Experience with PT Wave Guides [Archive]

Zilch

05-10-2006, 01:45 PM

The best assessment I can offer is that they (PT vs. 2344) are different.

Purchase either PT-H1010 or PT-F1010 (my preference) to mate with your 2431's and build up the HF section of AM6212/00 modified as per Q&D for the compensation. You can "shape" the curve by varying the values in the final LCR notch filter there.

Let them roll off a bit above 16 kHz. If I "push" the UHF extension very hard, the breakup colors the entire upper midrange.

Also purchase the JBL throat adapters. These allow you to use 2426H's exposed snoot to mount to PT waveguides, which combination also works with the AM HF filter. You can directly compare the two driver families that way.

Note that the total compensation "boost" employed with PT waveguides is less (2 - 4 dB) than 443x filters apply with 2344(A) horns. You may find that even 2426H sounds better on the PT's.

The "HF" tagged on to the PT part numbers indicates they are high frequency waveguides; using them at 800 Hz is dubious, but they seem to handle 1 kHz fine.

Another forum member (Stevem) is using 2450SL, as I recall, on PT's, with apparent success. I don't know how low he's playing them.

Finally, I found the 2352 horn to be the best of all those tried, and also developed a preference for the 50° x 90° dispersion pattern over the course. The PT-H version of that pattern, having a long throat, requires delay of the LF driver for optimum time alignment, however. When last checked, the PT-F95 was still not in production, but the $9.90 1" thread-on version is well worth trying with 2426H in comparison....

Edit: Frank mentions 2435. According to the published curves, they have better HF extension than 2431 before breakup. They don't "push" gracefully, either, tho. It's them playing on the 2352's here....

Earl K

05-10-2006, 01:56 PM

Hi Rob

Take a look at Mr. Widgets' CLIO studies of the "PT" waveguides mated to various 234x drivers. (http://audioheritage.csdco.com/vbulletin/showthread.php?t=6050)
IMO, they tell most of the story about these particular waveguides.

Notably :
- the PT-H, 100° x 100°, looks to have a loading limit of around 1500 hz .

FWIW :
- the 2344 spec. sheet mentions a 800 hz load limit for that horn bell .
- AC Impedance Studies using a 2344/2425 combo, bears this figure out .

Of Interest ( at least to me ) :
- The newer Array stuff seemingly uses the midhorn right down to the horns' load limit.
- This post strongly suggests the implementation of this loading practice . (http://audioheritage.csdco.com/vbulletin/showpost.php?p=102230&postcount=25)
- A quote from G.T. on the crossover point for his prototypes. (http://audioheritage.csdco.com/vbulletin/showpost.php?p=104924&postcount=52)

Conclusion :
- If you like how the 2344 performs in your listening space & if you already own the original 2 piece 2344 , I'd work with someone ( like Widget ) to make up a new 1.5" throat entry carved out of wood ( Widget has already published nearly all the pertinant info in various places ). Then, I'd lower the horns' crossover point to it's load limit and build a new crossover for it .

:)

Robh3606

05-10-2006, 09:05 PM

Thanks for the responses. I tried ordering from JBL Pro and of course they are out of stock on the 100x100's:rotfl: Sooooooooooo I will have to wait and see.

Rob:)

Zilch

05-10-2006, 09:31 PM

Alas, I've sold all of my spares. :(

[I can send you some funky-ass 1.5" P-Audio PH-316's to try with 2431H in the meantime, tho.... :p ]

Robh3606

05-11-2006, 05:00 AM

Hello Zilch

Thanks for the offer I want to see how long the wait is before I borough them. They are supposed to get back to me via email.

Rob:)

frank23

05-11-2006, 12:47 PM

Hello Zilch

Thanks for the offer I want to see how long the wait is before I borough them. They are supposed to get back to me via email.

Rob:)

ok, that might be the same reaon then that my importer also does not get his 100x100 order delivered

frank

Robh3606

05-11-2006, 04:30 PM

Hello Frank

I will let you know what happens with them.

Rob:)

stevem

05-12-2006, 12:19 PM

Another forum member (Stevem) is using 2450SL, as I recall, on PT's, with apparent success. I don't know how low he's playing them.

I cross over at 1 KHz, using 48 db/octave slopes. They sound real nice with the 2450SL, and have no trouble going flat out to 20 KHz (with some EQ or course).

Zilch

05-13-2006, 01:41 PM

OTOH; properly designed , true Constant Directivety horns ( which I suspect Bruno is not contemplating using down to 500hz ) offer a sort of builtin BrickWall Brake-effect for the loading of the lower frequencies. Ie; they seem to offer true loading right down to high side of the lowest, most significant impedance peak . At least that's my interpretation of all the various horn/driver impedance plots I've studied....

Notably :
- the PT-H, 100° x 100°, looks to have a loading limit of around 1500 hz .

WT2 impedance curve for 2431H on PT-H1010 waveguide.

First peak is at 560 Hz.

It's loaded down to ~750 Hz, then, no?

EDIT: Added the FR curves. First one says "PT-F," but it's PT-H. Bottom one is similar PT-F, which seems to play a little lower, and flatter from 2 kHz - 6 kHz....

Earl K

05-13-2006, 05:34 PM

Hi Zilch,

- The red lines in the picture below show where my words are drawn from .
- The operative words from JBL are : "on a true full-bandwidth constant directivety horn design ".

