So, filter 4. I wonder how many different filters I will end up making. hmm.. well, probably as long as I’m able to come up with interesting designs I’ll keep going. For this one, I know of only one single filter in the audio world which uses this topology, but since I am not omniscient I’m sure there are quite a few I missend. Nevertheless, I’m pretty sure isn’t a common design. Here is the story…
Once upon a time, I was working on my Sherman Filterbank 2 as it had developed some issues in recent years. I finally got around to contact Herman from Sherman and he was so kind to sell me some new potentiometers to replace the ones causing issues. It brought the filterbank back to full functionality which was great, but occupational curiosity got the best of me. So I started poking around, seeing what makes it tick. I was under the impression the Sherman Filterbank was based on some weird variation of the MS20 filter, given the character of its sound, but I couldn’t find the typical OTA for the filter. Instead I ran into an LTC1060 IC, which, after a google search, I noticed was basically a dual filter on a chip. Now, that peaked my interest. Could it be the source of the mighty filter sweeps character of the Sherman Filterbank, the one that made the Chemical Brothers signature sound? Well, there is A LOT more to the Sherman than this IC, but it did gave me an idea of buying one of those LTC1060’s and see for myself what it can do. I really love this filter and having something similar in Eurorack format would be fun indeed.
The LTC1060’s datasheet is quite comprehensive. It offers a really detailed explanation and plenty of examples on how to use and implement the IC. You can basically just copy the needed configuration out of the datasheet and you have yourself a working filter. So, figure 11. Mode 3: 2nd order filter providing lowpass, highpass and bandpass, 10 minutes later I had it working on the breadboard. Couldn’t be easier, but there were some issues to deal with. First hurdle was the IC needs bipolar 5V, but that’s an easy fix. Slap a 78L05 and 79L05 voltage regulator on there and we’re set, they can pump out 100mA which should be adequate.
Second hurdle is a bit more challenging. The filter frequency cutoff doesn’t get defined by a voltage or a current, or the phases of the moon. Nope. It needs a 50% duty cycle clock input — that’s a square wave peepz — Which relates to the cutoff point of the filter in a 100 to 1 ratio. So, the clock signal’s frequency needs to be 100 times the desired cutoff frequency. So, for a 10Hz cutoff we need a 1kHz clock, but for a 10kHz cutoff, we need a 1MHz clock, and that’s a freakishly wide range for some kind of oscillator to generate, since it’s one of the basic modular functions for a filter to be controlled by CV, I needed a CV controlled oscillator that could produce such a wide range. Luckily all it has to output is a square wave.
I don’t know how Herman from Sherman — gotta love that alliteration — tackled this issue, I didn’t reverse engineered the filterbank to find out, but I suspect the microcontroller in there is responsible for generating those clock signals. That would make sense given it also has a Midi implementation which kinds of requires some kind of microprocessor to be present anyway. I gave that idea some thoughts, but decided not to go that route as microcontrollers are unobtainium these days anyways. So instead I took a look at the all time favourite square wave oscillator, the 555. Google has no shortage on coming up with oscillator designs around the 555, so I did what any lazy engineer does, google a bunch and try out a few designs. Lo and behold, there was no need to reinvent the wheel as I stumbled upon a design that could be CV controlled and had the very wide range of output frequencies I was looking for. It took some convincing, some math and a lot of fidgeting around to get it in the desired range though and it didn’t had a 50% duty cycle, but you can’t have everything in life.
Converting a variable duty cycle signal to a 50% duty cycle isn’t all that hard. The CD4013 flip-flop could do just that with ease, although it would half the frequency. It made the demands on the oscillator a bit higher though as it would need to output double the frequency, so from 2kHz to 2MHz or something around that range, give or take, approximately. But it also gave me a bit of an ‘aha-erlebnis’. This is how the harmonics function on the Sherman filterbank works — and it also explained what all those flipflops and CMOS chips were doing in there. Remember how the frequency of the clock relates to the cutoff frequency? If you would halve the clock speed, you halve the frequency of the cutoff point and that’s an octave, a musically interesting interval. So, if you have 2 filters, you can have filter 2 follow filter 1 exactly one octave lower, or if you divide further, 2 octaves or, with some creativity 3. That’s an interesting feature. Sending the same clock signal to both filters will have them tracking at exactly the same cutoff point, not more or less, nope exactly. Great for chaining 2 12db filters to form a 24db filter. Having the 2 oscillators track in a musical way, like in an octave apart, allows for some interesting and unique filter options. Gosh, this is shaping up to a useful design after all.
Did I told you guys that the LTC is a dual filter chip? I guess I didn’t yet, but it is. It offers 2 2nd order filter cores, and since I haven’t build a dual filter before, it’s going to be a dual filter design with either serial or parallel configurations. The stereo Eurorack crowd will be thrilled, sales will soar and one can dream.
CV all the things!
