CHERRY - Innovating the World's Best Keyboards

[SpAmRaY]

https://www.cherry.de/cid/RealKey_Technology.htm?cpssessionid=SID-EECDB16A-B63BC481

[Bromono]

Accurate, fast, pure, what does that even mean 

not the best translation. Think they where going for clean. Like the look is clean.

[engicoder]

Analogue sensing for n-key rollover is not new. Microsoft has a patent for doing it with resistors. A guy who worked on it has posted here on GH about it. It is used in SideWinder X4.

Paul Dietz. Interesting fella.

The Cherry MX switch still needs to be debounced. It is not like there is an analogue curve from unpressed to pressed.

Well, if each switch is connected to GND and has a small capacitor in parallel, discharge (switch closes) could be almost instantaneous, recharge (switch opens) may take a bit longer (due to bounce, although opening is less bouncy.) Keeping track of the capacitors would be quite a task tho. Looking forward to the first teardown of a 6.0 board...

You'd  really need a classic RC debounce circuit spec appropriately for the debounce time. I noted earlier that they could be doing this and then using an ADC to read the voltage and detect switch press and release using and arbitrary point on the discharge and recharge curves. This could explain their separate line for each switch comment in the video.

[Bakgrund]

"up to 20ms", correct me if I'm wrong but don't many keyboards have 1000hz polling rate? Wouldn't that make them take 1ms to detect?

[Grendel]

You'd  really need a classic RC debounce circuit spec appropriately for the debounce time. I noted earlier that they could be doing this and then using an ADC to read the voltage and detect switch press and release using and arbitrary point on the discharge and recharge curves. This could explain their separate line for each switch comment in the video.

Very plausible, my guess is they added an ASIC (ie. the "analog controller" talked about in below's link) to do just that.

https://www.cherry.de/cid/RealKey_Technology.htm?cpssessionid=SID-EECDB16A-B63BC481

Looks like that page is in construction :/ Very vague too.

[Evo_Spec]

Details of the board here. $219 MSRP...

I wouldn't call that horrible pricing since it does have an aluminum top but I'm betting it has a plastic bottom, definitely no notches...

[FrostyToast]

All I got from that is that they remembered PS/2 existed...

[jacobolus]

"up to 20ms", correct me if I'm wrong but don't many keyboards have 1000hz polling rate? Wouldn't that make them take 1ms to detect?

They actually said "most conventional keyboards today need an average time of about 20ms", implying a scan every 40ms, i.e. 25 Hz polling rate. Anyone know what the typical scan rate is on a $10 rubber dome board or scissor switch board? Those are going to account for "most conventional keyboards today".

[Zeal]

Here's the 1:20 video showcasing the keyboard render. I don't think anyone posted it, but I had it loaded before it went private.

For those that can't see the embedded vid:

[Grendel]

"up to 20ms", correct me if I'm wrong but don't many keyboards have 1000hz polling rate? Wouldn't that make them take 1ms to detect?

1ms to detect that a key started changing, now needing debounce time. Sloppy firmware may do a 10ms debounce of the whole matrix when a key-change is detected which could lead to the worst case of up to 20ms for another key actuating at the beginning of that debounce to be send. Not the norm tho. Most higher end boards do individual key debounce taking 6-10ms.

[Grendel]

"up to 20ms", correct me if I'm wrong but don't many keyboards have 1000hz polling rate? Wouldn't that make them take 1ms to detect?

They actually said "most conventional keyboards today need an average time of about 20ms", implying a scan every 40ms, i.e. 25 Hz polling rate. Anyone know what the typical scan rate is on a $10 rubber dome board or scissor switch board? Those are going to account for "most conventional keyboards today".

Rubber domes are a lot more "bouncy" than other switch types and need a longer debounce time (scan rate is probably at least 200Hz even for RD's.)

[jacobolus]

Rubber domes are a lot more "bouncy" than other switch types and need a longer debounce time (scan rate is probably at least 200Hz even for RD's.)

That’s not the implication of their video, but I can totally believe that the message from the engineers got totally twisted around by the time the marketing people got to it.

Their video shows little red lines going across a scan matrix, and a clock ticking up, and basically directly implies that when a key is pressed it doesn’t get scanned for an average of 20ms. That’s quite obviously false, was my point.

* * *

Their longer video has a different engineer saying that the in the worst case it takes 20ms to get from keypress to the computer finding out about it, a much more reasonable statement. My guess is that there were 2 points from the engineers: (1) the matrix needs to be polled, and this can be slow, and (2) sometimes it might take 20ms to get a signal to the computer. Then when marketing got that, they mixed the two together into one silly sentence and animation, and never passed it back by engineering to fact check.

[IvanIvanovich]

It looks pretty nice... I would have to actually try it to buy into any improvements that may exist in functionality. They probably aren't doing themselves any favors with the price though. Especially not if they immediately let the Chinese OEMs start using that tech, they will undercut them by $100 out the gate even with aluminum top bezels.

[jacobolus]

The 9 minute video is hilarious. Some gems:

- dude thinks keyboard switches is "rocket science"
- "the mx module is absolutely the most precise module, and the mx module is very fast. digital technology is not good enough for it"
- "the material of the case isn't plastic, unlike any other keyboard we know of"

[jacobolus]

So in case anyone is still unclear:

They’re still using MX switches, just with some different scanning method, which is “analog”. There’s no analog position sensing of switches involved (sorry Parak).

I’m curious if anyone can dig up any recent patent filings from Cherry. Not sure if they’d file them in the US or Germany or what. But since patents now need to be filed before technical details are publicly announced, it seems like there’s at least some possibility they might have filed some already. Reading their patent applications should explain what they’re up to much better than the silly marketing videos.

[Zeal]

Interesting looking plate...

More: http://www.heise.de/newsticker/meldung/CES-Tastatur-mit-besonders-geringer-Latenz-2511214.html?wt_mc=sm.feed.tw.ho&hg=1&hgi=2&hgf=false

[tp4tissue]

Ok.... 

But someone please cut it in half..  and add thumb cluster...

[Grendel]

More: http://www.heise.de/newsticker/meldung/CES-Tastatur-mit-besonders-geringer-Latenz-2511214.html?wt_mc=sm.feed.tw.ho&hg=1&hgi=2&hgf=false

Now there is an interesting bit of info: "Jeder Taste und jeder Tastenkombination ist ein individueller Widerstandswert zugeordnet." -- "Every key and key combination is assigned an individual resistance value." Looks like Cherry licensed the technology from Microsoft (see) ! Also, there is of course a scan rate, 2500Hz (0.4ms.)

MS details:
http://www.microsoft.com/hardware/en-us/support/licensing-technology-overview
http://www.microsoft.com/appliedsciences/content/projects/AntiGhosting.aspx
http://download.microsoft.com/download/D/8/5/D85E2240-8688-4D49-BFD3-763E61A7D90E/AntiGhostingBrochure.pdf

https://www.google.com/patents/US20100066572

[Puddsy]

I'm not sure why cherry didn't put MX RGB switches into this board. Mistake.

Price point is too high, what's stopping me from getting a K70/95 RGB?

Congrats, your wrist rest is basically the one from the Steelseries 7G. It's a good wrist rest, though, so I'll let it slide.

Alu case is a nice touch, and probably a majority of the price as well.

The board looks fairly tasteful as well. Definitely gamer-y but you could totally bring it to some offices, as long as they're not ultra formal.

Overall, I think it's almost awesome.

My initial impression is 6.5/10.

I have no idea about the relationship between this and unikey.

[Grendel]

I'm not sure why cherry didn't put MX RGB switches into this board. Mistake.

From here: "Die analoge Abtastung und der M3-Controller sind recht stromhungrig: Samt Tastenbeleuchtung schöpft die Tastatur das USB-Budget von 500 mA voll aus." -- "The analog scanning and M3 controller are pretty power-hungry: including the BL the keyboard requires the full 500mA a USB port can deliver." Using RGB LED's the power requirement probably would grow above that. Making the board USB 3.0 aware would be a way around this (on USB 2.0 the board could limit the BL to a primary color), expensive solution tho...

[rainb1ood]

Ugly font

Hopefully the price would get lower

[IvanIvanovich]

USB 3.0 dosen't really seem like it would be a big deal on a +$200 board... I have to agree there they missed a trick by not using the RGB ones. Or... they could have done it the messy crap way with usb 2.0 and had it use it's own power brick LOL.

