scanspeak

OK, let's have a closer look.

- no extension
- no undercut
- thin full length copper sleeve, effective only in the midrange
- bevel on the top plate, will have to simulate that
- no lower outside shorting ring -> will have significant flux modulation

Seems that the motor is not really top notch.

[capslock]

Here's my contribution, a naked Dayton RS90. It's got a .5mm coper sleeve through the gap. I can't tell yet if it's undercut.

I've got a couple Excels I could hack apart and measure, standard and hex neo magnet. It seems that those are fairly well known inside though.

I'm just beginning to play around with the free version of Maxwell myself.

Any word from PE as to when that one is going to come out?

I started with Maxwell 9 SE, but the post processor turned out to be a nightmare. FEMM was only one hours' reading and works much better.

[dlr]

Here's what I can see

Yes, this seems to be the original SD motor. Dave, does the pole piece extend straight down below the lower end of the gap (so that the pole piece forms one long cylindrical column) or is it undercut, forming a T-pole?

The copper extends almost all the way down, so there's no undercut that I can see. The copper has a rolled out end that must be used for pressing it onto the P-P through the gap.

Is the top plane (the non-copper-clad steel around the pole piece bore) level with the top of the outer top plate or is it elevated by 1 - 2 mm?

The top plane IS slightly extended, at most 1mm, can't tell precisely.

The copper measures 0.5mm thick.

[Feyz]

The copper measures 0.5mm thick.

My guess is inside the air gap the thickness is lower, because again my guess the way that copper is shaped must be some sort of stamping or pulling whatever the terminology or procedure is from a plate. Can you give that shape from a cylinder tube? Even if so, again the thickness at the plane top section will be greater than the section of the tube that goes into the air gap.

Like already have been said, the extension of the tube on the top provides better symmetry of inductance as the coil moves upwards. Voice with only air core without the shorting ring can have higher inductance then with the shorting ring and some steel from pole piece. Extending it like that prevents that, and also will provide additional shielding of the upper parts of the pole piece and top plate from being modulated by the voice coil field.

As a side note, I don't think this that well known but the patent for using shorting rings above and below a T shaped pole piece belongs to McIntosh. Some interesting history:http://www.roger-russell.com/xr250.htm

Scan-Speak's patent, which is later, uses similar configuration but divides the lower ring into two shorter ones ending up with three rings.

[Feyz]

I'm just beginning to play around with the free version of Maxwell myself, but I'm nowhere near what Feyz and Capslock are doing yet.

As capslock said, there is not much to it. But I would also suggest FEMM for ease of use which has good manual's and tutorials for a free program. It even comes with a speaker motor example.

[jdybnis]

An image from Troels Gravesen's site http://www.troelsgravesen.dk/philips9710.htm

Philips 9710 with the pole piece covered with copper.

Circa 1980

[dlr]

I'd bet it's the same in the gap

I understand your point, but if you think about it, they can't make that shape from a flat piece. I suspect it starts out as a tube with an end "cap", so-to-speak, that has a hole in the center. Then it's stamped to press that capped end inward and create the rolled section. This would leave the center part largely unchanged and the main tube section totally unchanged. I'd still put my money on it being 0.5mm in the gap.

[Feyz]

I suspect it starts out as a tube with an end "cap", so-to-speak, that has a hole in the center.

How would they make the tube with a cap in the first place?

[dlr]

I wish I knew

How would they make the tube with a cap in the first place?

I'm not familar with casting/forming. But it's got to be a tube for the main section.

Copper is easily formed, especially when that thin. I would guess that it would start as a plain tube with a disc attached by some heat process onto the end if made that way.

It could even be purely a casting process. Thay may be the least expensive way to do it. The general shape makes me think of a bullet casing that's made of brass.

It would be interesting to learn exactly how they do it.

[wkhanna]

The copper 'blanks' (discs) can be pre-formed to have the thickness vary across its cross-section. This can compensate for 'necking' or reduced wall thickness along the length of the cylindrical feature when the part is die-formed. Also, 'tricks' can be designed into the die, to maintain an even wall thickness when required.

[Mark K]

Hi Dave,
I actually have the 8636 and 8640. I've just never gotten around to testing them. Or, maybe I sold the 8640's and just have the 8636 left. Hmmm, I'll have to check. ( know I sold some 8640's, but can't remember if I sold just two or all four)

[Feyz]

The copper 'blanks' (discs) can be pre-formed to have the thickness vary across its cross-section. This can compensate for 'necking' or reduced wall thickness along the length of the cylindrical feature when the part is die-formed. Also, 'tricks' can be designed into the die, to maintain an even wall thickness when required.

Thanks for the illimunating info. You obviously know a lot about these. How do you think the steel cup in the below picture is formed? The thickness at the bottom is roughly 1/4", and the side walls are ~ 1/8". My guess is this is also coming from a die press sort of forming



I did some more search after reading your post. I remember as a child I had watched some TV show where copper plate was being shaped into a vase on a lathe by spinning. It looks like there are two major ways of doing this, one is spinning a disc into the form, the other is punching a disc plate into a die that gives the shape. Aluminum beverage cans are made this way according tohttp://en.wikipedia.org/wiki/Beverage_can

I can now see especially in the case of spinning how using varying thickness of blank disc could be used to give same thickness on the cylinder wall and top cap. But I don't think those on the speaker motors have that, since there is no point of complicating the manufacturing process that yields no gain. If they want thicker wall, they would start with a flat thicker blank and at the end again the top section will be thicker than sides, but that doesn't cause no harm, actually it is better

[BobEllis]

I used to sell formed metal bellows, so I'll confirm that is is fairly trivial (with the right equipment) to draw a copper disc into a can shape punching it into a die. Need it longer and thinner? Do it again with less clearance between the die and punch. A setup and labor intensive process, but very old technology. Yep, beverage cans are made the same way, although with a more automated process.

[wkhanna]

You got it Bob!

As for spinning, it was a technique developed long ago, it is V labor intensive and therefore costly. Hydraulic presses and forming dies are used in most cases for most mid to high volume production.

As far as the steel component, that is a machined part, cut from a round bar of steel on a CNC lathe. As are the other components shown in your pic. Relatively cheap to mass-produce.

The frame in the pic, is most likely molded if it is Aluminum.

[capslock]

Yet in most drivers except Seas, Scan and some Vifas, you will find only cold rolled steel pole pieces.

[Feyz]

An image from Troels Gravesen's site ....

Circa 1980

I believe this one below is from 1970's, Yamaha NS-690 woofer's pole piece



Looks like there is a copper cap on top, and then a an alumium shorting ring below. I had seen a 70's patent with shorting ring covering the magnet's inside.

[dlr]

Any chance that you'll put in your list for testing?

I'd be very interested in the distortion tests. I think that the 13m8640 was too often passed up due to the upper end, but it's competitive as a midrange. the 8636 would be even easier to cross. The sensitivity is a bit limiting, but they're still very viable as high quality midrange drivers.

Seeing the quality of the motor has me looking to try my hand with them again. The 8640, that is. I never had an 8636, but with a re-cone job that 8640 pair will then be an 8636.

I'm interested in seeing how they compare to the 12m in the distortion area.