So I got my first pair of RS-28 tweeters...

[Sakura]

I see that they have stickers on them with the sensitivity?

One says 91.2 and the other says 92.0.

I ASSUME those are the sensitivity? I'll measure them tomorrow.

It's nice that they measure and label that for us, but, for being "reference" tweeters, seems to be a little disappointing to me that a pair should be off by almost a full db? Depending where in the frequency response they're off, might mean different crossovers for each speaker. Guessing it might not be noticable in listening tests though.

I do like the look and heft of the tweeters. I like the metal face plates, the sturdy gold plated terminals, the black anodized dome, and I even like the metal mesh grill protecting the dome (I have little fingers in my home that like to touch and prod speakers).

Can't wait to hear them

[Sakura]

Measured the impedances:

[Curt C]

I guess I'd consider this within the normal variation of any quality driver. Small variations generally can be compensated for in the tweeter network with a minor resistor value change.

For a 'matched pair' of RS28's, I'd suggest buying another pair. Odds are you will end up with at least one pair that are close relative to each other.

Alternatively, Hiquphon tweeters come in matched pairs with individual response plots, and are a very nice tweeter, IMO.

C

[Sakura]

Thanks for the feedback. I see the difference above 2Khz is mostly an equal offset and they both follow the same smooth slope very closely, I would agree, if any change [in the crossover] is needed, it could be handled by simply padding down the more sensitive one slightly more.

From about 700Hz to 900Hz the levels are closer, the FS on the the less sensitive one is slightly lower, and slightly sharper peak. I wonder what the most likely cause is, the most obvious one to me would be higher moving mass? The domes aren't hand coated, as far as I know, perhaps the glue is hand applied? Perhaps a different amount of ferrorfluid? I wonder what the effect of packing the stuffing material in the chamber more or less densely would have?

In the end, if I just used identical crossovers, and left the < 1db difference, I don't think I would hear a difference anyway? Out of curiosity, I will test some other tweeters I have which I suspect are more consistent, despite being less than 1/4 the cost of the RS28As...

[Dave Bullet]

Zaph is currently retesting the RS28, so you might want to compare your findings with his

[Curt C]

From about 700Hz to 900Hz the levels are closer, the FS on the the less sensitive one is slightly lower, and slightly sharper peak. I wonder what the most likely cause is, the most obvious one to me would be higher moving mass? The domes aren't hand coated, as far as I know, perhaps the glue is hand applied? Perhaps a different amount of ferrorfluid? I wonder what the effect of packing the stuffing material in the chamber more or less densely would have?

The impedance plot differences made me wonder about the ferrofluid as well. It might be enlightening to swap the diaphrams on the motors and see if things might 'even out' somewhat. While you are in there, check out the felt plugs for centering, excessive glue, etc, and that the phase shields are centered on the inside of the grill.

Dave (dlr) is the man to ask about tweeters.

http://www.speakerdesign.net/home.html

C

[dlr]

I guess I'd consider this within the normal variation of any quality driver. Small variations generally can be compensated for in the tweeter network with a minor resistor value change.

For a 'matched pair' of RS28's, I'd suggest buying another pair. Odds are you will end up with at least one pair that are close relative to each other.

Alternatively, Hiquphon tweeters come in matched pairs with individual response plots, and are a very nice tweeter, IMO.

C

Hi guys,

I've been following this. My early pair of XT25s had a 1db sensitivity difference. I attributed that to variation in magnetic field strength. I think that the impedance curves were close.

The RS28 impedance curves shown here are unequal down to the 100Hz lower limit. It makes me think that the problem is the Rdc of the coils. I suggest measuring the Rdc with a fairly accurate meter, being careful to get a solid connection to the terminals. I've had instances of poor connections to the terminals when testing causing a DC resistance, just a point to check.

The delta is a bit too high to be within normal variances IME.

Dave

[Sakura]

Switched the diaphragms, improved, missing ferrofluid

Hmm, no Ferrofluid, is it supposed to have Ferrofluid?

EDIT: Yep, from the web page: "Tuned rear chamber and high-quality ferrofluid for great low-end response"

I have the more sensitive one open, that one has no ferrofluid, will open the other one now to check that too.

