Devil Fish mods to the TB-303

TB-303 maintenance and simple modifications

Here are some modifications and maintenance tips.  Please also see the notes on spare parts in Q9 of the FAQ at the main Devil Fish page.

To the main Devil Fish page.


An update history is at the end: #updates .


This information is for people who already know about electronics and in particular about avoiding static electricity damage to electronic equipment.  If you don't already know about static electricity and if you did not already know that it can destroy components - including the CPU of the TB-303, for which there is no replacement - then please get a competent technician to work on your machine rather than doing it yourself.

Contents

1 - Replacing the tact switches with Omron B3W-4050 sealed tact switches

#Omron_switches

These notes apply to the TB-303 and TR-606, but there are also some notes on similar switches which are used in the TR-808, JP-8 and the original DX-7.

The TB-303 has 24 "tact" (tactile) switches which were made by the Japanese company ALPS http://www.alps.com who make the pots and rotary switches in the TB-303 and the miniature pots in the Devil Fish. These are good switches, but they are not sealed against dust.

Household dust, especially flakes of skin, get into the switch and contaminate the contacts, causing them to become unreliable. This can make the TB-303 impossible to program.  The original DX7 uses the same switches as the TB-303 / TR-606.  However, they seem to last very well in the DX7, where they are well protected from dust.


Until mid-2010 I replaced these switches with the same type, which by 2010 were known as SKHCBEA010, but which were originally known as SKHCAA.  To protect the switches from dust, I installed a thin plastic dust guard which covered the switches, but had holes for their stems and for the nearby LEDs.  This worked very well, but not perfectly.  Some machines which were used intensively needed their switches replacing again after about 10 years.


One alternative is to use ALPS sealed tact switches: SKQEAAA010.  Based on my experience with 100 of these I ordered from the ALPS distributor here: http://www.switchesplus.com.au I decided not to use these, because I found their click action to be much weaker than that of the original ALPS unsealed switches and of the Omron sealed switches.  Please see the above-mentioned comparison page for details 


The only other suitable sealed tact switch I am aware of is the Omron B3W-4050.  Omron, like ALPS, is a very highly respected Japanese manufacturer.  The Omron tact switch page had a very long and perhaps not very permanent URL in August 2010.  Likewise the PDF datasheet, so I have archived a copy of it here:  Omron-Tact-Switches-2010-B3W_1109.pdf . 


The B3W-4050 has a stem almost like the original ALPS switches, but the stem is unfortunately too wide for the TB-303 / TR-606 switches.  (It is specified to be the same size: 3.8mm.  However, in reality it is too wide.) To replace the stemless switches (originally known as SKHCAB) in the TR-808 or JP-8 the B3W-4000 switch would be one to use. 


These Omron switches are available from Farnell / Newark.  In mid-2010, the Australian Farnell system had two entries for the B3W-4050.  One was for Australian stock and the other was for US stock from Newark - with the Newark switches prices being about 1/3 of the Australian stock's prices.  The Farnell number for the Newark-stocked switches is1436527.  The Australian page has an image of another type of tact switch.

Modifying the Omron switches

The Omron switch stems are too wide for the TB-303 buttons.  One approach is to file them down - but this is very difficult, since the stems are made from nylon or something similarly resistant to filing.  Another is to use a small, fine, but very fast grinding wheel to shave a little off the four sides of the top of the stem.  Initially, this is what I did.  I built a fancy jig to enable me to do it reasonably reliably, but still I found it very difficult to get the final dimensions correct.  Doing this, it is important to shave evenly, otherwise the button will be rotated clockwise or anti-clockwise.  I had quite a few rejects.

Part of the problem is that the holes in the TB-303 buttons vary considerably.  This may be partly due to molding differences, but I suspect the main cause of variation is the different thicknesses of copper plating underneath the chrome plating. 

(By the way, the chrome plating on these buttons and knobs is absolutely first-class.  Many of these buttons and knobs have been in frequent use for nearly 3 decades now, and are showing little or no sign of wear.  I don't know of any other button or knob which lasts so long, and looks so good, no matter how much it is used.)

The solution is to cut the stem twice, at right-angles.  This doesn't reduce its outer dimensions, but makes it springy.  There are several advantages to this over trying to shave down the sides of the stems:


1. The cuts are non-critical in terms of thickness and placement.  Only the depth needs to be tightly controlled - just to the bottom of the outer part of the stem, and not any further into the neck of the stem.  Shaving the sides down is extremely critical, and requires even removal of about 25 microns from all 4 sides.  This is thinner than human hair.  The stems of the ALPS switches are about 3.775mm and the stems of the Omron switches are about 8.820mm.
   
   
2. The cuts don't alter the outside of the top of the stem, so there is no alteration of the button's rotational orientation on the stem.
   
   
3. The springiness which the cuts introduce into the stem makes the stem fit very nicely into TB-303 buttons no matter what size their holes.

2 - Replacing the 6 small pots from Tuning to Accent

#6_small-pots

Update 2013-11-24:


The pots described below are not the ones currently being supplied by Technology Transplant.  They are now supplying the same pots which are used in the Cyclone Analogic TT-303 Bass Bot.  These pots are similar to those described below, but their base is higher so the shaft height above the PCB is correct.  I don't know of any problems with the Resonance pots becoming noisy in Bass Bots, so as far as I know these pots can be used as they are supplied.

Soldering TB-303 and TR-606 pots is tricky.  Once they are soldered in and the machine is ready to be reassembled,  be prepared to heat the melt the joints and push the pots around a little so the shaft is close to the centre of the front panel hole.


Please see the page ../pot-wear/  for photos of how the original ALPS pots can fail due to being turned while downwards pressure is applied to the shafts.  This causes the outside of the rotor to cut into the conductive tracks.  Earlier generation replacement pots from Technology Transplant could fail in the same way.

In late 2009, Technology Transplant http://www.technologytransplant.com started selling another batch of pots, which I think was their third or fourth batch.  The first one or two batches (the first was in 2002s, I recall) were dull zinc plated.  The next batch (mid 2000s) was yellow cadmium plated, and quite hard to solder.  The cadmium plated batch had a difficulty with the Resonance pot shaft being too large for TB-303 knobs, so it was necessary to drill out the knob.  These Resonance pots were not linear - they had a log curve at both ends.  This is a minor problem, since it doesn't alter the range of sounds, just which knob positions the sounds occur at.  Most of the pots in this cadmium plated batch, and I think most of those in the earlier zinc plated batches, were also subject to the problem of downwards pressure while turning causing the rotor to cut into the conductive tracks.  This was due to the design of the pots which involved downwards pressure being taken by the outer part of the rotor - so it id a design fault in the ALPS pots and most of the earlier batches of TT replacement pots..

The late 2009, early 2010 batch of pots from Technology Transplant, as pictured here:



is completely different from earlier batches and from the original ALPS pots.  These are bluish silvery plated and very easy to solder.  They are 17mm in diameter, which is marginally wider than all the other types of pots - but this size is fine, they are not too big.

The good news is:


1. The shafts of all the pots fit the knobs perfectly.
   
   
2. None of the pots are subject to the failure mode of the rotor cutting the conductive tracks if the shaft is turned while downwards pressure is applied.  This was the primary mode of failure of the ALPS pots.  In these new pots, the downwards pressure on the shaft is taken by the very end of the shaft, which goes through a small hole in the phenolic board, and presses against the steel back-plate - about in the middle of the "A50K" text in the image above.  This means the outer part of the rotor never goes near the conductive tracks.

Here is a photo of the rotor and phenolic board of the Resonance pot:



The central part of the shaft goes through the hole in the middle and presses against the metal backplate.  This takes all the downward force applied to the shaft.  The outer part of the rotor is not close to the phenolic board, so it can never press against it or cut the conductive tracks.

The black epoxy is to protect the riveted connections between the conductive tracks and the terminal pins from corrosion.

1. The Resonance pot still has the log curve at each end, rather than being linear. As noted above, this nothing to worry about, since it just alters the knob position for a given sound, rather than alters the sounds which are available.
   
   The photo below shows how the three resistive layers are printed, for both sections of this dual pot.
   
   
   
   
   
2. The shafts are about 1.6mm too short.  See the discussion below about what to do about this.  The knobs on a TB-303 are already marginally too low.  I have been boosting them up about 1.6mm for quite a few years.  That technique - a small circle of leather inside the knob - could be applied to these pots to restore the knobs to the original height, but it is not possible to extend this technique to more than about 1.6mm.
   

I understand that later in 2010, Technology Transplant will have another batch of pots which solve the shaft length problem by having an extended frame at the bottom.  This should work fine.

In the mean-time, from August 2010 onwards, I am using the above pots, with their shorter shafts, and boosting their mounting position with some 1.6mm thick phenolic paperboard spacers I had laser cut.

