Gold plated USB connection

[Inertially]

Is it anything more than a gimmick?

I really doubt that it does anything significant, or give any bit of advantage during game play..

Would it even be measurable?

[ch_123]

Nope, there are no benefits. None at all. Not even slightly... You get the idea. xD

[keyb_gr]

Gold-plated connectors are just about better contact - specifically, better long-term contact (gold doesn't oxidize after all). They should be like this on both sides though. In any case you are not normally going to notice any difference, even if you leave the board plugged in for years.

Ironically the only connection that ever gave me trouble on a keyboard was the SDL on a Model M, with gold-plated contacts on both sides! A bit of tuner cleaner fixed that right away. Seems like I had neglected to take a look at the jack when originally cleaning the board.

[Manyak]

Yeah the only thing they got goin for them is that they won't corrode. But for conductivity, many other metals are actually better (like copper, silver, etc)

[timw4mail]

Gold plating is truly significant.  It has made Radio Shack and Monster Cable millions of dollars.  On the other hand kajillion electrons have cared less.

I think its more on the magnitude of many googolplexes of electrons...

[mmgoose]

it's very pretty.

like spoilers on rice rockets.

[Hak Foo]

No, those are sensible, mmgoose.  You can hang the car up by them and make extra space in your garage to set your body kit parts in.

[Red October]

Gold plating is truly significant.  It has made Radio Shack and Monster Cable millions of dollars.  On the other hand kajillion electrons have cared less.

LOL I work at a Radio Shack.  Actually our gold-plate cables are way cheaper than Monster.  Don't buy super wootsy-poot oxygen-free high-test magic fairy-dusted cables for digital signals.  This includes HDMI.  Gold plating does in fact inhibit corrosion and makes for a better contact, but that doesn't matter on digital.  About the only time to buy a better-than-average cable is analog Hi-Def like RGB.

[Manyak]

LOL I work at a Radio Shack.  Actually our gold-plate cables are way cheaper than Monster.  Don't buy super wootsy-poot oxygen-free high-test magic fairy-dusted cables for digital signals.  This includes HDMI.  Gold plating does in fact inhibit corrosion and makes for a better contact, but that doesn't matter on digital.  About the only time to buy a better-than-average cable is analog Hi-Def like RGB.

You've got the idea, but the details are a bit off.

Yes in almost every home theater or PC setup the cables are the last of your worries as long as they are halfway decent (as in, just don't do something dumb like trying to power 500W RMS speakers through 24ga wire). Even analog Hi-Def doesn't need that great of cables as long as it's a short distance and you don't have any high-power lines running near them (such as the wires used on your 500W speakers ).

But once you need to send signals across long distances, use high-power equipment, or have a signal that is high-frequency, the cables make all the difference - even with digital signaling. Like, say you wanted to connect a 1920x1200 100Hz monitor to a PC 100ft away. DVI won't get the job done in any way or form - not without putting a repeater in every 3-4 feet. On the other hand, a single 75ohm VGA cable with proper construction and shielding will carry it all the way with less than 0.5% loss (this sort of cable costs around $150). Then again, a cheap-o 50 cent VGA cable can give you the same signal loss in just 15ft.

[Ulysses31]

Gold provides a better contact due to its softness, and is less prone to generating whiskers that can make a connection unreliable.  Arcing is also reduced, although I doubt it makes much difference at the voltage levels used by a keyboard.

[talis]

I'd be supprised to see any proof that a few micron thick gold plating could deform by any demonstrable amount.  If it was soft enough to do so, you'd likely scrape it off simply making contact.

With connectors that you re-plug on a semi regular basis (even just a couple of times a year) you need to worry less about corrosion, as the act of plugging and unplugging the connector is usually enough to break through any oxide layer that forms.  If a connector stays plugged for years or decades, then I think gold becomes very important.


The claims they make here seem a bit off to me (and I'm assuming this is what the OP is referring to).  The gold plating should not affect latency in any demonstrable way.

