membrane keyboard with NKRO

[gcogger]

4+8+9 works fine.  I've been trying everything I can in the numpad (and cursor keys) area and still can't get less than 6.

Edit:
I'm off to bed  I'm happy to try any other combinations tomorrow.

[Mental Hobbit]

4+8+9 works fine.  I've been trying everything I can in the numpad (and cursor keys) area and still can't get less than 6.

Wow, I'm impressed. Haven't seen any rubberdome capable of handling keypresses in more than one row on the numpad - and I've tried that on any board I got my hands on. Of all keyboards I tried, it only worked on mechanical full-NKROs, i.e. Filcos, Steelseries and ancient Cherries.  

I'm left-handed, so I use [num] 7-8-9 as my home row in games, and R-arrow as default action (thumb-)key.

[Brodie337]

Hey, where in Australia are you, TS?

I've got a Chicony KB-5181 that you can have if you're interested.

[Paul Dietz]

Welcome to Geekhack!  We're always gentle (at least I am up to post count 10).

One keyboard that has a rubber dome yet claims 26-Key rollover is the new Microsoft Sidewinder X4.

Show Image

.

 Unfortunately they screwed up the design.  It has a problem called ghosting where keystrokes register that you actually didn't want.  

This is WORSE than key blocking so I don't recommend you buy it.

From the ExtremeTech Review.

Hey gang!

I am the researcher behind the Sidewinder X4 anti-ghosting technology. I wanted to clear up some misconceptions I've noticed in this forum.

The "ghosting" referred to in the quote above is not from the keyboard. That's a software bug on the PC side. You can see the same behavior with competitive USB keyboards that report large number of keys. For what it's worth, I have only seen this behavior when you exceed 11 simultaneous keys from a single keyboard, so it's probably not much of an issue for most people. This is all discussed on our web site:

http://www.microsoft.com/appliedsciences/SideWinderX4.mspx

On that page, you'll see that we fully document what X4 can and cannot do. The people in this forum seem to be unusually well informed about keyboard issues, so I hope you'll appreciate having all this information plainly stated.

In addition to this, there has been some misinformation about how X4 works. We add a fixed resistor in series with each key. Every resistor is nominally the same value, although this is not critical. Ultimately, the system is just making on/off measurements, checking to see if a current exceeds a simple threshold. I know this is quite an innovation in the keyboard space, but this type of technique is how many multitouch systems work and has been used for decades in industrial sensors (e.g. Tekscan).

In order to help cut through the confusion, we've put up a browser based keyboard tester. You can find it here:

http://www.microsoft.com/appliedsciences/KeyboardGhostingDemo.mspx

It's capabilities will depend somewhat on what browser and what operating system you are using. But it should be good enough to give most people a quick way to test their keyboards for obvious problems. Here's what I do when I want to show just how bad most keyboards are:

Press and hold ASDW. Now try pressing every other key while continuing to hold ASDW. You will probably find a number of keys that don't work. When you find these, try releasing some of the ASDW keys. You will probably be able to narrow it down to specific 3-key combinations that don't work.

Anyhow, I hope this is helpful!

[elbowglue]

The "ghosting" referred to in the quote above is not from the keyboard. That's a software bug on the PC side. You can see the same behavior with competitive USB keyboards that report large number of keys. For what it's worth, I have only seen this behavior when you exceed 11 simultaneous keys from a single keyboard, so it's probably not much of an issue for most people. This is all discussed on our web site:

Whatsup, thanks for the response!  I don't know about you other geekhackers, but ghosting with 11 simultaneous keypresses is a big problem for me!

[namelessguy]

Greetings, Dr. Dietz!

Actually I'm curious about the driver, rather than the SideWinder X4 hardware itself.
You just can't handle more than 11 keystrokes at a time with the standard USB HID driver (on Windows XP, at least), you will see ghosting problem when you try (but please don't ask me why, I know very little about how the driver or Windows OS itself works).
And, from what I've gathered, you can't use the brute force here if the hardware is only using USB 1.1 low-speed; to confirm key inputs, every USB keyboard on the market today continuously reports the "current condition" of the keys, while the PS/2 keyboard only sends changes in its state, so I guess 800 Byte/second on USB low-speed interrupt transfer is totally underpowered for NKRO with a decent report rate (though it may be enough for anti-ghosting up to 10-14 keys, not enough for 26 or 17 keys).
And my concern here is that no official article of this keyboard boasts that it uses the USB full-speed transfer for, like, say, faster response for gaming.

