resistor parallel over compression drivers

[frank23]

hi,

I have built a set with passive crossover that I like the sound of very very much. One of the key features in the crossover is the parallel resistor over the +/- from the 2420 driver. As this is only about 1.2ohm, this damps the impedance peaks of the driver/horn combo a lot. A series resistor of about 6ohm brings the nominal level to about 8ohms again so the amp doesn't see any low impedance. I can do this because I don't need the voltage sensitivity of the driver.

Yesterday I connected my newly acquired 2440/2382 combi to my active m552 crossover and it didn't sound as good as my passive 2420 setup by far. It might have something to do with the fact that I only had about 20uF of capacitors that I could put between the amp and the driver for DC protection. The 20uF makes a 1st order crossover itself in the 500-1500Hz range [about, I don't have my calculator at hand]

Now I have also done the "parallel resistor trick" with the 2440 and I like the sound much better. I think this has to do with the impedance peaks that the 2440/2382 combi shows at about 800Hz, which is right in the crossover range of between 500Hz and 1500Hz that was created by the 20uF capacitors for DC protection that I tried out.

And the 1.2ohm in parallel effectively shorts the driver resonances, like a bad tube amp would do, so it's also damped more.

The sound is much more sophisticated now, with much less "ssssssss" spitting out. And more balanced because the impedance peaks are no longer there because of the 1.2ohm in parallel. So the voltage driven crossover curves created by the active crossover, arrive at the 2440 as they are intended.

greetings, Frank

[Guido]

JBL often worked with 7.5, 10, 15 or 20 Ohms parallel resistors.

The 1.2 Ohms are a little low IMO.

[whgeiger]

F23,

The 1.2-Ohm shunt resistor is drawing a lot more current than the driver. All this does is load down the amplifier and generate extra heat (from the shunt resistor and power devices of the amplifier). To reduce driver sensitivity, use an L-Pad [1] with a series and higher shunt resistance. To suppress an impedance peak, place a Zobel circuit [2] across the voice coil terminals.

Regards,

WHG

References:

[1] Attenuation Networks (Rod Elliott)
http://sound.westhost.com/lr-passive.htm#6.0 (http://sound.westhost.com/lr-passive.htm#6.0)

[2] Mid Range and Tweeter Impedance Compensation (Rod Elliott)
http://sound.westhost.com/lr-passive.htm#3.1.2 (http://sound.westhost.com/lr-passive.htm#3.1.2)

[Ian Mackenzie]

hi,

I have built a set with passive crossover that I like the sound of very very much. One of the key features in the crossover is the parallel resistor over the +/- from the 2420 driver. As this is only about 1.2ohm, this damps the impedance peaks of the driver/horn combo a lot. A series resistor of about 6ohm brings the nominal level to about 8ohms again so the amp doesn't see any low impedance. I can do this because I don't need the voltage sensitivity of the driver.

greetings, Frank

Interesting ideas, it helps if people can read.

The possibility of direct amp connection and bi amp would yield even better dynamic damping without losses.

Ian

[whgeiger]

Interesting ideas, it helps if people can read.

The possibility of direct amp connection and bi amp would yield even better dynamic damping without losses.

Ian

IM,

The ad hominem ‘opener’ adds nothing meritorious to the conversation. However, the fact remains, that for a given acoustic output, less current is required if the shunt resistance is only 'seen' at driver resonance, or if an L-pad is required; a shunt resistor of higher Ohms should be used (for the same reason). This is also true if the Zobel is retained in the bi-amp setting. Also if this tack is taken, a series capacitor should be introduced for driver (tweeter) protection. So, it is not necessarily true that all low impedance passives will be eliminated by use of a dedicated amplifier even though level setting and crosover filtering functions may be beneficially moved out of the power delivery circuits.

Regards,

WHG

[Zilch]

Frank: I'd recommend you buy or build an RTA if you are going to work with this, and a reliable means to measure impedance, as well.

It can't be accomplished by whim or fiat....

Note: My M552 has nasty on/off transients.

[Ian Mackenzie]

I have recently done my own tests in this area and aside from the effects of damping and zobel networks, the effect of active crossovers and series capacitors in the signal path can have a marked effect on sound quality

It seems to me that while theory is a good starting point, an empirical approach to design is often what is required in reality. Less is more tends to be the outcome...

I have sent a pm to Frank 23 to discuss his experiment further.


Ian

[frank23]

I agree with the less is more philosophy.

Of course a zobel network can be used, but it would work for one frequency only [so you'd need more in parallel in case of the 2445/2382 combo, see pic, a 2360 horn is even more difficult], is difficult to tune, creates phase shift. And a resonance frequency can change with ageing and use, so you'd have to review the tuning periodically to be sure that it is still "spot on".

For me, the parallel resistor has much less components, easier to predict, easier to implement and can be used as a attenuation circuit at the same time.

----------

For now my passive setup still sounds better than the active setup, I think this has to do with 3 things:

1 I like 1st order [passive] crossovers, this gives less phase shift than 4th order [active] crossovers, and the crossover is less sudden between drivers

2 my passive setup has paper-oil capacitors and kimber cable, while my experimental active setup has to work with MKT left-over capcitors and bad cable and of course, more active components [the M552 is of course in between]

3 I optimized my passive setup over the years and this active setup only has been "in business" for 3 days

greetings, Frank

[aust-ted]

I have recently done my own tests in this area and aside from the effects of damping and zobel networks, the effect of active crossovers and series capacitors in the signal path can have a marked effect on sound quality

It seems to me that while theory is a good starting point, an empirical approach to design is often what is required in reality. Less is more tends to be the outcome...

