Driver impedances for AR3a?

[Guest]

What are the impedances of each of the three drivers in the AR3a? Not that I would ever consider replacing them with anything different ...

Thanks --

--Daniel

[Carlspeak]

>What are the impedances of each of the three drivers in the

>AR3a? Not that I would ever consider replacing them with

>anything different ...

>

>Thanks --

>--Daniel

I worked on a pair a while back with serial no's 47405 and 43070. With my V.O.M. I measured 4.0 ohms on the 3700-1 Woofer, 5.2 ohms on the HR 4500-1 mid and 2.7 ohms on the HR4600-1 tweeter.

Carl,The SpeakerDoctor

[Guest]

Carl -

Were those impedances or DC resistances?

Thanks!

--Daniel

[Carlspeak]

Hi Daniel:

Those were simple DC resistance measurements.

Carl

[Guest]

Anyone happen to have impedance specs for these drivers? I'm experimenting with replacements (just for fun -- I'm *not* destroying the cabinets) and I'd like to know whether candidates are good matches for the existing crossover, or if I'd need to change the effective impedance of the replacement (e.g. by placing appropriate resistor in series or parallel with the driver) to maintain the crossover parameters.

As mentioned in another post of mine, I'd be very interested in seeing an analysis of the AR3a crossover which explains how the crossover points are obtained given the component values in the network; for example, I understand that the woofer low-pass point is either 525Hz (according to the AR3a brochure) or 575Hz (according to the AR frequency response and specifications doc posted here). However, when I tinker with 2nd-order crossover formulae using the 150MFD cap and 2.85mH coil values, I get numbers like an impedance around 3 Ohms and a crossover point of 250Hz (Butterworth).

Can anyone shed some light on this?

Thanks!

--Daniel

[kkantor]

If you try to design crossovers on the premise that speaker drivers are ANYTHING like resistors, you are doomed. The whole concept of "4 ohm" or "8 ohm" speakers is anachronistic. You know those boxes they sell at Radio Shack that say things like "1000 Hz, 8 ohm crossover?" Actual measurements turn out to be 200 Hz, or 3000 Hz, or a strange bump that isn't even a true rolloff. Or whatever.

Speaker drivers are complex loads,("complex" in the mathematical sense). One must do the kind of simulations you are attempting using comprehensive electrical models, or specialized crossover software that takes into account the exact woofer or tweeter you are working with.

If you want to mess around with this kind of thing, I would recommend going to www.circuitmaker.com and downloading the free student version. Try several of their circuit examples, and spend a few days learning your way around the program. Once you know how to simulate basic filters and response curves, I can give you some advice about how to model actual speaker designs.

-ken

[Guest]

Ken --

Thanks for the help, and for the lead on the circuit software. I'll check it out. Do you happen to have the needed electrical characteristics you mention for the AR3a drivers so that I can perform the analysis accurately as you suggest?

Also, as you seem to know a bit about this (!), I'll ask you: would the level pots on the '3a shift the crossover points around as they are not L-pads and would seem to present a rather wide variation in impedance across their ranges?

Thanks again --

--Daniel

[kkantor]

Thanks! Easy question first: yes, turning the pots does alter the crossover point. However, the whole idea of there being a crossover "point" is quite an abstraction. The notion grew out of simplified spec sheets, and customers used to idealized response curves (smoothed, made in test chambers, made up totally, etc.) In the real world, there is a crossover "region." Every location in space has a somewhat different crossover "point." Hmmm... I guess this really wasn't the easy question, after all.

Onward. I don't think I can be of much help about the 3a woofer model, at least immediately. It is highly possible that it exists in my collection, or I can get it from a friend. Unfortunately, stuff that old is buried in boxes and boxes of dead storage, and isn't practically accessible until it turns during a random search. It isn't terribly hard for me to derive a new model, once I dig up an appropriate woofer. But, you can count on that being after the Consumer Electronics Show in January! (Unless I really an even worse case of procrastination than even this.) Additionally, the AR 12" woofer parameters changed a great deal over the course of the product's life. There are an infinite number of combinations of parameters that can yield the same effective bass response. Do you know the age of yours?

Lastly, do you have any electronics tools available for making basic measurements on your woofer? A test generator and decent voltmeter come to mind.

Ken Kantor

www.intelligentaudiosystems.com

[Guest]

Ken -

Thanks again for the info. I'm clearly in way over my head here!

