Saturday 13 June 2020

Modding my Roland TR-505 - 4 Drum ROMs plus Pitch Change

The first drum machine I ever built was a heavily modified Clef Drum Machine based on a design published in Practical Electronics (I think, this is a l o n g time ago!). Then I got a Kawai R-50 (in my Kawai period - I also got a K5 additive synth which is still gorgeous!) plus the extra ROM for alternate drum sounds. Then I moved to a Yamaha phase and got an RY-30, which was an unusual and amazing drum machine apparently designed by the 'SY' synthesizer people and so doesn't have an 'Rn' number, and which is half synth, half drum-machine - it has a mod wheel that can control pitch, pan, etc. plus it reads SY wave cards!

Sometime later, I bought a Roland TR-505 off someone that I knew at work, during my D-50, U-220, CM-32L 'Roland' phase, and it has been part of the 'live' rig ever since. But recently, I figured that this might be a good time to revisit some of my now-vintage gear and do some modding, updating, LCD backlighting, etc., and so I looked at what could be done to the TR-505. I wasn't looking for the OTT separate outputs, sliders, and other 'I want to be an 808' mods, instead I just wanted a bit more choice of sounds. I ended up deciding on a ROM expansion, and a Pitch Change mod. Here's what I did, and how I did it.

The 505 Rom Expansion is a kit from HKA Designs, and you get a circuit board with 4 ROM images from 'classic' drum machines combined into a single larger ROM, a socket that you replace the existing factory ROM with, and two switches to select between the ROMs. It is compact, nicely designed, looks very professional, has good instructions, and is very reasonably priced.

The Pitch Change is a bit more homebrew. The Circuitbenders LTC1799 module was 'out of stock' when I looked on their web-site, so I bought the much larger and more 'industrial' module from Electronics Salon. As it turns out, going 'boutique' wasn't such a good idea, and I should have waited for it to come into stock with Circuitbenders... Oh well...

Opening it up...

The first thing to do with any mod is to survey the device you are going to mod. Familiarity is key to get good results and avoid disasters. My TR-505 has the usual UV damage to the flame retardant additive in the plastic - yep, that nasty yellow colour affects the buttons, rotary controls and the lower half of the case. I haven't used any of the hydrogen peroxide based remediations - partly because that 'beige' look is part of the authentic 'vintage-ness'. I thought about this recently when I saw one of the 'white-top' Yamaha Montage synthesizers - what are they going to look like in 5 years' time when they are cream in colour instead of pearly white? Anyway, in sympathy, I'm going to have 'warm' images in this blog post...

At some stage in its life, my TR-505 has been subjected to leaky batteries, and so one casualty that I noticed was a rattle. it turned out to be one of the springs at the far end of the battery compartment that had corroded away and come loose. Getting at that looked tricky, so no batteries!

Anyway, removing three Pozidrive screws and separating three Molex connectors undoes the two halves of the case:

The lower part has the main PCB, with (reversed from normal, because the sockets are on the lower edge) analogue circuitry on the left, and digital circuitry on the right. Looking at the neat columns of capacitors and resistors just above the three audio output sockets, you can see why so many modders add separate outputs to the 505...

The upper part is the buttons, pads and the LCD. You can see some pencil marks where I've been planning out where the rotary control potentiometer might go... Looking around, the far right hand side has an interesting socket:

Which holds the red LED that is used for the beat indication. I did wonder about going fashionable and replacing this with a bright blue LED, but I decided that authentic red, tiny, and slightly dim was okay for now, and it looks like it would be an easy change in the future!

One thing that I had forgotten about was refreshed in my mind when the rear panel insert fell off. It seems that I had used Blu-tack to hold this in place at some stage... Because Blu-tack gradually hardens with ages and loses its stickiness, then I peeled it off and glued the panel on with some general purpose clear adhesive.

Since I've mentioned Blu-tack already, then I might as well reveal one of the techniques that I use when modding gear, and I need to find out exactly how much space is available inside a case. In this example, I wondered if there was enough room above the Tape I/O socket for the rotary control potentiometer that I wanted to use for the Pitch controller, and so I put a blog of Blu-tack in the place where I was hoping to put the control, and closed the case...

You can see the Blu-tack being squashed as I push the two halves of the case together...

And when you then separate the two halves again, you can see exactly how much space is available, and exactly where components are. In this case, you can see the outline of the Tape I/O socket, and the Blu-Tack is about 8 mm thick, so you know that whatever you want to fit into the gap has to be less high that that... As it turns out, space inside the TR-505 was quite scarce, and so I needed to use this technique quite a lot so that I knew where I could squeeze things in.