EDIT 2: Ok, BOTH tweeters are missing the ferrofluid, also, I've measured the impedance with the diaphragms switched, here are the results gentlemen:

[Undefinition]

Hmm, no Ferrofluid, is it supposed to have Ferrofluid?

EDIT: Yep, from the web page: "Tuned rear chamber and high-quality ferrofluid for great low-end response"

I have the more sensitive one open, that one has no ferrofluid, will open the other one now to check that too.

You can't really SEE the ferrofluid. It's down in the gap. And it doesn't like to come out, either, because it's magnetically attracted to everything around it.

There may be a small amount of its residue around the voice coil of the dome.

I have been following this thread since you started. Of all the different tweeters I've owned (which, compared to many grizzled veterans isn't that many) --say, 15 to 20 pairs-- the RS tweeters are the only ones that mark their exact sensitivity on them. But I have certainly had tweeters that clearly had differences in sensitivity between the two.

Could you hear a .8 dB difference? Well, there are people on both sides of the argument. Commonly held "rule of thumb" is that 3 dB is the minimum change in volume that a person can detect. In reality, however, I think this depends on the frequency range we're talking about. Personally, while voicing a crossover, I can definitely hear a difference of 1 dB if it's around the "critical area" of 1 KHz. Will you hear the .8 dB difference in your tweeters? Hard to say, especially depending on how you cross them.

If you CAN detect a difference, then it IS true that you can certainly even things out with minor changes to resistors. But personally I'd find that frustrating... sort of kills the symmetry of it all.

[Sakura]

The Emperor's New Ferrofluid

You can't really SEE the ferrofluid. It's down in the gap. And it doesn't like to come out, either, because it's magnetically attracted to everything around it.

In this case, I can't see it, because it's not there. I even stuck a piece of paper into the gap, no discoloration whatsoever. I've worked with Ferroflluid before. I know what you mean, sometimes there's hardly any left [on the voice coil] when you pull the coil out.

I have been following this thread since you started. Of all the different tweeters I've owned (which, compared to many grizzled veterans isn't that many) --say, 15 to 20 pairs-- the RS tweeters are the only ones that mark their exact sensitivity on them. But I have certainly had tweeters that clearly had differences in sensitivity between the two.

Could you hear a .8 dB difference? Well, there are people on both sides of the argument. Commonly held "rule of thumb" is that 3 dB is the minimum change in volume that a person can detect. In reality, however, I think this depends on the frequency range we're talking about. Personally, while voicing a crossover, I can definitely hear a difference of 1 dB if it's around the "critical area" of 1 KHz. Will you hear the .8 dB difference in your tweeters? Hard to say, especially depending on how you cross them.

If you CAN detect a difference, then it IS true that you can certainly even things out with minor changes to resistors. But personally I'd find that frustrating... sort of kills the symmetry of it all.

I added a picture to my last post, was doing that while you were replying I think - anyway, now they are closer to each other.

Overall, I am impressed with the design and build quality. However, I would have thought a pair of these tweeters would be a little closer to each other in level then they were when I got them. I actually don't want more than 90db sensitivity, so these should be fine for me, I just expected a little more precision than I got.

EDIT #2: Now, what to do about the missing ferrofluid, I'm not sure yet...

[dlr]

Could you hear a .8 dB difference? Well, there are people on both sides of the argument. Commonly held "rule of thumb" is that 3 dB is the minimum change in volume that a person can detect. In reality, however, I think this depends on the frequency range we're talking about. Personally, while voicing a crossover, I can definitely hear a difference of 1 dB if it's around the "critical area" of 1 KHz. Will you hear the .8 dB difference in your tweeters? Hard to say, especially depending on how you cross them.

The XO Fc will make a difference, of course, but last I recall reading the requirement for research into audibility studies is 0.5db. 0.8db across the usable spectrum of a tweeter should be noticeable, if nothing else than as a shift in the imaging that will depend upon the music content. I find lateral shifts due to this the most easily detected, probably largely due to my preferred equilateral triangle configuration.

Dave

[ThomasW]

Might call the PE toll-free tech support line and ask them about the ferro fluid situation. It's either a production mistake, or they've stopped using ferro fluid to keep costs down.

Also consider gently burning-in the tweeters for a few hours then remeasure them...

[Sakura]

Measured FR w/swapped diaphragms...

Might call the PE toll-free tech support line and ask them about the ferro fluid situation. It's either a production mistake, or they've stopped using ferro fluid to keep costs down.

