Motivation for buying this

Every now and then a take alook at the bay, although i should make music rather than soldering or finding something to solder. When i spotted the ML303 i could not resist. I recently made some songs with the 303-Sound which i made with the Phoscyon Softwareemulation. I once owned some TB303s, about 10 or 12 years ago but they are far to expensive and i like midi more than pushing those next-next-next-up-slide-Buttons and letting 1 or 2 patternlines accompany the music while turning the pots. I am much more interested in using this incredible 303-Sound for melodic lines. The ML has Midi and has a very high reputation considering the clonequality.

It was completely ready assembled which is a lot of work. Much more work is it, to get all the parts from different sources. These two steps i spared. So i got this one for about 160 €. It was flagged as "no sound, sequencer running" - after i got it i could see that the description was - well, true.

All the repairing in the following would have been MUCH harder if not impossible without the help of some collegues from the ML-community. A BIG thank you. They are very kind persons which extensive knowledge in electronics.

Step 1 - Fixing the Oscillator

If the powersupply is running and you wont get any sound out of the headphone- or lineout-jacks, the first place you should look for sound is the oscillator output. This MLs oscillator did NOT oscillate. The reason was not easy to find: obviously in lack of the original FET-transistor 2SK30 the former owner used a J201 which works quite well - if you keep in mind that it has another pinorder. In contrast to the 2SK30 which goes Drain-Gate-Source, the J210 has Gate-Source-Drain. After soldering the J201-legs in the right holes the oscillator worked - together with the already running sequencer.

Step 2 - Fixing the VCA

With runing oscillator i could follow the soundpath and recognized very luckily that the VCF and the VFC-/Accent-Envelopes also worked. But still dead on headphone- and Outputjack! BEFORE the VCA everything was ok. AFTER it ... silence. It seemed that the VCA was dead which is quite a pity because it is a special part you only get in japan. A very kind guy from the community helped me out with a VCA. First of all i had to desolder the old one which was VERY hard - like pulling teath. And like sometimes when pulling teath, the old VCA broke apart from my evil pulling. I dont have a specialtool for IC-desoldering, they are quite expensive. To get sure, i put a socket in the VCA-Place for easier replacing if the new one also wouldnt have worked. But ... - ... it worked :) Now i got a functional 303.

Step 3 - baking eggs on the voltageregulators

The voltageregulators got VERY hot after letting the thing run for an hour. The reason was a resitor with to less resistance ;). The display sucked to much energy and heated the regulators. I first thought, the coolers on the regulators were to small and replaced them with a bigger one. But the result was, that i had one BIG, HOT cooler instead of two SMALL, HOT coolers. Anyway, darkening the LCD-Display with a larger resistor lowered the temperature of the regulators.

Step 4 - Fixing the Opto-Chip for Midi

Now i was curious if the thing would work over MIDI. but the results werent satisfying. Horrible Timing, even for playing simple notes. The problem was the optokoppler CNY-17III. This chip has a little (kind of) LED inside and a lightsensitive device, it splits the midisocket from the microcontroller to protect it and transmits the bare serial datastream. There seem to be large differences between chips with the exactly same name "CNY-17 III". I bought some other ones of them and tried to replace the existing one on the pcb. I was lucky, after replacing the chip with another one the ML worked perfectly via Midi.

Step 5 - Modding

The original 303 is very limited. But the few pots allow to mudulate the sound in a very musical way. There are lots of mods for the 303. A mod is a modification of the existing schematic to bring some additional soundmanipulation-pots on the frontplate. I am not interested in making the 303-sound dirty or distorted or hard. But there are some mods which i found to be a definitive MUST to implement. Most of them are from the web, i tried them out, some didnt work as expected (slidetime with a pot). Here they are, without guarantee. Dont complain if your machine breaks because you tried them out.

1. Accent-Amount on VCF-Cuttoff (replace R46 with a 100k pot)

2. Accent-Amount on VCA (replace R119 with 100k pot)

3. Accent Decay (Wire up a 1Meg Audio (log) Pot with one end to Pin 4 of IC12, the wiper and the other end to the anode (+) D28. You will have to cut the existing trace between Pin4 of IC12 and D28.). This Mod is just awesome, a MUST.

4. VCA Decay (replace R123 with 2M Pot). Put a 10k resistor on eon leg of the pot to prevent a shortcut. there wont happen something dangerous if you dont but its not good to have the vca-time so short that you wont hear any sound. With 10k it is VERY short but you can go sure to maintain a musical signal.

3. and 4. are taken from http://www.ladyada.net/wiki/x0x/envelopemods

5. Accent switchable between Normal/all notes with Accent (R140/R141 switched to mass)/no notes with Accent (R140/R141 over 100k towards +6V)

6. Slide switchable between Normal/all notes with slidet (R59/R60 over 100k against +6V)/no notes slidet (R59/R60 against mass). I didnt find R60 in the 303-scheme. But on R59 is another resistor which is called R60 in the ML-sheet, dont know....

7. Amount for external CV on VCF-Cuttoff (left leg of the pot goes to mass, the right one to the socket and the mid-leg to the mod 8.)

8. VCF-CV switchable for positive modulation (over 100k to R73) or negative (over 100k to R69). With a simple cable CV-out can be patched to VCF-Cutoff-In to have +/- keytracking in the ML. Of course also a LFO can goe here.

9. SlideTime switchable. There is a mod to solder a pot between R89 and Pin 8 if IC12 but i got horrible noises in my signalpath with it. Instead of that i took a turnable switch which takes 5 different capacitors to switch them agaoinst mass, replacing C35. The middle position is the original capacitor, 2 positions faster and 2 slower. The shorter times take 47n and 100n, the longer times are 470n and 4,7 mycro. The last one is far to big and makes strange things on the tune-CV but not uninteresting

I will put up a MP3 to demonstrate the mods.

Step 6 - Assembling and a case

Now everything had to be assembled. I screwed everything to the frontplate which i had let manufactured from - guess - Schaeffer Apparatebau. 100 EUR! Not cheap but awesome. For the LCD-Display i cut out the window from a CD-jewelcase. 8 EUR appeared much to expensive to me for a simple LCD-Window. The case is MDF and was very cheap: 1 EUR. It was glued together and will be painted in near future.

The raw *bay-article

This FET is responsible for the oscillator-oscillating and was soldered with wrong pinorder

The end of a legendary VCA. Dont worry, the broken one was already braindead. To the left the new one with socket

The new bigger cooler for the voltage regulators.

The backside of the frontplate before assemblage

Another look inside

Closer view to moddingpoints

Closeup of the turningswitch with the capacitors for slidetime. Beneath you see the 10k-resistor on the VCA-Decay-Pot

Ready for casing ...

... case needs to get painted in silver and knobs will be replaced with cuter ones.

the finished machine

the switches quite dont fit perfect

modding section

midi-out socket