Metal vs Softy Shootout

[augerpro]

Finally made some progress on two of my projects today, the dipole and these bookshelf speakers. I wanted to test out the metal versus soft drivers myth to see if there was any validity, even if just for me. The winning combination will be used for the "big" project some time late this year. My aim is to eventually complete a 5.1 package good enough to replace my Rockets. But I need to test various things and see what my tastes are first, thus this little project. The contenders consist of the Dayton RS150S / Scanspeak 9800 versus a Peerless HDS 832873 / 810921 setup. Baffles are exactly the same to keep the same FR signature. I may also swap tweeters just to listen to that too.





Crossover work starts tomorrow. Another experiment I intend to do is using a 31 band EQ to dial in various contours that you could do in designing a XO, such as a BBC dip. So I figure start with a flat response and start tweaking and see if there is anything I really like or more importantly dislike.

PS Don't mention any cool new drivers so I don't get sidetracked testing stuff :rofl:

[Jed]

I'm really impressed with all the work you have put in with the distortion tests, and the comparison of these new mini monitors will be interesting as well.

[AJINFLA]

Brandon

If you can, listen to/compare the speakers outside on stands. Including your BBC curves, etc.
When you bring them inside.....
BTW, :T on your driver tests.

cheers,

AJ

[augerpro]

I'm really impressed with all the work you have put in with the distortion tests, and the comparison of these new mini monitors will be interesting as well.

Speaking of comparisons when are we going to hear some reviews of the last 2 or 3 projects you did this month using the quickie change baffle contraption :B You buy all the cool drivers and we're lucky to get a post of the FR :cry: We need details man!

[augerpro]

Brandon

If you can, listen to/compare the speakers outside on stands. Including your BBC curves, etc.
When you bring them inside.....
BTW, :T on your driver tests.

cheers,

AJ

Interesting. I'll try that.

[JonMarsh]

Should be an interesting experiment. Be sure to ruthlessly suppress the out of band stuff with the RS150, or I bet ET will give you a lot of grief over your crossover- I hear he's really fond of those RS drivers... being as he's a "hard hat" kind of guy, he likes his speaker domes the same way. :B

But to be a really fair comparison, it should be the phase plug peerless, not the one you're using here - differences in motor and copper in the gap.

[Jed]

Speaking of comparisons when are we going to hear some reviews of the last 2 or 3 projects you did this month using the quickie change baffle contraption :B You buy all the cool drivers and we're lucky to get a post of the FR :cry: We need details man!

Well, I'm sorta one of those fanatic perfectionist type people, so yes, I have a working speaker with the MHT12, RS52, and ER18RNX- but I just recently got the crossover looking the way I want it so I have to order some new parts to see what it'll be like.

https://www.htguide.com/forum/showpost.php?p=349264&postcount=338

I want to maximize the potential before I go all out with reviews, post FR plots etc.

As for the big 3-ways, I'll be honest I haven't started to cut wood yet on it because I really want to be VERY comfortable with LSPcad 6.0 before I can really do the expensive drivers justice. Since I decided to go with dual Aurasounds NS10s, the old box won't do anymore- have to start new so that speaker won't be done until the end of August. All the details really add up and are very time consuming. I don't have an electrical engineering background like some on this board either, so a lot of theory takes a lot of time for me to research, learn, and apply to my designs. Also, I'm not the type to just use someone else's design, although that is how I started out and I've learned a tremendous amount from those on this board.

I'm a bit of a rebel and like knowing how things work too. All that takes time. So, enough excuses... let's build some speakers!

Jed

[Paul Ebert]

But to be a really fair comparison, it should be the phase plug peerless, not the one you're using here - differences in motor and copper in the gap.

My thought as well, unless the comparison choice was made with cost factored in.

[Jed]

Peerless lists the nomex series with a shorting ring.

[chasw98]

Well, I'm sorta one of those fanatic perfectionist type people, so yes, I have a working speaker with the MHT12, RS52, and ER18RNX- but I just recently got the crossover looking the way I want it so I have to order some new parts to see what it'll be like.

https://www.htguide.com/forum/showpost.php?p=349264&postcount=338

I want to maximize the potential before I go all out with reviews, post FR plots etc.

