Thursday 11 February 2021

Synthesizer Build part-39: MOOG LADDER FILTER (YuSynth Design).

The famous Moog ladder Lowpass filter (YuSynth version) using two CA3046 transistor array chips so no need to hand match transistors. But this is not a project for beginners!

This is the second Moog Ladder Filter on this website. With the first one I wanted to avoid using the CA3046 chips because I was at the beginning of my synthesizer build project and at that time designing a layout including those chips was a bit too much for me but now, after more than a year of designing layouts from schematics I had confidence enough that I could make a layout that worked. Well, I was right. The layout I made worked perfectly first time.
This filter does sound a bit better than the first one I think. The resonance is better controllable and it just sounds like a professional filter allround.
 
This is a 4 pole Lowpass filter (24dB per Octave cut off slope.) 
Using the CA3046 transistor arrays means that we don't have to hand pick and match transistors ourselves. The transistors in that chip are all matched and even though the chips were obsolete, you can now buy them again. There are also new manufacturers like Alfa Rpar's AS3046 which costs about 6 UK Pounds and is available on Electric Druid's website.
If you match transistors yourself and you don't do a good enough job, you might get a filter that self oscillates on the top half of a squarewave but not on the bottom half. That actually happened to me once too. That's why, if you build the all transistor version posted on this website (chapter 7), you need to be accurate with matching and that's also why this project is easier because there's no matching involved.
This filter is meant to run on a dual 15V powersupply but it will work on 12V too, but you'll need to make three resistor changes:
For Eurorack R24 and R27 need to be changed from 270K to 220K and R26 needs to be changed from 1K2 to 1K. These resistors are situated on the righthand side of the stripboard below the second CA3046 (U2).

Here's the schematic by YuSynth. It's the same one as in chapter 7. In the layout the numbering of the TL072 chip is reversed. I used pins 1,2 and 3 for the input and pins 5,6 and 7 for the output, instead of the other way around like on the schematic:



In the picture below is the verified layout. I used this for my own build and it worked first time. I spent about 5 days making and checking and double checking it. It wasn't easy to fit everything on a 24 by 56 hole stripboard but I managed it luckily.  The first layout design had a jump wire in it but I made an updated version that got rid of the jump wire. Because of the complexity of the layout I would not advise this project to people who are just starting out in DIY synth building. It's not beginner friendly. There was no room left for any mounting holes so I connected the print to the panel using an L-Bracket and plenty of Hot-Glue. If you use a bigger piece of stripboard you will of course have room to use the extra space for mounting brackets.

Here's a Falstad simulation of the ladder filter I made myself: -- CLICK HERE --

There are three CV inputs on the schematic drawing and only one of them has an attenuator. In my build I gave two of them attenuators and the third CV input is un-attenuated. You can use that for a 1V/Octave input, but the self resonance won't track over the octaves. At least I don't think so. There's no specific 1V/Oct. input marked on the YuSynth schematic nor is there a trimmer for it. The Bill of Materials already has two 100K potmeters so you can have level control on two CV inputs. 

Wiring Diagram (This is a 24 x 56 hole stripboard):


(Last revised: 23-Feb.-2021: Removed jumpwire and added a wire bridge. All layouts updated. 02-March-2021: reversed wiring of Resonance potmeter. 25-March-2021: Added de-coupling cap to U3 and two 22µF el.caps on powerrails. 16-7-2023: Removed colour coding from resistors.)

Stripboard only. As you can see I re-used an idea I implemented in the first Moog Ladder Filter project namely to make the gain of the in- and output opamps adjustable with trimpotmeters. So I replaced R5 (56K) with a 100K trimmer and R32 (120K) with a 200K trimmer. Take particular care with the orientation of the transistors! As you can see on the layout the middle ones are rotated 180° in relation to the other transistor pairs. These transistor pairs do not need to be matched. The matched pairs are all contained within the transistor array chips. However, it can't hurt to check their Hfe values and choose transistors that have simular values. Again, this is not necessary but it's something I did myself.