[rainb1ood]

from the PDF

[Findecanor]

"up to 20ms", correct me if I'm wrong but don't many keyboards have 1000hz polling rate? Wouldn't that make them take 1ms to detect?

1ms to detect that a key started changing, now needing debounce time.

Yes. That means that the switch is sampled every ms. First after n identical readings is the switch considered on or off.
Cherry MX and Alps are both specified as needing 5 ms of debounce time.

Now there is an interesting bit of info: "Jeder Taste und jeder Tastenkombination ist ein individueller Widerstandswert zugeordnet." -- "Every key and key combination is assigned an individual resistance value."

I agree. There is another explanation in the document (slides?) that SpAmRaY linked to:

With CHERRY RealKey technology
there are different voltages at the individual
conductor tracks. When the circuit is closed
by the press of a button, the controller detects
how much current is flowing. So each key
on the keyboard is read analogously
in realtime by the controller.

[quake4mhg]

USB 3.0 dosen't really seem like it would be a big deal on a +$200 board... I have to agree there they missed a trick by not using the RGB ones. Or... they could have done it the messy crap way with usb 2.0 and had it use it's own power brick LOL.

this.

[jacobolus]

USB 3.0 dosen't really seem like it would be a big deal on a +$200 board...

Having a keyboard with features that stop working if your computer doesn’t support USB 3.0 sounds like a really bad idea.

[HendyZone]

Private Video ?

[Ngt]

Would it come with usual MX switches? Or will they bring a totally new type of switch as well?

[Zeal]

Would it come with usual MX switches? Or will they bring a totally new type of switch as well?

Just regular MX switches.

[tp4tissue]

sigh...

THis is such freaking bull ****...  NOT FAST ENOUGH?   

Even if you get 20ms down to 1ms...  are humans going to be able to take advantage of that ?   are games designed around 1ms input intervals...  what application requires human precision of 1ms...

Seriously.. this is an exercise in BULL **** g4m3r philosophy to suck money out of parents through their stupid kids..

[dante]

Screw Cherry; where are my Gaterons?

[willc]

Seriously.. this is an exercise in BULL **** g4m3r philosophy to suck money out of parents through their stupid kids..

Show Image

... But moooooooooooooom

[derezzed]

I don't know why y'all are hatin'.  I really like the design of this new Razer keyboard.

No delay in signal read?  Is Cherry using quatum entanglement?

The main problem with the MX Board 6.0 is that Cherry over-hyped it in both the teaser and the presentation.  Cherry's marketing has been painfully embarrassing lately.  Marketing savvy is not part of Cherry DNA.  Hyperbole is part of Cherry DNA.

[tp4tissue]

I don't know why y'all are hatin'.  I really like the design of this new Razer keyboard.

No delay in signal read?  Is Cherry using quatum entanglement?

The main problem with the MX Board 6.0 is that Cherry over-hyped it in both the teaser and the presentation.  Cherry's marketing has been painfully embarrassing lately.  Marketing savvy is not part of Cherry DNA.  Hyperbole is part of Cherry DNA.

Dude.. is -hyperbole- the new buzz word here on GH? hahahahaha

[derezzed]

I don't know why y'all are hatin'.  I really like the design of this new Razer keyboard.

No delay in signal read?  Is Cherry using quatum entanglement?

The main problem with the MX Board 6.0 is that Cherry over-hyped it in both the teaser and the presentation.  Cherry's marketing has been painfully embarrassing lately.  Marketing savvy is not part of Cherry DNA.  Hyperbole is part of Cherry DNA.

Dude.. is -hyperbole- the new buzz word here on GH? hahahahaha

Haven't you heard? Everybody's saying it.

[jacobolus]

Even if you get 20ms down to 1ms...  are humans going to be able to take advantage of that ?   are games designed around 1ms input intervals...  what application requires human precision of 1ms...

What makes you think keyboards are designed for regular humans?

As every nerd knows, Jedi knights can easily make thousands of keystrokes/second.

[derezzed]

Even if you get 20ms down to 1ms...  are humans going to be able to take advantage of that ?   are games designed around 1ms input intervals...  what application requires human precision of 1ms...

What makes you think keyboards are designed for regular humans?

As every nerd knows, Jedi knights can easily make thousands of keystrokes/second.

[BucklingSpring]

So basically NKRO... I guess?

And without the "lag"... But this thing has nothing to do with switch technology and is everything about controller and circuitry to scan depressed keys.

Innovation parts of Cherry DNA - Now that is funny. These guys innovate as much as crocodiles and sharks. But to their credits, they do have something that works well to begin with.

[bowji]

So...... thats it?

[spiceBar]

So basically NKRO... I guess?

I can't believe it.

Cherry has become totally pathetic.

I had no idea they were so desperate.

[tp4tissue]

Even if you get 20ms down to 1ms...  are humans going to be able to take advantage of that ?   are games designed around 1ms input intervals...  what application requires human precision of 1ms...

What makes you think keyboards are designed for regular humans?

As every nerd knows, Jedi knights can easily make thousands of keystrokes/second.

(Attachment Link)

Tp4 can't deal with this...

[Dihedral]

Here is the 9 minute private video, recorded and uploaded:

Almost fell asleep watching that...
Maybe that's why they took it down...
 

Yeah, It's just a ton of people doing the buzzword olympics.

[awts]

Just how many limbs do gamers have? Even on all fours, you can only attain 20-KRO at max. 

[Ngt]

Screw Cherry; where are my Gaterons?

Ordered the switch tester on Massdrop with the Gateron on it. I'll see what they are worth.

[Oobly]

MUAHAHAHAAH!!

So Cherry innovation is licensing a technology from another company (RK=resistive ADC matrix, from Microsoft)? Yup, that's true innovation for sure.

Cherry MX - imprecise and scratchy technology made for decades without any change at all except slowly deteriorating molds = more Cherry innovation!

What a load of BS.

[Dihedral]

MUAHAHAHAAH!!

So Cherry innovation is licensing a technology from another company (RK=resistive ADC matrix, from Microsoft)? Yup, that's true innovation for sure.

Cherry MX - imprecise and scratchy technology made for decades without any change at all except slowly deteriorating molds = more Cherry innovation!

What a load of BS.

The title now sounds like a bad joke, now we know what rk is.

[Oobly]

Not to mention limiting the number of MX types to 4 ("the choice of 4 individual switch types" showing Black, Red, Blue, Brown) when I know they currently make 8. Even more Cherry innovation.

Seriously, getting bought by ZF has made this company so lame.

[Oobly]

I just noticed something else I find a little disturbing. The ML datasheet on the Cherry website has the MX Brown force diagram in it instead of the ML one...

http://cherrycorp.com/product/ml-series/

"Innovation at your fingertips" - switches which magically feel like different switches...

At least the Force Travel Diagram tab shows the correct one. And they don't show the pinout of the switch ANYWHERE. Or the stabiliser dimensions. I know I seem to be Cherry bashing here, but I'm designing a PCB for ML switches and finding this very frustrating.

[dorkvader]

MUAHAHAHAAH!!

So Cherry innovation is licensing a technology from another company (RK=resistive ADC matrix, from Microsoft)? Yup, that's true innovation for sure.

Cherry MX - imprecise and scratchy technology made for decades without any change at all except slowly deteriorating molds = more Cherry innovation!

What a load of BS.

Cherry changes their molds regularly. I have examples of new switches that would be absolutely different were they made on a mold that processed 5,000,000,000 other switches (even if they did 100 at once, after 50,000,000 shots the mold would be completely unusable). Plastic injection molds are usually good for about 2,000,000 cycles max. after that the effects would be painfully obvious, so for a device requiring this level of mold precision, I would expect them to change them more frequently.

At least the Force Travel Diagram tab shows the correct one. And they don't show the pinout of the switch ANYWHERE. Or the stabiliser dimensions. I know I seem to be Cherry bashing here, but I'm designing a PCB for ML switches and finding this very frustrating.

Based on the marketing BS video and the part number this looks like they use a normal cherry MX switch and different resistive sensing technology (licensed from microsoft) instead of diodes and a more expensive controller

[CPTBadAss]

Screw Cherry; where are my Gaterons?

Ordered the switch tester on Massdrop with the Gateron on it. I'll see what they are worth.

Oh ****, can someone link me that? I didn't realize they had Gateron switch testers.