Also consider gently burning-in the tweeters for a few hours then remeasure them...

I will burn them in before taking measurements for crossover design use...

If they're advertised as coming with ferrofluid, then the customer should get ferrofluid, right?

Here are the responses - only accurate as a comparison between the two, not as an actual representation of their response in a speaker because they're not mounted in a baffle, just sitting on my carpeted floor on a marked spot with the mic held in a fixed position about 6 inches above. Also, there are some boxes/wall/other object less than a foot away... But it does show that the two tweeters (with the now swapped diaphragms) are VERY close.

[Sakura]

Just for fun, a pair of Vifa tweeters measured...

For comparison (of deviation from tweeter to tweeter of another brand) here are the FRs of a pair of Vifa D27TG-35-06 that I got in the same order with the RS28s.

One of these Vifa tweeters came with the dome crushed in, which I gently sucked out. For some tweeters sucking the dome back out works, for some, it's not good.

[TacoD]

I think that due to the re-alignment of the domes (you swapped the diaphragms) the tweeters are similar again. It seems that the difference wasn't in the components, but in the assembly.

[Curt C]

But it does show that the two tweeters (with the now swapped diaphragms) are VERY close.

I love it when a plan comes together...

C

[---k---]

If Zaph wants to open up the pair of RS28s that he has in his hands and check for ferro fluid, that is okay with me.

[---k---]

Zaph's test results are up:

This was my chance to measure a new RS28. As mentioned many times previously, this tweeter has had some production problems for a while. This is just a look at what current performance is. There has been some discussion on the forums about this tweeter not having ferrofluid in the gap anymore. The pointy nature of the peak pretty much says no ferrofluid. This is a design change that might have repercussions on designs currently posted using this tweeter. To be sure, I opened up one. Indeed, there was no ferrofluid.

Removing ferrofluid is a good call however. This tweeter is one that doesn't need it, thanks to a free flowing pole piece. It doesn't result in the "twin peaks" impedance curve and ragged low end that some other non-ferro tweeters have. Seas for example, pretty much needs ferrofluid in the gap. The current RS28 product description calls for ferrofluid however, so I can't say if it was forgotten in this batch or if this is the new standard. I'll wait until I find that out before I move this to the comparison page.

Performance is generally good, and close to some of the other good tweeters in this price range. Build quality is still excellent, with a metal flange and nice looking black mesh grill. I'd say it's a good tweeter to use, but only if you can measure it on your own to be sure of what you have.

[fjhuerta]

I guess this means the RS28 still has production problems

[---k---]

I didn't take what John said to mean that. I'll let John further explain what he meant.

[---k---]

Could someone post some pictures on how to take the RS28 apart and check for ferro fluid. I'm now curious which version I have in my Khanspires. I don't really care if it doesn't have the ferro fluid. I just want to make sure what is posted is representative of what is out there.

[dlr]

I have a page showing disassembly

Could someone post some pictures on how to take the RS28 apart and check for ferro fluid. I'm now curious which version I have in my Khanspires. I don't really care if it doesn't have the ferro fluid. I just want to make sure what is posted is representative of what is out there.

Go to the page Dayton Reference Series Tweeter RS28A-4 at my site. I've got photos showing how to disassemble it.

As for checking for FF, I'm pretty sure that if you have one with it, you'll see traces of it on the former on the inside and/or on the voice coil. It will be a faint, brownish liquid, in random spots. I would say that if it is not totally dry (no FF), you'll see it.

Dave

[---k---]

Thanks for pointing me in the right direction. I'll try and take one of mine apart this coming weekend. I'll also throw them on the Woofer Tester 3 that I have. It sounds like it is easy to look at the impedance graphs and see the difference. After that, we should know which version of the RS28 the Khans were designed with and anyone wanting to build them can proceed with confidence.

[Jonasz]

I wonder if all this fuss concern the rest of the Usher drivers as well? None of the seveninchers Zaph tested have any consistency-rating, probably due to only one driver tested. Would be interesting to see how Ushers own tweeter tests...