As for the big 3-ways, I'll be honest I haven't started to cut wood yet on it because I really want to be VERY comfortable with LSPcad 6.0 before I can really do the expensive drivers justice. Since I decided to go with dual Aurasounds NS10s, the old box won't do anymore- have to start new so that speaker won't be done until the end of August. All the details really add up and are very time consuming. I don't have an electrical engineering background like some on this board either, so a lot of theory takes a lot of time for me to research, learn, and apply to my designs. Also, I'm not the type to just use someone else's design, although that is how I started out and I've learned a tremendous amount from those on this board.

I'm a bit of a rebel and like knowing how things work too. All that takes time. So, enough excuses... let's build some speakers!

Jed

Jed:
Can you explain some more about the 'quickie change baffle thing' alluded to in a previous post or point me to a description?

CHuck

[Jed]

Jed:
Can you explain some more about the 'quickie change baffle thing' alluded to in a previous post or point me to a description?

CHuck

Basically, I've got a cabinet or 2 that I have tested a variety of drivers in by removing the front baffle (it just bolts on for now), or finding drivers with a similar diameter so I can plug, test, revise, and listen to as many drivers as possible without the need of new cabinet construction for each iteration.

https://www.htguide.com/forum/showthread.php?t=26030

https://www.htguide.com/forum/showpost.php?p=338964&postcount=30

[chasw98]

Basically, I've got a cabinet or 2 that I have tested a variety of drivers in by removing the front baffle (it just bolts on for now), or finding drivers with a similar diameter so I can plug, test, revise, and listen to as many drivers as possible without the need of new cabinet construction for each iteration.

https://www.htguide.com/forum/showthread.php?t=26030

https://www.htguide.com/forum/showpo...4&postcount=30

Very nice. I can see where the tall sloped cabinet with a seperate mid range section can be handy to just drop a driver in and check it. I am looking to build an infinite baffle with a quick change panel that will be large enough, portable enough, and solid enough to do up to 8 inch drivers.

[augerpro]

I have a problem and need some help. I'm making driver measurements and my phase doesn't look right. The rotation doesn't look normal, it's seems to reverse. Here is the measured plot (green line) overlaid by the HBT phase (blue line):





Normally I have to make a 50-140 dB/octave rolloff or "tail" on the highend due to the sample limit of 24k. This results in the HBT phase tracking exactly the measured phase. But the only way I get it to track this reversing phase to have a -63 dB/octave tail, which is really a huge rise in response. For the heck of it I used a more "normal" tail of 130db/octave here:





Now the HBT phase looks like it should but obviously doesn't agree with the measured phase. I'm at a loss. I haven't seen this in any speaker measurements I've done so far, including the BB project I did. I haven't changed anything since then, same method, distance, and equipment. :M

[augerpro]

BTW the mic is 1 meter from the baffle at a point mid way between the tweeter and woofer and the FFT start marker placed at 99.617 cm. This is a known method to capture driver layout and AC distance difference in the phase. I used this method just fine on my previous projects.

[Jed]

BTW the mic is 1 meter from the baffle at a point mid way between the tweeter and woofer and the FFT start marker placed at 99.617 cm. This is a known method to capture driver layout and AC distance difference in the phase. I used this method just fine on my previous projects.

I'm not familiar with SE, but maybe your soundcard is clipping? Did you check your mixer board settings and/or perhaps amplifier levels?

[dlr]

I think it's an issue with excess-phase and how you're dealing with it

I have a problem and need some help. I'm making driver measurements and my phase doesn't look right. The rotation doesn't look normal, it's seems to reverse. Here is the measured plot (green line) overlaid by the HBT phase (blue line):

Reversing of the phase has to do with the slopes that you've selected and the excess-delay. The proper way to model a driver is to use a bandpass to approximate the real conditions, 12db closed-box, 24db ported, etc. highpass. If you select something else, your injecting an unknown amount of excess-delay one way or the other. The lowpass needs to be roughly what should be the case for the driver, 12db for some, sometimes up to 18db for others (often for tweeters) since the breakup conditions are real and are usually different between drivers. These will still be approximate, so there will some small amount of injected excess-phase to start. Again, try to estimate the right frequency.

