Friday, 15 November 2019

Synthesizer Build part-3: TRIANGLE TO SINEWAVE CONVERTER

ARTICLE UNDER REVISION! New layout and schematic will be up soon.

After finishing the VCO build successfully I wanted to expand it's possibilities by adding a sinewave output. As I had used the LookMumNoComputer design for the VCO I thought I'd go with his design for the Triangle to Sine converter too. I didn't have a schematic so I did something I normally never do, I just used the stripboard layout that was available on the internet. Luckily it is a good design only I couldn't get a good sinewave out of it at first. I think the trouble was that I used the design 'as is' with the +12-0--12V power supply but the signal I put in came from a VCO that uses dual 15V power. Also that signal is multiplied by two in the output buffer opamp of my VCO to get to 10V peak to peak. The original design is for a signal that is 5V peak to peak. So I must have overwhelmed it a bit. Anyway, I decided to alter the design a bit to suit my needs.
This is the original stripboard layout from the LookMumNoComputer website:


What I got at first was a triangle wave that was cut off at the top and bottom and also not symmetrical. I tried different things to get the signal level down to get it to form a good sinewave and after some trial and error I managed to get it to work. I made several changes to this design and adapted it so it takes + and - 15 Volt because all my other modules use those voltages and so it can handle 10V peak to peak input signals.

A - First I made the gain of the input buffer adjustable by changing the 15K resistor for a 10K potmeter over the negative input and the output of the opamp. This makes for a very useful adjustment point as I found out when I tested this. I works really well.

B - Then there is a voltage devider over the inverting input of that opamp which in the original design is made up of a 30K resistor coming from +12V and a 10K going to ground. I changed the 30K into a 10K because after the first alterations I still had a wave that was flatter at one end than the other. I thought it would be a good idea to get the offset voltage to mid range to level things out. This did the trick nicely.
I had also made the input level of the triangle wave adjustable so I could test different levels but this proved un-necessary. The best setting was full level anyway so that idea was scrapped.

After having done all this I was able, by turning the potmeters, to get a fantastic looking sinewave which only got the tiniest little distortion at the bottom of the wave above 8.5kHz

C - The last alteration I did was to put 10µF electrolytic capacitors on the in- and output lines. This I did because I noticed that the in and output had a -5V DC voltage on them and that's not something you want to tolerate normally. The capacitors took care of that. So now it all works perfectly.

Take note: this converter only works well with a 0 to 10Vpp triangle wave on the input and it outputs a -6/+6Vpp sinewave. See the oscilloscope pictures below.

Oh and one final thing. Make sure to ground the audio in- and outputs to the ground of the power supply of this converter. I got 50Hz waves that were 100V peak to peak on my scope from only connecting the output to the scope. Adding a grounding wire from the 0 Volt rail to the audio ground turned that to zero!

I made a stripboard layout of the converter with all the alterations I added.


(Last revised: 13-July-2020: Changed 18K resistor from being connected to ground to being connected to negative voltage rail.)

This is a little sketch I made of the circuit with the alterations I made:


Note: 18K resistor beneath the two transistors needs to go to negative voltage rail instead of ground. This has been updated in the layout.

Here are some images of the sinewave from my scope and also an FFT image which shows how low the distortion is when you look at the harmonics which are negligible really.




The bottom image shows the tiny distortion at the bottom of the wave at frequencies above 8.5 kHz but it seems that only my own build suffers from this. Other people who built this report having no distortion what so ever.



When I build this, I didn't know where in the synthesizer I would put it but I had an off-cut of aluminium that was 3 cm wide and I thought, 'let's use that for the converter.' I added a noise generator made from the MM5837 noise chip (that is also used in the 'Prophet 5', my favourite synth.). The panel being so small meant I had to mount the pcb's at a 90° angle to the panel so I made some L brackets from some copper sheeting I have and I made some rings from plastic to make sure I didn't get any short circuits and so I put it all together. So now it looks like this:



I accidentally drilled one hole too many but luckily it was only a 3mm hole so I put a little LED in that to fill it up. Looks nice too so problem solved haha.

Okay, that's it for this one.
More synthesizer build articles to come. Stay tuned!  Oh and leave a comment while you're here, please!

13 comments:

  1. Een echt net plekkie is bij de AS3340 VCO, ben er nu 1 aan het bouwen

    ReplyDelete
    Replies
    1. Ja ik ben ook een VCO aan het ontwerpen met een ingebouwde triangle to sinewave converter. Dit was één van mijn eerste projecten en het schema heb ik destijds zelf aangepast maar het ziet er een beetje raar uit, vooral bij de ingangs opamp. Maar ik heb deze in mijn eigen synthesizer en hij werkt prima maar alleen met signalen van 0 tot 10Vpp. Bij -5/+5Vpp doet ie het niet. Dus ik ben op zoek naar een goed werkend exemplaar dat beide typen signalen aan kan.

      Delete
  2. Dat zal lastig worden, omdat je die transistor dan niet goed kan aansturen. Enige goede optie is denk een PLL, die volgt je signaal waarbij je de output constant kan houden.

    ReplyDelete
    Replies
    1. Nouja, eigenlijk bedoel ik ééntje die beide typen kan uitsturen. De AS3340 VCO geeft immers 0-10Vpp triangle waves dus die kun je in de converter insturen en dan moet ie 0-10Vpp afleveren en als je dan -5/+5Vpp wil kun je een 10µF condensator in serie met de output schakelen met een SPST schakelaartje. Dat is wat ik gedaan heb met mijn laatste VCO. Alleen ik moest een heleboel aanpassingen doen om het signaal goed te krijgen dus heb ik niet echt een kant en klaar schema voor wat ik gedaan heb. Daarom ben ik op zoek naar een goed ontwerp die 10Vpp aan kan. Het schema in dit artikel is ook een aangepaste versie van de LMNC versie dus ook niet ideaal. Maar goed, het is een kwestie van even wat opzetten op een breadboard en dan experimenteren.

      Delete
  3. ah, snap het. Wat je wel veel ziet is dat die drie driehoek naar wave omzetters allemaal een balanced transitor pair hebben gevolgd door een differential opamp. Ben nog aan het uitzoeken hoe het nu precies werkt

    ReplyDelete
    Replies
    1. Ja dat hebben ze allemaal behalve dat originele ontwerp in the Digisound VCO. Die gebruikt een CA3080 chip, maar die krijg ik niet aan de praat. Das jammer want die is ideaal als toevoeging aan de VCO maar ik vind een ciruit met een balanced pair beter. Die kun je tenminste ook afregelen met een aantal potmeters.

      Delete
  4. Eddy, ik denk dat je schema en stripboard layouts niet kloppen. Ik heb het nagebouwd en het werkte niet. Na wat meten etc, viel me op dat de 18k weerstand die vanaf de 50K potmeter naar GND loopt. Dat is raar omdat de basis via een 2k ook aan GND ligt, dus op basis-emitter staat 0 Volt. De 18K aangesloten op de -12 en bingo werken.

    ReplyDelete
    Replies
    1. Ik had even in mijn haast niet door dat je de converter wel aan de praat hebt gekregen. Dat is mooi werk. Ik ga het schema en de layout aanpassen met de 18K aan negatief voltage. Bedankt voor de info Rolf!

      Delete
  5. Graag gedaan.
    Heb een balanced pair 2N3904 genomen en de 10k en 2k2 als 1% weerstand. Werkt perfect van enige tientallen hertz tot iig 50 KHZ, geen vervorming bij 8 KHz. Mooi sinus signaal via de FFT op de scope.

    ReplyDelete
    Replies
    1. Hey Eddie
      Im building your digisound 80 VCO design right now and wanted to make one of these too. Right now I'm just wondering if I should follow your changes or build the original one, because I'm running my Synth on a +/-12V power Supply. Thanks for the Help
      Leon

      Delete
    2. Yeah that, s a bit difficult. I've had my problems with this design and it's one of the earliest projects I posted. Although it works fine in my synth but only with 0-10Vpp signals. Because you're running it on 12V I advise you to build the LMNC version posted at the top of my article and go from there.

      Delete
  6. I run the converter on +/-12 volt, works just fine. Balanced pair for the 2N3904 and 1% resistors for the 10k and 2k2, resulting in no distortion above 8.5 KHz.

    ReplyDelete