Here's an overview of the cuts that need to be made. Follow it with great accuracy! I'm giving you a view from both sides of the print because I personally always mark the cuts on the component side first and then stick a needle through the marked hole and then mark and cut it on the copper side where the needle comes through. That's my procedure and it guarantees that all the cuts are made accurately. It also helps should you need to troubleshoot later. The first image also has the wirebridges on the component side. Put those in after you finished the cuts.
Here are the cuts and wirebridges marked on the component side. The green ones are connections to ground, red are connections to positive voltage rail and purple to negative voltage rail:


And here are the cuts marked on the Copper Side:



Bill of Materials. 



BUILDING PROCEDURE:
Make sure you go about building this filter very methodically. It's certainly not one of the easiest builds on this website. Mark out the cuts accurately and then cut where indicated. Then put in all the wire bridges. There are 40 in total. Here's a picture of the wire bridges installed and the cuts marked on the component side (something I always advise to do). This is a picture of my print made using the old layout. I have now updated the layout, as mentioned before, and got rid of the jump wire in exchange for one extra wire bridge. Make sure you use single core copper wire, like transformer wire, or tinned copper wire for the wire bridges. If you use normal multistrand insulated wire it will get very messy very soon, especially because there are 40 of them. Break open an old transformer, get the wire out in as long pieces as possible, sandpaper the wire because there's a layer of insulating lacker on there that needs to come off. Then you can use it. 


Then put in the flat lying resistors first. In this picture I forgot the 680 Ohm resistor R34 and the 22 Ohm is a 220 Ohm so I had to replace that one. So beware you get these right ;)  Mark off every component you solder in on a paper print-out of the layout to keep track of your progress. This print also doesn't have the two 10 Ohm resistors on the powerrails because I left those out in my build.


After that put in the rest, I left plenty of room around the trim pots so you can accomodate different sized trimmers. For instance, I use a lot of old de-soldered trimmers I got from old circuitboards and they are usually a bit bigger than what you get these days, but it will all fit.



CALIBRATING THE FILTER:
The trimmers can be normal types. No need to use multiturn trimmers. You need to connect this filter to an oscilloscope and keep an eye on the output signal while inputting a squarewave from your VCO. Then set the trimmers to good resonance response and good signal amplitude over the full throw of the panel potmeters. Then connect it to the VCA and trim again for best sound. It's a matter of experience but it's really not that difficult. You just need to use your logic. I can't really give you a procedure for calibrating.
Beware that the Resonance only works at the last bit of throw on the potmeter. That is why you need a reversed logarithmic potmeter, to stretch out the resistance in that area of the potmeter and get more fine control over Resonance. This is a normal characteristic of the Moog Ladder Filter and there's not really anything you can do about it. All Moog Ladder Filters have this. You could say it's a design fault.

Here's a video of the filter in action, connected to my little sequencer so you have an idea of what it sounds like. You might see occasionally that the self oscillations on the lower part of the squarewave disappear, but this is due to the fact that I have the volume set quite low (itchy neighbours ;-) )  Play with this filter at full volume and you'll blow your mind. It sounds so good! ^___^
In case the video doesn't show up on mobile devices here's the link to the video below:  https://www.youtube.com/watch?v=rIlgBagLW-0



KNOWN QUIRCKS OF THIS FILTER:
This filter has some weird characteristics that occur in all Moog Ladder Filters. For one, if you turn the Resonance (sometimes called Emphasis) fully open, the volume drops a bit. You can see that happening in the demo video above. The second quirck is that the Resonance potmeter is only effective in the last little bit of the throw of the potmeter. That's why you need a reverse logarithmic potmeter. To stretch out that last fully clockwise bit of the potmeter. But don't worry, you can still dial in the Resonance very effectively and to compensate for the drop in volume there's a volume control potmeter on the audio input. So if you set that three quarters open, you can give it a little extra when you open up the Resonance. Anyway, if you set the resonance first then the volume will stay the same when you turn the Cut-Off Frequency potmeter. So it's not too big a problem really. That's probably why Bob Moog didn't address this issue.
There is an alternative schematic by Harald Antes that addresses this drop in volume problem but it is too complex to implement on this layout and this size stripboard but I'll give you the link (which a longtime member of the EB Facebook group kindly posted). There are Gerber files included so you can just order a PCB and build it that way.