For a tweeter, you may have to do it a tad lower due to the upper cutouff, but maybe not even then. It may look odd, but if you set the lowpass at, say 30KHZ (using the manufacturer's data), even though the measurement drops below that. It's the slope that is the most important as this determines the absolute amount of excess-phase added no matter the Fc. You could always edit the file to extend the frequency flat at to top end, then use SE to add the slope. The problem with this would be that you won't have the measured phase, so I'd stick with just setting the lowpass at the highest point prior to the upper cutoff.

Normally I have to make a 50-140 dB/octave rolloff or "tail" on the highend due to the sample limit of 24k. This results in the HBT phase tracking exactly the measured phase. But the only way I get it to track this reversing phase to have a -63 dB/octave tail, which is really a huge rise in response.

50-140 db on the high is adding a huge amount of excess-phase, additional delay. I see that you've set the HBT to NOT include distance in the phase. This is fine if you're using the minimum-phase and have modeled the driver's minimum-phase, definitely not the case as you've set it up. The high slope you select is adding excess-phase to the point of trying to include the error in the start time marker, whatever it is, plus the acoustic offset.

Now the HBT phase looks like it should but obviously doesn't agree with the measured phase. I'm at a loss. I haven't seen this in any speaker measurements I've done so far, including the BB project I did. I haven't changed anything since then, same method, distance, and equipment. :M

The best way to match the model to the measurement is to always start with the best SPL bandpass approximation you can determine, check the box that says "Include distance in phase plots", then, starting from zero, add delay slowly until the HBT overlays the measured phase above the low-limit of your measurement. Remember that the low end, especially for a tweeter, may start to turn up as yours does due to the noise floor in the room. In your case it appears to be about 200Hz. Nothing below that is valid data, it it's not even higher than that. The low limit is going be dependent upon the time markers as well as the noise floor.

Dave

[augerpro]

thanks dave.

[augerpro]

Well I've been working out some measurement issues but have made some progress on preliminary XO designs. I'm getting great phase agreement and flat power response out to at least 30 degrees. I even have the first breakup at 6k Hz on the RS150 down about 40 dB, so I'm pretty happy with that. Both XO's are at 1800 Hz, LR4.

One question though for those who have used the 810921: how low of an XO can you go with it at LR4? I wanted to raise it up some but just can't get phase agreement like I can when it's lower. I considered raising the RS150 XO up to but the breakup stopped me from trying. Then again ctc distance is equal to 2600 Hz so I'm not to sure I want to get real close to that.

[mazurek]

I cross mine at 1700 LR24 and get reasonable swept sine harmonic distortion performance. I did not try IMD measurements, which may give a different picture.

[JonMarsh]

Well I've been working out some measurement issues but have made some progress on preliminary XO designs. I'm getting great phase agreement and flat power response out to at least 30 degrees. I even have the first breakup at 6k Hz on the RS150 down about 40 dB, so I'm pretty happy with that. Both XO's are at 1800 Hz, LR4.

One question though for those who have used the 810921: how low of an XO can you go with it at LR4? I wanted to raise it up some but just can't get phase agreement like I can when it's lower. I considered raising the RS150 XO up to but the breakup stopped me from trying. Then again ctc distance is equal to 2600 Hz so I'm not to sure I want to get real close to that.

The 150 and 180 both start to get into some significant energy storage issues much above 1600 Hz, and I wouldn't recommend a higher crossover for that reason. CTC is also a limitation, as you observe, and makes good integration with a moderate roll off like LR4 difficult as you raise the crossover frequency. Two ways are generally a host of carefully managed compromises; you have "discovered" some of those by the natural design process of investigation. For a 6" midwoofer, beaming will start with a higher crossover, too.

[augerpro]

First cutup of the XO's is going very good. Here are teh Softy results so far. XO about 1850 LR4:







Onaxis and 30 degrees offaxis both to the inside and outside, and the reverse null:




And the XO and Z plots:








And the results for the Metal speaker, XO at 1750 LR4:










Amazing how the small baffle dominates the response. Here is the Metal vs Softy:




It will be interesting to see how these sound. Response is actually very close because of the baffle. My current concerns are:
1) Z plots. Greatest angle I saw was 47 and -52 degrees.
2) Topend of the 810921. I used a CR to drop it's level, but when I look at the offaxis plots it drops quite a bit anyway. Maybe I won't use the CR.
3) Phase integration looks good to me...what do you guys think?
4) Lowend of the ScanSpeak 9800 is pretty hard to drop, even after after the resonance trap. Might have to go third order, but I'll play some more and see.
5) BSC. Not sure if all 6 dB is really needed, it's a real sensitivity killer. The level on the plots reflects actual sensitivity at 1.37meter.