We are very fortunate that Technology Transplant has these pots manufactured, and sells them world-wide.  This is an obscure pot-shape, and ALPS haven't made this kind of pot since the early 1980s.

The values of the pots are:

• Tuning: 50K B (Linear).
• Cut Off: 50K A (Logarithmic curve on anticlockwise end).
• Resonance: Dual 50K B (Should be linear, but replacement pots have a log curve at both ends).
  
• Env Mod: 50K A (Logarithmic curve on anticlockwise end).
• Decay: 50K A (Logarithmic curve on anticlockwise end).
• Accent:  50K B (Linear).

The original parts are no-doubt unobtainable, but here are the Roland part numbers anyway:

• Tuning and Accent pots:  50K linear (Marked underneath as "50KB").  Roland part number 13219331.
• Cut Off and Env Mod pots: 50K log ("50KA"). Roland part number 13219340.
• Resonance pot: Dual 50K linear ("50KBX2").  Roland part number 13219775 (not the 13219774 mentioned in the service manual).  
• Decay pot.  1 meg linear ("1MA").  Roland part number 13219339.

To help with debugging, the following diagram shows the connections and function of the Resonance pot:

3 - General maintenance

Here are a few tips on keeping your TB-303 healthy, and in reassembling it.

3.1 - Never spray anything in the machine

Do not under any circumstances spray anything, such as WD-40, contact cleaner etc. into any item of electronic apparatus.

If pure water gets into the machine, due to rain or some other source, just let it dry.  If there is lots of water, pull it apart and mop it up with tissues before letting it dry in warm air. The ALPS 24 tact switches will not like being immersed in anything – but they may survive water.  But how could they dry properly, since the actuator is closed against the retaining ring except when it is pressed?  (Omron tact switches are supposed to be sealed against liquid and dust, so they should not be affected by liquids, unless they contain sugar and so gum up the movement of the plunger.)


If some sticky liquid gets into a machine, open it up and wash it away with warm water, using a toothbrush and maybe just a little detergent. Then let it dry in warm air.   Best replace the ALPS tact switches!

 

3.2 - Overheated Q45 power transistor

Sometimes, presumably due to application of more than 9 volts, the power transistor Q45 may be brown or defective. I replace them with a relatively common PNP transistor called a TIP30. (Farnell Electronic Components has them). Whatever you use, make sure it has the pin functions, left to right, of Base, Collector and Emitter.  The TIP30 has the same connections.  The TIP30's beta (hFE, AKA current gain) may not be as good as the original 2SB596 transistor.  The TIP30A spec sheet from Multicomp from Farnell quotes minimum beta of 40 for 0.2A at 4 volts VCE.  However the minimum is only 15 for 1A at 4 volts VCE.  The 2SB596 datasheet from: http://www.ortodoxism.ro/datasheets/WINGS/2SB596.pdf quotes a minimum beta of 40 for 1A with 5V VCE.

3.3 - Pinched wires during re-assembly

It is very easy to pinch various wires when re-assembling the machine – especially black wire 28 near the output socket. Replace any wires which have been pinched and exercise care when re-assembling

3.4 - Knobs falling off

3.5 - Noisy volume pot

3.6 - Batteries not connecting

Early TB-303s had a positive battery connector which was rather thin - too thin for modern batteries whose positive terminal does not stick out very far. Later ones were pressed differently and were a few mm more towards the positive end of the battery. Simply remove the old battery connector and form (IBM Selectric service instructions use the term "form" instead of "bend") the edges of the connector so the contact area sticks out a little more.

The negative spring connector can be a problem too – especially when it has been subject to corrosion from leaking batteries. Remove it from the machine, wash any corrosive materials and rust from it and file the end to expose fresh steel.

The wires to these connectors often fracture after several assembly-reassembly cycles, because the wire is impregnated with solder. Simply trim back the wires and re-solder them.

Technologytransplant.com now sell a kit with replacement battery contacts (very nicely made, including the negative contact spring) and a new battery compartment cover.
 

3.7 - Memory problems - corrosion of RAM chip pins

I describe this problem in terms of the TB-303, but the same problem would occur in the TR-606 as well.

On many occasions (probably 20 or 30, from the early 1990s to 2010), I have encountered an extraordinary fault: severe corrosion of pin 18 (power) and pin 10 (/WE) of the 1K x 4 static RAM chips.  This could cause all sorts of memory problems, including the inability to write new patterns correctly.