I have sent a pm to Frank 23 to discuss his experiment further.


Ian

Hi Ian

You got me thinking. I use a trick taught by an old sound engineer to protect my mids & tweeters (2450/ 2404) in my actively xover'd system - banks of polyprop caps, with a freq cut-off well outside the pass band in series with the drivers, to protect against DC and other stupid things I might accidently subject them to.

Do you think these could affect sound quality?

Regards
Ted

[frank23]

Hi Ted,

How large are your combined capacitors for each of the 2450 and 2405?

Polypropylene capacitors mostly are of a sufficient quality. I like paper-oil caps myself, but they can be very expensive and big [I've got some that are the size of a 2405 and provide only 4uF each].

If you need large values to keep them outside of the passband, polyprops can be a good alternative.

But DC protection can also be done through transformers I think, like the outputs of single ended [tube] amps that are also used to downconvert the tube's voltages. Then you wouldn't need caps.

Greetings, Frank

[Ian Mackenzie]

Very interesting discussion brewing here chaps.

Ted,

Yes, have you tried charge coupling those caps at this stage....nine volt battery with 2-3 megs of series resistance?..worth a try.

Frank,

I don't like scewing with Zobels in the mids and highs, they effect not only driver but amp performance as they are a reactive load. Low impedance damping is far better and this is well documented. If a driver has a bad resonance and its audible, think about an alternative driver/horn.

I have been playing around with the 4343 style monitors with Ed over the past few months and we have pondered over the active versus passive thingy.

Our conclusions thus far are that for results in terms of mid and top end sound the quality of the active crossover is critical. The bi amp thing offers 2 key benefits, high damping of the woofer and less clouding of midrange (where the active crossover offers AAA performance).

Unless you have AAA performing active crossover the passive crossover offers better absolute HiFi. We have found the popular Ashly crossover and the JBL 5235 crossover while clinically clean they loose vital fine details and transparency which can be the determining factor of mid fi and hi end.

The good idea is a simple Passive RC filter before the amp with a low source impedance pre out if you only need 6 db per octave filter slope. Otherwise discrete class A buffers are the go...I just do not think chips cut the mustard (hot english mustard!) ..particularly where pistonic devices like compression drivers are used...they are ruthlessely accurate and reveal all manner of harmonic distortions...

If you are careful and have muted started up and DC protecton on the amps, a direct connection to the amp is the best. Only in PA applications is a dc blocking capacitor an absolute must. Also its much easier to have a small value hot cap in the amp input to filter out of band frequencies

regards

Ian

[frank23]

If you are careful and have muted started up and DC protecton on the amps, a direct connection to the amp is the best. Only in PA applications is a dc blocking capacitor an absolute must. Also its much easier to have a small value hot cap in the amp input to filter out of band frequencies

Ian

living on the edge!

well I would sure like to try it that way, but I measured the DC offset on my amps and it wasn't zero, so how many millivolts of DC offset can these 2440 drivers work with? If it were a cone-mid-driver I would connect them directly without any caps, so maybe I am being too carefull. Of course direct connection would be the best, then the whole dampingfactor of the amp would be available.

Frank

[Ian Mackenzie]

I doubt if 50 millivolts would be an issue but is the amp ac or dc coupled at the input and does it have unity gain at dc or not? If it is dc coupled and has gain at dc it is unwise to dc couple the driver.

Hmm.... .I take no responsibility.

Frank just Looking at the big picture Are you utterly convinced the shunt makes this much difference..was the level exactly the same?

Or is it the active crossover slopes and extra electronics.

Even 0.5 db variations are quite audible .

Ian

[Guido]

.....it helps if people can read.....

Sometimes keeping it factual helps even more :bash:

[Ian Mackenzie]

Sometimes keeping it factual helps even more :bash:

Pity our ears are sometimes the only instruments sensitive enough to register the facts..what is actually happening in reality!

[Earl K]

Are you utterly convinced the shunt makes this much difference..was the level exactly the same?

- I've had similar results/experiences with conjugate resistors across the terminals of biamped JBL drivers. Like Frank, I too include a buildout / "series" resistor ( forming an Lpad ) to maintain the circuits' load impedance . To my ears , and dependant on the value of the conjugate used, the driver/resistor combo can start to sound like a form of applied "soft knee" compression with the threshold set to "always on" . I've accepted the conjugates' necessity in my passive circuitry when using JBL drivers. OTOH, I don't enjoy the resistors' mitigating effect ( on dynamics ) when applied in the same fashion , to Altec compression drivers.

- I feel the two differing experiences result from my observation that JBLs' "Sound Reinforcement" compression drivers typically sound underdamped ( to me ) while Altec drivers sound critically damped ( both used on smallish horns ) . So I feel, the conjugate can be used to somewhat tame the JBL driver ( but then , so can aquaplas dusted diaphragms +/or frequency specific; series type, LCR resonance traps ) .

- Just a personal opinion mind you :)

[aust-ted]

Hi Ted,

How large are your combined capacitors for each of the 2450 and 2405?

Polypropylene capacitors mostly are of a sufficient quality. I like paper-oil caps myself, but they can be very expensive and big [I've got some that are the size of a 2405 and provide only 4uF each].

If you need large values to keep them outside of the passband, polyprops can be a good alternative.

But DC protection can also be done through transformers I think, like the outputs of single ended [tube] amps that are also used to downconvert the tube's voltages. Then you wouldn't need caps.

Greetings, Frank

Hi Frank

I have 60 uF in series with the 2450s and 9 uF in series with the 2404. The active xover freqs are 800Hz and 5KHz. The caps I am using are polyprop fluoro caps which I acquired very cheaply.