The 2380 and the 2344 are obviously not "true full-bandwidth" and one can see the ambulations in their Z responses, as well as the attendant Frequency Response drops. I've identified the loading limits that I would use based on available info ( mostly the FR curves ) .

The impedance ambulations that you've shown has some aberrations that make it obviously deviate from a " true full-bandwidth constant directivety horn design " ( ie; there is no single loading peak at resonance). In a case like that one must trust a full bandwidth study of the driver/horn combo and choose the appropriate load-limit based on a 3 db down point . ( I find the 2380s' response interesting in that it shows a significant weakening in what the bell can support below where I drew the red line ) .

It's ( PT-H ) loaded down to ~750 Hz, then, no?

No . Reading an impedance plot from a horn like that , requires a different set of criteria ( since it's obviously not a "true full-bandwidth constant directivety horn design " ) . I don't know what causes that little promenance that's attached to the side of the most significant Z peak / but I would ignore it's relevance in this study .

Cheers

Zilch

05-13-2006, 11:19 PM

So, Earl, you posit that the zero-crossing null between the two resonance peaks in the impedance curve correlates with F3 of the driver/waveguide combination, and their loading.

The Fmin is easily ascertained by WT2, either via Fs or Vented Box tests, and the F3 may be found from CLIO MLS measurement.

Here they are for the same 2431H driver on PT-H1010 and PT-F1010 waveguides, an interesting case, since they are so similar. Can your theory discriminate between these two?

Indeed. ;)

PT-H:
Fmin = 1393 Hz
F3 = 1208 Hz (cursors)

PT-F:
Fmin = 1162 Hz
F3 = 1099 Hz

PT-H impedance (red, phase is blue,) top, PT-F next. Note that the FR curves (bottom) nicely track those Mr. Widget did a year ago, above.... :thmbsup:

Zilch

05-14-2006, 12:10 PM

How does the AM crossover "work" on the frequency responses of this driver on these waveguides?

I've normalized to the 2 kHz - 4 kHz area where the filter basically does nothing; the violet follows the red, and the blue follows the black in this region.

Am I operating "loaded" with these combinations?

Earl K

05-14-2006, 01:35 PM

Hi

How does the AM crossover "work" on these frequency responses?

I've normalized to the 2 kHz - 4 kHz area where it basically does nothing:

- I'm not clear about what you're actually asking here .
- I had to research ( just now ) what the AM crossover is. It seems to me that you , Guido and Giskard mapped all this out, a year ago. So, I'm confused regarding your need to query. Anyways ;

- I'll look at the schematic , then later I'll offer a verbal analysis , element by element .

:)

Zilch

05-14-2006, 02:27 PM

I'm not clear about what you're actually asking here .
You stated in #7, above:

- the PT-H, 100° x 100°, looks to have a loading limit of around 1500 hz.I'm seeking to refine that determination, as several of us are trying to use them closer to 1 kHz.

Toward that end, I have provided new quantitative information about the impedance and response curves, and have also shown how the filter I'm using is "working" (interacting) with those, in actual use.

You've recommended crossing down to the load limit of the horn/driver combination. I'm trying to ascertain more precisely where that is for these PT waveguides with 243x drivers. It seems to me it may be as much as 200 Hz lower than I am presently operating them.

The "How does the AM crossover 'work'" question was rhetorical, answered through presentation of the curves. I'm not asking what you thought I was asking here (sorry for confusing,) but any suggestions for improvement would certainly be welcome, as well as comments on the topology. It incorporates a series resonant LCR notch, which you have recently mentioned elsewhere (Re: Dawson).

[Looks like it could use another one @ ~13.7 kHz for 2431H, actually. :thmbsup:]

The question is whether or not I'm operating "loaded" under these conditions....

Robh3606

05-23-2006, 09:55 AM

You've recommended crossing down to the load limit of the horn/driver combination. I'm trying to ascertain more precisely where that is for these PT waveguides with 243x drivers. It seems to me it may be as much as 200 Hz lower than I am presently operating them.

I will be trying them out at 1.5K with 24db slopes. JBL runs them down to 1.1K the PTH that is. Let you know what happens. Take a look at this quote I grabed from Earl about his OS waveguide. The link takes you to the thread.

Length isn't a factor, but mouth area is. Down to 800-900 Hz, with absolute directivity control the mouth would have to be about 24" in diameter maybe a little more if you want to be safe. Now with that knowledge in hand, what is the angular coverage that you want, because that will then determine the depth. Wider coverage is a shorter waveguide. Larger throats also have shorter waveguides.

Horns and Waveguides operate from such completely differnt points of view that its hard to compare them dimensionally because each case is different. There is also the fact that there is no such thing as "cutoff" in a waveguide - so this entire concept is out the window. But, much to my delite, the OS the coverage angle is explicitly stated in the design equations. This then tells me directly what the tradoffs are for coverage angles and throat sizes etc. This does not exist in horn theory, since angle independence is assumed.

Horn theory deals with rates of change of areas, while waveguide theory is concerned with rates of change of the boundaries. The later gives you the former, but not the other way arround.
Earl Geddes

http://www.audioasylum.com/forums/hug/messages/107845.html

Rob:)

Zilch

05-23-2006, 10:22 AM

Hi, Rob.

Several of us have looked at and worked with them now, and I'd say that you're in good territory crossing at 1500 Hz.

