memory, lithium battery, individual
in to the cymbal and hi-hat circuits, accent button, snare sound modifications, noise level control, switch replacement and new
© Robin Whittle, Real World Interfaces firstname.lastname@example.org 19 December 2012../smem/ .
The pushbutton (tactile) switches of the TR-606 are open to dust. With months and years of use, the dust compacts in the contact area, leading to erratic behaviour. I replace the original ALPS switches with a functionally similar tact switch, made by Omron. The Omron switches are sealed against dust and liquids. I expect they will last for decades.
The switches should always be replaced, even if fresh switches have just been installed – unless they are these sealed Omron switches, or the sealed ALPS switches. Ordinary unsealed ALPS switches will not last very long.
This is functionally identical to the 32 bank system described for the Devil Fish here.
This memory system enables you to play beats from the same pattern number (1 to 16, A or B) in other memory banks while the pattern is running. This can be done by flicking the toggle-switches or pressing the pushbutton, and it can be done for one or more beats at a time, in the middle of patterns.
The result is a style of improvisatory live performance which involves dropping in beats in an intuitive manner and potentially dramatic.
There are 32 combinations of positions of the 5 toggle-switches. The pushbutton switch reverses the function of the lower toggleswitch. If the internal sequencer is playing a given pattern, such as 1A, then by moving the toggle-switches you can select pattern 1A in any of the 32 banks. Since the sequencer reads from memory for each beat it plays, on the beat immediately following the change of switch positions, the sequence will read the drum and accent information from the memory pattern 1A in the newly selected memory bank. So it will play these notes from the newly selected pattern a fraction of a second after the switches have been set to the new combination. Of course when you leave the switches in a given combination, the sequencer keeps playing its patterns from that bank.
In principle, you can play from Track mode when changing banks, but this is likely to confuse the sequencer unless you very carefully organise the Tracks in all the banks you are selecting.
This system also multiplies the total amount of pattern and track memory by 32.Please read the above-mentioned Devil Fish memory page. The simplest way of understanding this system is that the CPU in the TR-606 accesses one "bank" of memory – which is the total battery backed-up memory of the machine. The CPU accesses this "bank", but in reality there are 32 such banks, and you control which one is currently being read and written. A good analogy is that the drum machine has 32 heads, only one of which is active at a time.
It is best to install a Lithium battery to power the battery backed up memory. This is true whether or not the 32 Bank Memory system is installed, since relying on C-cell batteries for backup is unreliable. This is a non-rechargeable battery and cannot leak. Its life will probably be 20 years or more, since the current drain of either the standard memory or the 32 Bank Memory system is less than a microamp, and the batteries have a capacity of about an amp-hour. So, if not for the very low self-discharge rate of the battery, the battery would last more than a century. 20 years is a conservative figure. If the battery is fine in 50 years, I wouldn't be surprised.
Once the Lithium battery is installed, there's no need to install C-cell batteries for the purpose of powering the memory. If you don't need to run the machine from batteries, then there's never a need to install C-cell batteries. One significant danger with having the C-cell batteries in the machine is that it can be left on for days by accident, which will flatten the batteries and cause them to leak.
In the Devil Fish the Lithium battery is part of the main Devil Fish modification. For the TR-606, it is a separate modification.
I can install a red pushbutton switch, like that in the memory system and like the Accent button of the Devil Fish, located to the upper right of the volume control, which activates the Accent whenever it is pressed.
I install a 3.5mm socket on the rear panel, just to the right of the Hi-Hat volume knob. This is an audio input which takes over from the internal six square-wave oscillators which are used for the Cymbal and Hi-Hat sounds.
This is not an audio trigger. It is a sound source for being filtered, gated, distorted and filtered again when one of these sounds is triggered. If you put a bright, continuous keyboard sound such as a chord into it, then you get metallic, shimmering fizzy Cymbal and Hi-Hat sounds which are tonally and harmonically related to the chord. Low input levels result in clear tonality. Higher levels result in fizz and hiss – but the relation to the chord is still audible.
Here are some sound samples. The main long file is for those with patience. It gets pretty wild a the end. The 32 second excerpt dips into some mild-mannered and wild sounds from the long file. These are presented as SoundCloud pieces, but the compression used by SoundCloud does not allow the full sound quality to be perceived. I have provided a 44.1kHz stereo WAV files which have no such limitations and higher-bit-rate MP3 files:
SoundCloud Flash-player and link:TR-606-cymbal-hi-hat-mods-32-sec-excerpt by Robin Whittle
||Noise pulse contains more mid and low frequencies and so is somewhat louder.||Tone
has lower frequency.
||Noise pulse contains still more mid and low frequencies and so is significantly louder.||Tone
has lower frequency and is louder.
I was planning a MIDI In and Out system for the TR-606. However, in December 2012 I decided that other projects have higher priority.
Social Entropy in Texas is working on a TR-606 equivalent of their TB-303 Quicksilver 303. This will be a CPU replacement system with MIDI In and Out via DIN connectors and via USB. This is discussed in the thread: "Quicksilver 606!": in late 2012: http://socialentropy.com/quicksilver/forum/viewtopic.php?f=2&t=56
When this becomes available, I guess in 2013, this will be a good way to upgrade the TR-606 to MIDI In and Out, with extra memory and with new sequencer functions. As with the Quicksilver 303, I will be installing it with my own internal USB isolator (unless it already has one), with MIDI In via the Sync socket and with MIDI out via a small adaptor lead which plugs into a 3.5mm socket near the Sync socket. For more on my Quicksilver 303 installation, please see: ../dfish/midi-options/ .
The forthcoming Quicksilver 606 will not be compatible with the 32 Bank Memory system described above. However, all the other mods listed above will be possible.
I do not have any mods for the kick drum or other drums which I can install inside the machine. Various things could be done, but there is almost nowhere to mount the controls. The TR-606 has an extra circuit board just behind the front panel compared to the TB-303, so a Devil Fish style front panel cannot be contemplated.
I think the TR-606 kick is a unique and excellent sound – better in some ways than the TR-808. The TR-808 has a single Twin-T resonator which rings, after being excited by the trigger pulse. In the TR-808, there is a circuit to change the frequency of this Twin-T resonator in a simple manner so the start of the sound is at a higher frequency, before switching to the lower sound. This gives an initial click or pulse-like nature to the sound.
In the TR-606, there is no such switching. There are two Twin-T resonators. They both start in phase, due to the trigger pulse, and then go out of phase since they are at different frequencies. So there is a solid start to the first cycles of both resonators, with a big positive edge – then they resonate freely and generally won't be in phase after that.
For one customer in 2000, I did very extensive modifications to the TR-606's sound generation circuits. I explored the possibilities with him, and at each juncture he guided the exploration and approved the final changes. I did the raw electronics, leaving a bunch of wires with dangling pots and switches. He mounted these pots and switches in his own way – by installing them in the case of an old CSQ-600 sequencer. There was a wide variety of knob-shapes and types. Let's just say it didn't look at all slick and "Made in Japan"! It sounded great.
While I could do these modifications to another machine, it would be very time-consuming to do the raw electronics, and still more time-consuming to figure out a good way of mounting all the knobs and switches. These things need to be permanently wired into the TR-606, so it needs to be joined permanently to a larger box of some kind. I could replicate these mods in another machine, but the cost, including mounting a separate box, labeling the switches and knobs etc. would be at least AUD$2000. Due to the variety of the pot values, I wouldn't be able to use the small pots like I use in the Devil Fish mods, so they would be conventional miniature pots with shafts and knobs. I have some audio samples of this work, if you are really interested.
The modifications included the Individual Output and Cymbal/HH Audio Input mods as noted above, but there were separate outputs for each Tom. Some of the other mods not allow the creation of the original sound. But that's not such a problem since TR-606's are relatively inexpensive, and it is simple to keep a non-modified machine and use its sounds in addition to those of the modified machine. These extensive mods included:
- Altered one of the first of two bandpass filters IC15A and IC15B. The IC15A filter drives only the HH circuits, while the CY circuits are driven by both filters. The new bandpass frequency is controlled by a pot and can be from much lower to about twice the normal frequency.
- Pot for variable decay for OH: longer and shorter times than normal.
- Switch to disable the circuit which made the HH decay time shorter with high tempos.
- Pot for variable decay for CY: longer and and shorter times.
- Separate disable switches for the noise components of each Tom.
- Each Tom circuit was rebuilt to be a more complex arrangement with:
- Switch to double the capacitor values in the Twin-T resonators, to reduce its frequency by an octave, all other things being the same – which they are not.
- Independent sweep circuits to replace the original shared circuit. The "sweep" circuit shorts out a resistor at the start of the note, to raise the frequency of the Twin-T resonator. Its not a smooth "sweep", but it gives the impression of quickly lowering pitch as the average tension of the drum skin is reduced, when it is not vibrating so far.
- Pot for each Tom to control sweep decay time.
- Pot for each Tom control sweep drive intensity.
- Pot for each Tom to control the pitch when the sweep circuit is on, giving a range of frequencies including the normal one, and higher frequencies.
- Pot fore each Tom to increase gain towards 1.0, increasing the decay time of the resonator. This is the "Monster" pot and when turned far enough to the right, leads to ~2Hz continual oscillation.
