Monday, 9 December 2019

Synthesizer Build part-7: THE MOOG LADDER FILTER.

Okay, here we have a filter who's sound is iconic and instantly recognizable if you know the music by Jean-Michel Jarre, Giorgio Moroder or Vangelis and (thousands of others really). The Moog Ladder Filter produces that sharp, snidey, ripping the fabric of the universe synthesizer sound and also that wet, watery sound you sometimes hear (if you use it it with loads of reverb). That's why I had to include it in my synthesizer build project of course.

Since this article was written I have made a new version of this filter, this time including the CA3046 transistor arrays and it works very well and no need to match transistors with that version. If you want to build this filter I would really advise you to use that layout instead of this one (although this one works fine too of course). You can find it in Chapter 39 (click here).

I used the schematic from Yusynth's website.


Before we start. Most people always want to know if it works on 12V. I tested the filter on dual 12V and it works just fine.
In this schematic the top and bottom transistors are applied in the form of a transistor array chip, the CA3046, but I couldn't get hold of that quickly enough and this early in the build I didn't really trust myself to design a layout including those arrays, so I decided to use all transistors and that works just as well. It makes the layout a lot easier. It is always mentioned that you must use matched pairs of transistors for this filter but really, that's a throw-back to the early seventies when transistors were not as consistent and reliable as they are now so if you have transistors from the same batch they will probably be matched well enough but put them through the transistor tester on your multimeter and match them on hfe value. The only place where the transistors must be matched well is on the place in the schematic where they use the CA3046; the top and bottom of the filter and the output on the side. I personally matched all my transistors by using the Transistor Curve Tracer I described in an earlier article on this website.
I built a second ladder filter as a test for the layout below and I used all unmatched transistors. The layout works fine but using unmatched transistors did not turn out well. I could not get the resonance trimmed correctly and there were enormous differences in volume when using the resonance potmeter. I used a squarewave for testing and the top of the squarewave had an angle to it instead of being horizontal. So you must used matched transistors!
This filter has a few quircks that you need to know about but which are normal for this design.
- The Resonance potmeter has only a small area of influence. For most of the throw of the potmeter you will hardly notice anything. This is normal for this design. That's why we need a reversed logarithmic potmeter for Resonance. To stretch out that last bit of the potmeter.
- When the Resonance is fully open, the output volume drops. This too is normal for this filter and even the original Moog ones have this. Yusynth also talks about this on his website.

The build is quite straight forward but you need to be very accurate. The 50K anti-logarithmic potmeter for the Emphasis or Resonance control was an other thing I couldn't get a hold of so I made my own by using a linear potmeter with a 5K resistor between pins 1 and 2. This works very well, In the layout I used a reversed logarithmic potmeter and I show the alternative that I myself used, next to it.
The input level potmeters are 50K logarithmic ones but if you don't have those just use linear ones. They don't even have to be 50K. You can use 100K or 1M or even 10K if that's what you have available. They're just audio input level pots so they act as attenuators or voltage dividers and the value has no impact on the working of the circuit. This goes for all the level potmeters in all the projects on this website unless it is mentioned otherwise on the layout.
The Frequency Cutoff potmeter must be a 10K!

I made a layout for stripboard including the wiring. I used this layout to build a second filter and it worked straight away so this layout is verified. (All potmeters viewed from the front):


(Last revised: 24-June-2020: Corrected polarity of C3. 15-July-2020 added alternative for reversed potmeter.)