Zaphs update on the RS28 issue: http://www.zaphaudio.com/blog.html

[Curt C]

What PE says:

Originally posted on the PE forum:

Below is the response I received from PE: (slightly edited)

“We did not authorize a change in the RS28. We are sure it was a mistake on our manufactures end. We were unaware that the tweeters no longer contain ferrofluid, and have confirmed this with our current inventory. An urgent email was written to our manufacturer and we probably will have to remove these tweeters from stock. We are just waiting on their reply as to when we will receive a correct shipment.”

-"Whatever the reason it will be addressed and corrected."

I have requested if they can share with us when the suspected non-ferrofluid units were shipped from PE, or what lot numbers, were in the suspect batch. Hopefully PE can obtain and provide this information, so all who recently purchased RS28’s can either rest easy, or arrange for replacements.

I’ll suggest a bit of restraint at this time, until this additional information can be obtained from their manufacturer.

C

[---k---]

Zaph used my RS28 tweeters for his testing. Based upon his measurements and comments, I'm very happy with my purchase and intend to use the RS28s. He never suggested that I should return them or be unhappy with them. I don't want to speak for John, but he told me:

Here's the results. Looks pretty good actually, though a bit different from previous RS28's.

I'm very confident that the results I get from these will be great. I have absolutely no concerns about using these. I have no intentions of contacting PE and asking for a refund or any consideration. My only concern is posting the design and other people attempting to use my design with different tweeters.

[---k---]

Zaph's most recent blog:

http://www.zaphaudio.com/blog.html

October 15, 2008

RS28 Thoughts...
Everything posted here is my opinion. Here's my additional comments about the RS28 situation. I was going to comment in a forum discussion, but didn't get a chance. So here it is in my blog now.

Going back for some history first: The original internal QC done by Darren did not find all the lesser performing or defective tweeters. This is because a linear distortion test (such as an impedance sweep or response curve) doesn't always correlate to non-linear problems such as a crooked or rubbing voice coil, which was the major issue with the RS28.

Not mentioned in my first write-up is the fact the the voice coil was glued on straight in my last tested RS28 drivers. This to me was an even more important issue than the presence of ferrofluid. There is currently more quality put into RS28 assembly than there was in the past. If I had my way, all future RS28 tweeters would be exactly like the ones I just tested, without ferrofluid. They were better than the previous 8 I tested.

The real problem with this tweeter is unknown history. Nobody, including PE or even Usher, knows when the crooked voice coil issue was resolved. All we have is unsubstantiated subjective claims without any testing. Additionally, nobody knows when ferrofluid was no longer in the gap. This puts a big question mark in front of those who want to replicate one of the many posted designs. That question mark is probably the worst possible quality control failure that a driver sold to DIY'ers can have. The damage may never be undone. Even if all quality control issues are resolved, in the future kit builders will ask if a design was done with a "good" tweeter. Note that PE is not to blame for this, Usher is.
...

See his blog for more.

[---k---]

I personally think that Zaph is over-reacting a bit when he says that the RS28 should be discontinued. I think a path would be to rename the current tweeter. Then those of us with designs posted with this tweeter can check to see which one was used in the design and either pull the design or change the tweeter to the new name.

[Jed]

It is interesting to me that the RS28a has had some inconsistency issues and not the other drivers in the line. Maybe it is a more difficult design to produce?

[jkrutke]

I personally think that Zaph is over-reacting a bit when he says that the RS28 should be discontinued. I think a path would be to rename the current tweeter. Then those of us with designs posted with this tweeter can check to see which one was used in the design and either pull the design or change the tweeter to the new name.

Well I do understand that the whole issue with the RS28 is more important to those who use it in published designs, so the mere suggestion of discontinuation would seem like a overreaction. It's not though. In my eyes this tweeter's reputation is damaged beyond repair. It wouldn't really affect me either way if it was kept around or discontinued. I don't intent to use it in any designs.

It is interesting to me that the RS28a has had some inconsistency issues and not the other drivers in the line. Maybe it is a more difficult design to produce?

Various Dayton RS line drivers come from different manufacturers. The RS28 is Usher, the dome mid is Peerless of India and the woofers are another manufacturer too. (one that PE would probably rather keep secret)

I suspect the RS28 is indeed harder to manufacture. All that copper in the gap means the tolerance has to be tighter to maintain sensitivity. That's probably why there were rub and buzz problems early on.

