# Unusual Rotary Encoder Output

I'm trying to replace a rotary encoder with push switch from a Boss SE70 effects unit as the switch is broken; the same encoder is apparently used in a number of Roland products such as the JD-990, JV-1080, SC-880 and JV-880 (all '90s vintage I think).

This type of repair is normally straightforward: find something that is roughly the same mechanically and electrically. The part is listed as an RK09710WL, with Roland part number 13289196. I can't find any data on this part other than it was made by Alps (marked on the base). It has 20 detents.

I chose a Bourns PEC11R-4230F-S0024 as it was the right length, had a switch and had roughly the same amount of pulses (24 rather than 20 on the original) and detents. It even had pretty much the same footprint.

However it does not work. It seems that the encoder used by Roland gives a very different output to commonly used rotary encoders.

With the replacement encoder installed (that does not work correctly), the output from the two pins is as expected (direction reversed half way) for most rotary encoders:

With the standard (working correctly bar switch) encoder in place, the output is:

The stock encoder sends a short pulse (<50ms) on one pin only when turned in one direction and a pulse on the other pin only when turned in the opposite direction.

I've never seen a rotary encoder with this kind of output - does anyone know what it is and how to find a replacement? The dimensions are not too important as the encoder mounts on a small PCB (no other components on the board) and is chassis mounted so anything can probably be made to work, but that unusual output is needed.

Here is the excerpt from the schematic:

The rotary encoder lines connect to pull up resistors and debounce/filter resistors and capacitors, then to microprocessor pins as expected.

Can anyone provide any help, information or guidance?

• I'm noticing, but perhaps over-thinking it, that the width of the pulses correlates with the rate of rotation vs time -- slower rates yield wider pulse widths. This is perhaps useful information used by Roland. It might change the "feel" to always use the same pulse width. Do you notice the same here? Or is this just me? Can you consider a full break-down of one of these switch systems? – jonk Nov 16 '18 at 21:13
• The width of the pulse probably does change with the speed of turning, but the knob is only used to scroll through presets and so on, so is only used in quite a basic way. It is tempting to dismantle it, but it (the switch) works enough that I don't want to do that if I can't find a replacement. It isn't held together with tabs, so would be destroyed. – Duntle Nov 16 '18 at 22:01
• that is a rotary pulse switch ...... see comment from @Justme here electronics.stackexchange.com/questions/404734/… – jsotola Nov 17 '18 at 7:47
• Excellent information - does anyone know where such things can be bought or even anyone that makes them? – Duntle Nov 17 '18 at 9:53
• It seems these things are very few and far between and there is not necessarily any specific name for them. Searching for rotary pulse switch often finds rotary encoders. Alps sell a couple and call them rotary switches: SRBM1L0800 & SRBM1L1400, but neither have a push switch as well. – Duntle Nov 17 '18 at 14:56

Apparently, that encoder has built-in circuitry

1. to debounce the signal edges;
2. to detect the direction (by reading the state of the other line when an edge happens on the first one); and
3. to generate a short pulse.

Nowadays, all this can be done in a microcontroller, so it is unlikely that a direct replacement is still made.

It would be possible to build your own circuit to do this (with RC filters, logic gates, and a monostable (555 or 74xx123)), but a microcontroller board might be easier.

• It's a fun exercise to do that one with a 555, a JK flip-flop and some gates. I've done it on paper before but never for real. There may actually be a chip out there that does that, but I have no clue what name to look for (aside from "itty bitty microcontroller with custom programming). – TimWescott Nov 16 '18 at 21:36
• It does seem like there is some processing going on in the encoder, but is that possible given it is no bigger than a normal encoder and has no power. The debouncing is done outside the switch and it then connects to a microcontroller or microprocessor (custom Roland chip), which surely could also do all decoding of "normal" signals. It all just seems a bit odd. – Duntle Nov 16 '18 at 21:58
• Hmm, without power, all I can imagine would be some mechanical contraption. But that makes it even less likely to find identical hardware. – CL. Nov 17 '18 at 8:27