[keyb_gr]

I'd guess these were supposed to be advantages of the USB connection in general and not of the gold plated connector in particular. Bad wording.

[talis]

I would think that, by its polled nature, USB would potentially be higher latency then PS/2.

[Manyak]

Well, since you work at Radio Shack I'll stop making wisecracks about them (they are certainly convenient).  I will though mention the "In Praise of Monoprice" post has some alternatives and a nice cable discussion.

DVI can handle longer distances surprisingly well - I've got this 50 foot cable running from a wiring closet to the PC above running 720p.  I read the specs too (vague as they are about distance) but decided to ignore them.  Works flawlessly.

Another use for the RipOmeter!

Show Image

There are many advantages to moving the media PC out of the living room.  Chief among them for me it allows me to reuse an old PC (AMD X2).

Now try running it at 1920x1200 @ 100Hz and see if it still works!

Look at how much extra bandwidth is needed when you bump up the resolution and refresh rate:

720p: 1280 * 720 * 32 bits * 60 Hz = 1 769 472 000 bits/sec (216MB/s)

1200p: 1920 * 1200 * 32 bits * 100Hz = 7 372 800 000 bits/sec (900MB/s)


Trust me, there's no way DVI can handle that resolution and refresh rate at 100m without some sort of signal amplifier and seriously thick wires and shielding, while still maintaining less than 1% signal loss. I guess theoretically it can happen, but I highly doubt you'll ever see it done since VGA is still much more efficient.

[huha]

I would think that, by its polled nature, USB would potentially be higher latency then PS/2.

If it's polled faster than it takes to transmit data over PS/2 (PS/2 is incredibly slow, I don't know if USB polling rates are limited), latency is lower. Still, the whole latency business is horribly overstated in my opinion.

Speaking of gold plating, almost everything has already been said, so I'll just add one thing: Irrespective of its merits, gold plating the outer portion of a USB connector won't do anything at all. No signal or even current is transmitted over it, it's just to ground the shielding. Shielding needs to be grounded, but that's it basically. If gold-plating anything is especially useless, it's the shielding connector.

-huha

[Manyak]

If it's polled faster than it takes to transmit data over PS/2 (PS/2 is incredibly slow, I don't know if USB polling rates are limited), latency is lower. Still, the whole latency business is horribly overstated in my opinion.

Speaking of gold plating, almost everything has already been said, so I'll just add one thing: Irrespective of its merits, gold plating the outer portion of a USB connector won't do anything at all. No signal or even current is transmitted over it, it's just to ground the shielding. Shielding needs to be grounded, but that's it basically. If gold-plating anything is especially useless, it's the shielding connector.

-huha

USB is polled at 8ms intervals (125Hz). But you can increase it as long as the device plugged in can handle those rates as well.

It's not really bandwidth that anyone is worried about though, it's latency. And if the keyboard has to wait 4ms (on average) to send signals, it's therefore slower than PS/2.

Of course, I don't see how anyone could possibly tell the difference between an 8ms and 2ms key response time. You'd have to type over 125 characters per second, or about 1500 WPM.

[huha]

Latency is used as a marketing ploy to get gamers, i.e. teens/twenty-somethings to buy lots of overpriced crap.
I don't quite see the point in having gaming keyboards with such tiny latencies even audio applications don't give a damn. When your keyboard has less latency than tolerable for realtime audio (about 10 ms), you really ought to wonder.

That said, as far as latency in games goes, I think USB could be better off there. I've calculated the maximum key press/release rate of PS/2 but am too lazy to search for it; I think it was about 250 Hz. USB has the advantage of transmittting all keys in one single packet, i.e. each poll cycle, whereas PS/2 needs to transmit them sequentially. So if you want to do that duckjump while throwing a smoke grenade, ordering a pizza and reload, USB could well be faster than PS/2 because all keystrokes are transmitted at once.

