Wednesday 4 March 2020

Synthesizer Build part-21: ARP2600 LOWPASS FILTER (4072).

The famous ARP 4072 VCF. The best sounding filter of any I built so far! With verified stripboard layout.

A word of warning right at the start; this is an advanced project, not for beginners. You need to know your electronics and you also need to have a good oscilloscope.

The ARP2600 is my favourite synth from the early 70's. It's been used on so many iconic records and luckily Behringer did a very good remake of this synth so it is now in reach of the general public. I was lucky enough to get the limited edition Gray Meanie version which has through-hole capacitors in the filter and the sounds you can get from it are truly amazing. 
In any synth the filter is the defining factor in the creation of the sound and after tackling the ARP's Envelope Follower I thought it was time to try out the famous 4072 filter. ARP has had a number of well known filter types. The 4012 (4035 for Odyssey) which was a Moog type ladder filter over which they got in trouble with Moog for patent infringement. The 4023 two-pole filter of the early Odyssey synths. Then later came the 4072 (the one we're going to make) for the later ARP2600's. The ones with the orange labels with white lettering. These had a fault at first due to miscalculation, which limited the bandwidth of the filter to below 10kHz. This was later fixed with a few component value changes. And then there's the 4075 which was the filter used in the later ARP Odyssey's.
If you want to build this filter there's really only one schematic you can turn to and that's the Yusynth schematic. So I set to work making a layout. I first tried just starting at the lower left of the schematic and building the layout up from there. Within minutes it turned so complicated I couldn't make heads nor tails of it. So after an other unsuccessful try I came to version 3 of the layout and this time I decided to place all the semiconductor components neatly on the board first. All transistors in a row on top and the two chips in their own space underneath and wire it all up that way. This worked fantastically and after a days work I had a layout that looked really good and, more importantly, turned out to be faultless right from the get go.
I was blown away when I tested the finished filter. Of all the filters I built, from the Moog Ladder Filter to the Steiner-Parker, there is no filter that sounds as good as this one. Now I love the Steiner filter and it sounds awesome but this one just has more quality and better resonance control. More meat on the bone if you know what I mean. It sounds how a synthesizer should sound. But of course this is all a matter of personal perception. Mind you this filter, at least the one I built, has less output volume. It's a bit quieter than other filters which is why I suggested a upgrade of the gain in output opamp. More on that further in the article.

Like I mentioned at the beginning, this is not a project for beginners. It's reasonably complicated and you need to work very methodically and do things in steps. First map out all the cuts in the copper strips with a Sharpy and cut the traces accordingly. Then solder in all the wire bridges and then solder in the components. Keep counting the holes and make sure everything is placed exactly like on the layout, otherwise you will run into trouble with space on the board and things end up not being connected right. I worked from left to right soldering it all in and checking every connection with a powerful loupe. And in the end, of course, it didn't work straight away. It turns out I had forgotten to cut four copper strips near the 1V/Oct trimmer. After I cut those the filter suddenly sprung to life and started making sounds that instantly reminded me of the ARP2600.