These modifications lead to a machine with all sorts of unruly behavior in the Tom circuits. Note that despite the customer's enthusiasm for changes, he was perfectly happy with the Kick circuit in its standard form.
Please think very carefully before attempting electronic work on your own. The following information is not for people who are new to electronics. All the information you need is here, provided you have the required electronic skills. Please do not ask for further details! Nonetheless, if you find something here confusing and can suggest an improvement, please let me know.
The switch replacement is rather tricky, because the Omron B3W-4040 tact switches have stems just a little to wide for the TR-606/TB-303 buttons. Please see ../dfish/303-mods/ for instructions on cutting two slots in the stem so it becomes springy, and will fit nicely into the TB-303 and TR-606 buttons.
The B3W-4050 switches are available from Farnell / Newark. Please see the Devil Fish page for further information. If you would like to purchase kits of 25 Omron switches cut and ready to install, please see: ../dfish/#tact_switch_kits .
Someone has documented the individual output mod at Hyperreal, but I do not recommend this approach. ( http://machines.hyperreal.org/manufacturers/Roland/TR-606/mods/roland.TR-606.indiv-outs ). My approach involves 3.5 mm switch sockets, which take the signal after the resistor which is driven by the wiper of the volume pots. For instance this is after R57, R102, R108, R109 and R149 as shown in the schematic at: http://machines.hyperreal.org/manufacturers/Roland/TR-606/schematics/roland.TR-606.schem-6.gif . This way, the mixing bus consists of a wire running along the connectors themselves, rather than the need for a separate wire from each connector to bring the signal back to the PCB for mixing when nothing is plugged into the socket. So you have a common ground wire running to all sockets, a common mix wire, and then each socket has a wire coming from the resistor on the PCB. All you need to do is isolate the ends of those resistors from the mixing line. This involves a bit of thought, but it is obvious once you have the board in front of you. Just use a sharp knife to cut the copper tracks.
I mount the sockets between and to the left of the relevant volume control, with the socket for the Cymbal / Hi-Hat input mod to the right of the Hi-Hat volume knob. The centre of the holes should be about 8mm above the lower edge of the outside of the case.
The Cymbal / Hi-Hat input mod is straightforward. There are three wires to the socket. Ground, the signal to the socket (its "normally closed" switch contact) and the signal from it (the tip of the plug drives this, if a plug is inserted). When you plug a signal into it, the internal noise source is disconnected and your external signal drives the Cymbal and Hi-Hat circuits. You only need to cut the circuit board track at one point. Looking at the bottom right of the schematic (URL above), you need to isolate the mixing point of the six square-wave oscillators (that which is common to R227, R228, R229, R230, R231, R232 and R210) from the track which leads to C92 and C96. It is quite obvious when you look at the PCB.
There are some other mods, like making the trigger outputs be driven from the Accent and Cymbal rather than the Toms. This is obvious when you look at the schematic.
My Snare mods are as follows: Add a 0.1uF cap across C52, replace R107 with a 470 ohm and replace R110 with a 33k. This lowers the pitch of the Twin-T oscillator a little and boosts its mix level relative to the noise component.
If you are interested in weirdo snare sounds, perhaps you might like to boost the level of the snare white noise source, or provide an external input in its place so you can plug whatever you like into it. To boost the noise generator, put a resistor across R117 (say a 2.2k or a 1k) and then adjust TM3 to give a much higher level. With the TR-808, this can create really loud snare noise sounds and big rumbles in the Toms. At high enough levels, the noise breaks through even when no drum circuit is triggered.
The Accent switch is straightforward. Connect the normally open contacts of a quality switch such as a C&K 8125 SHZBE across the emitter and collector of Q10.
If you need tiny toggle-switches, I highly recommend the C&K Tiny Toggle range, such as the T101 SHZQE solder tail, thread mounting single pole changeover switch, the double pole T201 of the special three position changeover T211, which I use in the Devil Fish. http://www.ckcorp.com/ . Distributors and global parts searches are available from http://www.electronet.com/ . Some useful electronics distributors include: http://www.farnell.com/ , http://www.digikey.com/ and http://www.mouser.com/ .
The TR-808 is an excellent drum machine. I am now doing revised versions of the Sound Mods and 4 Level Accent mods which I have been doing since the early 1980s. Please see: ../tr-808 .
I don't have any sound mods for TR-909, TR-707, TR-727, TR-505 or MC-303, though I can install multiple banks of memory in all these machines. As with the TR-909, this memory (say 8 or 16 banks) is controlled by two push-button switches and a 15mm high 7 segment LED display. It is not possible to do the dynamic switching of banks with the MC-303, TR-909 or TR-707 which I can do with the TB-303, TR-606 and TR-808.