I highly doubt the average game logic could make anything out of it, though.

-huha

[Manyak]

Well, I've been trying to convince my wife to get a bigger TV but this isn't probably the winning argument.  What about HDMI - wouldn't I be OK with that and reasonably thick cables??  We're talking 50 feet.

HDMI cabling (and signaling) is identical to DVI. But I have a feeling that you'll be able to get away with it. I'm sure that if someone actually made a 50ft DVI cable it was made properly. Besides, that cable you've got is thick as hell for DVI.

Besides, even with a few bits of the signal lost you won't be able to notice the difference (especially not on an LCD, since the panels don't have the gamut to display true 32-bit color anyway).


If you find that after switching TVs you have problems with your current cable, double check to make sure you aren't running it parallel to any nearby power cables (unless you're absolutely sure you're running a very small load on that cable).

If you're 100% sure you're free of major interference and it still doesn't work, your only choices are to go VGA or get a DVI signal amplifier and stick it at around the 25ft mark.


Edit: Just noticed that one I linked won't do 1080p. But you get the idea.

[Manyak]

Latency is used as a marketing ploy to get gamers, i.e. teens/twenty-somethings to buy lots of overpriced crap.
I don't quite see the point in having gaming keyboards with such tiny latencies even audio applications don't give a damn. When your keyboard has less latency than tolerable for realtime audio (about 10 ms), you really ought to wonder.

That said, as far as latency in games goes, I think USB could be better off there. I've calculated the maximum key press/release rate of PS/2 but am too lazy to search for it; I think it was about 250 Hz. USB has the advantage of transmittting all keys in one single packet, i.e. each poll cycle, whereas PS/2 needs to transmit them sequentially. So if you want to do that duckjump while throwing a smoke grenade, ordering a pizza and reload, USB could well be faster than PS/2 because all keystrokes are transmitted at once.

I highly doubt the average game logic could make anything out of it, though.

-huha

Yeah it's definitely nothing but a marketing ploy. The only time it's worth it is for mice, because you really do want to be able to move the cursor across 1920 pixels in less than 1 second and still have it be absolutely precise. But that's another argument all together.

Either way, do you know what speed the PS/2 clock runs at?

And in all technicallity, even with USB the keystrokes are sent one by one. After all, it is a serial connection.

[huha]

Either way, do you know what speed the PS/2 clock runs at?

I think 10-16 kHz. Try searching the forums, I did some calculations on this. (Okay, you don't have to search: It's here). It amounts to about 290 Hz, i.e. 290 keys/second.

USB transmits 6 keys at once. If you keep hammering on the same keys, the controller needs to send a packet without keys to "reset" the keypresses. But then you'd still get 390 keypresses per second with 125 Hz polling rate.

And in all technicallity, even with USB the keystrokes are sent one by one. After all, it is a serial connection.

USB is serial, but the keystrokes are sent as a packet. Packets have checksums, so processing of the packet takes place as soon as the packet has arrived in its entirety; so technically, they're sent in packets of 6+mod. keys.

-huha

[Manyak]

But PS/2 boards only send key information upon a change of state (once when you press the key, again when you release it), while USB boards have to send all keys with every packet. That's why PS/2 supports full NKRO and USB doesn't.

Also, those measurements you did determine the bandwidth of the connection, not the latency....

Ok check it out. Lets assume that you press a key on a PS/2 board. It has to send the following information:

1 start bit
8 data bits
1 parity bit
1 stop bit
1 ack bit
Total: 11 bits

PS/2 boards send 1 bit per clock high, which means that a total of 11 clock cycles are needed to transmit the data. And while some extended keys require more than a single scan code (like the Pause/Break key), even the uberl33t mama's boy doesn't care about those. So let's ignore them.

At 16.7kHz, each clock cycle is 0.058ms.
11 Clock cycles comes out to a 0.643ms delay for a single keystroke, unless you somehow magically manage to press two keys at the exact same moment in time. In which case the second key is delayed.

