Victorian Boombox Mk III. A brand new start.

[Miranda.T]

Really interesting! Analogue 3D audio . You really don't shy away from a challenge, do you? Still, if anyone can do it, I'm sure you can.

All the best,
Miranda.

[J. Wilhelm]

Really interesting! Analogue 3D audio . You really don't shy away from a challenge, do you? Still, if anyone can do it, I'm sure you can.

All the best,
Miranda.

Thank you, dear Miranda!

But there's nothing to shy away from! Thinking is free! At least a lot cheaper than buying crapola amplifiers online and buying tons of insecticide or building beds (see Gaah! Thread). I like weird projects like these. When I was in graduate school I dreamed up a small neural network which could calculate derivatives of sinusoidal functions ("matrix multiplies" for Fourier transform based CFD solvers - Fourier Chebishev spectral DNS, to be more precise. My dream back then was to make a neural network based "blackboard" which could solve fluid flows in much-faster-than-real-time, by directly simulating virtual particles flowing within the networks. My supervisor wouldn't let me use it for research. It didn't matter, as I couldn't pass the qualifying exams. But this will tell you how crazy I am.

Now that the bugs in my rental abode took all my money, I have no choice but to retreat back to my mind and work with cheaper hardware (Op Amps), while I recover some of that money I spent. I just recovered about $100 out of $200 I spent thwarting those pests my landlord gifted me. Piece of advice: never accept free furniture from your landlord. Better yet, never live in the same house with your landlord.

Anyhow, perhaps with a truckload of maths experience behind me, I can crack some of the mysteries that thwarted me the first time I tried my hand at it when I was a junior student and my mind was so easily distracted. Though to be honest, I feel like the answer will be something deadpan simple. I'm not trying to simulate an exact sound source location. More like having fun the way those 1960s guys were.

[J. Wilhelm]

So I have laid out a strategy for creating a surround spacializer or virtual surround sound generator. I've briefly reviewed the maths on Fourier transforms and Laplace transforms, as well as how these  two relate to linear electric circuits. I forgot how much I learned in my undergraduate years in college!

The good news is that after forgetting all of this material (it was basically elective coursework for me in college), I finally remembered that it's not just possible, but rather straightforward to make a machine that will generate the HRTF of the human ear for a *monaural* sound source. The trick is to write KIrchoff's Voltage Law (KVL, basically enerrgy conservation for electric circuits) in terms of first principles (ie basic physics laws of each electric component ) to develop an equation to which you will apply either the Fourier or Lapace transform (read below).

To get to the frequency response of a physical circuit you first have to express
it's energy conservation law as an integro-differential equation

As an intro to the maths, I can just summarize in this way: basically what you're trying to do is to express the electrical energy conservation law (expressed as KVL), in terms of sine waves and exponential functions using complex variables (Euler's equation and related theory). The reason is that there is something called Fourier's Transform Theorem that basically says that any time domain signal, like music for example, can be substituted by an infinite number of sine waves of different amplitudes and frequencies, similar to the way an mp3 player or a 1980's graphic equalizer can break the signal into different bands of frequencies that you can control individually to your liking.

The advantage of expressing a signal in terms of sinusoids of different frequencies is that sine waves have special maths characteristics that allow you to simplify the calculations a great deal when you need to apply calculus or process differential equations. The first principle laws that govern the behavior of capacitors and inductors in a circuit turn from differentials and intergrals in the time domain, into multiplications and divisions in the frequency domain, and your KVL equation is converted into a single and a fairly simple algebraic equation.

So now you can perform the Laplace Transform (shown) or the Fourier Transform to express in the complex frequency domain



The equation turns into an algebraic equation in terms of the complex frequency "s"

So now you can start evaluating an algebraic equation that explicitly tells you how frequencies are handled by the individual components or the collective assembly of electric components in your circuit. In other words, you can "translate" between electric components like capacitors and inductors directly into a plot that gives you Transfer Function, or frequency response of the electric circuit. You can get the phase as function of frequency too (The two plots together are called a Bode Plot). And that is exactly what we want. A Transfer Function that matches or approximates the HRTF that I was talking about before, and a way to look at the phase of signals as well.

And solve for current (shown) or voltage behavior as a fuction of the complex frequency.
If "s" is set to have only an imaginary component (j) that is the same as a Fourier Transform and
all the functions in the plots are expressed in terms of frequency, "f"

The Laplace transform, is basically very similar to a Fourier transform, except it deals with functions of frequency expressed in the entire complex number domain as opposed to the Fourier transform that deals functions of frequency and imaginary numbers that represent rotating vectors called phasors. In fact, if you set the complex numbers in the Laplace Transform to have their real component equal to zero, then the Laplace Transform is exactly a Fourier Transform. The HRTF plot I compiled from the German researchers above is a Fourier Transform, but it's trivial to convert to a Laplace Transform.

In fact during the design process, you're not going to have to calculate too much other that put together a collection of simple RLC circuits that will produce quadratic equations that look similar to the HRTF curves, then you just work your way back to what the values of the RLC components are! Not very different than building a 1980s graphic equalizer - in fact that is exactly what it is, except that you need one graphic equalizer for every single "virtual sound source" you want to "hang" somewhere in 3D space.

The difficulty is that using analytical equations, you can only do this for point sources, because it's impossible (as far as I can tell) to find a maths transform that will map a time domain signal (eg music) to an infinite number of space locations (inside your room) with varying frequency response. In fact, I think there is an infinite number of ways you can arrange speakers in a room to give you a given time domain signal or equivalently a frequency response distribution in your room. This difficulty is referred to as an inverse design problem. You can try an infinite number of ways to give you the same result, and consequently there is no analytical method to give you a unique solution for a given asked question.

More specifically, you can't map the time domain signal to a signal expressed as function of azimuth or elevation around your head; that'd be like having an infinite number of  1980s style frequency equalizers for every possible angle around your head. And to make it worse, for a stereo signal, azimuth is rather arbitrary, because musicians are not necessarily encoding the sound location of say, an instrument or a singer, in their stereo signal. The latter, is in fact what quadraphonic systems did in the 1970s; these were methods to try to encode a sound location into the music and then decode it again. The music was manipulated in terms of phase and strength to make sure that you could extract a hidden signal using the "R-L" difference between left and right channels. That meant a finite number of channels.

So what to do? Well, you have to generate discrete virtual sound sources in space, each one should differ from the other one, even if by only an arbitrary parameter. Like having a finite number of speakers floating in your room. How many do you want? 4, 6, 8 virtual speakers? Because that is what you will have to do. Trying to do something more complicated would require that you encode the exact location in a way that a digital computer could separate that specific sound instrument from a sound track and manipulate its sound accordingly, for each and every single instrument. Only NASA's Convolvotron can do that.

But this makes my task much easier now. I can take that HRTF and use Excel to come up with a quadratic equation or set of quadratics to design a frequency profile for each virtual speaker around your head. I I have 2 real and 4 virtual channels, I need to build 4 frequency equalizers, basically. That is very straight forward. What is more nebulous, is how you will separate the 2 channel music into 6 different channels.

One approach is to leave front right and front left channels untouched (I'd prefer that) as opposed to using a center channel, and maybe add a virtual center channel (which I feel might not be necessary), then try to have "exaggerated" or weighted R-L signals. So say for example, immediately 45 degrees clockwise from the Front Right (FR) channel, you develop a "2R-L" signal that exaggerates the right side of the stereo music by a factor of two. Then another 45 degrees clockwise and you have a virtual channel that puts out a "3R-L" signal that exaggerates the right side even more, relative to the left side. But these are not real channels, bit only synthesized virtual channels. The frequency content of your music will be affected, inevitably - this is not for audiophiles, but for people who want to hear a special effect. This by definition is a stereo expander, and I suspect that was exactly what Sony did in the 3- and 4 speaker "Matrix Stereo" portables. The difference is that the channels are not going to real speakers, but rather they're being filtered through the HRTF circuit ("equalizers") and being pumped through the front stereo speakers.

The main hidden secret I see in this method is the manipulation of phase and time of arrival. In theory, the HRTF already has the right phase changes "encoded" as part of the frequency response (Bode Plot of HRTF), but I need to make sure that the phase distribution of the signals actually makes any sense when I try to reproduce the HRTF. There are very rough approximations to phase and time of arrival differences bewteen right and left ears that were measured in laboratories throughout the 20th C. I need to make sure that the circuits I build are actually obeying those basic rules, and not just the HRTF data, and if missing phase and time delay cues, I need to synthesize phase and time delays wherever needed. That will probably be the most difficult part, but I won't find out until I do the Bode plots.

A proposed 8- discrete channel Virtual Speaker System for Project PSSSIII
L is the Left stereo signal. R is the Right stereo signal.
H_n is an operator that filters the signal with the Head Related Transfer function for:
the real front speakers if n=0, and the n^th location of the virtual speaker if n>0
d is an additional time delay as a function of center front location to ensure a precedence effect for the virtual speakers

But does using the "R-L" signal even make sense now? We're not decoding anything at all! It's original purpose was to extract differences between right and left channels and use that to amplify the sound separation of sound sources, taking advantage of previously manipulated sound in phase and volume to extract create a new channel. In the event L=R all the other channels except the front go to zero, which is very dramatic.

I've been thinking that another approach would be to exclusively rely on the stereo expander "nR+L" method. The logic is as follows: towards the center of the room there is a lot of similarity between the R and the L signal. The sound clues are in fact identical according to the HRFT. So for all practical purposes the front of the room is an "R+L" signal. As the angle between the sound source and the front of your face increases, the signal difference in terms of intensity and phase increases (largely as a function of the sine of the angle, BTW), until you have maximum separation at 90 degrees or close to that.

If one is to "wrap" stereo around ones head, it stands to reason that the Left signal would be strongest and more "pure" next to your left ear and same for the right ear. The signal should be "less pure" in front of you.  So I'd propose to stop using difference signals to generate the virtual speakers, and just rely on exaggerating the left signals by varying their relative intensity only using the "nL+R" approach (equivalently "L+R/n," as such:

Another proposed 8- discrete channel Virtual Speaker System for Project PSSSIII
L is the Left stereo signal. R is the Right stereo signal.
H_n is an operator that filters the signal with the Head Related Transfer function for:
the real front speakers if n=0, and the n^th location of the virtual speaker if n>0
d is an additional time delay as a function of center front location to ensure a precedence effect for the virtual speakers

Each method has its pros and cons. Both of them are assigning weights toward one channel to create a new channel, gradually putting more weight on one side as the stereo signal leans toward that side; however the difference method cancels a signal with equal right and left content everywhere except for the front, whereas the sum method does not at all, which means that the sum method gives you a center signal that goes to the center of the room, not the front. The stereo separation could be much lower with the sum method. On the other hand the sum method avoids potential sound cancellation problems by not adding phase differences too abruptly as the difference method does. A signal with a right sided content only cannot interfere with a left signal that doesn't exist, so the potential for odd "echoes," common in Dolby surround is eliminated. The sum method also preserves the frequency content of the program by not lowering the volume of the centered sound sources. a common problem in SQ Quadraphonic. The delay operator would be forced on the two front speakers. To increase sound separation a sharp weight would be used for the two real front channels. The two front channels still should sound like stereo. Maybe use a golden ratio rule or geometric rule for the distribution of weights of all the channels to allow a perfect net energy conservation across the left and right side content.

One added perk of the discrete virtual speaker system (DVSS?) I outline is that you can always install jacks that tap into the front end of the  "R-L" circuits and actually if you have extra amplifiers pump the same stereo expander channels into real speakers (minus HRTF and phase change elements) to compare the performance of the virtual speaker to the real speaker! That would be something!

You could balance the sound a bit by introducing a center channel (R+L) and introduce phase changes - I suspect- as Sony did by delaying the  center channel. That would take advantage of something called the "Precedence Effect," where the brain perceives a sound which arrives early to one ear as being the original source of the sound, creating the illusion of sound location. The most advance quadraphonic systems, and the Sony Matrix most likely used this phenomenon. This could be a reason *I may have* to use a center channel - I will try to find out if I need the center channel or not, but the stereo separation between speakers is very low already, and that would reduce it even more.

As a last remark, the system, being pumped through two speakers, would be fully compatible with headphones. I see no reason why the DVSS system can't be used directly in headphones. In theory that is in fact the preferred method of reproduction for the HRTF filtered music!

All in all, this would be an exceptionally flexible project, and potentially complicated as well. The question is how complicated I want to make it. The complexity is determined by how many "humps" you are using to approximate the HRTF for a given virtual speaker, and obviously by the number of virtual speakers. This means potentially an awful lot of Op Amp circuits, depending. So I need to get creative to simplify the HRTF and work my way from there.

[J. Wilhelm]

As an aside today, for the first time I connected the active crossover/low pass filter to the black Sony's 5.1 channel input with the unfiltered signal sent by splitter to the front speakers and the subwooofer signal to the rear speakers. Much to my delight, the sound was *very* pure. Really unadulterated bass that sounds very different from the base made by the Lepai amplifier. Zero noise artefacts, or seemingly resonant humming, and very lovely sound all around, without having to use a second amplifier. Even though the volume of the subwooofers is fairly low because you only have a fixed amplification factor of two with the Op Amp circuit.

I can fix the amplification factor simply by using a potentiometer or resistors to lower the volume of the front channels and crank up the overall volume of the amplifier.

I was careful not to run the amplifier for too long, as the impedance of the subwooofers is only 4Ω, and this amplifier likes 8, otherwise it'll complain in the worst possible way. In fact, I have to adjust all speakers as the front speakers are 6Ω. The last thing I want is to fry my one good amp.

But this is very exciting because I get to bring the good subwooofers and finally hear music without artefacts or dangerous kraken mating sounds.

Tonight's test for the amplifier used Herbie Hancock's 1998 tribute to George and Ira Gershwin, an album called "Gershwin's World," which I highly recommend. I'm going to bring those speakers now before life takes all my money away (which seems imminent) and I can't do anything enjoyable again for the next 20 years.

https://m.youtube.com/playlist?list=PL8a8cutYP7fpt6kgfknQfP_ospmVm8Ftl

[J. Wilhelm]

I fineagled two high power resistors out of gate spring coils - the same type that I used to make inductors earlier, and came up with a 1.4Ω resistance to add to the subwooofers so at least they go over 6Ω to match the front Sony drivers. So with that, I conducted another test and I'm satisfied with the results.

A home made 1.4Ω high power stereo resistor. The inductance should be minimal (few micro Henrys) with no metal core.

I bit the bullet and I ordered a new pair of subwooofers. I figure this is my last chance to do it. The woofer units I chose are wide range mid - bass (sounds like an oxymoron, doesn't it?) woofers with an 80W max rating.

The reason I chose these woofers instead of dedicated subwooofers was the impedance rating, the size, and also the price. The frequency cutoff at - 3dB doesn't go down to 40 Hz like I wanted with the Tymphany 5½ inch subwooofers, but I just found out that the tymphany units will *not* fit the enclosure as the diameter is too large, and worse they're out of stock at Parts Express until September. God knows where I'll be in September, chances are I could be living in San Diego by then.