What you write about DC protection using transformers is correct. However I am also trying to protect the drivers against my own stupidity (eg incorrect connection where the bass channel ends up being connected to the tweeter or mid say after some change in gear).

Regards
Ted

[aust-ted]

[QUOTE=Ian Mackenzie]Very interesting discussion brewing here chaps.

Ted,

Yes, have you tried charge coupling those caps at this stage....nine volt battery with 2-3 megs of series resistance?..worth a try.

Ian

I have read the discussions on charge coupling in the forum with great interest and intend to try it out some time with passive xovers.

However in this case I just want to make sure you are aware I am using active xovers (currently M553, but am in the process of swapping over to a dEQX) and only use the caps outside the passband to protect the drivers.

If you think the caps could be affecting the sound, I could try a bypass switch which shorts out the caps after I am sure the system is working ok. That way I could minimise an accident.

Regards
Ted

[Ian Mackenzie]

- I've had similar results/experiences with conjugate resistors across the terminals of biamped JBL drivers. Like Frank, I too include a buildout / "series" resistor ( forming an Lpad ) to maintain the circuits' load impedance . To my ears , and dependant on the value of the conjugate used, the driver/resistor combo can start to sound like a form of applied "soft knee" compression with the threshold set to "always on" . I've accepted the conjugates' necessity in my passive circuitry when using JBL drivers. OTOH, I don't enjoy their mitigating effect ( on dynamics ) when listening to Altec compression drivers.

- I feel the two differing experiences result from my observation that JBLs' "Sound Reinforcement" compression drivers typically sound underdamped ( to me ) while Altec drivers sound critically damped ( both used on smallish horns ) . So I feel, the conjugate can be used to somewhat tame the JBL driver ( but then , so can aquaplas dusted diaphragms +/or frequency specific; series type, LCR resonance traps ) .

- Just a personal opinion mind you :)

But a valued opinion my friend.

I mean JBL has imho done much experiements of an empirical nature...none of which is printed but remains in their giant heads (to coin a phrase) ..the fine tuning and voicing...the networks ..hand selecting matched pairs of drivers..therefore mere engineering for thought is not enough...one must make use of the ear as they do ....then measure the result and compute scientifically..after the fact.

Macka

[Ian Mackenzie]

[QUOTE=Ian Mackenzie]Very interesting discussion brewing here chaps.

Ted,

Yes, have you tried charge coupling those caps at this stage....nine volt battery with 2-3 megs of series resistance?..worth a try.

Ian

I have read the discussions on charge coupling in the forum with great interest and intend to try it out some time with passive xovers.

However in this case I just want to make sure you are aware I am using active xovers (currently M553, but am in the process of swapping over to a dEQX) and only use the caps outside the passband to protect the drivers.

If you think the caps could be affecting the sound, I could try a bypass switch which shorts out the caps after I am sure the system is working ok. That way I could minimise an accident.

Regards
Ted

A very good idea.

Please keep us posted.....

[Ian Mackenzie]

Shall we call this the JBL tweakers asylum?:p

Ian

[Ian Mackenzie]

Hi Frank

I have 60 uF in series with the 2450s and 9 uF in series with the 2404. The active xover freqs are 800Hz and 5KHz. The caps I am using are polyprop fluoro caps which I acquired very cheaply.

What you write about DC protection using transformers is correct. However I am also trying to protect the drivers against my own stupidity (eg incorrect connection where the bass channel ends up being connected to the tweeter or mid say after some change in gear).

Regards
Ted

Ted,

I would like to wager that a Solen fast cap bypassed with 0.01 film foil and charge coupled would create the prefered JBL sound for your 60 uf .

Ian

[frank23]

I was going to post my passive crossover layout. Hereunder it is. I use only paper-oil capacitors as they sound the most real to me.

I am not sure about the level of the red curve [high] as compared tot the green curve [low], I think it should be a few dB higher and therefore in reality the crossover point moves a little to the left.

The blue curve is an alternative high crossover using only one capacitor as opposed to the two caps in the red curve [the red is the one I actually use].

By the way, I used an 6 ohm resistor for modelling the 2420 driver, which I know is not 100% accurate, but because there is such a low resistor in parallel, it doesn't make a big difference in the models results.

And the Davis 20klv8A almost has no self-inductance due to copper rings, so a single inductor suffices and no compensation is needed.

Earl, I am thinking about your remarks on the "compression knee", I think I'll try a layout with a somewhat lower parallel resistor once.

greetings, Frank

[Earl K]

Earl, I am thinking about your remarks on the "compression knee", I think I'll try a layout with a somewhat lower parallel resistor once.

- For the sake of clarity ; I didn't actually suggest that you tinker with your present scheme of "current dump" or try to lessen the apparent "dynamic-compression" effect .

- But if you are looking to hear a bit "less" dynamic loss / then raise ( not lower ) the value of the conjugate from your present 1.2 ohms to something like twice the rated AC impedance of the driver . ( JBL would typically choose in the 10, 20 to 30 ohms range ). The "effect" should become quite minimal , if present at all, when using these other, higher impedance choices .

Note : adding a 30 ohm conjugate in place of your present 1.2 ohm will raise the overall ( impedance ) load presented by the "driver/resistor combo" to everything in the circuit before them. The load will go up towards 8 ohms, ( created by paralleling 30ohms with @ 11ohms ). A new load impedance means you'll need to rework all your RC ( & L ) elements "source side" towards the amp / ( just to recreate the HiPass you presently enjoy ) .

:p

[frank23]

I've just built a simple 1st order crossover for the 2440/2382A combi instead of the active crossover setup.