My own experience is that they sound better crossed at 1200 Hz than they do at 1 kHz, and I have made no serious attempt at using them below that. JBL uses them down to 1.1 kHz, yes, but that's in SR, and may not stand up to Hi-Fi listening criteria.

Mr. Widget has listened to them here at 1 kHz, and categorizes them as "PA horns" used that way. From our first measurements, he concluded 1200 Hz was the lower limit.

Bottom line at this point is we each simply have to try them. To me, the PT-F sounds better than PT-H, generally, and I think some of the data supports that.

Admittedly, I have not conscientiously controlled the variables (slopes, filter curves, etc.,) but those are my impressions.

From Geddes, I conclude that "cut-off" does not compute with "Waveguides," and perhaps not "loading," either, but who knows WHAT these actually are (the design theory upon which they are based,) and how they relate to conventional horn theory? :dont-know:

Robh3606

05-23-2006, 10:44 AM

Hello Zilch

Yeah I hear you. I use the 2344 up there as well. I tend to be real conservative and don't push horns towards the lowend of their range. I thought Earls Geddes comments were interesting as they confirm the different depths on the PT wavequides and also felt that his comment about the lack of a cut off frequency compared to horns was something new to me at least. I like to keep as far away from the lowest recomended crossover point as makes sense. I am curious about how the 2431's sound around that peak in their response. I would like to see what the coated diaphrams look like up there to see if the coating helps that peak at all. I know in the new Array Series they crossover to a tweeter so that peak could get burried and notched and simply not be an issue in that application. Might do the same if they sound ragged up there.

Rob:)

Robh3606

06-05-2006, 08:05 PM

Oh well I got my horns from JBL and guess what!! They are the wrong ones:rotfl: So off I go tomorrow to give them a call. It took me a while to figure out what they where. They are the 75X50 352606 from the SRX722's. I ordered the PTH1010's. So there a bit different. I am actually glad cause I wanted to try a horn out in this geometry. This is the first time I can actually listen to the 2431's and they sound like they could be quite nice. I have not done any comps just a series cap and a notch to get them smoothed a bit. Zilch I think you looked at there little brother. These are decent size. They are aluminum and ring like hells bells un mounted.

Rob:)

Zilch

06-05-2006, 08:42 PM

Zilch I think you looked at there little brother.
Yup. Power response was poor. Above 10 kHz, they crashed off-axis.

Sounded good, tho, if I stayed put. :p

Yours, a narrower pattern, may behave better, tho.

CLIO'll tell the tale.

http://audioheritage.org/vbulletin/showthread.php?p=65528#post65528

[Pleasingly crispy-sounding, as I recall.... :thmbsup:]

Robh3606

06-18-2006, 08:12 PM

Well I got the right horns finally and a pair for 2435 diaphrams I plan on using in the 2431's. Have any of you compared the diaphrams on these, I got guite a surprise I expected them to be the same except for the materail. There not at least the examples I have. The 2431's are flat with no halfroll like the 2435?? I am wondering if I have the older version in these. I don't seem to have the peak at 14KHz as well??? The red and Green are the drivers on the my other horns. The dip at 3Khz is a measurement anomally. The violet is on the PH1010 with the notch filter across the driver. Attached are a couple of photo's of the 2435 and 2431 diaphrams as well as the phase plug on the 2431H.

Zilch

06-18-2006, 08:55 PM

According to the specs, 2431H is "A version of the 2430 with a narrower coil milling and flat surround for extended HF response."

http://audioheritage.org/vbulletin/showpost.php?p=54746&postcount=402

Red and green are what drivers on what horns, please, Rob?

Robh3606

06-18-2006, 10:52 PM

Hello Zilch

Those are 2431 on the 90X50 SR horns with a make shift network I threw on them while testing.

Rob:)

Zilch

06-18-2006, 11:16 PM

Thanks, Rob.

Though I suspect you're moving on past the SRX horns now, if you try to EQ them flat on axis, I found they'll still collapse above 12 kHz or so.

I put considerable effort into making them and 2332 work, with poor results.

It's noteworthy that, while JBL shows essentially the same result as you found in the on-axis FR curve, they omit DI and beamwidth data above that frequency range.

http://www.jblpro.com/srx700/PDF/JBL.SRX722.pdf

Bottom line: they cannot be effectively EQ'd over the specified horizontal coverage, in my experience working with them.

[They'd make good dinner bells, tho.... :p ]

Robh3606

06-20-2006, 07:14 PM

I installed the 2435 diaphrams today and was surprised to find they wouldn't fit:banghead: . I laid the 2431 and 2435 side by side to see that the 2431 had part of the terminal block and rivet removed. I went to the grinder and took the edge off the terminal block to match the 2431's and they fit:bouncy:. I used the provided filter paper to remove the old ferrofluid and cleaned the gaps. Then put in the new fluid. That was fun as the gap flux sucked it right out of the tube!! I lined up the diaphrams and reinstalled the covers. Did a quick sweep at home and then tried some measurements with a preliminary network on the PTH1010 horns. To my surprise they looked very good and had much more extension up top then I thought I was going to get from them. Matter of fact a little too much which really surprised me looking at Widgets horn study and Project May Data. Attached are the Sine files with harmonic distortion both 2nd and 3rd and an overlay done with MLS to see how well they match up. I have not listened to them yet or finalized the network so I still have a bit of work to do.