So I settled for a Dayton Sound mid bass woofer which goes down to 50 Hz, and has very similar characteristics with a resonance frequency of 57 Hz instead of the 50 Hz of the Tymphany. It should work well with the 60 Hz resonance of the horns. What I liked about this unit is that the woofers come in 8Ω impedance, which makes my life easier. I can use the 1.4Ω stereo resistor above for the Sony drivers to increase impedance from 6Ω to near 8Ω.

The control of the subwooofer' amplifier will come from a knob that will balance the front Sony drivers against the Dayton woofers on the sides. So it's a passive tonal control.

For the moment, the black Sony amplifier is working brilliantly, with zero noise. I can connect or disconnect any cable I want while the device is working without sending an amplifier into a frenzy. The difference is astounding, compared to the Lepai amplifier.

[J. Wilhelm]

The little monsters arrived. They're really heavy. The magnets are larger than the paper cone of the drivers.

The first step besides installing these two little monsters, will be to make a passive splitter adapter for a sound source. In this case, a headphone Bluetooth adapter that I modified. I want to adjust the level of volume of the Dayton subwooofer drivers, but because I'm starting with a fixed signal, I need to attenuate the front speakers first, and arbitrarily decide how much lower the volume of the front speakers will be relative to the maximum subwooofer volume.

The only way to do that is to use a potentiometer to try to measure an acceptable volume. I'll first try to raise the volume of the front speakers to a "high" volume, and then use a potentiometer to reduce the volume to a "comfortable level" (I'm sure my roommate will appreciate that   ), the voltage ratio between the "high" and "comfortable" levels will be the same as the fraction in terms of resistance between the feed of the active subwooofer crossover and the bypassed signal going into the front channels. This is all very arbitrary and non scientific, but it's the only passive way to do it if you're using only a a single amplifier (your Bluetooth receiver) to feed the circuits.

Building a volume control to figure out the max ratio between subwooofer and front signals

The other way would be to build two Op Amp buffers, with a volume control each. I'll see if I have enough supplies at hand to do that. But if not, I'll try the passive method first.

I just need to start getting real results from the subwooofers, because throughout the process, I've never had components able to pull their weight below 80 Hz. Firstly, the blue Sony drivers sharply cut off at 80 Hz, then the Panasonic sub drivers could only handle 8W, and then the Lepai amplifier wouldn't play nice with other components, suffered from poor sensitivity and low volume. So I've wasted a lot of time and money with components that couldn't handle the low frequency ranges, and my horns remain largely untested in that range.

I'm anxious to get some real bone rattling bass so I can start making adjustments!

[J. Wilhelm]

I have a bit less time to work on this nowadays, but I'm still making progress. I find that working on the project esases my nerves from my real world problems.

I've installed the speakers permanently, mounted on 1/2 inch pine blocks, and to do that I had to rewire the interior of the speaker enclosures. The two new drivers added a significant amount of weight to the horns, I'm afraid... But this is Steampunk, so weight is not an issue, right?   Then I added a set of speaker wire connector jacks, which saves me the trouble of dealing with long cables permanently attached to the speakers. That will also allow me to cut lengths of speaker wire customized for whatever test I'm performing. The idea is to eventually use those to connect to an amplifier case bolted in the back of the horns. I might reuse the carcass of the Lepai amplifier to house a 4 channel amplifier.

The new Dayton driver is smaller but has an 80 watt maximum power handling capability.


Speaker wire jacks for the left speaker enclosure. The connections on the left side are for the front Sony drivers, full range and tweeter. The two connections on the right side are for the Dayton Audio subwooofer driver at the end.

The results from the change are very positive. I'm able to indefinitely increase the volume on the subwooofers without any audible distortion. Perhaps an oscilloscope and a frequency spectrum analyzer might be able to pick up on high volume speaker distortions, but I haven't heard any so far at normal listening volumes. The horn case at very high volumes will rattle, however, but that's expected until I finalize the installation and use neoprene gaskets between the speaker joints. In this project you have to suspect that any two surfaces close together, even if seemingly abutting one another, will produce buzzing sounds at certain frequencies at high volumes. This is normal, and you won't know which joints will do that until you perform a frequency sweep, like I did earlier to measure the horns' frequency response - that is the Holy Grail to get to the point of measuring, after normal volume hearing tests are successful.

Initial tests with the potentiometer I posted on the previous post show that the woofer driver needs more amplification for you to positively get a "thump" effect from the subwooofer. The potentiometer allows me to adjust the voltage of the subwooofer channels to twice the voltage of the front channels. Multiply that by an amplification factor of two from the sub's active crossover (low pass filter) and that should correspond to an amplification factor of 4 for the subwooofer, before the Sony amp pumps it into the Dayton Audio drivers.

At an amplification factor of 4 you're getting a very subtle, yet audible bass that sounds like a passive crossover. This is what I imagine the "audiophile mode" of the subwooofer setting would be - no "thump" just supporting bass. But a listener will probably want to increase that level through a rich bass and beyond to a window rattling thump. So that means that I have to increase the amplification factor beyond 4 (factor of 2 before pushing through the low pass filter), but being careful not no increase the voltage too high, as the OP amps in their current configuration can only handle amplitudes of 9 volts maximum (18 v peak to peak). If a line level signal is on the order of 1volts, that means my amplification factor shouldn't be higher than a factor of 8.