Here's the Yusynth schematic. It looks a bit weird but the LM3900 really operates on negative voltage, in this circuit. I got a very useful comment which suggested some changes for running this filter on a dual 12 Volt powersupply: The R35, R36 U2d network is supposed to feed -7,5 Volts into R37 and the inverting input of U2c. With a 12 V powersupply this will only be -6 Volts. By replacing R35 with a 15 to 18 kOhm resistor you can get it to -7,5 Volts. This makes the filter sweep sound better too. You do have to re-adjust the trimpots if you do this modification after you built the filter. Also you can change the behaviour of the Resonance pot by adjusting the value of R44. If you replace that with a trimpot you can experiment with that. R44 is the 15K resistor that is hanging overboard in my layout, connected to the wiper of the resonance potmeter. You could find some room on the stripboard for that trimpot. If you make cuts at holes U48 to X48 you can create some room between there and the screwhead and make the necessary connections.
HOWEVER..... the above changes may not be necessary. Anton commented below and makes the case that these changes may not be needed. I'm going to quote his comment here for you take in consideration:
[QUOTE] I do not fully agree that -7.5V is better for 12 Volt builders. As Eddy did point out, the LM3900 runs on a negative supply voltage: its positive supply is connected to GND and its negative supply is connected (in the 12 V case) to -12V. This means that the output can swing between (almost) GND and -12V where the -6V is the middle (actually GND for the signal). This gives a headroom of (almost) 6 V up and down, which is optimal in my opinion.
Changing the -6V into -7.5V reduces the negative headroom to almost 4.5V, while increasing the positive headroom (but this cannot be used).
So, I think the issue that Tim had was something else. I see that the filter is totally DC-coupled, There are no couple capacitors (also not in the original schematic). If you drive it with a signal with a DC-offset. like the sawtooth from the AS3340 oscillator, this could cause clipping of the positive or negative side of the signal.
So. my advice would be to change the voltage back to -6V and add couple capacitors (e.g. 330nF) in the audio input at the hot (input) side of P1 and P2. [END QUOTE] 

If you're building for 12V and the above confuses you just continue building by the layouts below. If you get problems like clipping use Anton's suggestion and try capacitors on the audio inputs.

And here's the layout. Like I mentioned before, the layout is verified because it's the one I used for my own build. (All potmeters are shown from the front with shaft facing you). 
Addition: I've had confirmation from multiple readers that this layout has been used successfully. 
To increase the gain I strongly advise to change resistor R41 from 56K to 100K. R41 is the 56K resistor over pins 6 and 7 of IC-2 at the bottom left (from hole U-8 to V-8). Otherwise the volume will be a bit too low.

(Last revised: 24-Aug-2020: Corrected a mistake in the layout. The 10K resistor from pin 5 of the LM3900 to the base of Q7 was not connected right. 22-Feb.-2021: slight cosmetic changes to the layout. 11-May-2021: Removed colour-codes from resistors to make values better readable.)

Stripboard only:

Sometimes you'll see a cut in the copper strip overlapping a component in the layout above. I've done that on purpose so the cuts are easily visible. The layout is pretty complicated especially for beginners because there are so many cuts to be made, so I want things to be as clear as possible. 
To make things even easier, I have made a layout showing just the cuts you need to make in the copper traces and the wirebridges you need to put in. You must do all this first before putting in the other components. This is viewed from the component side:
Cuts and Wirebridges component side:

And finally a layout showing just the cuts that need to be made in the copper strips. This is seen from the COPPER SIDE!:

Bill of Materials:

If you don't trust yourself to build this on Stripboard then here's the PCB design for this filter. If you print this on A4 size paper it will be to the right scale. You can check by measuring the DIP14 IC package which should be 15mm from pin 1 to pin 7.

I refer you to the Yusynth website for the rest of the build information:

There are 12 transistors in this filter and they need to be matched pairs but you don't have to use a curve-tracer or anything. I simply matched them on Hfe value with the transistor tester on my multimeter and that was close enough. Officially they need to be matched over the value of Vbe, so if you measure the voltage drop over the Base-Emitter junction, and match them that way, that will be the best method but you'll need to set up a little test rig for that on a breadboard. 
The four 470pF capacitors need to be high quality and also closely matched in value. I used polystyrene caps for those. I even matched the 220 Ohm base resistors so they all had the same value. In my case they are all 216 Ohm.  The CV inputs all have 100K resistors on the inputs and a 150K resistor on the wiper of the Cut-Off Frequency potmeter. I didn't have room for them on the stripboard so I hung them over-board so to speak. In reality I soldered those resistors straight to the wipers of the potmeters and in case of the 1V/Oct. straight to the input jack. Then I put some heat-shrink tubing over them and after that I put some heat-shrink tubing over all the wires from one input together so there's never any tension on the resistor itself. This works fine. Of course, if you use a bigger piece of stripboard you can accommodate those resistors on the board itself. Or you can use a small piece of stripboard, solder the resistors on that and connect it to the main board with wires and then use some hot glue and a plastic spacer to glue it to the main stripboard. Lots of options :-)
The resonance potmeter needs to be a dual- aka stereo potmeter. I didn't have one but luckily my neighbour, who repairs audio equipment, had one laying around but it was a logarithmic potmeter. I put it in anyway and it worked like a charm. :) For the trimmer potmeters you can use a 50K for trimmer T1 if you don't have a 47K. In fact, it can be any value from 20K upwards because it's just connected between plus and minus 15V so the actual resistance isn't important for the working of the circuit. But don't forget there is 30 Volts across that trimmer so don't use a value below 20K to keep the current flow down. For trimmer T2 you can use a 2K instead of a 2.2K, but you must keep close to the recommended value for that one because it is part of the input bias for the transistor Q3. I used a 2K on my print and this works fine.