Stripboard only. Beware that some stripboards are sold with 56 instead of 55 holes horizontally. The layout is 55 holes wide:



Here are a few pictures of the finished circuitboard:



As you can see in the pictures, I added two trimpotmeters which are not on the stripboard layout above. These are two 200K trim pots and they go over pins 1 and 2 of each opamp, to make the gain adjustable. It says in the schematic to 'adjust the value of the feedback resistors according to audio level'. These trimpots make that possible without having to use the soldering iron. It's a bit awkward with the wires but I had to put the potmeters on the print where there was room enough to accommodate them and the wires. Plus I added them as an after thought, so after I made the layout. At least they are all neatly in a row. :-)
For clarity I made a second layout which includes these alterations. If you decide to replicate this then don't forget to remove (or not solder in) the original resistors over pins 1 and 2; the 56K on IC-1 and the 120K on IC-2 because these are replaced by the trimmers, as this layout shows. Lay-out is verified not only by me but I heard from many people who built this filter successfully using this layout.:


I lowered the Control Voltage input resistors from 100K to 5K6. In the schematic they are 100K but after installing the filter in my synth set-up I noticed that the CV hardly had any effect at all when I connected a LFO to it, so I lowered these to 5K6.

After finishing the build I tested the filter on the oscilloscope first and set all the trimmers to the right positions. It's easy enough to do, you just watch the scope for the best response. Then, after installing the filter you can adjust the trimmers to get the best sound out of it. This filter is self-oscillating, meaning that if you have nothing connected to the inputs and you turn Frequency Cut-off and Resonance all the way up the filter will oscillate of its own. There's no 1V/Oct. input though so resonance won't keep track with the notes on the keyboard. An other problem is that the self resonance only occurs at the top of the squarewave and not the bottom part. When you start it up it will self oscillate on both the top and the bottom but as soon as the transistors warm up, about 20 seconds, the bottom oscillations disappear. This is due to the fact that I have used transistors that were not perfectly matched. Using the CA3046 chip would solve this and that's why I made a second Moog Ladder Filter project using the CA3046 chips. You can find that in chapter 39
The first all transistor filter I built didn't have this problem so if you're careful with matching the transistors you should be fine. Please note that the input volume is also an influence on this. If the level is too low you can also have the bottom oscillations disappear.
Here are some oscilloscope images to illustrate what I mean:
Self oscillation when filter is just switched on:



And here's the situation after about 20 seconds. The bottom oscillations are gone. The filter still sounds pretty cool though:


This is something I only discovered a short while ago but well matched transistors or using the CA3046 chip should solve this issue. The first Ladder Filter I built using all transistors didn't suffer from this problem because I was careful to match the transistors accurately.

Here's a video showing the test results on the oscilloscope so you have an idea of what the waveforms should look like. In this first filter I built and which I demo in this video the transistors seem to be reasonably well matched because I do retain some self oscillation on the lower parts of the squarewave:

       


A demonstration of the sound of the filter:



Here's a look at my PCB for the filter. This is what is actually in my synthesizer and I build it from a different layout that I made earlier especially for the Eurocard format of stripboard:


Here's the panel I made for it:


As you can see in the picture, I've installed a bypass switch for this filter. It's great to have this filter in series with the Korg MS-20 or the ARP2600 filter but sometimes you want to be able to easily switch to one filter. With a bypass switch there's no need to constantly connect and disconnect patch cables so this is really helpful. The switch only works with audio input 1 and it sends the signal straight to the output jack and disconnects the in- and output from the in- and output jacks at the same time. Here's the wiring diagram for the bypass switch:




Okay that's it for this episode. Stay tuned for more synthesizer build articles and while you're here leave a comment please!

27 comments:

  1. Hey Eddy, Roland from Synth DIY (snd some other group) here. I am a tad confused by the vactrol circuit you added to freq cutoff. There is already CV for that in the schematic. There is none for the emphasis/resonance though. Isn't it the resonance you have put the vactrol on?

    ReplyDelete
    Replies
    1. Oh yes, you are right. I mixed that up. The vactrol is connected to Resonance and not cut-off. The one parameter that isn't already controlled by a CV input in the schematic. I'll correct that straight away. Thank you!

      Delete
    2. I have corrected it in the article now. Thank you for pointing it out Roland!

      Delete
  2. Hello Eddy, thanks for publishing this. However, I have a remark about the polarity of C3. In my opinion, the schematic is correct.
    The DC-level at the opamp U3 pin 7 is at 0 Volts, the DC level of the base of U1b pin 7 is determined by the resistor ladder from +15V to ground, i.e. 220/(220+4*150+330)*15 = +2.86 Volts. So the plus-side of C3 shall indeed be at the base of U1b pin 7.