I am thinking of building a new buffer section, with a built in active crossover (low pass filter) which includes a volume control. The new front end section performs a dual function as bass control and buffer for the amplifier. I might add a volume control for the front speakers as well, but I'm not sure that's needed. Effectively, this is a type of "brute force" control, where you are regulating the voltage of the drivers, themselves.  I remember back in the 70s, consumer hi fi speakers used yo have fine tuning controls to regulate the output of the woofer and tweeters. This is basically the same.

In the future the tonal control would be expanded to include the DVSS 3D virtual surround - if I ever get to build it.

So now I'm getting to the point where all these amplifier chips and tone controls will need to be housed in some way. I can see the amplifier case with its heat sink bolted on the back of the horns. But on the front there will be a need to install a panel box of sorts where all the control knobs and jacks would be installed.

I've talked about the wood inlay decoration before, but besides that, I had this strange idea of building a center panel, somewhat similar to an antique secretary desk with a roll top to hide the controls!  I've seen miniature novelty roll top desks sold at hobby shops before, and I'm wondering if this would be a good way to bring the steam back into the project?

Is a miniature roll top too crazy for a front panel?

[John Zybourne]

This is an incredible build and I wanted to say thank you for taking the time to explain your thought process and method of calculation. The digression of fourier and laplace transforms was particularly standout. Huzzah to audio and electromotive design!

[J. Wilhelm]

This is an incredible build and I wanted to say thank you for taking the time to explain your thought process and method of calculation. The digression of fourier and laplace transforms was particularly standout. Huzzah to audio and electromotive design!

Well, thank you so much! That's the nicest thing I've been told about the project. The truth is that most electrical, acoustics and linear systems engineering was elective coursework for me I accumulated in my undergrad years (acoustics was a graduate elective), so I can't be an expert at any level. I kind of feel like I'm re-inventing the wheel at every step. It all looks far more complicated than it is, because I have very little practice on the subject.

When I was in my late teens before college, I was very much into building electronic circuits and so when I started, I didn't have the math background (Fourier/Laplace Transform) to even understand how to convert a circuit diagram into a frequency response plot or viceversa. That came much later, and by the time I had learned that in the Linear Systems class (Junior level), I was very much into aeronautical engineering and fluid mechanics. As a last resort in 1997, as a Senior at UT Austin, when I was presenting the final project in Aircraft Design, I tabulated the data I gathered when I was still in San Diego in 1993 as a freshman at UCSD (the Head Related Transform) - because I knew I'd be using it someday  Well that time has arrived! Almost 30 years have passed!

[J. Wilhelm]

Very Nice

"Sizing" procedures, for lack of a better term are proceeding nicely. Yesterday was the first time that I managed to play louder volumes with the subwooofers on without any major issues. Auditive comparisons with the Altec Lansing of the Vic Boombox Mk. I show that bass reproduction has matched and even surpassed the performance of the Altec Lansing. I won't be able to give you exact numbers until I finish "sizing" the system.

Two things were necessary, to get to this point.

1. When I say "sizing" what I mean is a preliminary calibration of the impedance of the horn's input, relative to the sound source. Now that I have 4 equal amplified channels in the Sony, it's easy for me to determine how much I need to amplify the subwooofer channels to match the performance of the Altec Lansing, beyond what the blue Sony drivers can muster. Using the potentiometer and changing resistance values, I determined that the input voltage (for a given current flow) for the subwooofer should be around 3.5 times that of the front speakers, plus a factor of amplification of 2 for the active crossover. In other words, the Dayton Audio subwooofer drivers receive a voltage 7 times higher that the voltage received by the Sony full range speakers if all power amplifiers are identical. This means that I need to put a voltage amplifier with a gain of 3.5 after the active crossover. Or I build a new active crossover with a gain of 7. This amplification factor is very much dependent on the type of active crossover you build and the performance specific to the Dayton subwooofers, so this was a type of calibration procedure.

Bluetooth setup for listening trials. The Bluetooth receiver is on the bottom left.
The active crossover is on the center, and the voltage divider/crossover volume on the right

Naturally an amplification factor of 7 is just an average. In reality I used a 50kΩ potentiometer on top of a 10kΩ base resistor. In other words, out of a total resistance of 60kΩ, the voltage is divided between the 10kΩ resistance, which is 1/6 of the total voltage, and the 50kΩ variable potentiometer. While the front Sony drivers get 1/6 of the available voltage, the active crossover gets a variable voltage, ranging from 1/6 to 100%  of the available voltage or a maximum of 6 times higher voltage plus an amplification factor of 2 due to the active crossover. Maximum, the Dayton subwooofers get an amplification of A=6x2=12 times the voltage of the Sony drivers. In audio tests that means that the subwooofer control volume knob is exactly at the middle point when you reach that ideal amplification factor of 7. At A=12, the sub woofer output is not so strong that it distorts sound, but the bass is very heavy. That should give the user a nice control range without blowing up the subwooofers. Naturally avoiding sound clipping means that the Sony amplifier must be able to handle a voltage amplitude of 12 times the input voltage. This keeps me from trying out maximum volumes at 50 W per channel. It's still possible to clip and distort the sound, and right now the Op Amp that handles the subwooofer signal can only handle 9 volts (battery power) in amplitude. I may have to built a special power supply to increase the input voltage of the Op Amp driven active crossover. The Sony amps should handle at least 17 volts in amplitude, so this is something I need to consider going forward.

2. The second thing that I did was to add the 1.4Ω resistors to the input of the Sony drivers.
The last time I was listening to the horns, the sound coming from the Sony drivers was "too bright," in spite of the fact that the bass was very strong. It occurred to me that the Sony drivers were still operating at an impedance of 6Ω, and that will affect the frequency response of the Sony MOSFET amplifiers. 6Ω is 25% lower impedance relative to the impedance of the Dayton drivers, so you would hear that difference if the Sony amplified channels are all exactly the same. That explains the "brightness"  as a higher response in the mid frequencies of the blue Sony drivers. *face-palm * I should have figured out that earlier. That is something you would only hear if all 4 channels (R, L, subwooofer R and subwooofer L) are exactly the same. Anyhow, adding the 1.4Ω stereo resistor revealed a rich warm tone across the entire bandwidth. That made me very happy. It sounds as warm as it did when I was using the Lepai amplifier and the Samsung subwooofers at much lower volumes. Shame that the Lepai was such a b*#*& to work with and the Samsung drivers only handled 8W.

Now, finally, the Mark III Boombox can go head to head against the Mk. I Boombox , and in theory I could start using the frequency analyzer to take frequency response measurements like I did before and compare the Mk I against the Mk III quantitatively. However, before I do that, I'd like to build a new front end buffer circuit incorporating the active crossover circuit for the subwooofers and buffers for all the channels. Either way, the Mark III boasts 50 W per subwooofer channel, vs the total of 30 W for the Altec Lansing mono subwoofer. Real limits to the Mark III are the small size of the subwoofer drivers, which nevertheless are amplified by the hybrid horns. It will be an interesting comparison.

The buffers are very important, because the idea is to connect *any* sound source to the amplifier. The way the circuit is now you have only the volume knob control in between the active crossover and the amplifier. You'd like to isolate all channels equally from whatever device they're hooked to.

You need to protect the circuits as much as possible from other devices. For example, the Op Amp circuit is extremely sensitive to voltage spikes and loading. I just blew up the same LM 1458 Op Amp I used to make all the tests in my posts above. I inadvertently reversed the input voltage when hooking up a new 9V battery pair. At first, I heard a faint "pop"  sound, and hoping that it wouldn't be what I knew was most likely, I connected the battery pack properly and the chip began to produce copious amounts of smoke. Luckily, I had a chip replacement, but this is my last LM 1458 chip, and if I burn this one, I'll be weeks away if not months away* from continuing tests.

This is the reason I always mount the chips on sockets and not solder them directly to the circuit board. You need to be able to replace burned components without resoldering a good part of the old circuit board.

Short circuit left a little crater in the center of the chip.

*Because internet commerce is "better" than brick and mortar type shops   I'll tell you what. Bring the printed Sears catalog back, and the horse driven stagecoach back, and I bet you I'll get those chips in faster  )

[J. Wilhelm]

I was going through my local hobby shop before this weekend, and I stumbled on these knick knacks that could be useful for giving a finish to this project:

And an interesting wooden box, that may be useful for the electronic components.

I particularly like those laser cut wood pieces. I wonder if I could work them in as inlays of some sort....

* *

I've been performing "informal" tests on the horn's performance and I'm starting to discover the upper limits of power for the device. The volume it can reach is higher than the Altec Lansing system of the MkI Boombox and that is no surprise, given that the total power of the Altec Lansing ACS 340 is 45 Watts total power, whereas the Sony driven horns have up to 50 Watts x4 = 200 Watts. The only thing is that the horn will never reach that power level on account I have discovered a limitation on the mechanical behavior of the blue Sony drivers.

If you remember, in a past incarnation of the horn I placed the Sony drivers on the extreme ends of the horn. The pressure was so high that the cone deformed and produced strange noise. Moving the speakers to the front eliminated that problem, and instead, dedicated subwooofers were placed at the location. The system was then turned into a hybrid Bass Reflex + Transmission Line enclosure, and that successfully drove pressure down while promoting resonance at 30 and 60 Hz.

The Dayton Audio subwooofers now build up pressure at the right frequency to induce resonance in the horn and the Reflex enclosures. But but it seems that there is an threshold at which the Sony speakers repeat the old behavior, so they'll never get to their 100 W power rating. The good news is that you don't hear this until the volume is quite loud (higher than the Mk. I can produce) and only while playing special soundtracks where the bass is exceptionally rich. It really needs to be a particular type of soundtrack, featuring electronica music and even then, you must have the subwooofers amplified for the phenomenon to occur.

This is a special circumstance that I was warned could happen with inverted horn/transmission line speakers, but up until now I had not experienced the reason why experts in transmission line speakers insisted on very rigid cones. It's obvious that pressure built up inside the horn tests the limit of performance of any drivers used. The solution in this case would be to use polypropylene high power drivers, such as those used for automotive audio. Transmission Line experts also prefer the use of oval drivers instead of circular drivers, and I'm guessing that adds stiffness to the system while discouraging higher modes of oscillation.

I will not be pursuing change of design for these horns, because the performance of the system is quite acceptable, excellent really, for the kinds of volumes used for a personal listening device. Otherwise I would need to go back and build yet another pair of Bass Reflex enclosures.

The Altec Lansing in the MK I Boombox is a personal computer sized system and the new horns definitely surpass the Altec Lansing's performance in every way, I just need to prove that with measurements, since I have shown you those data before. Any imperfections or shortcomings of the Mk. III will be dealt with with tonal control. I pondering whether to implement a digital sound optimization method (the latest rage in speaker design), but that may not be necessary. I may however, go back to looking at the Sound Blaster Extigy sound card, and integrating that into the horns. The EAX virtual surround system of the Sound Blaster card is able to work even without a computer, so I'll be most interested to try it out.

Even as a television sound bar, the system is already much better than most systems found at your local shop, short of a high power 5.1 subwooofer system like a Bose entertainment system with satellite speakers and a big floor subwooofer unit (which my rooomate has in the house living room). Even then, I'm willing to bet the frequency response of the horns is much better (flatter and much cleaner sound) than that of the Bose, on account Bose systems are notorious for their unnatural frequency response, which is despised by audiophiles. As is, the hybrid horn pair make a killer sound bar for your TV.

Instead I will wrap up this prototype enclosure, and finish it aesthetically, with an electronics compartment to add tonal control, sound effects and such, as I see fit.

A higher power version of this horn will be made in the near future, after I compile all of the lessons I have learned. And next time I will pick all the speakers myself after designing the horns around the specs I find online, as opposed to picking stuff from second hand stores. In other words, if you want better power performance, you need more expensive components, and that is a next step. That system would need to be made with the purpose to be sold because it will be (for me, at least) a significant investment of money on my part. The goal of that "third generation" horn will be to drive the power up to the 3 digit range like as a mid-high range hi-fi replacement system, but not necessarily to develop a higher sound quality, which I believe I have already achieved.

With that said, I will continue working on the electronic side of the project, including exploring the Sound Blaster sound card, and of course, I will start paying attention to the aesthetic side of the project.

[Sir Henry]

[snip]
I particularly like those laser cut wood pieces. I wonder if I could work them in as inlays of some sort....
[/snip]

I wondered that a couple of years ago and started playing around. This led to discovering a technique for easily making fake inlay using nothing more complicated than a Stanley/craft knife. Or, for more complex designs, a laser cutter (available at most Hackspaces).

At the start of the first lockdown I realised that I was going to be stuck at home for quite a long time, being rather vulnerable to most of the effects of covid. So I spent 3 days manically designing a really complicated box, covered with fake inlay, to keep me busy. Having just spent a week vectorising a catalogue of inlay strip designs, that seemed the way to go. So each side is made up of the same pattern of strips but with a different basic geometric pattern - circle, square, diamond and hexagon. Lots of repetition so quick to do but visually very complex.
The fascias of the boxes inside were done a couple of weeks later, after I had become really bored with working in traditional inlay strip design, so they went in a very different direction.

To keep the costs down I coloured/stained the designs with felt tip pens - I bought a hundred different colours for about £5. The colours have since changed a bit (presumably due to the acids in the wood) so I wouldn't go that way again. Brusho ink powders are much more vivid and stable. Or, for a more traditional inlay look, there are always wood stains in a huge range of colours (https://www.wood-finishes-direct.com/product/manns-pine-wood-stain). The 100ml sample pots go a long way when you're using a fine brush. This is the way I'm going with the next box, a complex series of puzzle boxes that go together to make one large box.

For personal and sneaky/silly reasons I'm reluctant to explain the process in public further just yet, but if you want to know how to do it pm me and I'll fill in the details.

[Sorontar]

Wow, they look great Sir Henry. My problem is always having pens last long enough to do anything. Paint is harder to work with but lasts as long as there is something in the bottle.

Sorontar

[Synistor 303]

Wow, they look great Sir Henry. My problem is always having pens last long enough to do anything. Paint is harder to work with but lasts as long as there is something in the bottle.

Sorontar

I would second that, Sir Henry! It looks amazing.

[J. Wilhelm]

[snip]
I particularly like those laser cut wood pieces. I wonder if I could work them in as inlays of some sort....
[/snip]

I wondered that a couple of years ago and started playing around. This led to discovering a technique for easily making fake inlay using nothing more complicated than a Stanley/craft knife. Or, for more complex designs, a laser cutter (available at most Hackspaces).

At the start of the first lockdown I realised that I was going to be stuck at home for quite a long time, being rather vulnerable to most of the effects of covid. So I spent 3 days manically designing a really complicated box, covered with fake inlay, to keep me busy. Having just spent a week vectorising a catalogue of inlay strip designs, that seemed the way to go. So each side is made up of the same pattern of strips but with a different basic geometric pattern - circle, square, diamond and hexagon. Lots of repetition so quick to do but visually very complex.
The fascias of the boxes inside were done a couple of weeks later, after I had become really bored with working in traditional inlay strip design, so they went in a very different direction.