This filter has a 1 Volt per octave input connection to make the resonance follow the chromatic scale. The filter sounds better over all if you use that connection although it is not necessary for the filter to function. There's a trimmer (T2) for the 1V/Oct and the way I set it was to listen to the filter's response while going over the keyboard from low to high. If it is set wrong you'll hear the notes become all muddled up and out of tune at the higher end. If you set the filter potmeters in such a way that it self-oscillates, then the resonance pitch will follow the keyboard scale. So you need to tune the filter so that the self-oscillation is in tune with the keyboard notes if possible. I myself however did not tune it that way. I simply tuned it so the notes sounded ok over all the octaves and left it at that. That's good enough for me and the filter works fine. I don't think the self-oscillation of the filter will track well over multiple octaves anyway, but again, I didn't try that so I may be wrong. The filter is an Alan R. Pearlman design (ARP) and they are usually really good designs. Let me know in the comments if you managed to get self-oscillation tracking over the octaves, please!
The other trimmer is the Low Frequency trim-pot (T1). It needs to be set so that the output wave at the lowest end of the keyboard, and with the Cut-off pot turned all the way counter-clockwise, is a nice sinusoidal bass tone, at least, that is the way I set it. I'm not saying that this the way to do it. I'm simply saying, this is how I did it (and that goes for every article on this website too).
The frequency cut-off potmeter is wired up in such a way that it opens up and lets through the high frequencies when you turn it clockwise and when you turn it counter clockwise it cuts off more and more of the high frequencies making the sound very deep and low.
The values of the potmeters for CV IN and for the audio inputs are not critical and you can use anything from 10K to 1M for those because they are just level potmeters. For the audio potmeters the schematic says to use logarithmic ones but in reality linear will work fine too. It's log because it's audio. Like I mentioned before, I used a logarithmic stereo-potmeter for the Resonance control because that's the only thing I had but it seems to work very well eventhough the schematic says to use a linear type. It probably wouldn't matter what value you use for the Frequency Control either but I'd stick to the recommended 50K or 47K for that one. (I used 100K's for the CV level control potmeters.)
Don't forget to solder the 100K resistors, for the CV inputs, to the wipers of the potmeters or to the input on the stripboard, and don't forget either that the resistor on the wiper of the Frequency Control potmeter is a 150K and not a 100K one! (A mistake I initially made.)

Here are some pictures of the finished stripboard. This is an early version that has one more jump wire than the new layout. I realized I had a copper strip that was not in use so I used it to replace a jump wire. You can see I marked out the cuts in the copper strips with a black felt pen. I also marked out the 0V/Ground strip with a black line on the component side of the stripboard. Marking out the ground helps to prevent mistakes.

Here's a little video with a demo of what the filter sounds like, taken right after I built it in. Remember when I filmed this it was the first time I played around with this filter so this is just a simple demo of the sounds it produces but I think you'll agree it sounds very much like the original ARP2600.