I have used the same idea that I used on my 2420/2344A combo. It tops out at about 20kHz, so you compensate for the rolloff above 4kHz. Effective acoustic crossover frequency will probably be somewhere around 1000-1500Hz.

This is much more like it! I don't know if it is due to the fact that the active crossover [M552] isn't any good, or it doesn't compensate well for the 2382A rolloff above 4kHz, but the active setup didn't work. Even my wife didn't like the sound.

I am still bi-amping because I haven't come to the finetuning of the sound levels, so I am using the same two amps that I was using in the active setup, same cables and same quality capacitors.

Maybe I can only live with 1st order crossovers...

Now I have to try what Earl meant by altering the value of the resistor network.

greetings, Frank

[ralphs99]

Hi Frank,

I have also included a low valued resistor connected across my compression drivers to minimise impedance variations and as half of an L-pad attenuator. But I think there's even more to the story....

I have done quite a bit testing and feel, like you, that it makes a big contribution to the sound, especially in an active setup. It started me thinking why this might be so. I also tried some experiments with charged coupling capacitors and wondered why I couldn't hear much difference.I also tried experiments with different capacitor dialectrics and again couldn't hear the differences that others had been reporting.

Then the idea struck me that the observations might be related in some way....

Compression drivers used in home environments typically have milliwatts of power circulating through them. This is because they are used for high frequencies where less power is required and because of their intrinsically high effiiciency. (we all know that much) The level of the drive signal never gets much away from the zero crossing as a result. Driving a compression driver with an amp directly, forces the amp to operate in the range where it least linear- near the zero crossing point (at least for push-pull topologies, class A amps being the exception). With a passive network in between, the situation is improved somewhat.

By using a an attenuator with a very low value resistor across the diaphram and a high attenuation factor causes much more power to flow from the amplifier as it also has to drive the attenuator, and also causes higher currents to flow as back EMF. This then means that the coupling capacitors work harder as well. The non-linearities of the coupling capacitors and the non-linearities of the amplifiers are thus proportionately reduced.

This, I think, explains my observations.

Sure I waste power in the attenuator by driving the amp and the network harder, but at typical listening levels, it's only a few watts, and that's a small price to pay for the improved linearity.

Comments invited!

Cheers, Ralph.

[Tom Loizeaux]

My understandinmg is that the "rule of thumb" is that Zobel networks (resistors between + & -) on tweeters should be at least double the impedance of the driver.

But then "my understanding" is always subject to challange.

Tom

[Ian Mackenzie]

Ralph,


That certainly fuels the discussions.


I agree class b and a/b amps , particularly large amps are least linear with gross crossover distortion in the millwatt range compared to say 1 - 10 watts.


I use a Passlabs Aleph and its light and day in every respect.:blink: The first watt is the most important watt....now who's quote is that?


Its the bass and midrange frequencies that create 3rd, 5th,7th and 9th harmonics from large amounts of negative feedback that cause the cheese grater effect in the horns. No wonder people snarled at the slant plate horns and slots in the 70's and 80's. Then pump it all through dirty capacitors and crap terminations and crap x crap x crap = very seriously bad sound.:blah:

But back to your point about loading capacitors, Finch and Marsh state that D/A and other capacitors distortions are reduced with reduced load impedances. Unfortunately this also means larger value capacitors which tend to have a higher magnitude of such problems.


From my own observations the run of the mill active crossovers which use mass market chips in the active circuits and low quality capacitors are the real villains in our quest to make our horns sound nice.


If when using an active crossover if you thing the sound is sterile , scrubbed or processed sounding compared to a passive crossover its the active crossover that is partly to blame.


Most designs use several chip opamps with 100% negative feedback buffers (TL072s, Ne5532, LM833s) nasty electo coupling caps and nasty 3 pin regulators. You would never dream of using a crappy preamp would you, well that's about the status quo of active crossovers used in pro audio.


I have built a prototype class A discrete crossover ..the results have renewed
enthusiasm for active crossovers. (Most v.good preamps have a single ended discrete class A stage.....why do manufacturers skimp on active crossovers..because they assume they will be used in a PA system)

Anyway, all the above does leave latitude for a very veiled electronic sound that is miles from the original signal.


While the above idea of using a shunt resister is unorthodox and may get sniggers from the engineering textbook types who time and time again choose plagiarise chapter and verse all over these forums and have absolutely no capacity think outside the Sq 3, I have for one have no reason to believe this approach does not offer real benefits.


In fact some of the most respected designers in the audio business apply radical ideas in their products because it works....and they are the trade secrets...(aka Nelson Pass...John Curl)

Ian

[Earl K]

But then "my understanding" is always subject to challenge.

Okay, Thanks for an oppotunity to blab . ;)


My understanding is that the "rule of thumb" is that Zobel networks (resistors between + & -) on tweeters should be at least double the impedance of the driver.




Ummmm, ( that is just a JBL convention or "rule of thumb" ) & about Zobels', well,,,, :)

- a Zobel ( an "Inductance Compensating Network" ) is composed of a capacitor and a resistor wired in series with each other . These two elements are then wired parallel across the ( + and - ) load lines. The location of a Zobel is typically right next to the transducer ( electrically speaking ). If a conjugate load resistor is also present, the Zobel then takes up a new location one position closer to the source ( amplfier ) / leaving the conjugate resistor to occupy the space, closest to the transducer .
- Zobels are more commonly used with woofers since woofers can have quite high values of inductance with their massive voice coils .

- A simple conjugate or load resistor ( as Giskard said many times ) can be used to tame ( lower ) some of the wilder ( highish ) impedance peaks found in the ( AC ) impedance curve of a transducer/box/horn system . With many amplifiers this can translate into sonic benefits ( think simplified "tracking" of the impedance curve ) . Also, a flatter AC impedance curve greatly simplies the design process when creating passive networks .

