Built my subwoofer, now I thought I'd model it...

[Blueeyedfrog]

Hi everyone. I built my subwoofer following a design I found on the net - before I found this forum...

Anyhow, the sub uses a 12" Peerless XLS driver and 2x 12" Peerless passive radiators in a 50 litre box tuned to 20hz (I think).

However, I have two problems:

1. Having measured room FR with REW and an RS meter I have found the sub SLP drops dramatically around 24hz (not 20hz as expected)

2. I'm experiencing clipping even though I have the receiver sub volume set to -10db (the lowest is -12), and the other speakers around +1db to +3db, when I have the receiver master volume set to -8 (when playing WOTWs DVD). The sub 240 (8 ohm)/350(4 ohm) watt plate amp gain is set around 35-40%. I have checked the receiver output using the BFD input meter so as not to clip at the loudest passages (hence -10db setting). I have used no gain filters in the BFD.

My two main concerns are:

1. Have I tuned the passive radiators correctly? I have 335gms extra mass on each PR giving a total of 760gms mass each (not sure how to interpret the recommended mass in WinISD)

2. Is my receiver or plate amp underpowered and therefore producing clipping? The main speakers seem fine at much louder volumes. WinISD seems to suggest I need 310 watts of power near 20hz - could that be why the sub amp is clipping?

Could someone take a look at the attached WinISD file for the sub and check it looks ok? I inserted a 6db boost at 35hz as per the sub amp specs - not sure if that is correct. The response magnitude thingy looks awful (all wobbly) doesn't it? Is amp power, driver and PR cone excursion ok?

Blue's Peerless Sub.zip

For those who may have problems opening the file, the driver is already in WinISD as "Peerless XLS-12 308 SWR 51 147 NX..."

The Passive Radiator specs are as follows:
Vas: 169.60
Fs: 10.4
Xmax: 24
Qms: 15.2
Sd: 466


I'm thinking about rebuilding the sub - can anyone make any suggestions if that may significantly improve things (especially SLP/extension down at/to 20hz) - e.g. different tuning, bigger box, bigger amp, modifying the amp to get ride of the 35hz boost, etc.

My room is about 7 metres x 5 meters x 2.3 meters high. Sub is used for HT only.

Many thanks

[Mike P.]

Hi Blueeyedfrog, I can't model this at the moment, but a quick search shows that the 12" Peerless XLS driver and 2 12" Peerless passive radiators in a 50 litre box tuned to 20hz is a proven design. The only thing I came up with is the PR's are massed to 790 grams each, not 760grams. Not sure how much difference that would make, but it's a start.

I tried changing the .doc to .wpr and it doesn't work for me.

[Blueeyedfrog]

Hi Mike,

I tried changing the file extension to doc and then back to wpr and it seemed to work for me. So what's the secret to sending/attaching WinISD .wpr files then? I can't seem to attach them to the posts. The Peerless drivers are in the WinISD file but I can't remember their exact description (will post it, and the PR parameters when I'm home tonight). I know, I'll try to create a zip file of the WinISD file and attach that to the post.

Yes, the design I followed recommended 790gms but after trying to tune them in the built box I decided on 760gms - but still not sure if that's correct (the box is slightly different in size by about 1 litre approx). See these two threads:

http://www.hometheatershack.com/foru...peerless.html;
http://www.hometheatershack.com/foru...built-box.html.

I have another question - the lead weight I have used on the PRs has about a 0.5 - 1 mm clearance between the weight and the basket - it clears ok when I push the cone in and out manually but could it move sideways as it distorts and bang up against the basket?

By the way - my sub sounds like it's popping at high volume and it appears to do this when the BFD lights indicate clipping (either from the input of output) - so the sound I'm hearing is clipping right?

Thanks,
Blue

[Mike P.]

I got the file to work and I notice a couple of issues. With 6db of boost at 35hz and input power of 240 watts you have exceeded the subs xmax. You would have to drop the input power to 100 watts to reach maximun xmax. Spl at 20 hz would be 101db. Changing the boost to 3db at 35hz and the sub will handle 210 watts at maximum excursion with an output of 103db at 20hz. Also note you can only get so much out of a 12 inch sub with 12.5mm xmax. If you do decide to get a 18 inch sub, you won't be dissapointed!

[Blueeyedfrog]

Wow, thanks Mike! I really don't understand WinISD all that well. Would you mind explaining how you arrived at your conclusions? I found looking at the cone excursion confusing because there are two red lines. On which graph does it show how much power the sub can handle before Xmax is exceeded?

Also, from memory, WinISD recommends the PRs to be weighted at about 456gms to achieve a box tuning of around 20hz. Is this in total (228gms per PR or 456gms per PR?)? Also, does that account for the inital weight of the PRs of 425gms each? I.e., is WinISD recommending a further weight or should I subtract the intial weight from the recommended (456gms-425gms=31gms). Further, how close is the WinISD recommended weighting to the 790gms (425+365gms) recommeded by the box designer you mentioned earlier?