[Curt C]

Well I do understand that the whole issue with the RS28 is more important to those who use it in published designs, so the mere suggestion of discontinuation would seem like a overreaction. It's not though. In my eyes this tweeter's reputation is damaged beyond repair. It wouldn't really affect me either way if it was kept around or discontinued. I don't intent to use it in any designs.

I personally think PE has pretty good brand loyalty, and as long as they resolve the issue, all will be OK in the house of Dayton. I predict the RS28 will have a long and useful future.

Certainly in the past, other prestigious brands have not felt need to let anyone know when they make a change, and often fail to update their specs even with significant changes to the drivers in question. When you think about it, PE exemplifies the exception rather than the rule in even admitting there was a change, let alone admitting the change was not intentional. Generally we are left to our own devices to determine if a driver parameters have changed.

Vifa, Peerless, Seas and Scan Speak certainly all come to mind as MFG’s who either have significant specification changes over the lifecycle of a driver, or have had consistency issues from one production run to another. I remember several years ago when the Peerless CSX driver line’s T/S parameters and measured responses starting resembling their (then) newer HDS series. IIRC, Peerless never changed the published CSX specs or even acknowledged there was a production change, although almost certainly there was. I’m sure you can all think of other examples, most without part number changes…

To my knowledge, PE has not heard back from Usher. I'll post when I hear anything.

C

[jkrutke]

Good news, the non-ferrofluid RS28 is going to become the standard, and PE is going to take care of people.

Post on PE forum

If they are going to keep the tweeter around, as ---k--- suggested a rename would be a pretty smart course of action to avoid confusion. RS28B anyone?

[---k---]

Yeah, the do need to rename them. I'm not sure just calling it B is enough though. Too many people don't read that carefully. But, I'm not sure what a better name is.

John,
Could you please post an impedance plot of the FF and non-FF overlaid. I thought I read that it was obvious which one you had by looking at the impedance graph. Might be easier for people to check that way than disassembling them.

[jkrutke]

Unfortunately, I don't have any older RS28's with the close-up of the impedance to make it clear. But in summary, the non-ferro is a pointy 8 ohms and the ferro version is a rounded over 6 ohms.

As far as modifying designs, for most the impedance will be less of an issue than the response differences. Those crossing over high will probably not have to change anything, but those crossing low probably will.

[---k---]

That is good info.
Thanks.

[---k---]

I'm going to need help interpreting this:


I opened both up, and they look the same to me. I don't see any Ferro Fluid in either. But, to me, they look the same as what Dave shows on his website. So maybe I don't know what I'm looking for. I just pulled the diaphragm off.

Another thing that is weird, all the early ones I have all have the sensitivity label rated as 90db +0, +0, +0.2. The new ones that Zaph tested are all 91 +.6.


Finally, here is the impedance measurement of my RS28a (12/2007) out of my center. I've always thought there was some not perfect about the center. It looks like it. I also couldn't get the diaphragm off this one. It was pulling the motor out with it. I think I'm going to be contacting PE about it.


When CJD gets back, maybe him and I will discuss taking new measurements with all the tweeters I have.

[jkrutke]

Ugh, you've got a defective center. I've never seen one that bad though. In the couple of defectives I've come across, the Fs only went up by about 100 Hz, and the second bump was tiny in comparison.

It's a little disappointing to see a defective one that's such a late model. (12/2007) I would have thought they had that issue ironed out by then. Other than your new ones, I haven't tested any since late 2006 myself.

[---k---]

So, any thought on whether the June 07 tweeter has ferro fluid in it? Like I said, I didn't see anything, but the impedance isn't the same.

I guess I should just put the new tweeters in and see how it sounds.


As for the center, I'm not sure I didn't break the center tweeter getting it out of the speaker. I'm not sure how I did it, but it doesn't sound the same as it did before. I've listened to it briefly after I posted the above. It tested funky before I tried to take the diaphragm off, so somehow I must have damaged it getting it off the speaker. I just don't know how I did that.

[jkrutke]

Guessing by your pictures, I'd say none of those have ferrofluid.

Damaging the tweeter while removing it from the cabinet isn't likely, so you can be pretty sure it was defective. Making it worse after trying to remove the voice coil is certainly a possibility though. The tweeter has a long voice coil with tight tolerances and has to be pulled out very straight or it might wedge onto the sides, making things worse.