HKA ROM Expansion

So here's the underside of the HKA expansion PCB, and there's a notch in it. Looking at the upper part of the case, there's a support pillar that holds the two halves together, and this notch avoids that pillar. This is quite close to where I was hoping to put the Pitch controller rotary control, so I made sure that I knew where the board would be when the two halves were combined in the final assembly.

 You can see some shaky pencil marks on the upper part plastic where I'm roughly outlining where the board will be. There's room for the rotary control, and at the lower edge, you can see where I was hoping to put a switch and LED to show if the pitch was the original 'Factory' setting, or the variable one adjusted by the rotary control.

Whenever you draw on the inside of a case, it is a good idea to also consider where this will put things on the outside! Here is where the rotary control position turns out to be on the top of the TR-505. Putting it lower down, on that nice flat bit of plastic, instead of those decorative ridges, isn't possible because that's where the expansion boars is going to be (you need to invert the photo of the expansion board so that the socket clips are at the top...).

Putting a suitably tech and retro control knob in place tells me that I'm going to need to add a label or panel so that I can hide those ridges... I quite like a 'verbose' look to my mods, and so some graphics is fine with me. I'm not so keen on mods where there are just switches, knobs and jack sockets and very little labelling...

Here's the 'Roland' ROM chip that needs to be removed and replaced with a socket. To remove the PCB from the lower half of the case, there's two more Pozidrive screws that need to be removed, and another Molex connector needs to be separated. I'm sure that I don't need to remind you that you don't pull on the wires - you gently lever out the plastic connector block from the socket, using tweezers or a pair of tapered or snipe nosed or curiously bent pliers that cost a fortune and you wonder when you are ever going to use them... Tell Lindstrom that I sent you...

I'm still prevaricating about getting a proper solder rework station, although there are some amazing bargains to be had if you look around. So it was solder wick and a solder sucker to remove the solder on all those joints on the chip. My 'old faithful' solder sucker had a perished O-ring and I couldn't find anywhere that sold the replacement nylon tips, and so I bought a modern mostly plastic version which worked, but didn't have the style of a single piece of machined aluminium.

Here's the underside of the main PCB, with the chip that needs to be removed and replaced with a socket. The blue arrow is there because one of the 'interesting' things that often happens is that you flip a PCB over, find the pins, suck out the solder, and then when you turn it over again, you find that it was the wrong chip! So I always mark it first, so that if I come back to it, I know exactly which chip it is!

Here's the result of solder sucking and wicking - not exactly my favourite activity! Using the soldering iron to check that the pins move freely is essential at this point, and is even more important on plated-through holes (these aren't!). Tweezers again a this point, to lever out the chip gently, and swapping from end to end so that you don't bend the pins too much...

With this result, all ready for the socket. This is a good time to double-check that IC11 is the chip that is supposed to be replaced with a socket! Yep, been there, and got the T-shirt.

This is the socket in place. I prefer these round hole sockets to the straight versions! I have to say that there are a LOT of wire links on this single-sided PCB. Technology has moved on just a little since... Anyway, with the socket installed, the expansion board can be connected...

This photo really shows how the board cutout wraps around that hole in the main PCB for the support pillar... I worried about C17, the ceramic capacitor in the lower right hand corner, so I looked a bit closer:

And, yes, there's just a little bit of a gap there. The inside of the TR-505 is pretty full, and there's not a lot of room to play with, and I've got a pitch change mod to do as well. On which topic, there's the bright blue component on the upper right...

Pitch Change

When the TR-505 was designed, I suspect that just being able to replay samples stored on ROM chips was more than enough progress from the analogue noise-shaping and resonant filtering of the previous generation of drum machines. It also got the magic word 'Digital' into the marketing material. It is fascinating to consider how, in the 1980s, 'Digital' was largely seen as leading-edge and fashionable, whilst 'Analogue' was often seen as old, tired and a bit unreliable. All of which probably sounds unbelievably inverted in the 21st Century where anything and everything 'retro and analogue' is being re-imagined for a different generation.

So if I had been part of the team designing the TR-505 (I wasn't, just in case the rumour mill gets to work...), then I would probably have suggested making the sample playback rate variable... Or would I? Adding another rotary control would be quite an increment in price, and to ensure that it is possible to play back the samples at the intended rate, there's the problem of making either a quite stable clock oscillator, or having a switch, all of which adds to the price. Given that the main PCB has wire links all over it so that it is single-sided instead of double-sided, then I think that the designers were somewhat constrained by price! I suspect that, as in real life, many of my suggestions would have been added to the 'Nice, but not essential' list rather than the 'Must have' list.

However, more than 30 years later, it is now possible to revisit the design and adjust it. The HKA ROM Expansion shows what you can do when the cost of storage drops, although I suspect that the modern limitations are more to do with the availability of affordable programmers and a lack of a market for very big fixed ROM storage. The flexibility of Flash memory would be a big driver if this was a redesign exercise.

