Just to clarify, the "Assuming that the speed of sound is 344 metres per second:" statement should have been positioned just above the Lpt <= 1000 / fr formula.
If the speed of sound is 344 metres per second then a conventional transmission line loudspeaker would require the line to be at least 215 cm long, or 1/4 wavelength (WL) at 40 hertz.
The rule of thumb formula can be generalised to:
L = A / fs
Where:
L = duct length in cm
A = ROT factor
fs = system tuning frequency in hertz
If the speed of sound is 344 metres per second then:
If A <= 1000 (L <= 0.03 * WL) the loudspeaker will be a bass reflex (BR) system
If A >= 8600 (L >= 0.25 * WL) the loudspeaker will be a transmission line (TL) system
If A > 1000 and A < 8600 the loudspeaker will (perhaps) be a birtle (BRTL) system... 🙂
I have never tried, but I can't see why it should not be possible to come up with a reasonable model.
Someone is bound to have attempted it - Google is your friend...
I was thinking 7ft was the magic number due to Pro Audio FLH's are usually around that length.
Member
Joined 2009
Paid Member
I would be interested to learn more about the mathematical model behind HornResp, and even the history behind how it was first implemented. I read on the internet somewhere "The Hornresp program, written by David McBean and based on Olson's horn model, is a very easy to use horn simulation program. David wrote the original version in the early 1970's in Fortran IV and ran it on a room-sized IBM mainframe computer." which quite fascinates me. But I am not sure which paper by Olson I should read and I've no idea how a solution would be implemented in Fortran (I last used Fortran IV as a teenager and have a interest in old IBM mainframe technology). I understand some or much of this maybe deemed proprietary, but if it's not a secret I'd appreciate some pointers on where to find this information (I've searched this thread for 'formula' but that didn't give me much satisfaction).
For those who don't know where Country Music really came from.
https://en.m.wikipedia.org/wiki/Rock_and_roll
Go Bearcats & Buckeyes!
https://en.m.wikipedia.org/wiki/Rock_and_roll
Go Bearcats & Buckeyes!
The original version of Hornresp was developed using models and formulas provided in the book "Acoustical Engineering" by Harry F Olson.
A soft copy of the Olson book can be downloaded as a .pdf file from:
https://pearl-hifi.com/06_Lit_Archi...ec_51/4445_Acoustical_Engineering_3rd_Edn.pdf
See Post #8,764 linked below:
https://www.diyaudio.com/community/threads/hornresp.119854/page-439#post-5580591
The power response chart was generated using an extremely noisy 132-column IBM 1403 line printer, similar to the one shown in the attachment.
Attachments
Member
Joined 2009
Paid Member
I only recently learned that Hornresp had such a long history. It must be a labour-of-love for you. I hope you figure out how to ensure it becomes a legacy that outlives all of us.
I have been looking at the IBM1401 mainframe quite a bit, I would consider building my own computer with some of the essence of that old mainframe. I would want to start with SS memory and implement the variable length I-chain instruction set processing but with microcode (the combinatorial logic would be immense otherwise). That's a new hobby project for when I have more time to spend. But it is quite fascinating to see where Hornresp started life. That 1403 printer is legendary, using a high speed chain it was the fastest line printer in the world. I'd love to get my hands on an old line printer with that folded tractor feed stuff.
I've downloaded a copy of the Olsen book.
So, the basic program would 'fit' into a Fortran IV program on a stack of punched cards - the stack of cards in the image you posted looks about 3" high, at 0.007" per card that's around only 430 punched cards!?? If the source code from 1970 still survives and in fact was something you could make available ?? I should be able to figure out how run it under simulation.
Can you say any more about the computational methods you have used, as the book shows formula but that's really just a starting point ?
I have been looking at the IBM1401 mainframe quite a bit, I would consider building my own computer with some of the essence of that old mainframe. I would want to start with SS memory and implement the variable length I-chain instruction set processing but with microcode (the combinatorial logic would be immense otherwise). That's a new hobby project for when I have more time to spend. But it is quite fascinating to see where Hornresp started life. That 1403 printer is legendary, using a high speed chain it was the fastest line printer in the world. I'd love to get my hands on an old line printer with that folded tractor feed stuff.
I've downloaded a copy of the Olsen book.
So, the basic program would 'fit' into a Fortran IV program on a stack of punched cards - the stack of cards in the image you posted looks about 3" high, at 0.007" per card that's around only 430 punched cards!?? If the source code from 1970 still survives and in fact was something you could make available ?? I should be able to figure out how run it under simulation.
Can you say any more about the computational methods you have used, as the book shows formula but that's really just a starting point ?
Last edited:
This is F N AWESOME DMB!
I remember having to program a racing car game on the Commodore Vic 20 in the mid 80's. I think I repeated the program on the Commodore 64. I always got my father's hand me downs.
You're pretty close - it's actually just 397 cards 🙂.
By way of comparison, the last time I checked, the VB6 version had more than 100,000 lines of code…
It does survive, but I would rather not release it.
It a matter of setting up algorithms to calculate the results you want. For example, the statement printed at the top of the card shown below represents one line of source code from the subroutine responsible for calculating the throat acoustical impedance of a finite conical horn.
The code punched into the card calculates the first part of the denominator of the impedance formula given in the Olson book, as shown highlighted in red below.
In that case, you might be interested in the Lap Simulator I developed some years ago 🙂.
You probably wont believe this, but I swear that it is true. The fastest that I actually physically lapped the circuit in my 1969 Mini Cooper S in standard road trim was 1 minute 23.8 seconds, achieving a speedometer reading of exactly 100 mph at the end of the longest straight.
The "best guess" inputs that I initially entered into the simulator for the Mini Cooper S are as shown in the attachments. Amazingly, as you can see, the predicted lap time is identical to the actual time that I had achieved more than 40 years prior, and the actual and predicted top speeds are effectively the same also, taking into account that speedometers are normally conservatively calibrated to read slightly high.
Last edited: