It is hard to remember that scant decades ago, electronic magazines — the pre-Internet equivalent of blogs — featured lots of audio circuits based on analog processing. Music synthesizers were popular for example, because microcontrollers were expensive and unable to perform digital signal processing tasks in the way you would use them today. [Julian] has been trying to build a vocoder from that era from ETI magazine. Along the way, he’s making videos documenting what he’s found and how’s he resolving issues.
The circuit generates levels for particular input frequencies. It does so with a two-op-amp bandpass filter, a two-op-amp rectifier, and then an op-amp lowpass filter. That’s five op-amps for each band (there are 14 bands) plus the support circuitry. And that’s just the input section! Today, you would simply sample the signal and do a fast Fourier transform (FFT) to get the same kind of data.
We can sympathize with [Julian] as it looks like he’s worked on this off and on for a long time. The playlist has some preparatory videos going back to 2015. Part of what’s taking so long is probably that he’s making different subsections and testing them in various ways. So while the project isn’t quite finished, it is interesting to watch the description of the pieces, the testing, and the issues he’s encountering. While we’d be more likely to simulate and use a scope on the real hardware, there’s something charming about building real circuits and using VU meter boards to visualize their outputs.
Building something from old plans is bound to turn up some oddities, and this project is no exception. [Julian] notes that in the 14 bands, all but three use a pretty standard bipolar op-amp. But three of the bands use a FET-based op-amp. If you’d guess it was for frequency response, you’d guess wrong. The three bands are in the higher range, but there are two higher bands that use the bipolar devices. [Julian] was soliciting opinions about why this should be. We immediately recognized the need for the extra resistor on the bipolar op-amp was to counter input bias error. But it wasn’t clear to us why the three bands were “special” to start with. There are a few theories about that in the YouTube comments.
One thing we don’t miss is the lack of a microcontroller to handle a user interface. Look at all the controls on that front panel! Not only is that a lot of wiring, the panel work is substantial. Today, you’d have an LCD, a few buttons for a menu, and a rotary encoder.
If you want to brush up on your filter design [Elliot Williams] did a whole series on filters, as did [Bil Herd]. Both of these can give you some background to further appreciate this project.
12 thoughts on “A 38-Year-Old Vocoder Project”
“Today, you’d have an LCD, a few buttons for a menu, and a rotary encoder.” — Or you’d still have a bunch of knobs because everyone loves knobs!
Don’t forget toggle switches and blinky lights!
Everybody loves knobs, but, more importantly, they make control much more efficient and precise for may things. I use a number of oscilloscopes, for example, from a compact portable that runs on Li-ion to an mature Tektronix that takes plug modules.
The little portable is menu to menu to menu to get anything done, and you need to pay close attention to the screen when making any change. All with buttons. Not the worst (it is a high quality US brand), but still requires taking your eyes and concentration away from the job EVERY TIME a minor setting change is needed. Next is a bench/portable digital (PRC manufacture, but the name on the front is a major US based company). It has a couple knobs, but they change function based on context. Easier to work with, but it is still important to pay attention to the menus that take up about 1/3 of the screen space. Next is the daily user bench scope. Modern PRC manufacture with the name of the actual manufacturer on it (many people here probably have the same make). 7 knobs, IIRC, all of which are also press-for-select or press-for-mode, two of which are exclusively for scanning menus. The others are, mostly, dedicated to loogical things like timebase and channel gains (one for each channel– a primary criterion at purchase time). Much less distraction in use.
I am the only one that will use the Tek. “It’s too complicated” they say. I say: “No menus” Everything is a knob or button on the mainframe or a plugin. The input amps are here, the time base is over here, the delay unit is here, the trigger unit is here, the auxiliary timebase and trigger is over here, and so on. Push the selection or click the knob where needed and move on with life. The ONLY unit that requires reading a menu or using a keypad or distraction from the work is the measurement unit, where you are focused on the screen. If it wasn’t made of metal and 200Kg, I’d marry it.
For this application, I would choke myself unconscious in ten minutes if there weren’t independent controls for each and every band and property. Yes, I have used a compact mixing board with two knobs (a picker and a setting) and a few buttons. It was hell.
I hate “dig through 5 menus for everything” based UIs… What’s even worse is the “5 of these options are in the main menu, but these 3 are in this other menu with a different button, have fun reading the manual to find it” effect.
I don’t really mind the more modern high resolution color display interfaces as much, they can fit more on one screen with less time consuming navigation or obscure modal buttons, but I really appreciate a knob or a slider or a button dedicated to the things I’m going to use all the time.
I like encoders and displays more than real pots though. Especially when they sense how fast you’re turning them and move faster non-linearly.
Honestly, I enjoy the knobs mostly but don’t appreciate building them. Also, some things like scanning electron microscope had way too many knobs many of which you rarely used. I’m just as happy to have those on a menu.
One more problem with pots mounted on panels is that unless you spend a fortune on the pots, after just a couple years they’re going to get worn and noisy. That’s one reason why so much of the equipment from that era has been scrapped. It can be hard to get pots that match the original resistance curves, if you can figure out what those curves were.
… and if you want damn near identical knobs to the ones in the original design, they you could of course 3D print them, thus bringing the project bang up to date.
I’m old enough to have bought the original ETI magazines this project series was printed in, and I still kick myself that I threw them all out when moving house about 15 years back.
For those who don’t want to roll their own http://www.cylonix.com/windows_vocoder.html
Also Rakarack for Ubuntu Linux Studio. I have not tried vocoder yet but the rest is super cool. It’s current unlike Windows 95.
DIY – Elektor Vocoder:
Ha I totally recognize this :) My uncle was one of the ‘three stooges’ (their nickname) that designed it along with Becker. He had a couple of more iterations at his house, one of which had a rudimentary beat sync that just looked for amplitude spikes to trigger the filters or envelope. Good find and fun read HaD. Will have to pass it along to him.
I saw this design in ETI as a teenager. One of the other boys at my school, slightly older than me, actually built the vocoder and I saw it once at his house. Very impressive high-tech bit of kit at the time!
Please be kind and respectful to help make the comments section excellent. (Comment Policy)