[gcogger]

Here's what I do when I want to show just how bad most keyboards are:

Press and hold ASDW. Now try pressing every other key while continuing to hold ASDW. You will probably find a number of keys that don't work. When you find these, try releasing some of the ASDW keys. You will probably be able to narrow it down to specific 3-key combinations that don't work.

Works fine on the Razer Tarantula

Of course, yours is half the price, has backlit keys and is probably nicer to type on.  I may have to buy one to see if I like it

[Mental Hobbit]

Thanks to Dr. Dietz, interesting insights. I'm also glad we can finally put a name to this familiar face.

[ohmage]

hey all im back

was just curious.. seeing as the X4's ghosting "issues" were  caused by the USB HID driver... does that mean if ya plug it in via a PS/2 converter it should be full nkro or? just out of curiousity tho, i wouldn't need any more than 6kro ^_^ need somebody to test this!

[JBert]

Actually, no: PS/2 converters are notorious for their 6KRO.
It doesn't matter my model F is virtually NKRO, any convertor that hasn't got a special driver will only pick up the first 6 keys.

[Rajagra]

And he may take away my mug.

It isn't your mug. It remains Microsoft's property at all times and you merely have a licence to use it. :smile:

Those technical details are interesting. Putting resistive tracks directly on the membrane is quite neat. And since it doesn't suffer from the "variable resistance" issues I referred to before, I think it can work well.

I can't quite visualise how the scanning algorithm would work, but I'm inclined to believe it is workable.

[Paul Dietz]

Thanks to Dr. Dietz, interesting insights. I'm also glad we can finally put a name to this familiar face.

Show Image

Cool. That's a much better picture than my official MS one.  The hair is way nicer. And it's much less gray!

:wink:

[Paul Dietz]

Greetings, Dr. Dietz!

Actually I'm curious about the driver, rather than the SideWinder X4 hardware itself.
You just can't handle more than 11 keystrokes at a time with the standard USB HID driver (on Windows XP, at least), you will see ghosting problem when you try (but please don't ask me why, I know very little about how the driver or Windows OS itself works).
And, from what I've gathered, you can't use the brute force here if the hardware is only using USB 1.1 low-speed; to confirm key inputs, every USB keyboard on the market today continuously reports the "current condition" of the keys, while the PS/2 keyboard only sends changes in its state, so I guess 800 Byte/second on USB low-speed interrupt transfer is totally underpowered for NKRO with a decent report rate (though it may be enough for anti-ghosting up to 10-14 keys, not enough for 26 or 17 keys).
And my concern here is that no official article of this keyboard boasts that it uses the USB full-speed transfer for, like, say, faster response for gaming.

The product team did the new USB interface, so I can't claim to be an expert on that. But it is full speed. Ironically, when X4 was in the planning stages, the faster interface was going to be one of the major features listed on the front of the box. But when we showed up with our resistive matrix solution, the marketing team latched onto it as the signature feature. So now, the faster interface is barely mentioned. But it does have it.

For what it's worth, I suspect that a lot of the marketing on faster response is confusing. Just because reports are being sent every ms doesn't mean that the latency is anywhere near that small. Mechanical switches bounce, so keyboard firmware has to do some filtering and this can add many ms of delay. In addition to this, you have to pass through the whole HID stack before the app sees anything. Finally, most screens introduce their own latency, which can easily be over 10ms. And this isn't counting any delay in the application itself. So the time between hitting a key and something changing on screen can be surprisingly long. This is particularly evident on touch screens. If you drag an object around in a circle, look at how far behind the image lags.

[Paul Dietz]

I was going to mention to Paul Dietz that his employer might want to look into that problem of the OS ghosting keyboard input.  Keyboard computer input has been tackled by multiple companies over many computing platforms over 50 years so I would have thought that they would have figured out how to do it by now.

But that would be rude.  And he may take away my mug.

Your mug is safe from me. And yes, we are very definitely looking into it...
:wink:

[Paul Dietz]

Haha!  Good to see you have a sense of humor.  Have some wallpaper for your PC and send my apologies to the Microsoft R&D crew.

Show Image

The brass rat is a very nice touch! I haven't seen mine in many years. And the hair is sadly closer to the truth...

Thanks!

:smile:

[kishy]

Looks more like a beaver...

[Phaedrus2129]

Shhh! The rat knows all!



You seem to be everywhere, Mr. Dietz. Sorry for TwoCables over at OCN, he gets stuck on things sometimes. And also sorry for my lack of EE knowledge! Wait, you aren't the Pope? Oh.