I sometimes get the remarks that my designs don’t offer enough CV control. I listen to my customers – albeit mostly ignoring it all – but this time I’m going for it and it’s CV control on everything, whatever the cost may be. So break out the box of VCA chips, I’m going in heavy. We tackled the CV control of the cutoff already, so just double that up for good measure and let’s go ahead and do the resonance. Filter resonance is a rather simple thing, you feed the bandpass output back into the input, slap a potentiometer in there to control the amount and you’re good to go squealing like there is no tomorrow. If we replace that potentiometer with a VCA, you have it voltage controlled. Easy peasy. I have never felt the need to CV control the resonance of a filter in 20 years of using synths, but why not, it might come in useful some day. My VCA IC of choice is the AS3360. It’s a dual VCA, easy to work with and since I use them in various other modules I have a few on hand. A bit of tinkering with the resistor values and you can get your feedback loop just right so it nicely self-oscillates at high resonance settings.
Now, to the filter modes. On VCF-3 I provided a separate output for each filter mode, the metal-o-tron allowed for switching, but this time I wanted to tackle the problem slightly differently. The Mutable Instruments Blades offers one knob to sweep from LPF over BPF to HPF – the Sherman does too btw – and that seemed like a great way to handle the filter mode selection process. Small front panel footprint and a lot of flexibility.
But how? I looked into some ways to solve this problem but came to the conclusion the only viable option to achieve this was to build VCA’s. 3 of them for each filter and use control voltages to mix between the filter’s different output modes. It’s a whole lot of components to add to the design, but I really don’t see another way to do it. On the plus side, adding CV control would be easy, and I did say I wanted to go all in when it came to CV control, so better earn up to it. Driving the 3 VCA’s from a single pot comes down to a little arithmetic and tinkering with opamps. A bit of inverting, summing and subtracting later I converted a continuous signal to 3 signals able to drive the different VCA’s. If you take a look at the schematics you’ll notice it’s not very complicated to generate the control signals for the VCA’s as we can fall back on our dual rail setup and hook te pot up from -5V to +5V. I’ve added some protection for the VCA CV in the form of a green led — yes, Zeners were another option, but leds look cool and mysterious — which has a forward voltage of around 2V, which is perfect for the range of voltage the AS3360 likes to see on its CV input.
Is there anything else… hmmm. yes there is. Based on the Sherman (yet again) I did add the -BP +BP function which allows you to add or subtract the BP filter as you please. I always like the sound of the LP+BP filter on the NordLeads and this will allow you to do just that. It’s just a little more flexibility when it comes to beefing up the sound and adds a nice way to confuse people who try to use the bandpass output while having that knob turned to -BP, which results in silence off course.
So, how does it sound
Not unsurprisingly, it does sounds a lot like the Sherman filterbank. It shares the same bit crusher kind of effect on low filter settings which is introduced by the switching capacitor filter’s control signal getting in the audible range. You can see on a scope how ‘steps’ are introduced in the signal at low cutoff points due to the capacitors switching speed. It also tends to do wild things with high resonant settings and being able to space the resonant peaks exactly an octave apart is great fun. However, it’s far from a Eurorack version of that filter, it’s not a copy of the circuit nor was it ever intended to be a port of that filter to Eurorack. I can only recommend the Sherman filterbank as something to get for your studio as it’s a darn wild beast. The similarities stem from the use of the same, off the shelve IC at the core of the filter. ( and I don’t want to get into trouble with them for claiming this is some kind of clone )
Apart from that, it’s quickly becoming my weapon of choice. There are really some cool tricks you can do with it which are impossible to do on other filters. Having the first filter on low pass, link the 2nd filter 2 octaves lower in high-pass gives quite a wide bandpass effect that’s incredibly useful for acid like sounds. I’m going to need to write some kind of manual for this one I suppose.
Buy one
I haven’t really taken price or size into account when developing this module, I wanted it to sound great and be fully featured. It kind of turned out on the big and expensive side though. Sorry for that, but you can’t have everything.
Schematics
If you’ve read the above, you probably won’t be surprised if I’d told you that this is the biggest module I’ve designed so far. It has 306 components spread over 3 boards.
Build notes
This is a more advanced build. It’a all through hole, but at certain places the components are densely packed together and debugging is difficult since the sandwich design of the boards make board 2 – the one where all the filter stuff sits – pretty much inaccessible when the filter is assembled which makes troubleshooting a bit more of a challenge.
Boards are logically assigned. board A is where the UI lives; input, outputs, jacks. Board B has all the analog filter stuff and board C has all the digital clock and logic stuff.
A scope capable of going up to 2MHz is highly recommended for this build.
Links
Build Notes
If you plan on going the DIY route and build this module from kit or PCB/Panel, read the following carefully. This is not an easy build due to the high frequency clock signals and the 3 board PCB sandwich (which makes access to Board B impossible when the module is assembled, and thus hard to troubleshoot ). I’ll add to these notes whenever new information roars it’s head.
- C24, C27, C21, C22, C25 and C26 should be poly capacitors ( WIMA or the likes ). Ceramics will work, but due to a lot of boring reasons the clock signal won’t be as stable.
- D9, D10, D11, D12, D13, D14 are green Leds. Color isn’t important at all, but the Forward voltage should be around 2.6V.
Tweaks
- You can change the value of R27 and R28 to 10k instead of the 5.6K used. This will add a lot more aggressiveness to the filter as the input signal will distort much easier.