I used to think this is only a problem if there has been traces of corrosive battery material in the case for long periods of time.  However, I have now seen this corrosion in machines where there is absolutely no evidence of battery leakage.  Therefore, I think the corrosive atmosphere is created by the batteries themselves - but probably only if the machine is left in a tightly sealed container for a long period of time, such as several years.

The corrosion only occurs on those pins which are at +5 volts from the battery supply. (Actually, it can be more like 5.5 volts!).  Perhaps not every type (manufacturer) of memory chips is affected.  The pin is typically visibly "rusty", but in some cases, it looked to be in quite good condition, despite being completely corroded inside the plating, and not making reliable contact.

The worst corrosion appears to occur inside the pin at about the level of the phenolic PCB - just inside the hole.  Pressing firmly, sideways, on the pin with a screwdriver will reveal the corrosion: the pin will bend and buckle, or perhaps fall away.

These low-power CMOS 1K x 4 chips have not been manufactured since the early 1980s.  Please do not replace them with 2114s which have a high current drain and are not suitable for battery backup.

(I can supply these chips, as removed from working machines, and checked to ensure they have no pin corrosion.)

The functions of the three memory chips, left to right, are:


• IC3: Patterns of Pattern Group I and II.
• IC4: Patterns of Pattern Group III and IV.
• IC5: Pattern edit buffer and Tracks 

3.8 - Buzzing due to LED activity

4 - Modifications

4.1 - Increasing the range of the Cut Off and Env Mod pots

To provide a lower limit to the range of frequencies, without altering the high limit, short out R47 - a 10K resistor just to the left of the Env Mod pot. Do this short with a small length of insulated wire on the rear side of the PCB.

To extend the range of the Env Mod pot to include zero envelope modification, then short out R61 (10K) which is located logically and physically to the anti-clockwise pin of the Env Mod pot.
 

4.2 - Improving the bass response

The standard TB-303 has a weedy bass response. The culprits are C20 and C21 – both 0.01 uF. Take these out and replace them with 0.1 uF capacitors. I use monolithic ceramic caps because they are compact. It does not matter that mono caps have a wide capacitance tolerance and a lousy temperature co-efficient. This is a non-critical value and anything around 0.1 uF is fine.

This gives a worthwhile improvement in bass response. If you want to restore the original TB-303 sound, just do it externally with EQ.
 
 

Update history:

• 1996 September 15:  Original version.
• 2001 January: To cover replacing the pots.
• 2001 March 13: Reliability in general
• 2001 April 22: Fixed mention of resonance pot - should be linear.
• 2002 April 26 and 28: Mentioned the newly manufactured replacement pots, some other replacement parts and corrected my mention of the Decay pot to be log, rather than linear.
• 2002 May 2: More information about the shaft-length of the six pots, and the difficulty of fitting the knob on the new Resonance pot.
• 2002 May 26: Updated Rob Cyborgzero's email address.
• 2002 August 24: Updated Rob's web site, report on his translucent knobs and wrote some more about replacement pots.  (He only did this briefly, so I later deleted this material.)
  
• 2004 August 5: Updated contact details for http://www.technologytransplant.com .  Updated reliability information on dust-guard protected switches. Various other updates.
• 2005 August 25: Added note about restricted active rotational range of replacement Resonance pots.
• 2007 August 21: Linked to Mouser rather than Farnell for SKHCAA.
• 2007 November 20: Notes on beta of TIP30A.
• 2008 January 27: Pointer to new page on the 6 small pots.
• 2010 January 3: Updated links to Technologytransplant.biz (for a while they were not at .com) and deleted information which is either obsolete or better covered by the pot-wear/ page.
• 2010 August 27: Complete revision to cover the 2009 - 2010 batches of Technology Transplant 6 small pots, and mods to Omron tact switches.  Technology Transplant URLs changed back to technologytransplant.com .
• 2010 September 2: Added initial measurements of tact switch hysteresis.   
• 2010 September 5: Replaced these initial measurements with a new page of much more detailed force and displacement measurements: ../tact-switches/ . 
  
• 2013 November 24:  Added  note that the pots from Technology Transplant are different now.  Added section on reducing LED noise.  Added note to the Noisy Volume Pot section on DeoxIT Fader Cleaner.
  
  

© Robin Whittle 1996 to 2013 – First Principles and Real World Interfaces
Return to the main Devil Fish page.