- JBL circuit design ( of a certain vintage ) does make pretty consistent use of conjugates that are twice the average AC impedance of the transducer when applied to horn drivers.
- One can also find quite consistent evidence of load resistors on woofers that are five to ten times the nominal AC impedance . ( These are found in the newer designs ).
- These approaches are "JBLs" particular design conventions, for use with their specific drivers.
- That doesn't mean that there's some electrical rule dictating those same values be used with other manufacturers drivers ( or even JBLs ).
- As Frank found out, using other values is doable ( with JBL ) / but with some sonic consequences.
- Exploring those "consequences" can be fulltime work/fun for the tweaker .

- I play around with Altec drivers more than JBL drivers and I'll assert that some of JBLs' design conventions shouldn't be applied to Altec drivers .
- For instance; my ears have told me that I don't like conjugate resistors wired across the terminals of Altec drivers.
- OTOH, I feel they are sonically beneficial with many JBL drivers.

- Unfortunately this forum, lacks an Altec "seer" or "oracle" who is on par with the retired Giskard. Without some comparable insights we may never know the reasoning behind some of the interesting passive design approaches found in Altecs' legacy crossovers ( they are not all boring & straightforward 2 pole butterworth designs ).

- For instance ; here's the schematic for Altecs' model 19 network. I feel it is a good example of Altecs' best design thinking for the day. It is roughly comparable in electrical function and vintage, to JBLs' N3131a network .
- These two network examples underscore two very different approaches taken to deliver listenable sound from vintage 2-way systems ( both using compression drivers ) .
- In the N3131b, one will find the inclusion of a typical Zobel on the 2231h in the 3131a network . Also present, is a 20 ohm conjugate that conforms to the mentioned 2 to 1 impedance ratio ( JBL "rule of thumb" ) . A 2420/le85 will be an @ 10-11 ohm driver ( according to Zilches measurements ).
- The N1201a has neither a Zobel, nor a compression driver "conjugate" .

- Let the ears be the final arbriter over the inclusion of these so-called passive parts and/or the implementation of these different approaches to network design . :blink:

:p

ps ; I'll try to edit this later for brevities sake . :snore:

[Ian Mackenzie]

Earl,

I am not a horn or compression drive guru by any means but I wonder what mechanical and magneting damping is at work with a typical driver / horn?

Also in some cases absolute damping is desirable in some types of drivers ie the 2405 is apparently meant to work far better if directly coupled to amp amp (ref Giskard). Mind you that is a ring radiator, not a compression driver per say.

Ian

[Harryup]

I had 2450 with 2380, 2382 and 2402, 2404. Tried first with 553, to much noice. Switched to Tractrix and just the 2450 and a DCS280. Got myself an BSS366T and came closer to Nirvana. Bought myself a pair of 2405 used 3106 filter in between the 2450 and the 2405. So much better. Measured a bit. Disconnected the 2405 and the sound became even better. Now I have just adjusted the BSS filter a bit around 2kHz to avoid some resonaces and letting the 2450 together with a SET play from 675Hz up to about 16-17kHz without any additional treble unit. And this is sofar the best sound according to both my ears and all my friends ears too.

[Ian Mackenzie]

Does not sound un reasonable....:barf:

Enjoy

[frank23]

Hi, since I couldn't get the active setup using the M552 compensation to sound right, I spent my evening trying out a new passive crossover for my 2440/2383 combo. Here it is. It sounds very nice, maybe even a little too much treble for my liking.

Very spacious!

[whgeiger]

...if applied to the personal transportation function, we would all be riding unicycles.



When the signal coming out of a high pass filter section is declining at a rate of 6 dB per octave, diaphragm displacement of the connected driver will remain on the increase even when the signal frequency drops into the filter stop band. At 12 dB per octave, diaphragm displacement remains constant under these conditions. Only at higher slope rates, is a compression driver ‘protected’ from its displacement limitations at frequencies below its minimum recommended crosover point. When compression drivers, with their ‘hard’ displacement limitations (due to ‘very close’ phase plug proximity) are operated through wimpy-sloped filters, the signal they produce will be modulated (distorted) by lower frequencies that are essentially inaudible due to masking effects of the ear. If these lower frequencies approach or extend below the horn cut-off frequency, then the audible distortion products generated are increased further due to the loss of the acoustic load on the diaphragm. In this setting “minimalist design” dogma does not ‘hunt’ very well.



Regards,



WHG

[frank23]

so it is a bit dangerous to go with 1st order crossovers, but I like the sound and as I am only using this setup in my living room at non-pro-studio-band-impressing-levels, I think all will be well

I know excursion is an issue with bas-reflex loading where the cabinet provides less and less damping where you provide the driver with a frequency below the tuning frequency of cabinet / port

does this work the same then for compression drivers?

frank

[frank23]

I had 2450 with 2380, 2382 and 2402, 2404. Tried first with 553, to much noice. Switched to Tractrix and just the 2450 and a DCS280. Got myself an BSS366T and came closer to Nirvana. Bought myself a pair of 2405 used 3106 filter in between the 2450 and the 2405. So much better. Measured a bit. Disconnected the 2405 and the sound became even better. Now I have just adjusted the BSS filter a bit around 2kHz to avoid some resonaces and letting the 2450 together with a SET play from 675Hz up to about 16-17kHz without any additional treble unit. And this is sofar the best sound according to both my ears and all my friends ears too.

hi harry,

could you show the voltage drive that comes out of your BSS that you use for playing the 2450 up to 16-17kHz?

by the way, are there voltagedrive curves for the 2380-series compensation in the M552? In my experience they do not provide enough compensation to flatten the response curve.

frank

[frank23]

By using a an attenuator with a very low value resistor across the diaphram and a high attenuation factor causes much more power to flow from the amplifier as it also has to drive the attenuator, and also causes higher currents to flow as back EMF. This then means that the coupling capacitors work harder as well. The non-linearities of the coupling capacitors and the non-linearities of the amplifiers are thus proportionately reduced.