But for sample replay, then my thoughts on a switched clock oscillator are now realisable. The LT1799 chip is a very nice way of generating suitable digital clocks over a wide range, although it tends to be packaged in surface mount, which makes it awkward for modding. Plus I couldn't get the LTC1799 module that I wanted, and had to go with one from a more hobbyist-oriented supplier: Electronics Salon. This isn't a criticism: the module is very nicely put together for its intended market, with a terminal block to make wiring easy, and four solid mounting holes. unfortunately, for my purposes, this rugged, industrial-ish construction makes it large in comparison to the Circuitbender module, and size matters inside a small box.

So I removed the terminal block, which reinforced everything that I said about solder sucking and plated-through holes: not straight-forward. I also removed one of the two multi-turn presets so that it could be replaced with an anti-log potentiometer. You can see the remaining blue one of the right hand side of the module. Just to the left of it, where the shadow ends, is the LTC1799 - which is very small! One of the things that data sheets often seem to mention is that 'wide range of operation' does not necessarily also mean 'linearly-controlled', and in this case, even a log pot doesn't give the right feel. I reckon that a modern design would use a rotary encoder and do the clock division inside the microcontroller chip, and so avoid using the LTC1799 completely.

The blue component that was mentioned earlier is the key frequency-setting component for the sample rate clock generator, and is a ceramic resonator: a low-cost alternative to a quartz crystal when you don't need high levels of frequency stability. But what I wanted was huge levels of instability of frequency, but switchable back to the 'Factory' sample rate, and so I used the resonator and the LTC1799.

Oscillators usually have a resonant filter in a feedback loop around an amplifier, so one side is the output of the amplifier, whilst the other side is the input. What I needed to do was break the connection to the ceramic resonator, and inject the clock generated by the LTC1799 into the input. A switch chooses between the LTC1799 clock and connecting the resonator back in circuit.

You can see the lifted leg of the ceramic resonator in the photo below:

The yellow wire is the input to the amplifier, an the blue wire is attached to the lifted leg of the ceramic resonator. These wires go to a tiny 'Switch' board which has a latching push-switch that does the resonator/clock switching, as well as lighting a blue LED when the clock oscillator is selected.

I then planned out where the switch board and the clock module could actually be fitted. There is a small space to the side of the ROM expansion board, and this is just to the right of the LCD on the top panel of the TR-505, so that's where I positioned the switch board. The earlier test with Blu-tack told me that there was space by the Tape I/O socket, so that's where I positioned the clock module - the multi-turn trimmer makes this board very high, so choosing the right space is important.

These boards need to be fastened onto the main PCB securely, so I added standoffs made from hot melt glue, and used hot melt glue to hold them in place:

You can see these standoffs better with an oblique view from the other direction:

I then put the switch PCB onto the standoff by the side of the ROM expansion, holding it in place with Blu-tack, and used another blob of Blu-tack on top of the switch so that I could see where the switch was positioned on the top panel. There's a sequence of operations that need to be followed when you are doing mods to equipment:

- Plan what to do (circuit, PCB, wiring)
- Disassemble the case
- Find where it will fit
- Test if it fits!
- Temporarily put it in place (Blu-tack!)
- Assemble the case to check that it doesn't touch anything else (Blu-tack!)
- Disassemble the case
- Locate where holes need to be drilled (Blu-tack!)
- Drill the holes
- Fix it in place (hot glue in this case)
- Assemble the case again to check that the positioning is correct.
- Disassemble the case again
- Wire it up
- Test it
- Assemble the case
- Seal the case

Here's the switch board position being tested (you can't see the Blu-tack underneath the board):

The position for the holes in the top panel is then known, so the holes for the switch and LED can be drilled:

Then the top panel placed on top of the lower half of the case to check the alignment of the switch and LED:

This all looks good, so the boards can now be held in place on the standoffs with hot glue, and the wiring up completed:

There are a couple of wire links on the main PCB that have +5V and ground on them, so that's where the red and black wires go to... The two yellow wires from the clock module go to the rotary control on the top panel. I did some labels for the top panel rotary control and switches, and put a piece of paper over the LED because it was too bright!:

(The 'SW' (footswitch) and Headphone labels are available from my store...)

And here's an oblique view of the finished project:


You can hear the pitch-shifted sounds on my SoundCloud...


HKA Design for the TR-505 ROM Expansion Board for the LTC1799 module that I didn't use... (Cheaper and smaller...)

Electronics Salon for the LTC1799 module that I did use... (More expensive and larger!)

My store for labels, etc.


If you find my writing helpful, informative or entertaining, then please consider visiting these links:

Synthesizerwriter's StoreSynthesizerwriter's Store

1 comment:

  1. This comment has been removed by a blog administrator.