Daskeyboard Pro 4 and Matias Tactile Pro with Bouncing Issues

User avatar
vvp

Unread post by vvp » 27 Dec 2018, 21:34

Muirium wrote: I’d like to see someone take an oscilloscope to some badly chattering Matias switches. What I expect to see is a lot of white noise at the individual switch level.
I do not have any broken switches at hand :)
I have only Cherry MX and Gateron switches and all of them are good.
Here are some scope files from a key press event of a well working Cherry MX Blue switch:
500us.png
500us.png (2.21 KiB) Viewed 408 times
200us.png
200us.png (2.49 KiB) Viewed 408 times
20us.png
20us.png (3.22 KiB) Viewed 408 times
You can see that the switch stabilizes in 2 ms. My keyboard uses 4 ms debounce time now and the polling time is 2 ms. It has about 8 ms delay at worst. I debounce both key-down and key-up events.

A csv file is attached if somebody wants to look at it more.
Attachments
cherryMx.csv.zip
(1.96 KiB) Downloaded 1 time

User avatar
Muirium
µ

Unread post by Muirium » 27 Dec 2018, 21:37

Cool. And so noisy! Now I really want to see a chattery Matias comparison.

User avatar
Inxie

Unread post by Inxie » 27 Dec 2018, 21:42

I could get you a sample from my specimen maybe next month. I'll be attempting to retrieve it, since I do have a scope, but it's an older Tektronix from the 80's. It'd work though..

squizzler

Unread post by squizzler » 28 Dec 2018, 09:44

Image
Inxie wrote: I kept that email, because that was absolute proof that Matias doesn't care about your keyboard, they care about your money rather than deliver a quality product.

I am more than happy to forward that email to anyone who wants proof that this happened if proof is every required. Matias failed to honor the repair on my keyboard, and forced my hand.
That I'm not entirely sure. The keyboard is 3 hours away at my old house I was evicted from. I could try to get it back maybe in a month or so, but the keyboard was manufacturered around 2012.
I would like to see the email exchange that led to this response from Mattias before I think ill of them as a business. Or indeed to back up your version of events: as a new forum member I don't know the reason behind your being evicted from the previous address and the email conversation to this point would hopefully put our minds at ease you are not a habitual troublemaker :)

User avatar
vvp

Unread post by vvp » 28 Dec 2018, 13:56

Expecting some kind of warranty on a keyboard from 2012 is unreasonably optimistic :shock:
I'm surprised the support team did respond at all.

Edit: Actually their response is reasonable and to be expected from a good sales guy. They are trying to sell you a new keyboard; hoping to placated you with a discount and still earning some income. Good approach. Better than not responding at all.

User avatar
Inxie

Unread post by Inxie » 28 Dec 2018, 14:04

vvp wrote: Expecting some kind of warranty on a keyboard from 2012 is unreasonably optimistic :shock:
I'm surprised the support team did respond at all.

Edit: Actually their response is reasonable and to be expected from a good sales guy. They are trying to sell you a new keyboard; hoping to placated you with a discount and still earning some income. Good approach. Better than not responding at all.
vvp, why not look at the date of the response, that was back in 2013, the keyboard was only a year old (+a couple months). THAT'S unreasonable? No, that's very much rude and telling the customer that they know their product only lasts a couple years before people buy replacements.
Last edited by Inxie on 28 Dec 2018, 14:13, edited 1 time in total.

User avatar
Inxie

Unread post by Inxie » 28 Dec 2018, 14:11

squizzler wrote: Image
Inxie wrote: I kept that email, because that was absolute proof that Matias doesn't care about your keyboard, they care about your money rather than deliver a quality product.