To keep the costs down I coloured/stained the designs with felt tip pens - I bought a hundred different colours for about £5. The colours have since changed a bit (presumably due to the acids in the wood) so I wouldn't go that way again. Brusho ink powders are much more vivid and stable. Or, for a more traditional inlay look, there are always wood stains in a huge range of colours (https://www.wood-finishes-direct.com/product/manns-pine-wood-stain). The 100ml sample pots go a long way when you're using a fine brush. This is the way I'm going with the next box, a complex series of puzzle boxes that go together to make one large box.

For personal and sneaky/silly reasons I'm reluctant to explain the process in public further just yet, but if you want to know how to do it pm me and I'll fill in the details.

That looks incredible, Sir Henry. And to color it with markers takes an extra amount of patience! I'd probably.. No, I'd definitely be using wood stain (though a felt tip will give you unprecedented control. Controlling wood stain flow in the natural fiber is extremely difficult, speaking from experience. Wood stain is meant to penetrate, and it will do so for one or more millimeters.

I've seen these before. Perhaps do some trials beforehand. My fear is the woodstain penetrating too far, even completely separate pieces just touching each other.

There's a great many varieties of wood stains available. Water based wood stain attains those greens and blue hues, and it very much looks like someone used watercolors on wood. Oil based stains are always shades of yellow, red and brown.

I'm assuming the pattern is a trivial thing to do with laser, and a not so trivial with a hobby knife, though. I'll have to do some experiments beforehand. There's also some hackerspace / makerspace type places around, I've just never inquired before.

[Sir Henry]

One of the joys of this system is that it uses thin plywood. This means that the stain doesn't need (in fact can't) penetrate beyond the top layer because the glue is water-resistant. It also means that the stain spreads sideways much better, so you don't need to be that accurate, the stain bleeds up to the edges. As long as you cut through the top layer, the stains don't bleed into the adjacent pieces.

There is another technique that uses the laser cutter to 'etch' the plywood. This removes the entire top layer of ply in the design but no more. This can be left or filled with softened wax to add colour.


eta: This can also be done, depending on the design, by hand with a craft knife to cut the outlines and small chisels to remove the infill. Effective but slow.

[J. Wilhelm]

I wood need to find the patience to do it by hand though. Getting the craft grade plywood is easy enough, and I've done great things with an X-acto knife, but scrollwork, you could say I'm still green on that. It's basically a new branch for me.

[John Zybourne]

The Honorable J. Wilhelm, I must inquire as to the progress of this incredible build!

Regards,
JZ

[J. Wilhelm]

Oh dear. I didn't see your post! My apologies!

Status: I'm hunting for old Class B amplifier chips like the vintage 1990s Samsung chips I presented earlier. I had a little stash of cash earlier this year, and I felt confident I could built a 4 channel power amp to be integrated directly into the wooden enclosure. The idea is to match a "digital" square wave power supply with the analog amp chips, the reason being that a switching power supply is much smaller and lighter than a traditional transformer, so it's a hybrid of new school and old school where it counts (I want low harmonic distortion)

Changing the subwoofer speakers to a higher power rating was the concurrent plan. The limit in performance was the power rating of the Sony amplifier in the subwoofer channels (clipping) which require an amplification factor of 7 over the front channels, and also the power rating of the subwoofers themselves (more clipping), both of which meant that in some movie scenes you would suddenly hit a "hard limit" and the sound would be severely clipped. Otherwise the sound was beautiful throughout the whole movie (Lord of the Rings).

So I was going to do that and perform the final audio frequency response test to formalize the victory of Mark III over Mk. I. Unfortunately, life had other plans. First my roommate got sick with Covid in June, and then my employer lost contracts and in the space of 3 weeks I lost over $1K, leaving me little headroom over my monthly expenses. So went my tax return from last year.

I've recovered my income so far in the last 2 weeks, but I'm still down $1K, so I can't proceed as planned.  Oddly trying to monetize Mk. III is not a bad idea, but I face a fairly negative market with high inflation due to political events we all know. That doesn't mean I can't sell anything online, but it does mean I can't sell anything expensive now. Which means Mk. III can't be monetized directly as planned.

I'm exploring the possibility of creating a low budget spinoff of Mk. III, carrying over the "Energy Cascade™ method (harmonic overtone cascade) and the Ocarina ™ method (multiple frequency Helmholtz resonator) to make a portable low power boombox, possibly with a class D amp, like so many Bluetooth boomboxes you see out there.

So this would be the development of a new online business. Not necessarily Steampunk. Is there a market for that? Maybe. There's a hard psychological limit to how much people can spend.. Especially nowadays, with people talking about a coming recession. Realistically, in post recession times (eg 2010, two years after the global meltdown of 2008) I can sell $100 items several times a week and $500 items once or maybe twice per month online. But items over $1000 are once a year affairs. So I have to choose what I want to build based on that. And we're going into a recession, not getting out of one, so that gives me some pause.

But it's a thought.