Western Electric 32A and Altec 32 variants.

So, went down quite a number of rabbit holes with this project.... I really wanted to be able to create a model from scratch (using existing design features and dimensions) to understand better the fascinations surrounding the 32x horns. Spent quite a bit of time on an Altec 32C. Took careful measurements. Did numerous expansion calculations. Made models after models. Not one was satisfactory. Could not make the throat dimensions and geometries of the Altec 32C to fit Figure 5 in patent 197723 with regard to the throat expansions. (One of the two patents casted on WE 31A and 32A horns).

1613760073037.png

A few days ago, went back and read Chuck's post again.

Chuck made two important observations: 1) The original throat diameter is 0.76", and 2) For the 32C, Altec engineers lofted off the throat to where the cross section was the equivalent of 1" drivers. BAM, my earlier models were all too short since they were based on the 32C dimensions. There was not sufficient distance for the throat expansion from a circle (M) to a pie shape (F).

Went back and made some calculation as to what the "missing" length would be to expand from a 0.76" driver throat to an 1" driver throat. Turned out to be a little over 1". Added this distance back into the model. The extra distance allowed for much smoother transition from a round shape to a pie shape per Figure 5.

1613760140171.png
1613760215553.png
Cross section views of the transitions.

1613760404338.png
1613760437208.png
1613760342571.png
1613760498905.png

Was not easy. But, learned a few things. Hats off to WE engineers!
 

Attachments

  • 1613759503639.png
    1613759503639.png
    74 KB · Views: 7

Thermionics

Post Whore In Training
So is your hypothesis that if the 32C had been designed with a longer 1" section (rather than lopping off the length where Altec simply reamed out the throat from 0.75" to 1"), the 32C might be even better than it is (already excellent)?
 
So is your hypothesis that if the 32C had been designed with a longer 1" section (rather than lopping off the length where Altec simply reamed out the throat from 0.75" to 1"), the 32C might be even better than it is (already excellent)?
Do not know the answer to that question. I think the most forward approach is to create a model from scratch starting with the 1" throat and with the WE 32A parameters to conduct side-by-side shoot out.
 
Just to put it out there. My interest in modeling the 32A is purely academic. I do not have any sonic experience with any of the 32x horns. Most WE/Altec/Lansing horns I know are straight exponential expansions with the cross sections being mostly "rectangular" even while going through the bends. WE 31A and 32A are different. The throat expansion goes from a circle to a pie shape. The pie shape is actually created from revolving a bent exponential profile around an axis.

The really really hard part was how to come up with a profile such that the 90 degree revolved cross sections comply with the exponential expansions. Took quite a bit of time to figure it out even with my fancy spreadsheets and 3D modeling software. How did those WE engineers do it? I am still in awe.

1613769203012.png
 
Maybe you can answer a question for me. How close are the WE horns to a true exponential horn? I have a book and a few other articles that conflict. Some say WE used the exponential horn equation, some say that they actually used a variation of the exponential horn equation, some people say they did something completely different.

I am trying to design a horn from scratch (a mini KS-6368 horn if you will) and I would love to get your take on what sort of expansion equation I should use to get as close to the WE style as possible.
 
Maybe you can answer a question for me. How close are the WE horns to a true exponential horn? I have a book and a few other articles that conflict. Some say WE used the exponential horn equation, some say that they actually used a variation of the exponential horn equation, some people say they did something completely different.

I am trying to design a horn from scratch (a mini KS-6368 horn if you will) and I would love to get your take on what sort of expansion equation I should use to get as close to the WE style as possible.
Hey, I've modeled a number of WE horns. Most of them are true to the exponential expansions including the KS6368. (The KS6368 has a clever feature to accommodate the internal volume of the inner horn body.) The odd balls are KS12024 and KS12027. The pre-bend and post-bend are exponential. But, they are two different expansions slightly different from each other.

The other odd horn is WE 24A. The other walls are exponential. But, the inner cells are not. WE 32A is also exponential. But, the "bow tie" at the mouth is lofted off to create a rectangular shape.

Have fun with your mini-6368.
 
Last edited:
Hey, I've modeled a number of WE horns. Most of them are true to the exponential expansions including the KS6368. (The KS6368 has a clever feature to accommodate the internal volume of the inner horn body.) The odd balls are KS12024 and KS12027. The pre-bend and post-bend are exponential. But, they are two different expansions slightly different from each other.

The other odd horn is WE 24A. The other walls are exponential. But, the inner cells are not. WE 32A is also exponential. But, the "bow tie" at the mouth is lofted off to create a rectangular shape.

Have fun with your mini-6368.

Awesome. If I can use the basic S(x) = S(o)*e^mx equation, that makes life a lot easier.

With regards to the KS-6368 "trick", I kind of figured they did something to compensate for the throat. The only question I have is did Racon make the mouth larger than it had to be?

My goal is to make a 300hz version of the 6368 and I am planning on shrinking the KS6368 dimensions down to achieve this. With some simple math I am coming out to a horn mouth that is just a hair over 12x8. I am not planning on running the throat through the mouth, so if racon in any way oversized the mouth, I feel like I could build right at 12x8 and be happy.
 
Awesome. If I can use the basic S(x) = S(o)*e^mx equation, that makes life a lot easier.

With regards to the KS-6368 "trick", I kind of figured they did something to compensate for the throat. The only question I have is did Racon make the mouth larger than it had to be?