I am more than happy to forward that email to anyone who wants proof that this happened if proof is every required. Matias failed to honor the repair on my keyboard, and forced my hand.
That I'm not entirely sure. The keyboard is 3 hours away at my old house I was evicted from. I could try to get it back maybe in a month or so, but the keyboard was manufacturered around 2012.
I would like to see the email exchange that led to this response from Mattias before I think ill of them as a business. Or indeed to back up your version of events: as a new forum member I don't know the reason behind your being evicted from the previous address and the email conversation to this point would hopefully put our minds at ease you are not a habitual troublemaker :)
Sure.

https://i.imgur.com/CZFA994.png

At the time I thought the keyboard was a few years old as I had forgot the receipt (and bought it on Best Buy .com), but I did later find it and saw it was purchased August of 2012. Being now this is 5 years ago, I do not expect a replacement at this time, but what I want shown is how they treated customers.. Their site before I bought it, was indeed saying "Built to last", which I brought up to them. After this email correspondence, that line was removed from their site. I believe Best Buy .com's version still said it for a few more months before that changed?

User avatar
vvp

Unread post by vvp » 28 Dec 2018, 14:20

Inxie wrote: vvp, why not look at the date of the response, that was back in 2013, the keyboard was only a year old (+a couple months). THAT'S unreasonable? No, that's very much rude and telling the customer that they know their product only lasts a couple years before people buy replacements.
Good point. I did not realize we may discuss here something which happens in 2013. That sales guy probably does not even work for Matias now. It's a too old incident. The only point is that their support sucked in 2013. It may be OK now.

Well, if the keyboard was less than 2 years old and you are in EU then you should have contacted your local consumer protection agency about Matias' failure to fix or replace the keyboard. The agency would solve it for you and you would not have problem with a broken keyboard. If you are in USA then you may be unlucky because the minimum warranty there may be only 1 year.

User avatar
Inxie

Unread post by Inxie » 28 Dec 2018, 14:29

vvp wrote:
Inxie wrote: vvp, why not look at the date of the response, that was back in 2013, the keyboard was only a year old (+a couple months). THAT'S unreasonable? No, that's very much rude and telling the customer that they know their product only lasts a couple years before people buy replacements.
Good point. I did not realize we may discuss here something which happens in 2013. That sales guy probably does not even work for Matias now. It's a too old incident. The only point is that their support sucked in 2013. It may be OK now.

Well, if the keyboard was less than 2 years old and you are in EU then you should have contacted your local consumer protection agency about Matias' failure to fix or replace the keyboard. The agency would solve it for you and you would not have problem with a broken keyboard. If you are in USA then you may be unlucky because the minimum warranty there may be only 1 year.
Yea I'm in the US. We have Attorney General's here which usually can help with these kind of situations actually, but at the time in Florida, we didn't exactly have one that was all that good, so I never bothered, I wrote it off and got a Das Keyboard.

Granted, while the Matias is still super crummy to type on today as per my YouTube video demonstrating 2018 usage, the Das Keyboard, while it does have some bouncing issues, is still tolerable, and can be used day to day, and seems to change based on how I type on it. I can't do any typing tests with it though, I'll get errors when I usually don't have errors. My AT101W nets me at almost 90wpm with no errors (that's if the cheap PS/2 to USB adapter doesn't reset in the middle of my test, apparently I type "too fast" for the silly thing).

User avatar
Inxie

Unread post by Inxie » 28 Dec 2018, 14:37

squizzler wrote: as a new forum member I don't know the reason behind your being evicted from the previous address
I suffered a pulmonary embolism last December in 2017, and lost any job I had as a result and no longer could afford the place, since you so kindly wanted to know.

User avatar
Muirium
µ

Unread post by Muirium » 28 Dec 2018, 16:55

Inxie! Every single one of your keyboards would drive me insane. I literally cannot take failure like that, and would (probably figuratively but then again maybe not) smash them to a shredded plastic pulp if the fuckers acted up like that on me. How does anyone stand guff like this? I don’t mean condescending interactions with “support” but rather the keyboards themselves. Gimme keyy chhattter and I shall sunder unto you almighty pain and just furry. Gaah!

Replace that dud converter with a rock solid Soarer’s converter, such as one from Orihalcon. You’ll thank yourself when you do. And, yeah, torture the failed Matias and Das for the insidious demons that they are.