Hi ralph,

I only have full class-A amps, so it couldn't be the crossover distortion, but maybe biasing capacitors would be a good idea for reducing capacitor non lineairities. I have never tried it though

frank

[Robh3606]

"by the way, are there voltagedrive curves for the 2380-series compensation in the M552? In my experience they do not provide enough compensation to flatten the response curve."

Hello Frank

The M552/553 does indeed have compesation curves for both the 2360 and 2380 horns. This compensation does not work with the 2344. There is too much rool off above 10K. You are much better off using the compensation from a 4435 with your active set-up. Works great!

Rob:)

[frank23]

"by the way, are there voltagedrive curves for the 2380-series compensation in the M552? In my experience they do not provide enough compensation to flatten the response curve."

Hello Frank

The M552/553 does indeed have compesation curves for both the 2360 and 2380 horns. This compensation does not work with the 2344. There is too much rool off above 10K. You are much better off using the compensation from a 4435 with your active set-up. Works great!

Rob:)

Hi Rob, I was now trying a 2440/2382A combo, so the 2380 curve from the M552 should work, but I still didn't like it.

Passive crossover and passive compensation seems te best way for both 2344 and 238x horns

greetings, Frank

[Earl K]

Ralph


By using a an attenuator with a very low value resistor across the diaphram and a high attenuation factor causes much more power to flow from the amplifier as it also has to drive the attenuator, and also causes higher currents to flow as back EMF.

- Well okay, the good news is that you are one of a select few who is actually thinking about this conjugate resistors' sonic effects ( produced through loading the amplifier ? ) . Unfortunately, I think your logic ( and therefore conclusion ) is flipped 180° . The extra current/energy going to that load resistor goes into a nice, stable non-reactive circuit element. That means this non-reactive element ( with the load resistor acting as a voltage divider ) has in effect " proportionally lowered the ratio of EMF products that the amplifier might have to deal with .





This then means that the coupling capacitors work harder as well. The non-linearities of the coupling capacitors and the non-linearities of the amplifiers are thus proportionately reduced.

- What's the basis for the assertion in the last sentence ? It does seem counter-intuitive to me. ie, my perspective ; The harder some reactive component is worked ; the more it, as an reactive component will get out of control, possibly producing non-linearities .

- FWIW ; partially filling a cap with a DC voltage doesn't work it harder, as would increasing the AC voltage. A partial reduction in it's AC working voltage capacity is accomplished by partially filling this vessel ( the cap ) with stable, non-reactive DC . Think anchoring effect here .

:)

[Ian Mackenzie]

...if applied to the personal transportation function, we would all be riding unicycles.



In this setting “minimalist design” dogma does not ‘hunt’ very well.



Regards,



WHG

In theory I agree and JBL's filters tend to be 3rd order however I recall in another modification thread a 1st order version of the 4343 network where the member claims vast improvements in audio performance.

While the home diy tweaker likes to hear improvements after a tweak perhaps the reality is that at" normal home listening levels" with a low powered amp such sensitive horn loaded drivers are hardly raising a sweat.

Assuming 13db network shelving attenuation about 1200 hz of the compression driver with a peak program input of say 10 watts to the loudspeaker system what diaphragm displacement do you think equates to the thredhold of audible distortion in this situation?

Ian

[whgeiger]

so it is a bit dangerous to go with 1st order crossovers, but I like the sound and as I am only using this setup in my living room at non-pro-studio-band-impressing-levels, I think all will be well

I know excursion is an issue with bas-reflex loading where the cabinet provides less and less damping where you provide the driver with a frequency below the tuning frequency of cabinet / port

does this work the same then for compression drivers?

frank

Frank,

It is always challenging to be brief when addressing matters acoustic (as well as other physical phenomena), as all generalizations carry the needed baggage of an assumption set. And the briefer the prose, the larger the ‘load’ to be carried becomes. But, it is the only way to get a vivid view of “the big picture” of the matter.


For bass reflex system operated below its principal resonance, the air in the port oscillates back and forth out of phase with the driver cone and the masses of both may be considered as moving in unison with each other. Under these conditions, cone oscillations may become quite violent. Other resonances occur at much higher frequencies where box and port dimensions become comparable to or larger than signal wavelength. Most of these may be successfully addressed by manipulating enclosure geometry and proportion, and through the liberal application of acoustic resistance materials. Back-to-back fared port geometry may be used to reduce the shedding of vortices and suppression of attendant noises.



For the horn and compression driver arrangement, principal system resonance lies typically above [1] the horn cut-off frequency, [2] the first longitudinal standing wave mode, and [3] the region of high mouth reflectance. So, while the consequence of operating below the resonant frequency of the former has severe consequences, in the latter the consequences are less severe until [1], [2] or [3] is approached and the acoustic load is essentially lost or at least materially compromised. This assumes that the displacement limitations refered to in an earlier post have not yet become an issue due to use of a larger front cavity volume [Vf]. Note for the horn system there is a second resonance that is determined by this volume. In some cases, [Vf] is made small to extend high frequency response at the expense of displacement capabilities. Other resonances of interest include those radial standing wave modes that arise in the front chamber when acoustic energy emanating form a comparatively large diaphragm is compressed and ‘extruded’ out of a small driver throat. Typically the first few of these are nulled by judicial placement of annular slits of prescribed area in the phase plug. To cancel the first [n] modes requires [n] annular slits. Thus, the number of these is a very good indicator of the high frequency extension of the driver.