    ReplyDelete
    Replies
    1. Ah thank you for noting that. I had wondered about it. This is one of the first filters I built and I'm no expert by any means. Thanks, I will alter this in the article.

      Delete
  3. If building the version with added 200k trimmers and removing the 56k resistor on on IC1, I should also leave off the 10uf capacitor that would've been on the other side of that 56k resistor? Or should it be soldered in a way that connects to the 200k trimmer?

    ReplyDelete
    Replies
    1. You should replace the resistor with the trimmer, that's all. Don't remove the 10µF. That's the input capacitor. It needs to be there.

      Delete
  4. Goodmorning, this project work on +12v/-12v?
    Thanks :)

    ReplyDelete
  5. Hey, Eddy. Great work, but before i proceed to build this, will it work on +15/-15v?, if not, then could you please point out what i have to change in order for it to work correctly? I really need this to run on 15v!

    Thank you,
    -David

    ReplyDelete
    Replies
    1. Yes the project is meant for +/-15V but will also work on +/-12V so no problem. It is however very important to match the transistors well. So take care with that.

      Delete
  6. Thank you for posting this! Looking at the bypass switch, would you connect pins 2 & 5 (input & output) to the switched tip component of the jack or just the tip? Also, could/would you just connect pin 4 (to 1k resistor) to the tip of the output jack?

    ReplyDelete
    Replies
    1. The pins 2 and 5 should be connected to the tip of the socket not the switched bit. The socket switches are not used. The input that goes to the switch comes from the tip of the audio input jack (so it bypasses the level potmeter so you get the full audio signal) and the output goes to tip of the output jack. That way the filter is completely bypassed and the audio signal only goes through the switch.

      Delete
  7. I love this bypass switch idea. Simple and brilliant! I think I will make a module just of bypass switches. Recon I can get 5 or 6 in a slim 3u panel..

    ReplyDelete
  8. Hi Eddy, I've put together one of these modules and I'm wondering if you might be able to help me to get to work properly. What is happening is the output is only a very faint sound of the original input, and no filtering seems to occur. I was able to get it to come out a bit louder by adding the extra trimpots, but that is it. I have checked over everything as I can but it's obviously quite complicated and I'm having trouble figuring out where the issue could be. Thanks.

    ReplyDelete
    Replies
    1. Hi David, sorry you're having trouble with it. I can't troubleshoot it for you. You need to compare your component placement with that of the layout very closely and see if there are any discrepancies. Also measure the voltages, see that the chips are getting power.

      Delete
    2. Would you be able to provide some test points along the signal path? I'm trying to figure it out myself but I'm having trouble comparing the schematic to the layout. Thanks.

      Delete
    3. I'm sorry but I can't provide any test points. I'd have to remove the module from my synth and I have it in use at the moment. But you can start with the IC's. They have positive voltage on pin 8 and negative voltage on pin 4. Then see if the base of the transistors get voltage through the 150 Ohm resistors. The voltage should go down the further you go down the string of 150 Ohm resistors.

      Delete
    4. I couldn't find any out of place or missing components so I just started tinning all of the connections and that seems to have resolved the issue. I think I will build one with CA3046 in the future for the better precision, but this will do for now. Thanks again!

      Delete
    5. Oh cool! I'm so glad you got it working. Yep. it only needs one bad connection to spoil everything. Enjoy the filter!

      Delete
  9. I wanted to briefly announce again that my built filter works excellently, I operate it with +/- 12V. Thanks Eddy !!!

    ReplyDelete
  10. Replies
    1. i built this today and works perfectly. i cut the board in half and added jumpers. so i could fold it and fit it in a eurorack module.

      transistors were pretty easy to match with just hfe reading, i think i got a very consistent batch.

      Delete
    2. Glad it worked out so well! Good work! ^___^

      Delete