[augerpro]

BTW Jon, do you think the RS150 breakup is sufficiently attenuated?

[cjd]

RS150 looks good to me. heck, the whole mess looks good. nice work

[Jed]

I'd investigate that bump at around 2k for the RS150. That's right where there might be a peak in the power response or off axis response, which might give the RS an overly bright presentation. Just a thought.

[augerpro]

Thanks for the comments Chris and Jed!

Jed the hump is baffle diffraction, both speakers have it since they use the same baffle. I have to work on it some more. The good thing is, at least to 30 degrees offaxis, the bump doesn't change and power response is very even. I have to do more off axis measurements to see how far out this holds true. But smooth to 30 degrees is promising.

[cjd]

Yeah, I think that bump is going to contribute, but in a minor and consistent manner.

C

[dlneubec]

First cutup of the XO's is going very good. Here are teh Softy results so far. XO about 1850 LR4:







Onaxis and 30 degrees offaxis both to the inside and outside, and the reverse null:




And the XO and Z plots:








And the results for the Metal speaker, XO at 1750 LR4:










Amazing how the small baffle dominates the response. Here is the Metal vs Softy:




It will be interesting to see how these sound. Response is actually very close because of the baffle. My current concerns are:
1) Z plots. Greatest angle I saw was 47 and -52 degrees.
2) Topend of the 810921. I used a CR to drop it's level, but when I look at the offaxis plots it drops quite a bit anyway. Maybe I won't use the CR.
3) Phase integration looks good to me...what do you guys think?
4) Lowend of the ScanSpeak 9800 is pretty hard to drop, even after after the resonance trap. Might have to go third order, but I'll play some more and see.
5) BSC. Not sure if all 6 dB is really needed, it's a real sensitivity killer. The level on the plots reflects actual sensitivity at 1.37meter.

These look very good. If it were me, I would drop the FR a db or two in the 2-4khz area, if I could do so, especially with the metal drivers. YMMV!

[augerpro]

Hello Ladies and Gents. I'll just throw up the graphs and then add my comments. This was great learning experience. I went for more of "no compromise" approach than my first couple diy projects. First the Softy, XO at 1750Hz.

SPL on axis:




SPL on axis (red) and 30 degrees to the left and right (blue):




Reverse null on axis (red) and offaxis (blue):




Impedence:




Transfer function:

[augerpro]

And the MetalHead, XO at 1650Hz:


SPL onaxis:




SPL on axis (red) and 30 degrees L&R (blue):




Reverse null on axis (red) and off axis (blue):




Impedence:




Transfer function:

[augerpro]

This was a great project. I'm REALLY happy with the results and most importantly it met the design goals I set out, namely:
1)All metal versus all soft drivers
2)Use the same baffle for the same FR signature
3)Try to keep the XO point and final response of both fairly similar
4)Aim for flat response. Since I don't really know my tastes I wanted both flat so I could use a 31 band eq later to play around and see what I like and dislike for future reference
5)Even power response and good phase integtration both on and off axis
6)Try to design a true "reference" speaker

A lot of lessons learned too. For one, the Peerless Nomex woofer is a dream to work with. Any beginner could use this driver and make it adhere to just about any slope very closely. The breakup is very mild and it's pretty much dead flat before that. Bass response is very close to the Dayton but in a MUCH smaller box. The Dayton may really shine with a vented box, but for sealed the one or two extra decibel at 50Hz just isn't worth the big box, at least when making bookshelf type speakers.

The Dayton on the other hand tried my patience. I learned alot developing the XO but it still gave me headaches. Still the first little breakup is -45 dB and the main breakup is -50dB so I'm satisfied. Getting that breakup down while also getting good phase agreement with the tweeter and getting a smoother response about the XO region was PITA. Every little tweak to help one aspect would diminish another. At one point I had about six versions of the XO that were acceptable-ish. I kept trying to narrow it down and tweak tweak tweak and the final one is the best compromise of all those aspects. Note to noobs: notches for metal cones are NOT fun. I know the Dayton is fine performer for the price, but they are just not for the beginner making his first speaker. Spend the cash and get the Peerless Nomex or Seas paper drivers. You'll have a much better chance of making a successful design for you first couple projects.