Regards,

WHG

[ralphs99]

Hi Earl,

Thanks for your comments.



As WHG points out, explanations of things acoustic tend to be complicated. I guess I was being lazy and only offering a brief explanation of my meaning. Thinking through it again I realise that some of the logic is off….



There are two distinct effects due to the addition of the resistor across the voice coil that I can see. With the second effect having two further consequences:

1. The voice coil of a compression driver is largely inductive. Together with the coupling capacitor feeding it, it forms a resonant circuit. The circuit would ring forever except for the power lost as heat through the resistive component(s). The Q of the circuit is directly determined by the lossy resistive element. The lower the Q, the better, in this case. A low value resistor across the voice coil aids in achieving a low-Q circuit. This might be particularly so for an amp with a poor damping factor like some tube amps.

2. The attenuator formed with the resistor across the voice coil, works the amplifier harder.

2A. Non-class A amplifiers typically have their poorest distortion at small signal levels. Increasing the signal level required of the amplifier helps it to be more linear.

2B. The discussion around biasing of capacitors concerns one point. That a non-biased capacitor has to change it's polarisation at each zero crossing of the signal driving it. If the biasing voltage is larger than 1/2 the largest peak-to-peak signal swing expected, the capacitor will never have to change it's polarity. In other words, this type of distortion occurs at the zero crossing of the signal to the capacitor. This appears to be due to dielectric absorption causing the capacitor to have a memory of it last polarisation. My thinking was that this distortion mechanism was constant like crossover distortion and thus as the signal level increased, distortion would be comparatively reduced. Wrong! The voltage retained due to the dielectric absorption is always proportional to the applied signal. The distortion due to dielectric absorption is therefore independent of signal level. Scratch 2B….



Cheers,

Ralph.

[B&KMan]

Frank: I'd recommend you buy or build an RTA if you are going to work with this, and a reliable means to measure impedance, as well.

It can't be accomplished by whim or fiat....

Note: My M552 has nasty on/off transients.


Hey Why you do not put a gold plated relay with time delay on circuit ??? for nasty transient ???

:D

[B&KMan]

hi,

It might have something to do with the fact that I only had about 20uF of capacitors that I could put between the amp and the driver for DC protection.




well I read the developpement of this interesting question but maybe I completely out of field but manys friend of me protect speakers by " fuse "
for really overload transient DC created by default vacuum electronic tube... and this you save to add others more intrusives device in audio path...

this is my 2 ¢ commentary....

:cheers:

[Earl K]

Hi Ralph !

Thanks for your reply . :)


-There are two distinct effects due to the addition of the resistor across the voice coil that I can see. With the second effect having two further consequences:

1. The voice coil of a compression driver is largely inductive. Together with the coupling capacitor feeding it, it forms a resonant circuit. The circuit would ring forever except for the power lost as heat through the resistive component(s). The Q of the circuit is directly determined by the lossy resistive element. The lower the Q, the better, in this case. A low value resistor across the voice coil aids in achieving a low-Q circuit. This might be particularly so for an amp with a poor damping factor like some tube amps.

- This makes sense to me .


2. The attenuator formed with the resistor across the voice coil, works the amplifier harder.

2A. Non-class A amplifiers typically have their poorest distortion at small signal levels. Increasing the signal level required of the amplifier helps it to be more linear.

- I'll also buy into that concept when using pure class B amplifiers . Though with some other models of class AB amplifiers ( implementing a very heavy class A bias), I would prefer to keep those amps running within the class A zone ( if that is doable for the desired listening level ) .

- So I'd add; this resistive loss approach, ( of forcing an amp into a more linear operating range ) , should be targeted to specific amplifier models ( not just broadly applied to a class type ) .

- ie ; let the ears be the judge if this approach is a sonic improvement .

regards <>

[B&KMan]

The voltage retained due to the dielectric absorption is always proportional to the applied signal. The distortion due to dielectric absorption is therefore independent of signal level. Scratch 2B….


Hi ,

your point is true but if you put a film and foil ex: you decrease this factor by over 1000. In fact the good the film and foil is less 1000 dielectric absorption than standart polyester... and if you build in Dc charge probably you reduce at zero the dielectric absorption...

for horn it is possible to build filter with film and foil at not too much money.


:cheers:

[Ian Mackenzie]

Nice to see some members thinking along these lines.

I myself have been an advocate and active user of pure SE class A since 2000 and exotic film foils as of 2004. They go hand in in hand and yes, it does make a real difference, so much so that a reversal to prior art is unacceptable.

Ian

[frank23]

Nice to see some members thinking along these lines.

I myself have been an advocate and active user of pure SE class A since 2000 and exotic film foils as of 2004. They go hand in in hand and yes, it does make a real difference, so much so that a reversal to prior art is unacceptable.

Ian

Can you direct us to a site where we can read about these "film foil" capacitors?

I use Sangamo [rated at 2000Volts max !] and Cornell Dubelier paper-oil capacitors from the 70's. Are these "film foil" capacitors a replacement for these high voltage devices?

frank

[Ian Mackenzie]

Well it would appear not all film foils are created equal and as a result there is a boutique niche industry making audio film foils.

Try the search function Frank, there are plenty of juicy threads on this stuff.

My first encounter was cheap metalised polypropolyne, better than un bypassed mylars but lacking detail as raspy.