Rob:)

Zilch

06-20-2006, 09:57 PM

I used the provided filter paper to remove the old ferrofluid and cleaned the gaps. Then put in the new fluid.Well, there's how ignorant I am -- didn't know 2431Hs had ferrofluid in them. I've never pulled a diaphragm on one.

To my surprise they looked very good and had much more extension up top then I thought I was going to get from them. Matter of fact a little too much which really surprised me looking at Widgets horn study and Project May Data.Same surprise here:

http://audioheritage.org/vbulletin/showpost.php?p=66244&postcount=115

They measure better on PT waveguides than the specs would dictate.

Attached are a couple of photo's of the 2435 and 2431 diaphrams as well as the phase plug on the 2431H.Didya see what was "two-stage" about it?

Guido

06-21-2006, 03:37 AM

I have not listened to them yet or finalized the network so I still have a bit of work to do.

Rob:)

Glad that the 2435 Diaphragm deal worked out.

Please keep us posted with the crossver design. In what combination do you plan to use those nice drivers?

Robh3606

06-21-2006, 04:25 AM

Hello Zilch

They measure better on PT waveguides than the specs would dictate.

Thanks for posting the link. First thing I did is drop in some pink noise and look at the RTA. I got very similar results. You have more resolution than I do with my older 8024.

They measure better on PT waveguides than the specs would dictate.

Yes they do. It was a pleasant surprise.

Didya see what was "two-stage" about it?

No I tried to back light the driver through the phase plug. Maybe Steve will jump in on that one.

Rob:)

Robh3606

06-21-2006, 04:41 AM

Hello Guido

Glad that the 2435 Diaphragm deal worked out.

Yes they did and thank you for them:D

In what combination do you plan to use those nice drivers?

Right now I am going to swap out my 2344/2425 combo in my main active set-up. I am hoping for an easy swap. My main concern is the PTH1010 horn. It sure looks good but I won't know until I get it all working. That will team them up with E145's and 2123 midranges. I am planning on crossing them over at 1.5K.

Rob:)

Robh3606

06-21-2006, 04:26 PM

Just tried some capacitor value changes. I was trying to find the value where I wouldn't depress the low end response as much and still maintain a relatively flat curve from about 1.5k and up. The 1.6 Uf in the original curves actually give you a rising response relative to the low end of the drivers range. I wanted to get a slightly rolled off upper octave response. I tried 3uf and 5uF series caps and you can see the sensitivity and balance changes with each value. The measurements are all at the same level from the same set-up so the level changes are due only to capacitor value changes.

Rob:)

johnaec

06-21-2006, 05:39 PM

Rob - could you identify which values go with which colors in your charts? It looks like green is a different value in the lower one?

Thanks - John

Zilch

06-21-2006, 06:33 PM

Hi, Rob.

We need to know the circuit you're working with.

If you're going to cross at 1.5 kHz, I'm not sure I appreciate the point of trying to flatten between 1.5 and 2 kHz, though I believe it can be done.

Do you have Spice to assist with tweaking?

Robh3606

06-21-2006, 07:57 PM

Rob - could you identify which values go with which colors in your charts? It looks like green is a different value in the lower one?

Hello John

Sorry I didn't even realize I changed up the colors on the traces. On both graphs the top is 5uf middle is 3uf and lowest is 1.6uf. The larger the series cap value the higher the sensitivity.

Hello Zilch

It is a modified AM621200 crossover, with a series cap and the notch filter. The notch filter values are L4 .05uh R1 2.5 and C2 is 6uf

If you're going to cross at 1.5 kHz, I'm not sure I appreciate the point of trying to flatten between 1.5 and 2 kHz

I am not going to at this point. I want to try it first and see how much of an anomally I get through the crossover point using the active crossover. I can always bring it up 100Hz or so.

Do you have Spice to assist with tweaking?

No but I should spend some time to learn to use one of the programs. As off yet I have been putting it off but down the road I will need it. I have Micro Cap I have played around with and Cross Over Pro 3 so I could have but starting with JBL's crossover and having Clio makes it real easy to just change the series cap values and take actual measurements. Clio is a real pleasure it's so easy to use.

Rob:)

Zilch

06-21-2006, 08:35 PM

It's not changing the area of interest.

You're missing a key component.

The series cap and shunt inductor in combination determine the highpass frequency and "Q".

Adjusting the notch can alter the "Q" somewhat.

I'd set 1 kHz where I want it, then contour the rest with the notch, if possible.

I'm not familiar with MicroCap, but you can do this in Xover3P.

You'll need to put in the response curve of the driver and horn, tho.

With some assumptions, Spice sim of your voltage drives:

Robh3606

06-22-2006, 04:26 AM

It's not changing the area of interest.

You're missing a key component.

The series cap and shunt inductor in combination determine the highpass frequency and "Q".

Adjusting the notch can alter the "Q" somewhat.

I didn't change the notch. I felt it was fine. Overall the response it is smoother than the 2307/2425 to 2405 integration in my 4344's. Take a look at the response of the 4344 over the same range. The top end of the 4344 sounds good warts and all. I used my experience dealing with the 2344 before I was biamped. When you have a single series cap on a compression driver you can manipulate the curve by changing the value. If you look at the curves the relative level in the 10K-20k region gets lower as you increase the value. That was the region of interest to me. I wanted to lower the energy in that band a bit relative to the rest of the curve. You also gain sensitivity as well which never hurts. It's hard to see but it also help raise the level in the notch at 1K. To try to get the 1K-2K region flat won't work cause of the wave guide loading. All of the graphs for just the 1010 show a hinge around 1.5K where the loading changes. We are looking at this from 2 different angles. I would rather adjust the cap as I gain sensitivity, get a little more energy in the 1-2K region and roll off a bit above 10K which is what I wanted to do.