User avatar
Inxie

Unread post by Inxie » 28 Dec 2018, 17:01

I have a feeling the Das can be saved, despite one key losing its click (so that's a thing apparently MX blues can stop clicking and being tactile--and a key I never used either, right-alt key). It's chatter is far less of a problem compared to the chatty kathy Matias. The Das I think can be saved, because it's isolated down to just a couple keys that do it. The Matias, EVERY key does it, except caps lock ironically.

User avatar
Muirium
µ

Unread post by Muirium » 28 Dec 2018, 17:08

ThE IROny iS STRong wITh tHIS One.

You’re even seeing it on the other garbage keys like scroll lock?

User avatar
XMIT
[ XMIT ]

Unread post by XMIT » 28 Dec 2018, 17:22

vvp wrote: [...]for some things analog switches would be great. E.g. they could work as analog joystick axe when some modifiers are pressed.[...]
You're not the first with this idea. Have a look at HaaTa's KLL Spec.

User avatar
Inxie

Unread post by Inxie » 28 Dec 2018, 17:26

Muirium wrote: ThE IROny iS STRong wITh tHIS One.

You’re even seeing it on the other garbage keys like scroll lock?
Well yea, those are still standard switches. The caps lock key on my Matias is not clicky and is linear as it has the LED on it, and has a different construction. I'd go as far as to say the clickyness is what's causing the chatter itself on that board.

squizzler

Unread post by squizzler » 28 Dec 2018, 18:07

Inxie wrote: Sure.

https://i.imgur.com/CZFA994.png

At the time I thought the keyboard was a few years old as I had forgot the receipt (and bought it on Best Buy .com), but I did later find it and saw it was purchased August of 2012. Being now this is 5 years ago, I do not expect a replacement at this time, but what I want shown is how they treated customers.. Their site before I bought it, was indeed saying "Built to last", which I brought up to them. After this email correspondence, that line was removed from their site. I believe Best Buy .com's version still said it for a few more months before that changed?
From the previous conversation I agree you have not said anything especially provocative in your discussion to warrant the apparent tone from support. I suspect that the latter is due to cumbersome wording than any animosity towards you and would be wary saying it is indicative of their attitude to customers generally.

The situation is unfortunate, and the reason to use a reputable firm like Matias is a belief that the product will prove durable in the longer term. You pays your money and makes your choice. I hope my Mattias switches last better, but at the end of the day I guess if it becomes unusable in the same way as yours has done one has to just put it down to experience.

For whatever reason I found this technique worked with mine (at the time of writing:). I also shared this on the other post relating to matters switches. I would be interested if others having problems found it successful:
Sorry to hear you're having problems, I would fold the keyboards legs
away and try giving the key a good massage.

Press the key down and push the key firmly towards the back of the
keyboard then towards the front, then side to side, also do this in
circular motion, do this a lot.
Last edited by squizzler on 28 Dec 2018, 18:20, edited 1 time in total.

squizzler

Unread post by squizzler » 28 Dec 2018, 18:18

I think a way of addressing the problem with current switches might be to make the board with user adjustable anti-bounce. Have a small potentiometer (possibly screwdriver headed) hidden under the board where you can choose how much anti-bounce (would this come at the expense of response time?) you require. If the keys started to chatter, simply turn the dial up a little.

Another thought Matias might look at would be to adapt the design to make it capacitive. I was enquiring about Varmilo EC switches (before addressing the chatter on the Matias ones) and was advised they were like the Cherry designs, only with leafs that are made to work contactlessly. If Matias switches could be similarly reworked the feel that many of us enjoy could be retained but the bounce issue could be eliminated. Would it be difficult to re-engineer the switch in this manner? According to Keyboard Company of England:
What they did, which is clever, is took a Cherry MX and made it capacitive, simply re-designing the contact to 'not touch'. So it's not really that different, and a Cherry MX will work forever anyway.
Adding to the automotive metaphors relating to Matias already used on this thread, they say of the Matias switch:
...we like to keep it going because that is such a tactile switch and people really love them.
It is mechanical in the extreme, and like a Morris Minor, it has it's faults. But we love them.

User avatar
vvp

Unread post by vvp » 28 Dec 2018, 18:19

XMIT wrote: You're not the first with this idea. Have a look at HaaTa's KLL Spec.
It is an obvious use case of an analog switch. It would be depressing if I would have been the first one :D

Is an analog switch only in a spec or is there some prototype and user feedback how well it works as well?

E.g. another alternative is a switch which could be tilted as a whole. The tilt extend could be sensed. The switch was dual role: a simple switch and a track point as well. This did not catch on. It is important to get also some real use feedback as well.

User avatar
vvp

Unread post by vvp » 28 Dec 2018, 18:38

squizzler wrote: I think a way of addressing the problem with current switches might be to make the board with user adjustable anti-bounce. Have a small potentiometer (possibly screwdriver headed) hidden under the board where you can choose how much anti-bounce (would this come at the expense of response time?) you require.
It is easier to do this in firmware. Trying to do it in hardware would mean a LOT of potentiometers or multiplexers/de-multiplexers. That is an expensive option.

I added this in my firmware. Chrod Prg-Tab increases denounce time by 2 ms and Prg-Esc decreases it by 2 ms. But as I mentioned before this works only so far. If the switch is really bad then the debounce time would need to be so high that one would not by able to type quickly. E.g. you need debounce time below about 17 ms to be able to type 70 wpm. Otherwise the keys may not register at all when typing quickly.

User avatar
XMIT
[ XMIT ]

Unread post by XMIT » 28 Dec 2018, 21:06

vvp wrote:
XMIT wrote: You're not the first with this idea. Have a look at HaaTa's KLL Spec.
It is an obvious use case of an analog switch. It would be depressing if I would have been the first one :D

Is an analog switch only in a spec or is there some prototype and user feedback how well it works as well?

E.g. another alternative is a switch which could be tilted as a whole. The tilt extend could be sensed. The switch was dual role: a simple switch and a track point as well. This did not catch on. It is important to get also some real use feedback as well.
I'd put an accelerometer in the keyboard, and have the "shake keyboard" gesture be the same as Backspace. Sort of like erasing an Etch-a-sketch. :lol:

User avatar
Muirium
µ

Unread post by Muirium » 28 Dec 2018, 21:50

squizzler wrote: Adding to the automotive metaphors relating to Matias already used on this thread, they say of the Matias switch:
...we like to keep it going because that is such a tactile switch and people really love them.
It is mechanical in the extreme, and like a Morris Minor, it has it's faults. But we love them.
The trouble with this metaphor is what people really love is vintage Alps. Matias switches are dodgy clones of the late stage Alps design that had been through several generations of cost cutting and quality reduction. I know nothing about Morris cars, so I can’t adapt that analogy, but I can say that making users suffer so that Matias can simply continue to ship defective as designed switches is neither cute nor heroic. It’s shitty, pure and simple.

As for a redesign to switch over to analogue sensing: that’s a lot of work. May well not be tenable at all with the Alps shell and slider constraints. Just because marketers toss it off as a simple stroke of Turns Out design genius, does not make it a true description. They write to promote desire, not to spread the inside truth. That’s their job. Now if only Matias’s’s switch factory and designers would do theirs instead of leaving it to the hype department to tell us how to eat shit and like it.
Last edited by Muirium on 28 Dec 2018, 21:51, edited 1 time in total.

User avatar
depletedvespene

Unread post by depletedvespene » 28 Dec 2018, 21:51

Muirium wrote: You’re even seeing it on the other garbage keys like scroll lock?
Careful with those kind of comments, Mister... :roll:

User avatar
Muirium
µ

Unread post by Muirium » 28 Dec 2018, 21:53

Scroll lock is as useless as F14 for me*. Which is in fact what it maps to…

*This subjectivity being the most relevant part. We all use different keys. For me, GUI is essential, right up there with Shift, yet many people here rock winkeyless boards without a fuss or a remap. (Several of my own favourite boards are winkeyless, but I absolutely have to remap their mods for my use.) Caps Lock is the one we mostly do all sneer at. And Scroll Lock, well, is an oddball with its small group of fans: among which, I am not.

User avatar
vvp

Unread post by vvp » 03 Jan 2019, 18:01

Files from a faulty Gateron linear switch.
When pressing the switch quickly then it is almost stabilized in about 20 ms. Although some disturbance to the signal is still there after 55 ms.
bad-5ms.png
bad-5ms.png (2.42 KiB) Viewed 179 times
bad-2ms.png
bad-2ms.png (2.56 KiB) Viewed 179 times
When pressing it slowly then the situation is much worse. There are still major errors after 45 ms.
bad-slow-5ms.png
bad-slow-5ms.png (2.47 KiB) Viewed 179 times
When pressing it even more slowly then there are still major errors after 80 ms.
bad-slower-10ms.png
bad-slower-10ms.png (2.46 KiB) Viewed 179 times
You want a switch to stabilize in 15 ms at worst. Otherwise debouncing would interfere with typing. The maximum debouncing time is 5 ms typically.

User avatar
Muirium
µ

Unread post by Muirium » 03 Jan 2019, 18:28

Ouch! A whole order of magnitude worse than your MX sample.

For those following at home, 5ms = 1/200th of a second, 10ms = 1/100th, 50ms = 1/20th, and 80ms = 1/12th. Typing needs much higher resolution than tenths of a second. The white noise that’s pouring out of these switches is swamping the signal and wreaking havoc. What’s a controller to do when the switches can’t do their job?

User avatar
Inxie

Unread post by Inxie » 03 Jan 2019, 18:49

Yea that's absolute garbage.

User avatar
vvp

Unread post by vvp » 03 Jan 2019, 21:07

Muirium wrote: What’s a controller to do when the switches can’t do their job?
The only thing a controller can do is increase the debounce time (easy) or improve the debounce method.

The most simple debounce method just requests that N sequential reads of the switch return the same value. E.g. when reading the matrix is done every 1 ms then after 5 reads of 1 from one switch (that means after 5 * 1ms = 5ms) the firmware assumes the switch is pressed. This will not work well with the example of faulty switch.

One can implement some more sophisticated debouncing e.g. using a finite (or infinite) response filter to get an analog value from the signal and then comparing the analog value to an upper and lower treshold to add some hysteresis to it. This wold ensure that small faults(*) at the later stage of debouncing would not matter. It is possible (but not likely?) that the faulty switch I posted can be made to work well enough with proper tweaking of the response filter and the two tresholds. Although I think it is not worth the trouble. It would require a quicker MCU. Replacing a switch is not such a big deal.

(*) the small spikes down later in the posted samples

User avatar
Inxie

Unread post by Inxie » 03 Jan 2019, 22:32

"Replacing a switch is not such a big deal" : It is when the whole board is full of them though. When I say my Matias has ALL keys failing with this issue, I'm not joking.

squizzler

Re: Daskeyboard Pro 4 and Matias Tactile Pro with Bouncing Issues

Unread post by squizzler » Yesterday, 18:02

Have you tried the method from UK's Keyboard Co. in my earlier post (re-quoted below)? I have found it to work for me and would love to know I'm just lucky or whether it really works in most cases. I'm surprised if it were an effective method that nobody else in a forum of keyboard enthusiasts has mentioned it thus far (either here or in another long Matias thread).
Sorry to hear you're having problems, I would fold the keyboards legs
away and try giving the key a good massage.

Press the key down and push the key firmly towards the back of the
keyboard then towards the front, then side to side, also do this in
circular motion, do this a lot.

User avatar
Muirium
µ

Re: Daskeyboard Pro 4 and Matias Tactile Pro with Bouncing Issues

Unread post by Muirium » Yesterday, 18:13

Well, see if it *keeps working* every time the switches degrade. There’s more going on than something turning wobbly. They break down electrically, which no amount of manual wiggling will ultimately reverse.

Post Reply

Return to “Keyboards”