After a long period of soul searching, I went with Hovlands, they change my attitude about certain horns and what JBL's are capable of...I woke up.

More recently I tried Auricaps, a metalised but highly evolved film foil, a very accurate capacitor.

But for value for money the bypassed & charge coupled Solens have won favour with JBL's.

However unless you are running class A amps, much of the transparency is missed.

Ian

[whgeiger]

In theory I agree and JBL's filters tend to be 3rd order however I recall in another modification thread a 1st order version of the 4343 network where the member claims vast improvements in audio performance.

While the home diy tweaker likes to hear improvements after a tweak perhaps the reality is that at" normal home listening levels" with a low powered amp such sensitive horn loaded drivers are hardly raising a sweat.

Assuming 13db network shelving attenuation about 1200 hz of the compression driver with a peak program input of say 10 watts to the loudspeaker system what diaphragm displacement do you think equates to the thredhold of audible distortion in this situation?

Ian

I-Man,

For a 10-Watt input that is attenuated –13 dB, the net power seen by the driver will be in the neighborhood of ½ Watt. If we assume 50% efficiency then ¼ Watt acoustic output becomes the expectation. For single frequency test tones at this acoustic output the following distortion products have been measured and presented in [1] and reiterated here for convenience:



MFG/Model: TAD TD-2001

[Fc] = 350 Hz

[Sd] = 3.1 in^2

[St] = 0.39 in^2

[Xmax] = 0.019 in

2HD @ 2 kHz = 2.2 %

2HD @ 5 kHz = 4.5 %

2HD @ 8 kHz = 7.1 %



MFG/Model: Altec 288

[Fc] = 225 Hz

[Sd] = 6.3 in^2

[St] = 0.63 in^2

[Xmax] = 0.025 in

2HD @ 2 kHz = 2.2 %

2HD @ 5 kHz = 4.5 %

2HD @ 8 kHz = 5.0 %



MFG/Model: JBL 2440/2441

[Fc] = 225 Hz

[Sd] = 12.5 in^2

[St] = 1.25 in^2

[Xmax] = 0.025 in

2HD @ 2 kHz = 2.0 %

2HD @ 5 kHz = 4.5 %

2HD @ 8 kHz = 6.3 %



MFG/Model: TAD TD-4001

[Fc] = 225 Hz

[Sd] = 12.5 in^2

[St] = 1.25 in^2

[Xmax] = 0.025 in

2HD @ 2 kHz = 1.8 %

2HD @ 5 kHz = 4.5 %

2HD @ 8 kHz = 7.1 %



Note: [Fc] & [St] are for phase plug slits, not the attached exponential horn.


Based on these modest circumstances, it is reasonable to expect the generation of elevated distortion products when a signal of mixed frequency content is reproduced, particularly when diaphragm excursion continues to rise with a lowering frequency that is only moderately suppressed by a wimpy high pass filter. At these levels, such distortion becomes audible to almost any listener. In the reproduction chain, transducers particularly when abused, become the principal sources of distortion generation. Thus, such abuse should be avoided by use of appropriate signal filtering, conditioning and distribution to the loudspeaker drivers used.


Regards,

WHG

Reference:
[1] Title: Handbook for Sound Engineers
Editor: Glen Ballou
Publication: Book, Second Edition, Butterworth-Heinemann, Newton, MA, (1998)
ISBN 0-7506-80331-0

[frank23]

hello WHG,

Thanks for your thorough explanations. It seems to me like the data you posted is related to frequencies the compression driver is fit to produce. Do I understand your data correctly that these are 2nd harmonic distortion levels for a single signal at 2, 5 and 8kHz?

Can it be calculated at how much power at say 100Hz, 250Hz, 500Hz and 1000Hz the maximum excursion of the membrane is reached? I have software that calculates this for bass drivers, but the needed values are not available for compression drivers as far as I am aware of.

Frequencies below 1000Hz is where the problems arise with the "wimpy" high pass filters for the most common compression drivers.

greetings, Frank

[whgeiger]

hello WHG,

Thanks for your thorough explanations. It seems to me like the data you posted is related to frequencies the compression driver is fit to produce. Do I understand your data correctly that these are 2nd harmonic distortion levels for a single signal at 2, 5 and 8kHz?

Can it be calculated at how much power at say 100Hz, 250Hz, 500Hz and 1000Hz the maximum excursion of the membrane is reached? I have software that calculates this for bass drivers, but the needed values are not available for compression drivers as far as I am aware of.

Frequencies below 1000Hz is where the problems arise with the "wimpy" high pass filters for the most common compression drivers.

greetings, Frank

Frank,

Answers follow:

1) All test tones are sine waves of single frequency.

2) The formula you are looking for follows:
[W] = 4*([pi]^2)*([f]^2)*(X^2)*[p0]*[c]*([Sd]^2)/[St]

where

[W] – Acoustic Power Output {acoustic Watts}

[f] - Frequency of interest (above cutoff) {Hz}

[X] - Diaphragm excursion {m}

[p0]*[c] (Specific Acoustic Impedance of Air)
= 407 mks Rayl (after Lord Rayleigh) {N*s/(m^3)}

[Sd] – Projected Diaphragm Area {m^2}

[St] – Area of Phase Plug Slit Apertures (Adjacent to Diaphragm) {m^2}

Will leave the algebraic manipulation to you for finding the value of [X].

Like the JBL 2485J for bridging the speech range to a TAD 2000 series driver for the top end.

http://www.jblpro.com/pages/pub/components/2485.pdf (http://www.jblpro.com/pages/pub/components/2485.pdf)

Regards,

WHG