I understand that the driver characteristics are extremely important, if not outright dominant, in the frequency response of the crossover network + driver circuit, and that will probably explain why using a fixed impedance value and assuming otherwise flat frequency response when computing crossover component values will get me nowhere fast.

I did try to get the student version of CircuitMaker as you suggested, but it won't install on my XP Pro machine; sent a note to their tech support about it. I did get the demo version of SuperSpice, but that leaves something to be desired in terms of stability and overall polish (and I can't for the life of me figure out how to view voltage magnitudes in the output graph - I only saw currents...).

Back to my original quest: I am curious about trying out alternative drivers for the tweeter and mid (sometimes the [second?] best way to learn is to screw something up and then figure out why!). Bouyed by my satisfaction with a tweeter replacement, I attempted to swap the mids, but I found that the new mids were (hyper-) active well above 5K when I was expecting a crossover-induced dropoff. I then suspected that the replacements were of higher impedance that the originals, which would tend to raise the (non-existent!) crossover point, so I considered lowering the effective impedance on a test basis by using a shunt non-inductive resistor. Further perusal of the crossover schematic, however, gave me pause, as it seems that the lowpass action of the mid crossover is being provided by a first-order filter using a series coil of 0.044mH, which by formula (I know, I know...) would appear to yield a very high (20K+) crossover point. At that point, I stopped mucking around and composed a post to the classic speaker AR forum...

Anyhoo, I continue to welcome your input and advice ("stop perverting your speakers!!!"). My '3a's are a late pair with rear contacts on the mid/tweet (blew the tweets -- too embarrassing to say how, but did you know that just 'cause you can't hear a certain frequency no matter how much you turn it up doesn't mean your speakers aren't trying to produce it?).

Best --

--Daniel

[kkantor]

Screwing around is a very educational way to learn about speakers, I think. Not everyone would pick a 3a as their test bed... but, hey, it's only a speaker. Trial and error has its limits, though, and I would recommend a few things that might help you gradually turn the experience into knowledge.

1- Keep looking for a freeware modelling program, or keep trying Circuitmaker. I got the "pro" version working on an XP Pro machine without incident.

2- $100 bucks worth of test generator and voltmeter might prove a revelation. Details when/if your are seriously considering. There are many alternatives for doing measurements with a soundcard, too.

3- A copy "The Loudspeaker Cookbook," an elementary explanation of how speakers work that can be digested with or without math.

Other than that, midranges are tricky to design good crossovers for, since the circuits used for the LP and HP sections interact. And because the driver impedance changes so much over the operating band. And because of the range of wavelengths involved. And the sensitivity of the ear. Etc. Not a task I would wish on even an inexperienced pro.

Not sure what your goals are yet, so I can't say much else in specific. Think about using reactive components for your shunt elements, perhaps in series with lower value resistors.

-k

[Guest]

Thanks again for the guidance. Believe it or not, I managed to get SuperSpice running and I modeled the complete AR3a crossover circuit, though as a starting point I used purely resistive loads for the drivers. I swept frequencies from 10Hz to 100KHz and plotted voltage magnitudes across the driver loads as well as phase angles for the three drivers. SuperSpice doesn't seem to want to plot calculated waveforms based on more than two raw values, so I couldn't plot a compound response curve for the whole thing, but I tinkered with adding the woofer and midrange curves (accounting for the phase relationship) to observe the phase-induced dip inthe crossover regiod. Because the schematic has all the components on the '-' side of the drivers, I couldn't just plot the voltage magnitude of a circuit node (the '+' side of the driver always equals the source voltage, and the '-' side represents the inverse of the response in this arrangment), so first I re-drew the circuit with the operative series bits on the '+' side, which was successful, then I went back to the original arrangement and plotted the magnitude of the voltage drop across the loads by using a calculation, which also seems to prove correct.

Very educational!

Best --

--Daniel

[Guest]

Ken -

SuperSpice was a bit, um, unstable, but I'm working with Beige Bag's free Lite version of their B2 Spice A/D and it's much better (and with a good help system). No potentiomenters in the stripped-down component set, though. Still trying to figure out how to add signal plots accounting for phase relationships. It does seem that the positions of the mid and tweet pots have a *very* significant effect on the response curves of the crossovers, at least when the drivers are simply modeled as resistive loads; not sure why they chose this design vs. an L-pad arrangement. Thoughts?