My goal is to make a 300hz version of the 6368 and I am planning on shrinking the KS6368 dimensions down to achieve this. With some simple math I am coming out to a horn mouth that is just a hair over 12x8. I am not planning on running the throat through the mouth, so if racon in any way oversized the mouth, I feel like I could build right at 12x8 and be happy.
My calculations using the existing KS6368 dimensions put the cut-off frequency at around 225Hz. But, typically the practical crossover point is about one octave higher than the horn's cut-off frequency, so, 500Hz is a good usable crossover for this horn. If you want to design a 300Hz horn and use it at 300Hz, then, the design parameter is 150Hz. If this is the case, then, your 6368 would have to grow larger than the actual 6368.

I don't quite understand the one octave rule. I believe it's related to the wave refractions back into the horn from the mouth. JMC's horns were designed to minimize/eliminate such horn mouth refractions.

Racon did not oversize the mouth. It's dimensionally correct.
 
My calculations using the existing KS6368 dimensions put the cut-off frequency at around 225Hz. But, typically the practical crossover point is about one octave higher than the horn's cut-off frequency, so, 500Hz is a good usable crossover for this horn. If you want to design a 300Hz horn and use it at 300Hz, then, the design parameter is 150Hz. If this is the case, then, your 6368 would have to grow larger than the actual 6368.

I don't quite understand the one octave rule. I believe it's related to the wave refractions back into the horn from the mouth. JMC's horns were designed to minimize/eliminate such horn mouth refractions.

Racon did not oversize the mouth. It's dimensionally correct.

I thought that part of the allure to the KS6368 (and the 32a for that matter) is that WE/Racon supposedly got their horn to play down pretty darn low.


On page 46, you can see that there is a frequency response measurement showing that the horn with the 555 driver does pretty darn well at 200hz. If memory serves, the 32a in the WE753 speaker is also crossed over at a lower point than the math would suggest.

That being said, a lot of these frequency response graphs, articles, and so on are all pretty darn old. If you have the opportunity, I would LOVE for you to measure your speakers and get some modern data. That would just be so cool.
 
Had to do some thinking on this one. Trying to update the 32x horn to an 1" throat turned out to be not a simple thing. The horn itself is fairly short - about 11". A true 1" expansion would increase the "thickness" of the horn to where it would not be recognizable. I now understand why Altec engineers simply bored out the WE 32A throat to 1" in the case of the Altec 32A or cut off the horn to where the cross section is 1" round equivalent in the 32C. Anyway, decided to maintain the dimensions and the proportions of the horn. Instead of an exponential expansion from the 1" throat, linear expansion up until where the horn revolves starting with a pie shape. Let's call this the 32D horn.

Side by side with "Humpty Dumpty".
IMG_5774.jpg

Front view.
IMG_5775.jpg

IMG_5777.jpgIMG_5776.jpg
 
Last edited:
So, I had to make a few guestimates in modeling the 32 horn from Humpty Dumpty. Used the model to attempt to create an 1" throat version. Okay, but not entirely satisfactory. So, reached out to @hifitown and asked to borrow one of his WE 32A's. Early was very kind in offering me his best sounding unit!

Humpty Dumpty on the left.
IMG_6142.jpg

The features on the WE 32A were much more defined. Was able to take much better and more accurate measurements. Also, was able to determine with better precision the effective length of the horn and the distance from the throat where the revolving features start. These two critical dimensions were "guesses" in previous models.

Dimensionally and expansionary correct model of the WE 32A. Working on the 1" version next....
1625848016598.png
1625848069586.png
 
Last edited:
I struggled for days trying to model the 1" throat version of the 32 using the best guess dimensions off Humpty Dumpty because the dimensions did not quite go together. The resulting 1" version was not wholly satisfactory.

Using the dimensions off the WE 32, only took a little over an hour.... The model has the exact same effective horn length and mouth dimensions as the WE 32A. Also, the location of the bend transition is exactly the same.
1625858288358.png

The WE 32A wall thickness is about 0.20". A bit thin for my taste (IMHO, the main contributor to the characteristic ringing of the 32A horns). The new 1" 32 horn wall thickness is 0.25". The flanges are 0.30". The throat and the bend are much thicker to accommodate the 1" driver and for proper expansion. I am calling this one the 32D. :)
1625858517630.png
 
Last edited:
The 32 horn seems to have some special quality to it - the bend and transition is supposed to have less beaming than a straight horn would have. What was the usual lower crossover frequency for these? I will have some free time during a business trip, so I might try as an exercise to model my own version. There is lots of useful information in this thread on how to actually do it. Thanks for sharing.
 
Because of the very short horn length, the mathematical cutoff frequency is around 440Hz. Published WE data is 500Hz (for the original intended drivers). I like to observe the one octave rule, no lower than 800Hz.

The bend in the original WE 31A horn is in accordance to patent 197723. The Altec variants either reamed out the throat opening or chopped off the horn to where the cross section is about 1 square inch.
 
I am using a 700 Hz second-order crossover on my WE 32A which has worked out fine. This is what DejaVu Audio uses for the 32A and a wide-band driver like a 713, 555 or my YL 5500.

When I was using an Altec 32B (the plastic horn cut short for a 1” driver), I had an 802-8G driver and that combination sounded better to my ears with a higher crossover. I settled on 1200 Hz after trying several other frequencies as low as 850. Each of these was second-order too.

When I used a metal Altec 32A, I simply continued with the same crossover I used with the 32B but it’s possible that horn might have been able to use a lower frequency due to the longer length.

I have now used some version of the 32 for the past 13 years, and I am pretty sure I will continue to do so until my last breath. I just love the sound of a good compression driver on a 32! It has just the right combination of sonic qualities—smooth tonal balance, excellent detail and great dynamics—plus it blends beautifully with a fast vintage woofer with a coherency rivaling a single speaker. And the frosting on the cake is that it doesn’t have any horn “shout” coloration, something that I had assumed was inherent with horns until I heard a 32.
 
Top