Would changing the tuning of the box, and so weights on the PRs, elleviate the driver excursion? While the amp can be modified, I'm not sure how confident I'd feel soldering/changing resistors (although I'd like to do it if I can) and I'll have to check if I have the correct modification instructions for my amp. So it would be much easier to change the PR weights.

Finally, do you think changing the amp will make a very big difference (is it worth it)? It sounds like being able to deliver twice as much power (210 vs. 100watt) would make a big difference but you say it's only 2db more at 20hz. I still plan on building the 18" sonotube - so maybe I should just leave this sub as is and just accept that it's the best I've been able to achieve with this design (my first build)?

Thanks again for all the help.

[Mike P.]

We'll start with WinISD. Open the program and load your file. From the menu at the top scroll down to "Cone excursion (PR)" under the Passive Radiator heading and click on it. A red line will be seen at "24". This is the maximum excursion of the passive radiator. From the same menu scroll down to "Cone excursion" under the Driver heading and click on it. A red line will be seen at 12.5. This is the maximum excursion of the sub. In the modelling box, click on the "Signal" tab. Enter 100 in the "sytem input power" box. On the graph you'll see a line with a hump in it. The hump is just below the 12.5 mark. This is telling you that with 100 watts of input power you have reached the limits of the sub with the boost that you have. In the EQ/filter tab, deselect the 35hz 6 db boost that you added. Add another filter that is 35hz 3db boost. make sure only that filter is selected. In the Signal tab change the 100 watts to 208 watts. You'll see the excursion of the sub is at 12.5mm. The end result is by changing the boost from 6 to 3db, you increased the input power from 100 to 208 watts, without exceeding the xmax of the sub. Play around with different combinations so you get the feel for it. Remenber the "hump" can't exceed the xmax of the sub.

The PRs weigh 425grms each and you need to add 465grms each (not 456gms) for a total of 890grms each to tune the box to 20hz, according to WinISD. The 790 wieght would change the tuning frequency to 21hz.

Changing the tuning of the box will not alleviate the driver excursion problem. The problem is too much boost. Find out if there is a spec sheet for your amp that has the values needed to change the 2 resistors.

Is it worth it? It would be to me if the values for the resistors could be found. You have everything already. Just swap out the resistors.

[brent_s]

Quick note about WinISD and boost. WinISD appears to assume that the amp being used can provide the necessary power to cover the boost above the value on the input tab. Meaning, if you enter 100 watts on the input tab, but apply +3dB of boost in the EQ tab...your excursion graphs will correspond to the same alignment with no boost and 200 watts...+3db = 2*power.

I haven't looked at Blue's project file, but if it's got +6dB dialed in with a 240 watt input, WinISD is actually simulating the excursion graphs with 960 watts at 35hz. Reading Mike's summary, sounds like excursion limited power is around 400 watts...100 watts with +6dB = 400 or 208 with +3dB = 416.

Blue, if you've dialed +6dB into a 240 watt amp, but the driver and PR have the excursion to handle 400ish watts, then it is likely that the popping you're hearing could be the amp clipping since the input signal could be calling for more output than the amp has to give after the boost is applied. I don't have a BFD, but if it has clipping indicators for the line level signal as you describe, that could probably produce a popping signal if a clipped signal was sent to the amp...somehow you'd need to reduce the signal level before the BFD.

And this all may be nonsense rambling on my part...quick reading and replying when I should already be asleep could be dangerous.

-Brent

[Mike P.]

The problem he is having is 6db of boost at 240 watts is way past the subs xmax. You can only do so much with 12.5mm.

[brent_s]

I don't think I made my point clearly enough.

With no EQ active, above 20hz, Blue's model is driver excursion limited to 400 watts at 29.54hz...ok, it exceeds Xmax by .37mm. Unless the amp can deliver more than 400 watts, it will clip before the driver bottoms, no matter how much boost is applied.

If you engage the +6dB EQ, driver excursion now peaks at 19.1mm @ 33.94 hz. Disengage the EQ, change the input signal to 960 (240 + 6dB) and excursion at 33.94hz is 19.259mm. For excursion/spl calcuations, WinISD is not backing out the power increase demanded by the EQ, it's "adding" it to the specified input power.

-Brent

[brent_s]

Ok, having reread Blue's original post, I'm confused by the model. Blue, you added 335g per PR, but the model says 465g. It only changes the overall tuning about 1hz, but still a difference.

I believe the boost in your amp comes from a rumble filter. To model that in WinISD, you set the filter type to Highpass, not PEQ. Hopefully, the manfucturer supplied the Q and Fc for the filter or a chart similar to PE's boost tables for their plate amps, so that you can reverse calculate the Q/Fc needed to get +6dB at 35hz.