I would appreciate a recommendation on a multimeter that has useful audio-range AC functions (frequency counting? RMS measurements? dB display? External mic jack for using it as an SLM? RTA? [kidding!]). My $7 Radio Shack model's just not cutting it...

Still wanting to know the nominal impedance (or better yet, impedance curve) of the stock AR3a midrange and tweeter!

Thanks --

--Daniel

[kkantor]

You nailed the problem on the head.

Accurately modelling speakers at the listening position is a substantial technical challenge. Even without the room, it only became practical fairly recently. Add in the room and driver radiation patterns and cabinet effects, and it is state-of-the-art. That's why speaker designers still have jobs. Such as they are....

There are software packages that do pretty well, and most big speaker companies have their own creations, but I doubt a circuit simulation could do it. It's really a wave problem, and lumped parameters don't illuminate much. What simulators are for is taking the trial and error out of crossover network design, in terms of achieving a previously defined response. But what you want to feed to the drivers is another matter. This, I will opine (perhaps controversially), is why the very best speakers rarely are done with active crossovers. Knowledge and DSP are changing that, but it is still a valid observation of the past.

Years ago, when I was doing R&D at International Jensen, we wrote a free-space acoustical modeller (no room, but driver and cabinet geometry) that we subsequently released into the public domain. I will try to locate a copy and post it here.

Ken Kantor

www.intelligentaudiosystems.com

PS- Meter? How much do you want to spend? EBay is overflowing with superb meters for not so much money. I purchased two HP3400a voltmeters for about $40 each. Not a multimeter, but not long ago a gold standard in audio testing. Most of the used bench multimeters from HP, Keithly and Fluke are total bargains. What kind of soundcard do you use?

[Guest]

I'm in the computer biz, so I've got a few machines to play with -- what soundcard(s) do you recommend for testing/analysis? As for meters, I'd like to get a hand-held multimeter anyway, so if you had a recommendation on some units that would cover the needed audio functions, I'd be most appreciative. Barring a specific recommendation, what should I look for in such a meter?

I'm still plugging away with circuit simulators -- BeigeBag's my current favorite, though it frustrates me in some ways (where are the darn branch currents? Why no simple device power graphing). CircuitMaker's working, but if you want to reference voltages to anything other than ground (as one might want to in the case of the mid a tweet crossover in the 3a, where the crossover is on the ground side of the circuit), you can't have more than one non-ground reference active.

Still having fun, and still wondering how the woofer crossover's supposed to cross over at ~500Hz!

Thanks --

--Daniel

[kkantor]

Ideally? A "professional" card that accepts balanced inputs and high voltage levels. I have an older Antex StudI/O card and DAL Card-D that are troopers. A Frontier lightpiped to a Sonorus that is really excellent.

Realistically? These days, there are a boatload of affordable pro-sumer cards sold by music stores that are far better than you need, especially since you clearly know enough to build a simple, passive front end that does a bit of signal conditioning.

www.musiciansfriend.com is sure to have something on sale.

As an affordable, speaker-centric measurement suite, I happen to like www.purebits.com. There are others. Honestly, we use Audio Precision's and HP's when it comes to "real" work, because none of the freeware and cheapware is truely bug-free and fast.

In terms of Circuitmaker, you can just plunk in an "ideal" differential amp, and use that for your non-ground signals. Or, you can choose two test points and display A-B, though that might be exclusive to the "pro" version.

Ken Kantor

www.intelligentaudiosystems.com

[kkantor]

Used, I'd look for a Beckman 320. They are really flat over the audio band. New, Fluke 80 Series III.

[Guest]

Karl -

I notice you've got the DC resistance of the AR3a midrange at 5.2 Ohms, but apparently it's supposed to be a nominal 4 Ohm unit with a DC resistance between 2.5 and 3.5 ohms, according to Tom; the physically similar but electrically different AR5-type unit is a nominal 8 Ohm job with a DC resistance of 5.0 to 6.2 ohms per Tom. Are you sure the mid you had wasn't an incorrect replacement?

Thanks --

--Daniel

[HastyCroke]

Hi....I had AR-5 speakers. 8Ω, 10-inch woofer otherwise similar to the AR-3a. Probably more sensitive but I was able to get them real loud with a Dynaco Stereo 120 with 60wpc. The Rotel is rated 35wpc into 4Ω and would be more than capable of driving the speakers for most listening. When one wants it loud, you need more power. Having more power will allow the use of more bass or the loudness if so desired.