Now that I think about it, I can understand why your defective tweeter's impedance looks worse than the defective ones I tested... Yours had no ferrofluid and so you lost all of the lubrication and self centering that ferro offers. Maybe it's easier to see the defective ones without the ferro. The harmonic distortion would surely look bad, but I bet the response curve looks like hell for that tweeter too. Probably some sharp dips here and there.

[---k---]

cool.

I'll try and talk CJD into getting his measurement rig out and test with both my old and new tweeters and see if we can come up with something.

Any clue on the big difference in sensitivity. I mean, the printed a new label with 91 on it versus the 90 that the old ones has. Where did it gain the 1.6 db?

[dlr]

As far as modifying designs, for most the impedance will be less of an issue than the response differences. Those crossing over high will probably not have to change anything, but those crossing low probably will.

That FR difference looks to be more than FF to me. The higher Q of the knee, the steeper rolloff below it and the unevenness characterized by the dip around 3.5K and the peak above it looks more like a difference (reduction) in the chamber damping. I've seen similar changes when damping in a chamber was a bit thinner and/or did not fill the chamber up to the rear opening of the vent. The pipe resonance isn't fully damped in that case.

There's more than a 2db change in part of the low end. I don't see how that can due to lack of FF, it's more like the difference between drivers with/without a chamber, a decrease in apparent volume. I'd be interested in details of the internal damping.

Dave

[jkrutke]

Any clue on the big difference in sensitivity. I mean, the printed a new label with 91 on it versus the 90 that the old ones has. Where did it gain the 1.6 db?

Kinda hard to say. Some of it could be just the higher Qts from lack of ferro, some of it could be magnet charge and like Dave says, some could be from damping differences. Or all of that, who knows.

[fjhuerta]

In my eyes this tweeter's reputation is damaged beyond repair. It wouldn't really affect me either way if it was kept around or discontinued. I don't intent to use it in any designs.

I agree with this. Unless you have the necessary equipment, replicating a design with the RS28 is now even more difficult than before (I remember MarkK stating that some RS28's may be able to work with his RS225/RS28 design, and some wouldn't - that was a couple of years ago. Now there's even another variable to consider!).

I like the RS28 a lot, but I'd take a 27TBFC/G any day of the week over it, if only because of its consistency. And lower price

[Mark K]

I agree with this. Unless you have the necessary equipment, replicating a design with the RS28 is now even more difficult than before (I remember MarkK stating that some RS28's may be able to work with his RS225/RS28 design, and some wouldn't - that was a couple of years ago. Now there's even another variable to consider!).

I like the RS28 a lot, I'd take a 27TBFC any day of the week over it, if only because of its consistency. And lower price

Yes, and that still stands.

I've never been completely satisfied that the qc bugs have been worked out of this tweeter, as great a tweeter as it is. The original problem with the RS28A's were some VC were not aligned as well as others, and the felt discs also seemed to have alignment/placement issues.

I know that it was "fixed."

But I never got a good explanation as to how it was fixed. Was a production or QC method changed? Certainly the QC issues did not prevent the tweeter from being sold without FF. Now, it may in fact be better. The Seas 27TDC tests marginally better than the Seas 27TDFC. Still, not posting it, or renaming it to make it clear the FF had been removed is not ok.

Bottom line is that I would like to see the tweeters individually tested. That's probably not realistic. Still, you could hire a community college student and plow through distortion and impedance curves for 50 pairs of tweeters in a typical day. Not sure how many PE sells, but hiring a summer intern for two weeks is 500 pairs. (OK, so it would be a bit boring...beats shoving sardines in a can!)

It's one of the reasons I went to the Seas 27TBFC/G. The Seas units are clearly more consistent.

For what it's worth, my 9950 Usher tweeters show some variability in the low end distortion pattern, but no real duds and the xover point I used was higher.

[---k---]

Just a note. I sent PE an email today about my one RS28 with the funky impedance. They're sending me a new one without any argument. So, I think they are trying to do the right thing, it is just most things are outside of their control.

I wonder if the User tweeter that the RS28 is based on has similar issues and if they are shipping commercial speakers with that tweeter in it.

We'll see how the new one does. I'll probably have to add some padding. I need to swap the other ones out and listen, but I'm headed to Iowa this weekend, then Jacksonville for a week, then to St. Louis. I'm not going to see my speakers for a long while.