ADDING A 24dB/12dB MODE SWITCH:
Making the above mentioned version would be a whole other project in itself. Maybe something for the future. However, if you take a look at the schematic he does suggest a way to switch between the 'Poles' of the filter. I tried this and installed a DPDT toggle switch (ON-ON) and connected it so that I had the choise between 24dB (4 pole) and 12dB (2 pole) and all I can say is, spare yourself the trouble. It adds absolutely nothing to this filter. The only difference is that in 12dB mode the filter's self oscillation on top of the input wave disappears. It still sounds pretty cool but in no way different from the sound you can get in the normal 24dB mode. So don't bother. I just tried it so you don't have to.
But just in case you want to try it yourself here's how to connect the switch. Compare it with the original layout to see which two wire bridges you need to remove and replace with the switch. Do not mix up the wires. The dotted wires go to the bottom side of the ladder and the normal coloured ones go to the top side of the ladder. Don't mix them up! Be accurate. Btw, if you want to make the filter switchable between  24dB and 18dB instead of 24dB and 12dB connect the dotted green and the pink wires to the dots with the same colours. Each of the poles can be tapped off from the collector of the transistor in that pole.
[EDIT] I recently changed the connections to the ones I indicate on the layout turning it into a 18dB filter and this sounds a bit better. You get a little bit of self-oscillation in the signal this way but I just know I will only use this filter in 24dB mode because that sounds the best.



Here are some pictures of the finished module:


A view behind the panel:




Okay, that's it for now. If you have any questions please put them in the comments below or ask on the special Facebook Group for this website.

If you find these projects helpful and would like to support the website and its upkeep then you can buy me a Coffee. There's a button for that underneath the menu if you're on a PC or Mac. Or you can use this PayPal.Me link to donate directly. All donations go towards the website and projects. Thank you!

DISCLAIMER: The author of this article does not accept any responsability for the correct functioning of this, and any other, module/project on this website. What you build, you build at your own risk. All project layouts are thoroughly tested before publication, it's up to you to replicate them and the author can not be held responsable for any mistakes made.

21 comments:

  1. On the video I don't see resonance on the lower square

    ReplyDelete
    Replies
    1. That's true in some parts of the video. In other parts it is there. That is because I sometimes clipped the filter with too high level input but I assure you the resonance is there on both top and bottom of a squarewave because of the matched transistors inside the CA3046's. The oscillator pictures (which I have not posted yet) show this very clearly. I will add them soon.

      Delete
    2. I did some more testing and the main reason why the self oscillation didn't show on the bottom end is because I had turned the volume of my VCA way down low, because of my neighbours. When I turned it up it was all there, no problem.

      Delete
  2. The Yusynth schematic shows that the central pin of the resonance potentiometer should be connected to the 10uF cap, whereas yours has pin 3 attached to it. Is this intentional? And if so, what is the purpose of doing this? Thanks!

    ReplyDelete
    Replies
    1. You're right. That's an oversight on my part. I'll correct it.

      Delete
  3. Really good, detailed build instructions - thank you very much for posting this. One question: did you consider using a transistor array such as MPQ2222 https://www.mouser.de/ProductDetail/Central-Semiconductor/MPQ2222A-PBFREE?qs=l7cgNqFNU1hcO2ukcAJBdw%3D%3D instead of discrete transistors? It might give you a bit more room on the board.