Off axis response and phase integration was very good I think. I paid more attention to what was going on off axis with this project. Even off axis phase integration is very good. I think that is where a lot of commercial speakers go wrong and end up just sounding like speaker in box, no more no less. The XO region should be just as smooth off axis as on to from sounding unnatural I think.

Anyway this is really just the beginning for these speakers. After I listen to them and start to get a feel for how this particular response actually sounds in room I'll hook up the eq and start playing. Any comments or critizisms would be helpful. I want to make sure understand the things I did right and the things I did wrong.

BTW anyone use the new Jantzen Z series caps at PE? The Z Standard aren't much more than teh regular Crosscap ones. Also I wanted to use good caps for these since I think they deserve it. I was looking at Claritycap SA's. Now if the budget just gets way to high could I use both a Claritycap and say a Jantzen cap to split values adn keep costs down? Do you really get the benefit of the better cap when doing this? And should the split be 50/50 for best results?

[Nathan P]

Is one crossover coming out more/less complex than the other or are the pretty similar?

[augerpro]

Hi Nathan. I posted that last reply after you did. I think the you'll find the answer there.

For the most part the HDS combo was very easy to work with. You could cross at a wider range of frequencies easily. Taming the upper end of the 810921 was about the hardest part.

The Dayton/Scanspeak combo was much more limited to where you could cross. Too high and the cone breakup wasn't down enough. Too low and the tweeter couldn't handle the work. The 9800 is nice but distortion rises pretty rapidly under 2k so I didn't want to cross it too low. Balancing the effect of the notch filter with smooth response and good phase agreement was not easy. It seems to have made for a very component sensitive system too. I'm going to have to test every component to make sure it's value is VERY close, or bad things start happen in the XO region. I'll see how a big of deal this is when the actual crossover is built and measured.

[Nathan P]

So lets hear some listening impressions! Great work dude.

[augerpro]

BTW my current "reference" speaker are the Onix Rockets. Very nice speakers and very highly reviewed. But there are some limitations to them and some things I don't like. But in their price range they were the best that I listened to. Anyway it's those limitations that started me on a journey that ended up at this fine forum! ;x(

When i'm done with these bookshelves they will compared directly to the Rocket 250MK2 bookshelf speaker. I measured it the other night to see how my creations measure up, at least on paper.

Grill off, 1.37meter, 6ms gate:



Grill on:




Tweeter wired out phase:




Interesting yes? Things I don't like about this speaker: laid back treble but still harsh, midrange is kind of murky, and it has kind of an unnatural sound, sort of boxy.

I found out the XO is 3800Hz! Nice for the XT25, but REALLY bad for teh HiVi woofer. So that's probably where the harshness comes from. After looking at the poor phase agreement I start to wonder how the power response looks. Maybe this where the murky and unnatural midrange comes from? The high cross over probably isn't helping the off axis response in the midrange either...

And I don't think I'm going to sweat my less than flat response anymore. Anyway I thought it was pretty interesting to see the actual measurements of this speaker.

[PoorboyMike]

Looks like you are really starting to figure this speaker design stuff out Brandon. I'm hoping that once I get moved, I can start getting a measurment set up put together and give this a whirl myself. And after reading your post, I wont start with a Daynon RS driver!

BTW, I got the boxes built for the Bose Buster set up. I ended up going 2 1/2" inches wider and 1/4" taller on the center channel due to the high qts on the woofers and the rather large xover. Hopefully everything still sounds OK once I get it running. I need to get some more veneer and buy a new hot glue gun (my old one must have walked off because the kids haven't use it :M ) before I go any farther. Also need some new terminal cups/binding posts, as the ones I have now are only 1/4" smaller than the little mtms. :lol:

[dlneubec]

Hi Brandon,

FWIW, the overall FR looks pretty good to me, as does the reverse null and off axis FR. :T Interesting that the little peak on axis for the metal tweeter at around 16khz is propogated as a larger peak, relative the the FR level, in the off axis plots. I wonder if that will stand out off axis?

Playing a little with felt around the tweeters might have some effect on the baffle and/or woofer cavity induced response anomalies, if you are trying to eek out that last bit of performance.