This is already faster than just the USB polling rate alone, without even calculating in the protocol overhead (which is extremely large once you factor in all the different layers).

[huha]

Latency and bandwidth are somewhat intertwined. If you run out of bandwidth, latency will suffer. USB has enough bandwidth to bear the large protocol overhead, so latency is just influenced by polling rate. PS/2 has enough bandwidth for normal purposes, but bandwidth does become a limiting factor for latency when doing theoretical calculations.

That said, you miss one thing in your calculations: PS/2 needs some "rest" in between the keystrokes. Search for "PS/2 protocol" or something like that on the web, there are a few sites with a bit of in-depth information.

-huha

[Manyak]

Latency and bandwidth are somewhat intertwined. If you run out of bandwidth, latency will suffer. USB has enough bandwidth to bear the large protocol overhead, so latency is just influenced by polling rate. PS/2 has enough bandwidth for normal purposes, but bandwidth does become a limiting factor for latency when doing theoretical calculations.

That said, you miss one thing in your calculations: PS/2 needs some "rest" in between the keystrokes. Search for "PS/2 protocol" or something like that on the web, there are a few sites with a bit of in-depth information.

-huha

Ok so according to this, there must be a 50 microsecond (0.05ms) period of time where the clock must remain high before data is sent.

Still, that comes out to 0.693ms. Not even a whole millisecond.

So even if USB transfers were instantaneous, the fact that the board has to wait an average of 4ms before anything can be sent already makes it slower. A PS/2 board can send anywhere up to 11 keystrokes before a USB board can send it's 6.

[keyb_gr]

In practice the differences are not likely to be that large since key scanning typically occurs in about 20 ms intervals (much larger and you may get key transposition problems, below 10 ms and you may get in trouble with key bounce), so that alone means an average delay of about 10 ms (or 50 ms for the Das III). Ghost key detection has to be performed as well.

The main problem for USB boards (at least some of them) seems to be running out of processing power which forces key scanning to be slowed down.

[Manyak]

In practice the differences are not likely to be that large since key scanning typically occurs in about 20 ms intervals (much larger and you may get key transposition problems, below 10 ms and you may get in trouble with key bounce), so that alone means an average delay of about 10 ms (or 50 ms for the Das III). Ghost key detection has to be performed as well.

The main problem for USB boards (at least some of them) seems to be running out of processing power which forces key scanning to be slowed down.

Well key bounce isn't really an issue even if it does happen. It's very easy to program the controller to ignore state changes in pressed keys for 10ms or however long the switches need. Like, really, really easy. In fact I don't see how it would even increase costs. All it needs is a few extra gates and flip flops which I'm sure that their current ICs have room for - and that's only if they choose to program it in hardware.

[huha]

Well key bounce isn't really an issue even if it does happen. It's very easy to program the controller to ignore state changes in pressed keys for 10ms or however long the switches need. Like, really, really easy. In fact I don't see how it would even increase costs. All it needs is a few extra gates and flip flops which I'm sure that their current ICs have room for - and that's only if they choose to program it in hardware.

You don't want to ignore state changes due to RF noise. The keyboard matrix is essentially a fancy antenna and can pick up all kinds of wonderful stuff. If the controller's not programmed to ignore short state changes and rather ingegrate state over time to check whether the key has been pressed, you might get random keypresses, which is somewhat undesirable.

-huha

[Manyak]

You don't want to ignore state changes due to RF noise. The keyboard matrix is essentially a fancy antenna and can pick up all kinds of wonderful stuff. If the controller's not programmed to ignore short state changes and rather ingegrate state over time to check whether the key has been pressed, you might get random keypresses, which is somewhat undesirable.

-huha

Ok wait. First you say don't ignore state changes, then you say you have to if you don't want random keypresses?

I'm confused

[keyb_gr]

The point is, it may be helpful to scan faster than required just for input. If a key is being registered for just a single 10 ms interval or so, it probably was a fluke - hopefully not so much due to RF noise but rather key bounce or somesuch. Good point, forgot about that one.

[bsvP585hUO2Y6]

Manyak writes:

> Quote:
>
>     Originally Posted by ripster View Pos
>     Well, I've been trying to convince my wife to get a bigger TV but this isn't
>     probably the winning argument. What about HDMI - wouldn't I be OK with that and
>     reasonably thick cables?? We're talking 50 feet.
>
> HDMI cabling (and signaling) is identical to DVI. But I have a feeling that you'll be
> able to get away with it. I'm sure that if someone actually made a 50ft DVI cable it
> was made properly. Besides, that cable you've got is thick as hell for DVI.
>
> Besides, even with a few bits of the signal lost you won't be able to notice the
> difference (especially not on an LCD, since the panels don't have the gamut to display
> true 32-bit color anyway).

This is only true for DVI, where the bits transmitted correspond to
actual color values of the supbixels.  HDMI OTOH transmits an encrypted
stream which will go out of sync as soon as single bit flips.  This
results in a blanked screen for fractions of a second, which is much
more noticeable than a single off-colored subpixel out of millions for
1/50th of a second.

[Manyak]

Manyak writes:

> Quote:
>
>     Originally Posted by ripster View Pos
>     Well, I've been trying to convince my wife to get a bigger TV but this isn't
>     probably the winning argument. What about HDMI - wouldn't I be OK with that and
>     reasonably thick cables?? We're talking 50 feet.
>
> HDMI cabling (and signaling) is identical to DVI. But I have a feeling that you'll be
> able to get away with it. I'm sure that if someone actually made a 50ft DVI cable it
> was made properly. Besides, that cable you've got is thick as hell for DVI.
>
> Besides, even with a few bits of the signal lost you won't be able to notice the
> difference (especially not on an LCD, since the panels don't have the gamut to display
> true 32-bit color anyway).

This is only true for DVI, where the bits transmitted correspond to
actual color values of the supbixels.  HDMI OTOH transmits an encrypted
stream which will go out of sync as soon as single bit flips.  This
results in a blanked screen for fractions of a second, which is much
more noticeable than a single off-colored subpixel out of millions for
1/50th of a second.

Ah you're talking about HDCP. Yeah, didn't think about that :doh:

Still, you have to consider the setup ripster's using. HDMI coming out of a computer doesn't use HDCP (it uses standard DVI signaling) unless playing copy-protected video content. But of course I have no idea what ripster's playing on it.

I guess DVI still ends up a better choice just to be safe.

[Manyak]

No HDCP with my setup - I have a PS3 upstairs connected to the TV.  And the copy protection??  Well, that mysteriously disappeared somehow from my DVDs.  AnyDVD might have had something to do with that.

Then as far as your PC is concerned, HDMI == DVI.

[keyb_gr]

The datasheet of the Elan EM78M611 microprocessor found inside the i-Rocks KR-6230 allows shedding some light onto the process of key scanning. Apparently its A/D inputs can be scanned between 47 times and 1500 times per milli(!)second. Should be plenty for a keyboard.
Even in the old VT100 terminal keyboard, a key has to be registered for at least 2 scanning intervals in order to be considered a valid key press.

[talis]

The datasheet of the Elan EM78M611 microprocessor found inside the i-Rocks KR-6230 allows shedding some light onto the process of key scanning. Apparently its A/D inputs can be scanned between 47 times and 1500 times per milli(!)second. Should be plenty for a keyboard.
Even in the old VT100 terminal keyboard, a key has to be registered for at least 2 scanning intervals in order to be considered a valid key press.

Most Micro controllers can read a whole 8 bit port in a single assembly instruction in digital mode.  For mechanical key switches there's no need for using the A/D converter.  For a 4MHz+ controller, they should be able to read the full 8 bit port once every 1uS or so.