Just for fun, using PE's boost chart for the BASH plate amp, I plugged in 1.8/27hz...that's a +5dB boost in the 31-35hz range...most boost the table has. With 240 watts and 335g per PR, the model doesn't get close to Xmax down to 10hz (limit of WinISD)...max driver excursion is roughly 8mm at 35hz. It also roughly confirms your room measurement by getting much steeper below 24hz...the rumble filter is now combining with the natural rolloff of the driver/PR...above 24Hz, the driver/PR are relatively flat so you only "see" the rumble filter.

The rumble filter is likely the source of your woes. With the rolloff that it induces starting around 40hz, you're only capable of 97dB @ 20hz. Take the filter off, and you get 103dB @ 20hz while still staying well under Xmax anywhere above 18hz. The overall SPL curve becomes much flatter as well. My advice would be to find the proper resistor changes necessary to drop that rumble filter to 18ish Hz and close to a .7071 Q. That's pretty much the default for PE's SA240 and BASH-300...I think their Q might be 1.00 which will provide a very mild boost just above 20hz.

Depending on your listening habits, you could even use a traditional power amp with no rumble filter...240 watts will quickly drive you past Xmax below 18hz, but if you don't listen that loud, it won't be a problem. FWIW, I don't run a rumble filter on my ported DVC15...tuning is 17hz so there's plenty of material below tuning that could cause the driver to unload and I do have enough power on hand to overdrive it below tuning. However, our typical listening volume is -10/-15 to reference or around 95dB peaks with all channels engaged. Even at reference level, I've never heard the drive struggle...granted I sit relatively nearfield at 7' to the sub. YMMV.

-Brent

[Blueeyedfrog]

Okay, thanks guys.

Not sure if I have a rumble filter or not - the information supplied with the amp is scant - only that which is posted on the net by the supplier here: http://www.mwave.com.au/newAU/mwaveA...Y&sku=49070010

I have found modifications at the following two threads, however, I'm not sure which model amp I have (because I brought it from a different supplier):

Amp 1 mods: http://www.customanalogue.com/critical_q_sub_av_v.htm (older amp - I think I have this amp)

Amp 2 mods: http://www.customanalogue.com/critic...and%20Comments (newer amp)

I think with close inspection I should be able to tell which amp I have (this weekend project).

Brent, can you tell if either of these two amps have a rumble filter? Any suggestions as to which parts to change would be greatly apprecitated! I have no idea about electronics except I have done a bit of soldering, sometimes even successfully...

These mods are recommeded for an audiophile set-up rather than HT because I think it results in lower SPL. (thread here: http://www.customanalogue.com/sub_index.htm - note, this is not the design I used). Still, I think my problem is extension really - I want to avoid the drop off I have at 24hz (extend down to 20hz). But you still both recommend changing or removing the boost?

Also, I have noted that one thread states the driver Xmax is 12.5 liner one way while Max Linear Excursion is 28.75mm (e.g. the above audiophile sub build thread). Hope I haven't put you all wrong with the Xmax figure?

To answer your question re my PR tuning Brent, I built my sub following a build thread which recommended 790gms total per PR and I reduced it slightly whilst trying to tune the PRs after building the box - because my box was slightly different in size. I think this weekend I'll try adding the 465gms to each PR as WinISD suggests. And modding the amp if time permits...

[Mike P.]

Yes, definitely remove the boost. You'll have to find the proper values for the two resistors that need to be changed. As Brent mentioned, if you do have a rumble filter, changing the values to take out the boost at 35hz will also change at what frequency the subsonic filter takes effect.

[Blueeyedfrog]

Is a rumble filter a sub-sonic filter? If so, I think I may have one. Do you recommend removing all the boost or reducing from 6db to 3db? I think I may have to find an electronics repair firm to do your recommended mods, especially the changes to the rumble filter, I don't think the mod sheet stipulates how the change it. Would they know how to compute the values for the resistors do you think?

Cheers,
Blue

[Mike P.]

This is an example of what to find for your amp. Here's a link to a Dayton plate amp.
http://www.partsexpress.com/pe/showd...number=300-804
Here is the spec sheet for that amp showing the boost table.
http://www.partsexpress.com/pdf/300-804boost.pdf

[Blueeyedfrog]

How's this for an idea before I mod the amplifier?

Set a 6db cut at 35hz with my BFD? I'm not sure what bandwidth to dial in though...

I don't think I'm going to be able to find out what resistors need changing for the rumble filter - unless I go to an electronics engineer which I imagine means $$$. The modification details I have list 2 sub-sonic filter capacitors, and their existing values, but recommends unaltering them (are sub-sonic and rumble filters the same thing?). Is there any way to calculate what they should be changed to?

Cheers,
Paul

[Mike P.]

Subsonic and rumble filters are same thing. I don't know if the values you would need can be calculated. I'm unfamilar wo the BFD. I'll be ordering one next week. You might consider posting that quesrion on the BFD forum.

[brent_s]

Ok, I feel a long one coming on. You've been warned.

Nearly all plate amps available to the DIY market contain a rumble/subsonic filter in the form of a Sallen-Key filter circuit. By playing with the Q (slope) of the filter, one can induce a "boost" just above where the rolloff starts. Even DIY amps that claim to have no boost probably still have the filter circuit present, but it's simply populated with values that create zero boost and a very low Fc (-3dB point of rumble filter).

Sallen-Key Q/Fc are usually adjusted by changing two resistors in the circuit, assuming the two capacitors are the same value. Since the actual Q/Fc are based on a ratio, you can find more than one combination of resistor values that will produce the same response curve. Neil Davis is kind enough to have created this simple spreadsheet that comes *real* close to accurately matching every published boost table that I compared it to. You'll want to change both C10/C11 to .1. Column A should probably be R24...the lower of the two resistors. The actual Rxx/Ryy will change depending on the amp as this refers to specific resistors on that particular circuit board.

You appear to have copied or come pretty close to copying this project. Notice at the bottom, he mentions using Excelsub to calculate the Sallen-Key values. Earlier today (when I was supposed to be working) I found a copy, but now I can't located it again. Neil's spreadsheet will serve our purposes or simplye use a boost table published for a similar amp.

Using the R23/R24 values from the CustomAnalog link, plugged into rumble_boost, I get R23=12K, R24=180K, Q=1.94, and Fc=34Hz. Should be close enough to the stock Jaycar/Mwave amp you've got. I'll post some graphs in a minute.

Reading between the lines and comparing rumble_boost to the published boost table, you should be able to use these instructions for PE's 300-793 amp to do the mod yourself. Looks like it's the same filter/preamp board with a more powerful amp section.

In the attached graphs, yellow is what I think is the Jaycar/Mwave default and green is a Q=1, Fc=17hz. Guess which one I think is better?

[brent_s]

BTW, since it falls off my WinISD configured scale. Stock SPL at 20hz with 240 watts is 93.5dB...my suggestion is 101.5dB. That's the equivalent of having 2.5 of your current unit. Naturally, the advantage diminishes as the frequency increases.

With an f10 of 20hz, my way has a good chance of being flat with room gain. Stock is -18dB at 20hz.

If you step up to a 500 watt amp with the same filter, you'd pick up another 3dB across the board and should still be safely inside of Xsus for the driver.

-Brent

[Blueeyedfrog]

Brent, yes, that design link to the Peerless sub is the design I used


Brent, this is fantastic, I think we're getting somewhere. Using the Rumble_boost spreadsheet, if I put the resistor values given in the modification sheet I have (which might be for a slightly different amp model than the one I have), I achieve a Q of .74 which you initially suggested. This is with R24=82k, R25=180k, C1 & C2 = .1uF. Both yourself (I think) and the modification sheet recommend not changing the sub-sonic filter C1 & C2 (presently both .1uF). However, using the Rumble_boost sheet, changing the resistor values changes the Q, but Fc stays at 48hz. Should I change C1 & C2 to .33uF? That appears to give me Fc of 15hz?

Alternatively, following the Parts Express 300-793 resistor values chart, if R24=47k & R25=180k, then Fc=17 & Q=1. Any suggestions as to which values to use - your last post seemed to suggest this latter suggestion?

However the modification sheet I have does recommend changing two capacitors in the Frequency & Volume module (I presume this refers to crossover frequency and gain). Could this be changing the Fc?

Mod sheet is here (amp module AA-0508): http://www.customanalogue.com/critical_q_sub_av_v.htm

Many thanks - off to buy the parts now...

PS. What is "F10 at 20hz" (or F3 etc.)? I.e., what [] is F? (excuse me) - Edit - I think I figured it out - the drop in transfer function (whatever that is) from zero to Fx

Cheers,
Blue

[Blueeyedfrog]

I just noticed this statement at the bottom of the information sheet for the Parts Express 300-794 (found here: http://www.partsexpress.com/pdf/300-794.pdf).

"As it can be seen in the graph, even with the "no boost" setting, the rumble filter
is still in effect. The low frequency F3 is roughly 19 Hz."

Note: my amp looks like the PE 300-793 and not the 794 which this quote comes from

Is this anything to worry about? Does it relate to the Capacitors that may need changing as suggested in the Rumble_boost spreadsheet? I also looked at the excelsub.xls spreadsheet and couldn't figure out how to use the capacitor/resistor cells...

Again, thanks for all the help - I'm starting to see light at the end of the tunnel (a train?).

By the way - I massed the PRs t