OMRON microswitches increasing failure rate and falling mice durability

[ElectronicSupersonic]

While fixing my ~2y old Logitech mouse with random double click issue (on both the left button and the middle button/wheel) came across this (a bit dated) video on the subject of how less and less durable modern mice are getting.

Part one


Part two


The first video is quite long, so TL:DR is, that ever increasing failure rate is due to a combination of lower logic voltages (3.3V vs 5V used in the older designs), lower currents switches deal with and software/firmware failing to account for the above issues.

How legit his explanation of the issue is in your opinion?

[radar_macgyver]

I can confirm that I've had two different Logitech Performance MX mice die within a year of purchase. I replaced the switches on both with ones I bought from Digi-key, and at least one has started to stutter again. These are wireless, and likely use very low power pull-ups.

[dmills]

Switches (If you find a proper datasheet) usually have a lower as well as an upper switching current limit (Also relays, which is a trap for young players doing small signal RF or audio).

You can design switching contacts for microamp currents, but that involves gold plating, so the accountants usually nix that in consumer products (The other way is to use mercury wetted reeds, which has some fairly obvious issues these days).   

Regards, Dan.

[thm_w]

This was linked in a long thread by blueskull, its a very comprehensive and well made video.

Someone correctly pointed out in the comments that the switch used in mice D2FC (omron china) is indeed a "micro-load" switch, it is only rated for 0.1A of current. You can find the datasheet of that switch if you like and verify. So they did not choose the "wrong" switch with respect to that spec.

Still, his main points are all valid:
- They are operating below the minimum current and voltage specification of the switch.
- The difference between the internal construction of the D2FC (china - silver plate) and the D2F-01 (japan - gold plated) are very different. The D2FC clearly being a lower cost to produce part.
- Software polls the switch at 1000Hz which is very fast (but I believe this would only cause the double click issue, not the drag and release early issue).

Some of the middle mouse buttons use D2F style microswitch, some do not (razer uses panasonic EVQ-P0D07K), so might want to check inside if you haven't already.

Didn't realize there was a Part 2, will watch it later.

edit: I looked up spec for EVQ-P0D07 and its rated 2V 10uA minimum switch current (a 330k pullup resistor at 3.3V, so any mouse will use less than that). 1Mil cycles. No mention of contact plating.

[David Hess]

I have also noticed that switch reliability has gotten worse.  I still repair mine.

[Marco]

Maybe it would be better to make microswitches with insulated contacts. Contact failure is far more common than mechanical failure. Measuring switch activation capacitively with AC excitation would be far more forgiving.

[David Hess]

There are all kinds of non-contact switch technologies which could be used with a snap action mechanism for tactical feedback.  But snap action switches are awfully inexpensive and they *can* be reliable if the bean counters do not cheap out.

[all_repair]

There are good ones rated for few million clicks, same Omron.  I was replacing mine at 1.5 year interval, and then sourced and replaced with a good quality one, and no replacement for many years now.

[ElectronicSupersonic]

Thanks everybody for the input!

Some of the middle mouse buttons use D2F style microswitch, some do not (razer uses panasonic EVQ-P0D07K), so might want to check inside if you haven't already.

In my case (Logitech M150 mouse) D2FC are used for left/right buttons and some unknown 4.3mm high tactile switches are used for middle button/scroll wheel. The wheel can both down click and side click in both left and right direction - function that I don't use and which pisses me off, as it makes the usual wheel click less "sharp". I've replaced all the switches with new ones - japan made Omron D2F (though I should have bought D2F-01 instead, oh well...) and Panasonic EVQPAD04M respectively. I'm pretty sure those are all genuine parts since are purchased from TME.

I've also replaced all electrolytic capacitors with Nichicon ones (not that was really needed, but since the other modification I've made required to reposition one of the caps, it was somewhat reasonable to replace all of them). While at it I've decided to remove the side clicking "feature" by replacing the switches with 4.7mm high rubber pads.


Since the right side axel of the wheel broke off (another sign of cost cutting) it had to be fixed. I've used a spring pin from an old notebook ODD. By cutting it to the length and then placing it inside the existing thorough hole in the wheel's axel (had to widen it a bit with 1.6mm drill bit).


Also, since the wheel need to be removed to fix it, I've decided to add an additional spring to make detent action more pronounced. Black spring is the original one, silver one is the additional one.


Another modification that was done a while ago is adding more weight to the mouse (although I'm not in to gaming, but I like heavier mice). This was accomplished by adding more metal plates from recycled Logitech mice. That was an easy mod since everything more or less fit together. Though one of the electrolytic capacitor was in the way and had to be put sideways (the original cap had it's legs cut too short hence new one was needed).


As a side note pads on the PCB are too easy to lift (especially NC pads - this could be easily made large in cost of the copper fill). My soldering skill aren't that high, but nonetheless... Had to make few fixes after switches were removed. As it turns out the best way to remove components was heating the solder joints with hot air, removing the part, then clean the pads with solder wick. To bad I've discovered that method when damage was already done (oh well). In the pic bellow bottom left switch was removed with solder wick, while the bottom right one was removed with hot air. The difference is obvious (IMO).