In this second video the filter CV-1 input is connected to the little 7555 AD/AR with the big arcade button. This kicks up the cutoff frequency of the ARP filter as soon as a key is pressed and then releases it pretty quickly thereafter. The AD/AR is set to trigger mode so it gives an Attack/Decay response.  The filter is fed with a single squarewave from the VCO. I think you'll agree it sounds amazing. Like a synth should sound. With apologies for my poor keyboard playing :p 
This is a new video posted on the 12th of November 2020:

This filter can also produce those helicopter sounds that you can hear in the beginning of 'Apocalypse Now'. (Francis Ford Coppola had an ARP2600 himself.) All you have to do is turn the cut-off frequency counter clockwise and connect an LFO with a sawtooth wave to the CV input, set to the frequency that the rotor-blades of the helicopter would have and turn the resonance counter-clockwise too.
One little attention point you must remember when using this filter. It's possible to overload this filter with audio in so much that the resonance won't work at full capacity. I had this happen to me where the resonance wouldn't produce the famous whistling sound and I had been trouble shooting for a day changing out the IC's, checking transistors, replacing the capacitors until I finally found out I had the input level set too high (The Audio-1 level potmeter on the front panel). I turned it back by a quarter and everything was back to normal. I'm telling you this so you don't make the same mistake. ^___^

This is what the panel looks like now. I've touched the lettering up a bit because it was all crooked (and it still is I guess, LOL) so it's good enough for me. What's important is what this panel represents; the best friggin' filter I've ever built!! :)

Okay that's it for now.
To finish off this article here's a fantastic documentary about the history of ARP Instruments by YouTuber Alex Ball who has the best synthesizer channel on YouTube in my opinion. Enjoy!

That's it for this article. I hope you liked it.
If you have any questions or remarks please put them in the comments below or on the 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!


  1. Hi Eddy,
    Congratulations and thank you for sharing your effort with everybody.
    I have a question, what's the difference between the two 22uf caps close to the tl074?

    1. The difference between them? I used two 22µF caps from different manufacturers but other then that there's no difference. They are in the resonance part of the circuit and exactly what their function is, I'm not sure. In the layout I used one radial cap and one axial cap just for convenience but there is no difference in function and you can use either type as long as they are 22µF. The other two 22µF caps in this circuit are just for noise suppresion on the power rails and they can be anything from 10 to 100µF each but 22µF is recommended in the schematic. Are you building this filter? If so, please let me know how you get on. If you have any questions just ask ;)

  2. Thanks Eddy.
    I'm planning to build it,now sourcing components and looking to bulk order for some other modules I'd like to build. It'll take some time to receive them, so patience will be needed.
    Now I'm busy enough with my soma lyra 8 diy build....

    By the way, will your designs work well with +12 - 12, just out of the box with no changes?

    1. I can not say with certainty if it will work on +/-12 V but I'm planning to build a little adapter so I can switch between 15 and 12V because I get asked that question a lot. So, I'll get back to you on that one. Btw, no worries, I realize it'll take some time before you completed the build but do inform me when you've finished. ;-)

    2. Okay, I did some tests and I'm happy to report that the filter works just as well on 12V as it does on 15V. So no problems there. :)

  3. Hallo Eddy,
    ich habe den ARP zusammen gelötet,der Cutoff/CV/Oct 1/v scheinen zu funktionieren bei mir,mit dem Resonanz Potentiometer
    kann ich nur laut/leise machen,die Audio Eingänge zeigen keine Reaktion !
    Ist das Stripboard Layout Aktuell ? Du hattest im März etwas aktuellisiert,sind diese Erneuerung im Stripboard Layout enthalten ?
    Hast Du eine Idee wo ich auf Fehler suche gehen kann ?

    1. I'll reply in English Maik. I'm sorry you're having trouble with it. There's only one thing I changed or updated and that's the connection of the 220 Ohm resistors to pin 14 of the TL074. In the first version this went through a wire bridge, but then I saw I had one copper strip left over on the stripboard so I used that and did away with the wire bridge. I also had trimmer T1 connected to +15V with a wire before I realized the top copper strip was carrying +15V anyway, so it was double connected at first.
      I used this layout to build my own filter and I had it working almost immediately. I had forgotten to cut a few copper strips at first like I explained in the text but after I cut them the filter worked fine. The only thing I can recommend you do is to print out the schematic drawing and to take your filter and check every single component connection on the stripboard against the schematic. It is very tedious work but I wouldn't know what else to recommend. I have gone over this layout so many times I can't count them all and I can not discover any faults in it and like I said I used it for my own build so I know it's good. The +/-12V shouldn't make any difference. I tested it on 12V and mine works fine on it.

  4. Ich habe es mit +/- 12V angeschlossen (Eurorack) !

    1. Are you sure the resistor from the Base of Q2 to Ground is 220 Ohm and not 220K? I just noticed the Ohm symbol is missing from that one on the layout.

    2. MAIK I caught a mistake in the layout. The 10K to the base of Q7 is not connected! I'm correcting the layouts now!

  5. Hey jetzt funktioniert alles,Danke !!! Es ist etwas leise,welche widerstände könnte ich etwas ändern !

    1. You can change R41 into a 100K to make the gain of the output opamp bigger. It's the 56K between pins 6 and 7 of the TL074.

  6. Does it make a difference whether the power supply is hooked up to the left side or the right side of the stripboard?

  7. Is it possible to substitute each PNP pair with something like the AS395 ( )

    1. I really can't say. I'd stick with the transistors.

  8. Hi Eddy, love what you do here. Dankzij jou heb ik een geweldige nieuwe hobby. I just wanted to warn people who bought a component tester for this filter like me to match vbe values. MATCH YOUR TRANSISTORS ON HFE VALUE FIRST!!! Don't be stupid like me and build this thing twice by matching on vbe without looking at hfe at all. Get a big batch of 3906's and group them by hfe first. THEN match pairs on vbe with 2 mV tolerance. If the transistors aren't matched correctly the filter's output will fade in and out when you turn the resonance up, which can also be cool, but its probably not what you want from this module.

    1. Dank voor de complimenten en de waarschuwing Tim! Ja het is belangrijk om de transistors goed te matchen. I doe dat met de Hfe meter op mijn multimeter en laat de transistors goed afkoelen nadat ik ze beetgepakt heb..

  9. I might have a tip for 12v builders. The r35, r36, u2d network is supposed to feed -7,5 volts into r37 and the inverting input of u2c. with a 12v power supply this will only be -6 volts. By replacing r35 with 15-18 Kohms you can get it to -7,5 volts. In my module this made the filter sweep sound better imo. You do have to readjust your trimpots if done after building. Also, you can change the behaviour of the resonance pot by adjusting the value of r44. If replaced with a trimpot you can experiment with that

    1. Thank you for that excellent feedback Tim. I will make a note of that in the article. This will be useful to many I think.

  10. I do not fully agree that -7.5V is better for 12 Volt builders. As Eddy did point out, the LM3900 runs on a negative supply voltage: its positive supply is connected to GND and its negative supply is connected (in the 12 V case) to -12V. This means that the output can swing between (almost) GND and -12V where the -6V is the middle (actually GND for the signal). This gives a headroom of (almost) 6 V up and down, which is optimal in my opinion.
    Changing the -6V into -7.5V reduces the negative headroom to almost 4.5V, while increasing the positive headroom (but this cannot be used).
    So, I think the issue that Tim had was something else. I see that the filter is totally DC-coupled, There are no couple capacitors (also not in the original schematic). If you drive it with a signal with a DC-offset. like the sawtooth from the AS3340 oscillator, this could cause clipping of the positive or negative side of the signal.
    So. my advice would be to change the voltage back to -6V and add couple capacitors (e.g. 330nF) in the audio input at the hot (input) side of P1 and P2.

    1. That's a very good analyses! Thanks for commenting Anton.

    2. I quoted your reply in the text above because I think you make a very important point that can save a lot of people unnecessary work. Thanks again for sharing it with the community.


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