    ReplyDelete
    Replies
    1. I did use transistor arrays. The CA3046. I just kept to the schematic.

      Delete
    2. Sorry, I should have been clearer. On the Veroboard diagram (and in the parts list as Q2-Q7) you have 6 discrete BC 547s. That was the space I was thinking about. I suggested the MPQ2222 because it is four separate transistors, but as you need two the CA3046 would probably work just as well. But the layout may be so tight that you can't fit another transistor array, I suppose that is really what I am asking. Sorry for the confusion.

      Delete
    3. Well you know, those transistors don't need to be matched. Only the top and bottom ones in the ladder and those in the output need to be matched. But yes you could use an array for those too but I think that would make the layout more complicated to design. At least, as it is now, I know this filter to work very well. But you can always try it of course. :)

      Delete
  4. The 5.1k resistor over terminals 1 & 2 on the resonance pot does not seem to have any effect to me. The resistance seen over terminals 2 and 3 is still linear isn't it? Putting a resistance between 2 & 3 also wouldn't help, just making the response for low resistances even steeper. Unfortunately, it does not seem to be possible simulating a rev log taper with only terminals 2 and 3 connected, or am I wrong?

    ReplyDelete
    Replies
    1. I came across this solution on the Yusynth website somewhere but I agree it's better to just use a reverse log potmeter.

      Delete
    2. Thanks for the design layout for this circuit. Saved me time having to breadboard it and make my own layout. For the resonance pot, instead of the 50k reverse log, I used a 250k linear plus a 68k resistor across the middle wiper pin and the 3rd pin. So the wiper pin (2) goes to C11, 3rd pin goes to T3 and a 68k resistor goes across the middle wiper pin and pin 3. Pin 1 of the pot is not connected. Fully turned clockwise, the pot is a short; fully turned anticlockwise the resistance is 68k || 250k = 53.5k (close enough to 50k), and the characteristic as it turns through clockwise is close to an antilog.

      Delete
    3. I built three units of this filter but it was the Eurorack version. I've been looking for a way to improve the resonance potentiometer performance all over the internet. I would like to know if this solution using the 250K potentiometer together with the 68K resistor can be better than using the 50k antilog potentiometer or if it is just a way to make it work without having the 50K antilog in hand.

      Delete
    4. @CONS AUDIO This is for @shean to answer really but my guess is it's a slight improvement over having the 50k antlog pot but I can't believe it solves the difference in volume you get when turning up the resonance. If you check my article, under the section 'Known quircks of this filter' there a link at the bottom to a eurorack version that solves all the problems with this filter.

      Delete
    5. My use of the 250k linear pot plus the 68k resistor was as an alternative to a 50k antilog pot. I find that the resonance knob doesn't have a lot of effect until almost fully turned clockwise, then the resonance kicks in. But maybe I haven't tuned the circuit very well.

      Delete
    6. @shean No this is normal behaviour for this circuit. Resonance only kicks in at the last bit of throw of the potmeter. That's why it needs to be reverse logarithmic, to stretch out that bit of the potmeter.

      Delete
    7. Thanks for confirming that :)

      Delete
  5. A question about the second CA3046 (U2). The CA3046 datasheet says that the substrate (pin 13) must be connected to the most negative point in the external circuit. Although the Yusynth schematic does not show this connection, the component layout figure has pins 12, 13, and 14 of this IC connected to the negative rail (through R2). I just wanted to verify that you haven't had any problems with U2, pin 13 floating?

    ReplyDelete
    Replies
    1. The component layout figure on the Yusynth website has pins 12, 13, and 14 of this IC connected to the negative rail (through R2). Sorry for the confusion.

      Delete
    2. Hi David, no there's no problem. pins 12,13 and 14 are just from one transistor in the chip and it's no problem not using it and leaving it floating.

      Delete

Note: comments are moderated and do not appear straightaway. Your first comment is not allowed to contain any links.