I'm anxious to hear your listening impressions. You'll should try them straight ahead and then toed in to see whether you prefer the on axis of off axis FR.

[Dennis H]

Nice work Brandon! :T

[augerpro]

Looks like you are really starting to figure this speaker design stuff out Brandon. I'm hoping that once I get moved, I can start getting a measurment set up put together and give this a whirl myself. And after reading your post, I wont start with a Daynon RS driver!

BTW, I got the boxes built for the Bose Buster set up. I ended up going 2 1/2" inches wider and 1/4" taller on the center channel due to the high qts on the woofers and the rather large xover. Hopefully everything still sounds OK once I get it running. I need to get some more veneer and buy a new hot glue gun (my old one must have walked off because the kids haven't use it :M ) before I go any farther. Also need some new terminal cups/binding posts, as the ones I have now are only 1/4" smaller than the little mtms. :lol:

Very cool! I don't *think* you'll have much problem with the slight change you made to the baffle. I'm excited to hear your impressions of them!

[augerpro]

Hi Brandon,

FWIW, the overall FR looks pretty good to me, as does the reverse null and off axis FR. :T Interesting that the little peak on axis for the metal tweeter at around 16khz is propogated as a larger peak, relative the the FR level, in the off axis plots. I wonder if that will stand out off axis?

Playing a little with felt around the tweeters might have some effect on the baffle and/or woofer cavity induced response anomalies, if you are trying to eek out that last bit of performance.

I'm anxious to hear your listening impressions. You'll should try them straight ahead and then toed in to see whether you prefer the on axis of off axis FR.

Hi Dan. Both of the Scanspeaks have that little peak. I'm not *too* concerned at the moment, I may not even have the hearing to really hear those little peaks! We'll see. Once I get the eq hooked up i'll try to dial it in and out and see what I hear.

I'll try that felt treatment. I'm curious to see what it does with the slightish peak around 2.2k hz.

These will not only be my reference speakers, but also my control for a lot of future experiments I plan to do. I think they should do quite well in that role.

[augerpro]

Nice work Brandon! :T

Thanks D!

[augerpro]

I'd like reiterate some of my questions since they are kind of lost in my big rambling posts ops:

1) What do you guys think of the overall response being slightly tipped down offaxis?
2) Is using teh CR shunt really a good way to shape the top octave of the tweeter?
3) Anyone use the new Jantzen Z caps at PE? The Z Standard is just slightly more than the regualr CrossCap. Wonder what the difference is?
4) I intended to use ClarityCap SA's for the most part, but I do have a budget I'd like to stay in. So is splitting values with a cheaper cap really a valid thing to do? Is a 50/50 split the best?

Thanks everyone for the advice so far :T

[augerpro]

At some point I was thinking of swapping tweeters on these just for the heck of it. The cutout is the same for the 9800 and 810921. So I did just a quick cut and paste of the networks from the Softy and MetalHead just to see how far off I am. Here are the results without changing a single value, first the RS150 and 810921 combo:





And the 832873 and 9800 combo:




Ha! I don't think it will take much time at all to develop these alternative XOs! Probably some interest around here for the RS150/810921 combo I think...

[mazurek]

I'm curious what you feel about the 810921 after you have been listening to it for a while. I have difficulty making the audiophile judgement, all I can do is real measurements; I often rely on visitors for critical evaluation.

[augerpro]

I haven't listened to them yet. If no one here suggests any changes soon I'll order teh XO parts and should have these finished next week. Then the critical listening begins.

Are you using the 810921 now? Have your visitors liked it?

[mazurek]

I have been using the HDS tweeter. I have a DCX simulated system with it right now, and have been enjoying it. On occasion, Mazeroth has been over to listen to some of my efforts, but I made some mistakes in the versions he listened to. I wired a driver in one cabinet out of phase. The other mistakes I made were due to the assumption that my signal chain was good. I realized my Rane consumer -> pro transformer had some bad high frequency roll off, don't know if it is the product or my configuration. I also figured out my second and third pre-outs on my preamp had bad frequency response, don't know if I killed them.

I had to shelf the project for a while, I just finished my thesis and will be defending it Monday. After that, I hope to have stereo playtime with friends.

I really like the HDS tweeter's low distortion, and highish sensitivity. I've played around with various felt patterns around it, I think the abberations are diffraction related. I just can't get past the ugliness of the felt I used though.