Synthesizer Key Contact Resistance Measurements

I have been fixing up a Kawai SX-240, which was missing its keybed. I bought one from a Poly61 (essentially the same keybed), removed the rubber contact strips and cleaned the carbon pills and the gold contacts on the pc board with Q-tips and isopropyl alcohol. With the keybed installed in the SX-240, I discovered that several keys did not trigger notes. So I ordered some silver keypad repair paint. While reviewing some youtube videos on key contact repair, I found a post claiming good results from taking fine steel wool to the pcb gold contacts instead. I decided to do a test, measuring contact resistance before and after giving the gold contacts a better clean, so with my Fluke DMM on the diode test setting, I measured and recorded the "key contact plus diode" voltage across all 61 key contacts with each key pressed. (Diodes are used in the keyboard circuitry to prevent "phantom keys" from appearing when multiple keys are pressed, so if you measure at the connectors, you always have one diode in series with the closed contact) Then I removed the rubber contact strips, keeping track of where they had been. I examined the gold pcb contacts under a microscope and noticed some very small specks of red-colored crud on some of them. So I used a gum eraser to clean all of the gold pcb contacts, then vacuumed away all of the little bits of eraser that had rubbed off. I did not re-clean the pcb contacts with isopropyl, and did not clean the rubber contact strips at all. After re-installing the rubber contact strips, I re-measured the "key contact plus diode" voltages.

Before the cleaning, the minimum voltage I measured was 715 mV. Here is how the voltages (in mV) were distributed:

700-749 9 keys

750-799 18

800-849 10

850-899 10

900-949 8

950-999 2

1070 1

1090 1

1180 1

1360 1

After only cleaning the gold pcb contacts, here is the distribution:

690-699 4 keys

700-709 14

710-719 16

720-729 17

730-739 7

740 2

760 1

So all of the contacts except one were less than 741mV after cleaning. A surprising improvement. So what sort of contact resistances are we talking about here? Using the same DMM, I determined the resistances needed in series with the diode to get particular mV readings:

82 ohms 714 mV

100 ohms 722 mV

150 ohms 743 mV

243 ohms 784 mV

301 ohms 808 mV

470 ohms 875 mV

750 ohms 978 mV

1500 ohms 1201 mV

2320 ohms 1403 mV

So it seems that after the cleaning, most of the contact resistances were around 80-100 ohms. Before cleaning, the four worst keys were over 1K resistance, with one around 2K. In the SX-240, I measured the key contact threshold as around 1275 ohms. Values below this allow the key press to be detected, higher values prevent the key from working. (This threshold depends on the pullup or pulldown resistor used in the synth. SX-240 uses 3.3K. Poly6, 61, and Multi-Trak use 10K instead of 3.3K. The Prophet 600 and Six Trak use 22K. Larger pullup resistor values are more forgiving of excessive contact resistance)

My best guess is that very small clumps of crud on the pcb contacts can prevent the carbon pill from sitting flat on the pcb contact, which reduces the contact area and raises the resistance. But it's just a guess. Anyway, it may be worth cleaning the pcb gold contacts with something more abrasive than a Q-tip and alcohol before resorting to the silver paint.

Postscript: A few days after I wrote this, I discovered that one key was no longer triggering a note. So I re-measured about half of the key contact voltage drops and found that they were pretty much the same as I had measured the previous time. When I got to the key that was not working, the voltage was much higher than the other keys. Picked up the rubber dome and found a big flake of something white right in the middle of the carbon pill. So I removed all of the rubber strips and re-cleaned them, using the vacuum to suck up any little specks of crud that I saw. Anything on the surface of the carbon pill can make the key not trigger. Probably if the carbon pills were covered with the silver paint, their resistance would be much lower, so a much smaller contact area would be required for triggering, and more dirt could be tolerated.

First time installing new contact strips:

I recently got a later-version OB-8 which uses a similar "ESK" keybed to the SX-240, Poly61, etc. Most of the keys did not trigger. I cleaned the pcb and the rubber dome carbon pills with isopropyl, which made a slight improvement, but most of the keys were still silent. Measurement of the "contact plus diode" voltage with a 4.7K pullup (as in the OB-8) showed many keys to be in the 3-4 volt range when pressed. This time I decided to replace the rubber key contact strips with new ones from LA Synth Co. With the new ones installed, I repeated the voltage measurement and found all but a few of the voltages to be less than 1 volt with the key pressed. There were two with voltages between 2 and 3 volts. I removed that contact strip and cleaned the pcb again with isopropyl, pressing a little harder this time. That fixed those two keys and got their voltage down to around 1 volt. So even with the new contact strips, it's important to get the contacts on the pcb clean and free of anything that will keep the carbon pill from making good contact. Hopefully the new contact strips will work reliably for many years.