You mentioned that the ghosting issues (not key blocking ) are due to an OS-side problem, correct? Well, you do work at Microsoft. Has this been passed on to the Windows programming team, and could we see an update that fixes this issue some time in the future?

[Paul Dietz]

Looks more like a beaver...

That's the MIT ring, affectionately/mockingly called the "brass rat". It features the beaver mascot...

http://en.wikipedia.org/wiki/MIT_class_ring

[kishy]

Lol thanks, I had just Googled it myself.

I still say it looks more like a beaver (meaning the name is misleading).

Beaver:



Rat:

[Paul Dietz]

Shhh! The rat knows all!



You seem to be everywhere, Mr. Dietz. Sorry for TwoCables over at OCN, he gets stuck on things sometimes. And also sorry for my lack of EE knowledge! Wait, you aren't the Pope? Oh.



You mentioned that the ghosting issues (not key blocking ) are due to an OS-side problem, correct? Well, you do work at Microsoft. Has this been passed on to the Windows programming team, and could we see an update that fixes this issue some time in the future?

I'm definitely NOT the Pope. But I am very curious to see how people are reacting to the X4. I think it has a bit of an uphill battle because most people don't understand the ghosting/blocking/simply-not-working problems that many keyboards have when pressing some combinations of 3 keys. I know that there are a lot of factors that go into choosing a gaming keyboard. But I would hope that having the keys you press being accurately reported would be pretty high on everyone's list...

As for the PC-side issues that happen when 12 or more keys are pressed simultaneously on a single USB keyboard, rest assured that we have people looking into this. The problem is both intermittent (may or may not happen) and inconsistent (different problems arise with the same key presses). As best as we can tell, this is a very old issue that went undetected because we didn't have keyboards that would report that many keys...

[Phaedrus2129]

One more question, and you may not have an answer. The popularity of mechanical keyboards has been going up. Not quickly, but it is on the rise. Even Thermaltake is releasing a Cherry keyboard, and a few other keyboard manufacturers have quietly entered the mechanical ring, like Adesso. There are also a few specialty boards being released soon, like the new TrulyErgonomic (http://www.trulyergonomic.com/) and Mini-Guru (http://www.guru-board.com/). I've also noticed that certain models of mechanical keyboard on ebay have gotten harder to find, even in the time that I've been following them.

Some of this interest, especially among enthusiasts, is probably due to Manyak's thread on Overclock.net, which is now the top Google result for "mechanical keyboard".

So with this renewed interest in mechanical keyboards... Are you aware of any plans on Microsoft's part to release one? At the very least, it's probably much easier to make a PCB-based mech board NKRO than a membrane board.

[Paul Dietz]

One more question, and you may not have an answer. The popularity of mechanical keyboards has been going up. Not quickly, but it is on the rise. Even Thermaltake is releasing a Cherry keyboard, and a few other keyboard manufacturers have quietly entered the mechanical ring, like Adesso. There are also a few specialty boards being released soon, like the new TrulyErgonomic (http://www.trulyergonomic.com/) and Mini-Guru (http://www.guru-board.com/). I've also noticed that certain models of mechanical keyboard on ebay have gotten harder to find, even in the time that I've been following them.

Some of this interest, especially among enthusiasts, is probably due to Manyak's thread on Overclock.net, which is now the top Google result for "mechanical keyboard".

So with this renewed interest in mechanical keyboards... Are you aware of any plans on Microsoft's part to release one? At the very least, it's probably much easier to make a PCB-based mech board NKRO than a membrane board.

It's straightforward to make mechanical keyboards that support large numbers of simultaneous key strokes - use a big printed circuit board and put a diode in series with each mechanical key. The problem is that it is rather expensive. Of course, if enough people really wanted these things AND they were willing to pay for them, I imagine MS and the other large keyboard manufacturers would respond appropriately. But by their nature, most of the big players have too much overhead to get involved in small market products. In some ways, this is good for the industry because it gives smaller players a place where they can succeed and grow before going toe-to-toe with the big guys.

The X4 is an interesting product because it attempts to bring a rather high-end keyboard feature to the mass market. My hope is that it will be a huge hit. If it is, it will send a message to major keyboard manufacturers that making sophisticated, core functionality improvements pays off. However, if it only sells modestly, it will be a huge red flag, indicating that we are having trouble explaining the value proposition...

[Phaedrus2129]

This may be stupid, but I just had an idea. What about a compromise between a mechanical and a rubber dome? Not like a Topre or anything, no. What I mean is...