Rob:)

Zilch

06-22-2006, 09:20 AM

Ahh, yes. I see it now. The 2344A is down somewhat in that region as well.

You're apparently far enough along with this it's fair to ask, "How does it SOUND?" :)

Robh3606

06-22-2006, 09:25 AM

Ahh, yes. I see it now. The 2344A is down somewhat in that region as well.

You're far enough with this it's fair to ask, "How does it SOUND?" :)

Exactly!! I didn't want a voiceing change when I dropped this new horn and driver combo in place.

How it sounds is the $64,000 question. I plan on building network tonight and getting them in on the weekend. I have my fingers crossed I will like it better than the 2344/2425 combo.

Rob:)

Earl K

06-22-2006, 10:54 AM

Fascinating ! :D

- The resonance point for that LCR notch-filter looks to be centered about an octave lower than what the text-book formula would predict ( 4600 hz vs around 9190 hz for the standard formula) .

- Can anyone offer an explanation to this ? ( plus maybe offer up a new formula that includes the modifiers / this has me stumped )

:)

- "Known Formula" is given as ; Notch Frequency = 1 / ( 2 * Pi * sqrt ( L * C ))

- L is in Henries, C is in Farads, F is in Hertz .

Zilch

06-22-2006, 11:33 AM

Hiya, Earl.

Notch in the sim is at 9.152 kHz, graphically.

I'd say your known formula is correct.

[It rings quite nicely when I lower R to 0.025 Ohms.... :p ]

Earl K

06-22-2006, 12:17 PM

Thanks Zilch for the clarification !

- I now realize that I was mis-reading your pics' log scale .;)

Zilch

06-22-2006, 11:19 PM

Per Rob's request, 4 X 2435HPL on PT-H1010, AM filter:

http://audioheritage.org/vbulletin/showpost.php?p=111803&postcount=19

2435HPL on 2352 uses same circuit, but with 0.08 mH in the notch.

#9 and #10 need their gaps cleaned and ferrofluid recharge, looks like....

Robh3606

06-23-2006, 04:18 AM

Thanks Zilch

That's a sanity check. Glad to see they are all in the family with the dip between 1 and 2K.

Rob:)

Ian Mackenzie

06-23-2006, 08:52 PM

Rob,

What horn driver and horn lense did you use in that 4344 measurement?

Robh3606

06-24-2006, 12:17 AM

Hello Ian

That was a 2307/H94 serpentine. There is a dip at 3K that shouldn't be there as well. One of my first measurements with Clio.

Rob:)

Robh3606

06-24-2006, 12:13 PM

I got them installed last night. I used a charge coupled network with a 5uF in series to matched the roll off of the 2344/2425 combo. It was a simple drop in. I didn't have to change the crossover point at 1.5K or modify the cabinet. The only change I had to make was lower the drive level to adjust for the higher sensitivity of the 2435/PTH1010. My initial impressions are I like them. I want to live with them a while and then try 3uf to bring the last octave up a bit and see how they sound. I have attached a couple of RTA shots that show the 2344 and the PTH1010. The one with the line is the 2344.

Rob:)

Zilch

06-24-2006, 12:32 PM

May we see a pic or two of the setup, please?

Also, MLS per my post #45 for comparison?

I used "Export Graphcs" JPG "small" to produce that directly, 1 pixel in print line width. Resizing in a graphics editor messes up the text, so I don't do any processing with the image, rather, just upload it. It comes out a good width for viewing on the forum.

Seems we both favor rolled-off UHF with these....

Robh3606

06-24-2006, 01:01 PM

May we see a pic or two of the setup, please?

Sure what do you mean:blink: ???

Also, MLS per my post #45 for comparison?

I am using the nework in the previous post 34 top curve . Latter on I will try to get one with them in the active set-up. That's a PITA cause of where Clio is hooked up.

Seems we both favor rolled-off UHF with these....

Well we will see. I want to get used to these horns before I add any top end. If I do I would go with the 3uF curve.

Rob:)

Zilch

06-24-2006, 01:18 PM

Thanks for the pics, Rob. Looks good! Being that similar in frequency response, I'd suspect any differences you hear will be largely attributable to the Be diaphragm and waveguide.

Sure what do you mean:blink: ???

We was typin' at the same time again, apparently.... :p

Well we will see. I want to get used to these horns before I add any top end. If I do I would go with the 3uF curve.I can see that in the sims. 5 uF is delivering ~10 dB of compensation, whereas 3 uF will get you more like 14 dB and pump up the very top end. :thmbsup:

[I'm using 2.15 uF and 0.76 mH as highpass....]

Earl K

06-25-2006, 10:48 AM

Zilch,

(a) Do you feel like firing up your Spice Siming program ?

(b) If so, please run 3 curves ( with the AM networks' primary series capacitor value ) being either 20, 40 or 60 uF in value ( keep the values of the LCR components the same as your initial run ) .

- For those who haven't guessed the purpose , this will provide a nice pictorial showing ( along with Zilches' original sim ) how the obtainable notch depth ( in a series LCR / strapped in parallel ) is ultimately influenced by the " Effective Source Impedance" . "Source Impedance" here , essentially being just the "Capacitance Reactance" created by the inline capacitor, with a bit of cable resistance thrown in ( when using modern DC-coupled, SS amplifiers ) .

:)

Zilch

06-25-2006, 12:06 PM

Well, not very deep, apparently. :p

[Green is 20 uF....]

Earlier sim for comparison, bottom, 1.5 (green), 3, and 5 uF.

Earl K

06-25-2006, 01:38 PM

Thanks Zilch !

From my perspective ( What Series LCRs "need to see" for a source impedance ) ;

- It sure "appears" that series LCR filters ( when strapped in parallel ) are happiest, or most effective, when acting/resonanting in a "Constant Current" topology .

- This ( "best practice" for placemennt ) will have relevance if anyone decides to migrate/ or adapt this basic notch concept for use into other circuits ( ie; maybe one that uses a "fixed" Lpad while still offering a similar voltage-drive )

:)

Zilch

06-25-2006, 02:18 PM

Yup, in my Minis, I finally resorted to a parallel LCR notch 'cause I didn't know what I was doing; no source impedance, no notch :banghead:

I've got a new approach, though: gonna bury the response anomaly over there in an 18 dB/octave filter I just "discovered" for them.... :p

Earl K

06-25-2006, 02:39 PM

Yup, in my Minis, I finally resorted to a parallel LCR notch 'cause I didn't know what I was doing; no source impedance, no notch

Yep, that's about where I am on these sort of things .

- But at least now we know why a series LCR ( strapped in parallel ) will become a " frequency-specific impedance equalizer " / unless it can sense some sort of source impedance greater than the final load R ( at notch ).

- In the case of the AM filters you guys are playing with, that "source impedance" is created by the basic capacitance reactance generated within the first series cap.

- Capacitance Reactance ; Xc = 1 / ( 2 * Pi * F * C )
Xc given in Ohms , F stated in hertz , C stated in the basic unit of Farads .
eg; Xc = 1 / ( 2 * Pi * 9500 * .000005 ) = 1 / ( .298451302 ) = 3.35 Ohms .

- So sticking a 5 uF cap in front of the LCR filter is the equivalent of sticking a 3.35 Ohm resistor in front of the LCR ( but only at 9500 hz ) . This added circuit resistance is of course frequency dependant . :D

:)

Earl K

06-25-2006, 03:10 PM

,,,, snip ,,,,,, "Minis" network ,,,,, I've got a new approach, though: gonna bury the response anomaly over there in an 18 dB filter I just "discovered" for them....

- For some insight; take a look at the lowpass section from the AM6212,00 network. The two series LCR filters ( strapped in parallel ) do in fact "see" a source resistance. The source impedance would come from the first or second "inline" inductor ( 2.2 mH or 1.0 mH in this network / I'm speculating the 1.0 mH coil dominates this concern ).

- Inductive Reactance ( XL = 2 * Pi * F * L )
- XL given in Ohms , "F" stated in Hertz, "L" stated in Henries
- So; XL = 2 * Pi * 590 * . 0022 = 8.15557 Ohms . ( 590 hz is the frequency of resonance found in the first LCR .) The point ??? , within that frequency area , the LCRs' have the equivalent of either an 8 ohm ( or 3.7 ohm ) resistor in front of them ( so they'll resonate and the LCRs' load resistor can drop a few Db in the specified frequency range ).

- So you might be able to follow this same practice with your minis .
- FWIW ; there always seems to be some "peaking/ringing" produced from these setups. That ringing is certanly present in the plus 10K area above the notch frequency ( plus 2 dB ) .

:)

Zilch

06-25-2006, 09:39 PM

C'mon, Earl, get a pair of PT waveguides and play along here. :D

Entry's cheap -- the $9.90 1" thread-on 90° x 50° version (338800-001) works nicely on 2426H snoot using the same HF circuit. :thmbsup:

[That ringing is one of 4430's recently disclosed secrets.... ;) ]

Earl K

06-26-2006, 09:42 AM

C'mon, Earl, get a pair of PT waveguides and play along here.

- Later in summer I'll likely buy 4 , 60°x40° waveguides for a retrofit into some of my SR stuff. The horns' number escapes me at this time / though it is a model that you auditioned ( somewhere ) in the Q & D thread .

- I'm already playing along ! / just in my own way ( with a couple of E.E. Math Books, plus a calculator in hand and too many math formulas ). :D

- Here's a customization of your Voltage Drive sims . I've approximated the notch frequency with 2 newer blue lines . The red lines approximate the amount of attenuation ( at notch ) that I calculate out ( with my limited formulas ). You can see, my calc.s are always off by some margin.
- My calcs. don't account for the effects of that inline series capacitor ( or the inductor for that matter ) . Plus I don't know how to adjust upwards, for the top-end ringing/gain . :p

:)

Zilch

06-26-2006, 01:10 PM

Rob, is your active crossover easily adjustable?

If so, how low can you dial the frequency before the 2435/PT-H combination starts to evidence undesireable "coloration?"

Also, does CLIO Lite have the 1/6 octave RTA like CLIO Standard does?

Robh3606

06-26-2006, 02:56 PM

If so, how low can you dial the frequency before the 2435/PT-H combination starts to evidence undesireable "coloration?"

Well if I try to go below my 1.5K point the bottom drops out. I end up with a rather large hole. I can't run them with this waveguide lower than this without the driver unloading. You can see in the plots it's going south after the 2K mark. If I had a horn that supported them to say 800Hz I would give it a whirl. Me just trying to go lower really won't tell you much without a horn to load it properly.

Also, does CLIO Lite have the 1/6 octave RTA like CLIO Standard does?

I have the full package I updated it a couple of months back. So yes I do.

Rob:)

Zilch

07-01-2006, 11:55 AM

Sim voltage drives, all with 6 uF notch capacitor,

2.15 uF and 0.76 mH HP:

Blue = N3134 LP for reference

Orchid = 2.5 Ohms, 0.08 mH

Cyan = 2.5 Ohms, 0.04 mH

Yellow = 4.0 Ohms, 0.08 mH

Green = 4.0 Ohms, 0.04 mH

Robh3606

07-02-2006, 07:01 PM

Hello Zilch

You have driver response curves for them??

Rob:)

Zilch

07-02-2006, 07:17 PM

Hiya, Rob.

Well, the green curve does this:

http://audioheritage.org/vbulletin/showpost.php?p=115215&postcount=45

Lemme see how the holiday goes, maybe.

[I wanna do 2431s, too, which means horn-swapping; you know the drill.... ;) ]

Robh3606

07-02-2006, 07:28 PM

Hello Zilch

Just curious! I don't want to put you to work on the holiday!

Rob;)

Zilch

07-14-2006, 08:49 PM

For 2435HPL #7 on PT-H1010:

Robh3606

07-14-2006, 09:24 PM

Thanks

So changing the notch filter has no effect at all on the relative amplitude of the 14/16K peak these driver have relative to say the 2-10K average. If anything it looks like it can actually make it worse if you were trying to attenuate it compared to the average level.

Rob:)

Zilch

07-14-2006, 09:36 PM

Hiya, Rob,

With these, I'm seeing the peak at 16.5 kHz (cursor), and looking at green versus violet (both 0.04 mH), for example, calling them equal (within 1 dB) at that frequency, the average for green (4 Ohms) between 2 and 10 kHz would be somewhat higher, looks like. The dip is at 14.5 kHz, BTW.

Clearly, 0.08 mH plays too "hot" above 10 kHz, but your 0.05 mH seems like it might be a better value than my 0.04 mH for this combination, though I think we both like it rolling off a bit up there rather than maximally flat.

Do you have an update on your listening tests? Still like the PT waveguides?

[You've been playing them at LEAST a week, now.... :p ]

Edit: Perhaps this compressed version illustrates it better; they're all exaggerated by the higher sensitivity I used above to show the differences in filter behavior. The green and violet curves, for example, are both within +/- 3 dB through the range:

Zilch

07-14-2006, 11:58 PM

1) FR of 4 "eBay" 2431Hs @ 2.5 Ohms, 0.04 mH. Generalize with caution; illustrative of variability of such acquisitions. They should all be serviced to achieve better performance uniformity. There's no knowing what's "right" here.

2) AM crossover filter tunings using driver #14. Behavior of the notch filter variations is the same as with 2435HPL, but VHF extension of these 2431s requires 3 - 5 dB less "boost" than provided here, it would appear. I believe Earl has suggested how to achieve that reduction. I'm thinking this may be why I've always favored the "F" version of the 100° x 100° waveguide. "H" always sounds too "crisp" to me on this filter, in comparison. Followup: See bottom.

3) Compressed version of same data.

4) Test setup

5) PT-H (Red) vs PT-F (green), 4 Ohms, 0.4 mH.

Robh3606

07-15-2006, 07:24 AM

Hello Zilch

Edit: Perhaps this compressed version illustrates it better; they're all exaggerated by the higher sensitivity I used above to show the differences in filter behavior. The green and violet curves, for example, are both within +/- 3 dB through the range:

I like your original curves. You can really see what the differences are.

Do you have an update on your listening tests? Still like the PT waveguides?

Well I have had them set up for a couple of weeks now and aside from trimming the series cap value they are as they were originally. I ended up with a 3.8Uf cap up front when all was said and done. I built the networks with the provision I could trim out one of the pairs so I could raise the relative level in the last octave a tad. That's what I ended up with. I like them. They are now permanent in my mains and I have retired out the 2344's. I like the 2435 quite a bit. Both the 2431 and 2435 are a bit ragged above 10K but they are no worse than 2405's so I can certainly live with it.

Rob:)

Zilch

07-15-2006, 11:31 AM

I like them.I'm most pleased to hear this, of course. :thmbsup:

Perhaps this work will encourage others to try these drivers, waveguides, and networks.

Even within the AM product line, there's other suggested filter topologies to try.

Guido and I grabbed the simplest one for manageable experimentation; it works well enough, but there's still plenty more there to explore.... :bouncy:

Earl K

07-15-2006, 03:50 PM

Hi Zilch / Rob,

Behavior of the notch filter variations is the same as with 2435HPL, but VHF extension of these 2431s requires 3 - 5 dB less "boost" than provided here, it would appear. I believe Earl has suggested how to achieve that reduction.

- I'm not sure what I suggested previously ( must be an age thing :D ) ,,, but ;

(i) I think it's time that you all expanded your focus beyond the "elegant" simplicity of this network . It's nice and simple / but / it hog-ties a user in many fashions.
- I suggested you run the curves on the large size caps to educate whomever was paying attention, that the final attenuation one gets from this network ( in the HiPass ) derives from a-typical sources. ( Thanks for running the curves, BTW ! ) ie ; Rather than a two resistor "Lpad" providing over 80% of the attenuation / this so-called "simple" network gets most of its' attenuation from the complex interaction occuring between the LCR notch-filter and the first inline capacitor ( the coil more or less, fulfills its' typical role ) .

(ii) I'd suggest redesigning the network , but with the inclusion of a "fixed 2 resistor Lpad".
- You are going to need to discover exactly what that first capacitor is giving you as a F3 . You can make that discovery empirically if you want. Simply remove the cap and put in its' place a resistor ( say 8 to 10 ohms ) and take a look at what happens to the FR. This will give you an idea what the cap contributes to the overall curve .
- You can do the same with the coil ( no need to replace it with a resistor ).
- By doing these substitutions, you'll also discover what's interacting with the LCR to produce that 2db of "peaking" up in the UHF. That 2 db may not be useful to you guys since you seem to be trying to get rid of most of it .

(iii) By redesigning this filter along more traditional lines ( & using a fixed Lpad ) you'll gain back the flexibilty necessary to address the rest of those issues in the posted FR curves . For instance ;

- For laughs, you could put the L & C in the more traditional position "before" the pad / & choose values to promote a mild "bump" just below the lower FR cutoff ( to stabilize the FR weakness between 1200 to 1600 hz ). Thats one idea based on the JBLs' usage of bump filters in the lower reaches of the 4430 horn and all those other 434x products using 2307 horns ( though those are 3-pole bump filters ).

- Another idea ; bury a .2 to .3 mH coil under the Lpads shunt resistor to initiate a mild boost in the UHF ( you may want this after you get rid of the UHF "ringing" that the standard AM filter produces . The "buried coil", essentially gives you a frequency dependant Lpad. It was effectively used in the N3100 mkII and N9800 networks .

- You might also trying choosing a smaller value for this "buried coil" if you want to initiate some mild rolloff from the coil / say up in the 5K region to replace to lost effect of the first inline cap . ( As this coil gets smaller, the Lpad behaves more and more as a typical Lpad ) .

- Once a Fixed Lpad is in place , you could easily add a second LCR with a much higher Q, to notchout that bothersome UHF frequency, which you've both referred to as (15 or 16K ?? ) .

Anyways, some ideas <> Earl K ( I think it's time to punt this simplistic AM hipass ) , IMHO

PS ; if a person is biamping / & only using the networks' filtering for the HF contouring it provides / then ignore the Lpad advice .

Zilch

07-15-2006, 09:44 PM

Thank you, Earl, I'm understanding most of that. ;)

One element of this analysis that I've noticed is that the preferred contours with this AM filter place the "ringing" peak out past 20 kHz, where the driver response has fallen off precipitously. I ran the sim out to 100 kHz to show what's happening up there.

Not a good thing, sonically, I don't believe, pouring even more energy into the driver where its response has collapsed; it'd be better not to have that happening, as you suggest.

As more users work with these drivers and waveguides, I expect we'll see efforts toward achieving an improved filter intensify. Some members have crossover optimization software that could move this a major step forward. I can easily provide response data files on multiple drivers for such purpose.... :yes:

Robh3606

07-16-2006, 07:33 AM

Hello Earl

- I suggested you run the curves on the large size caps to educate whomever was paying attention, that the final attenuation one gets from this network ( in the HiPass ) derives from a-typical sources.

I thought we already had enough up here to show that but looking back I see your point. The higher the capacitance value you use the higher sensitivity you have with less effect on the contour. Eventually you are going to see something that closely match's the Plane wave tube measurements assuming the PTH1010 is a CD horn. The useful contouring is only over a relatively narrow range of capacitance values. If you look at the 3 curves I ran you can see that the 5UF red curve has already got you real close too the value where you won't have enough energy in the last octave. Take a look.

Hello Zilch

One element of this analysis that I've noticed is that the preferred contours with this AM filter place the "ringing" peak out past 20 kHz, where the driver response has fallen off precipitously. I ran the sim out to 100 kHz to show what's happening up there.

Not a good thing, sonically, I don't believe, pouring even more energy into the driver where its response has collapsed; it'd be better not to have that happening, as you suggest.

Why are concerned if the peak is above 20Khz?? How much information is in music above 20K where you would excite the peak??? For all intents and purposes the peak is out of band on the driver and for most source material unless your running DVD-A SACD or Vinyl.

Can you run the curves that Earl is asking for with the large value series caps?? I don't have a spare horn or driver at this point. I have another pair of 2435 in transit too me. I only have the SR horns for the time being.

Rob:)

Zilch

07-16-2006, 12:17 PM

The useful contouring is only over a relatively narrow range of capacitance values. If you look at the 3 curves I ran you can see that the 5UF red curve has already got you real close too the value where you won't have enough energy in the last octave. Yes, that's what I am suggesting as a way to "dial in" the high end of the 2431s.

Your red curve looks like it'd be nearly optimum for them.

[Geez, our curves are quite similar, which tells me we're getting real characterizations here.... :thmbsup:]

Why are concerned if the peak is above 20Khz?? How much information is in music above 20K where you would excite the peak??? For all intents and purposes the peak is out of band on the driver and for most source material unless your running DVD-A SACD or Vinyl.Another option for 2431H (here on PT-F,) then, would be 4 Ohms, 0.02 mH (green):