Here are few pics of switches used (original vs new ones; tactile switch with white actuator is new one - brown one is original).


Mulling other the linked video, it had occurred to me, that by using a simple circuitry consisting of few cheap components a fix can be devised. In essence (as I see it) what is needed, is to increase the voltage/current value to match the values recommended in the datasheet. If we connect the circuitry in between the switch an the pad coming from controller IC, it should provide the recommended values (about 1mA at 5V; for micro load models), thus increasing usable life of a device. The proposed circuitry is given below. Note that I'm not an engineer, thus it might be a non working "solution" (read utter BS).

[thm_w]

Thats almost better to see a generic 4 pin tactile switch, there are tons of alternative options for a replacement.

Mulling other the linked video, it had occurred to me, that by using a simple circuitry consisting of few cheap components a fix can be devised. In essence (as I see it) what is needed, is to increase the voltage/current value to match the values recommended in the datasheet. If we connect the circuitry in between the switch an the pad coming from controller IC, it should provide the recommended values (about 1mA at 5V; for micro load models), thus increasing usable life of a device. The proposed circuitry is given below. Note that I'm not an engineer, thus it might be a non working "solution" (read utter BS).
(Attachment Link) (Attachment Link) (Attachment Link)

Sorry I'm not completely sure what your schematic is showing, as it has the MCU pullup going to ground.
I think adding the resistor to USB Vcc is a viable solution as you've shown.

The question is, is it actually the problem or not. Until your switch fails, then you perform this pullup mod, we won't know for sure. Its a hard task as once the mouse is open I would be too tempted to replace the switch..   

[ElectronicSupersonic]

Sorry I'm not completely sure what your schematic is showing, as it has the MCU pullup going to ground.
I think adding the resistor to USB Vcc is a viable solution as you've shown.

Sorry it's indeed confusing. That resistor (MCU pull-up) is meant to show an internal pull-up/current limiting resistor inside the MCU. My understanding is that there has to be something inside the MCU to limit the current when the corresponding signal is pulled low by clicking the switch. Since the circuit was drawn in a simulator, in order for the simulation to work, some current limiting resistor had to be present. I have no idea how to show that internal MCU resistor hence it's shown like this.

Basic idea behind the circuitry is very simple. Since these Omron switches should work better/last longer when certain conditions specified in the datasheet are met, we need to somehow provide these conditions. These conditions (AFAIK) are current/voltage ratio. MCU might not provide the above conditions.

Say this MCU uses 3.3V logic and can only provide 0.5mA(random number), which is not what the switch is expected to handle properly. If it's an USB device (wireless mice is whole another story) when it has ~5V supply. According to D2F datasheet 1mA@5V is within recommended conditions (for D2F-01 variant). Thus if we configure the switch to use 5V instead of 3.3V MCU provides (presumably) and limit the current passing through that switch to ~1mA, we can potentially get better reliability/longevity.

Since the switch is no longer directly connected to MCU and both are now using different voltage levels, we need to somehow make the two communicate with each other so the device would function properly. In order to do that, a P-channel MOSFET is used. Gate of the above P-MOSFET is controlled by the switch, while its source is connected to the same MCU pin the D2FC switch was connected originally (but no longer is) and lastly the drain is connected to GND. When the switch is in its normal condition, it is open and thus does not pass any current, therefore the gate of the MOSFET is driven HIGH at about 5V (via 3.3K resistor) thus turning it OFF and hence it does not pass any current (apart of the leakage current of course). When one clicks the switch, the gate of the MOSFET is pulled LOW thus turning the MOSFET ON and therefore pulling the corresponding MCU pin LOW this subsequently should register as a click. In other words this MOSFET would perform the same function D2FC switch did (i.e. pulling the corresponding pin LOW).

To put it simple D2F switch would now work with 5V and by controlling the gate of the MOSFET, would control the corresponding pin of the MCU.

[ElectronicSupersonic]

well i opened up mine, and found the leaf spring not centered ha !

It seems that the root cause of my mouse double-clicking was the same as in your case, because after smashing (lightly) few times the left side of the mouse it appeared to be working again (for a while), but then problem returns.

...if i understand correctly, most marketing materials are stating mechanical life, not electrical life. ...

It's true pretty much for all datasheets (regardless of the component type). Usually only the best parameters are stated in the banner specs (i.e. in the first page of a datasheet). Although these banner specs are mostly true, those parameters rarely represent most usable/useful operating conditions (say max current in MOSFETs where the banner specs may be valid for switching mode only when the MOSFET is OFF for most of the duty cycle).

... to reduce the click noise. i do suppose it could be done to D2F.

These new D2F switches aren't that loud to be an issue IMO. Although clicking noise can be an issue with mechanical keyboards. Especially when you have few people in the room (even more so if they all use the same noisy mechanical keyboards and are typing all day long). For this reason I use keyboard with the same type of switches they use in notebooks - an Apple aluminium wired keyboard (arguably the most durable computer related product Apple has ever made).

[amyk]

If I was going to take apart my mouse to replace the switches I would probably try to mod it with an optical switch (phototransistor/photodiode-based) and never have problems again.