In a typical membrane keyboard you have three membranes: top one with the rubber domes, middle spacer membrane, and bottom membrane with the contact surface. What if you replaced that bottom membrane with a PCB? You'd then be able to put a diode after each rubber dome switch, thus giving NKRO. It would be more expensive than a typical rubber dome, but less expensive than a mechanical keyboard because you don't have to use mechanical switches.

You would be able to provide an NKRO rubber dome keyboard in the $50 range, compared to the cheapest NKRO mechanical keyboards like the $70 Adesso MKB-135B.


There's probably some reason this hasn't been done yet, but nothing that stands out to me. Please point out my stupidity.

[Hak Foo]

It's straightforward to make mechanical keyboards that support large numbers of simultaneous key strokes - use a big printed circuit board and put a diode in series with each mechanical key. The problem is that it is rather expensive

BULL****.
In a quantity of 1,000 pieces, my local electronics parts shop sells diodes for THREE CENTS each.

Likely, a Logitech or Microsoft would be buying in quantities of 100,000 or more, so I'd expect a steeper drop.

104 units for 104 keys = $3.12 extra.  Even adding extra tooling and assembly cost, there's no reason the retail boost can't be $20 or less.

Note the delta of price on Filcos is $25 for NKRO.

[ricercar]

Hak, you seem to be oblivious to the realities of marketing (as well as enthusiastically rude). OEMs look for ways to cut 1/2 cent in manufacturing costs even if it compromises functionality for 20% of the consumer base. Adding $3 for a feature recognized by fewer than 10% of the customer base simply Ain't Gonna Happen without strongly established evidence of 30 dollar profit per unit (10x mfg cost increase in profit, not revenue).

I'm basing this on professional experience in NVIDIA graphics board and motherboard reference designs. Keyboards are different, but not much different. The OEMs are much the same.

[TexasFlood]

Hak, you seem to be oblivious to the realities of marketing (as well as enthusiastically rude). OEMs look for ways to cut 1/2 cent in manufacturing costs even if it compromises functionality for 20% of the consumer base. Adding 3 dollars for a feature recognized by under 10% of the customer base simply Ain't Gonna Happen without strongly established evidence of 30 dollar profit per unit (10x profit, not revenue).

Sadly I've seen this as well, both in manufacturing and jobs.

[Mental Hobbit]

BULL****.
In a quantity of 1,000 pieces, my local electronics parts shop sells diodes for THREE CENTS each.

The question was about a mechanical keyboard. I don't think Paul Dietz was saying the diodes would be expensive. Rather the mechanical stuff around the diodes.

[Hak Foo]

Well, we know what a mechanical board can be sold for: $50 (Scorpius M10.  So an NKRO M10 at 80.00 is feasible.  At 120, it's a scam.

[Phaedrus2129]

Well, we know what a mechanical board can be sold for: $50 (Scorpius M10.  So an NKRO M10 at 80.00 is feasible.  At 120, it's a scam.

Adesso MKB-135B is basically an NKRO M10, and it's $71 at Provantage.

[Paul Dietz]

This may be stupid, but I just had an idea. What about a compromise between a mechanical and a rubber dome? Not like a Topre or anything, no. What I mean is...


In a typical membrane keyboard you have three membranes: top one with the rubber domes, middle spacer membrane, and bottom membrane with the contact surface. What if you replaced that bottom membrane with a PCB? You'd then be able to put a diode after each rubber dome switch, thus giving NKRO. It would be more expensive than a typical rubber dome, but less expensive than a mechanical keyboard because you don't have to use mechanical switches.

You would be able to provide an NKRO rubber dome keyboard in the $50 range, compared to the cheapest NKRO mechanical keyboards like the $70 Adesso MKB-135B.


There's probably some reason this hasn't been done yet, but nothing that stands out to me. Please point out my stupidity.

I've heard of companies trying pretty much all the variants - buckling springs with membranes for electrical switching, domes on top of printed circuit boards, etc.

Not to be redundant, the resistive membrane technique is the answer you've been looking for. It really does let you see every key for a very modest cost increase over traditional membranes. On the X4, the max number of simultaneous keys is limited by debounce processing and USB communications - not by the matrix. And we have been mocked for the level of overkill we've already implemented in X4. I believe it's more than ANY other USB keyboard on the market. Oh, and it lists for US$59.99, with Amazon selling it for under US$50 during pre-orders.

So, short answer - SideWinder X4 is a membrane-based keyboard that easily meets your simultaneous key pressing needs, and sells for about the price you want. Problem solved!
:smile: