Post by: trance4jim on June 07, 2004, 01:04:00 PM
Post by: opjose on June 08, 2004, 07:46:00 AM
This is -NOT- the way to do it, it's not quite this easy.
You need to ISOLATE the two devices so that they NEVER appear on the same bus at the same time.
If this makes no sense to you, don't attempt it.
Post by: carmoniousmonk on June 08, 2004, 09:36:00 AM
of course not switching devices while the xbox is powered is important.
but would that satisfy the requirements to not have more than two devices on the bus at the same time?
Post by: opjose on June 08, 2004, 10:58:00 AM
Even a disabled device, if it is physically attached to the bus, draws down the line levels.
Remember it's still grounded and it has filtering capacitors, etc. on it which remain active even when it's unpowered.
That is why this is so problematic.
The device must be physically disconnected from the BUS.
The ProjectXdeck guys knew this when they started looking at their design.
Unfortunately they fell off the radar and have disappeared, their project not to be seen.
Anyway they were talking about optically isolating the drive, which is one GREAT way to do this, another is with a 40 pin mechanical or relay activated contact switch.
Needless to say it all takes quite a bit more electronics than merely dropping stuff onto a cable.
Post by: Palmore on June 08, 2004, 11:21:00 AM
And ProjectDeck is far from dead just an FYI, our prototype was pushed back a week (everyones getting out'a school etc) but the issue we where addressing, is drive swapping for cheaper then buying X amount of relays $$$. That and reliabilty of relays etc. We wanted push button and remote controlled drive swapping, that would not be cost effective from a relay/power standpoint so we where looking at other methods of drive switchhing.
All in all. News to me, never had or seen a problem with it.
Post by: opjose on June 08, 2004, 11:40:00 AM
Actually I have been admonishing against this since I joined this board and I had a good discussion about it with the Xdeck people when they first announced their project.
They recognized this problem even before I mentioned it and had already designed a way to overcome it using isolation..
"If this is such a detrementle problem then why is it a all that do multiple harddrives use this method, as well as various vendors sell cables for these devices."
Your mistaken. Vendors do NOT sell or produce multi-drive IDE cables. You may be confusing these with SCSI cables. IDE only supports two devices on the bus (e.g. a single cable). It's part of IDE design from day one.
" Doesn't windows power down Drives that currently aren't in use? "
Yup, but this has no bearing on this whatsoever. The powered down drive is only one of TWO devices on the IDE chain.
It still draws power, which is why Windows can bring it back up from the power saving state.
I believe you are confusing low power with presence on the IDE bus which is not the same thing.
"I've never payed much attention to the details, and it may make me look like a noob, but windows powers down drives not in use after X amount of minutes, wouldn't this create issues with other drives?"
See above, power savings modes have nothing to do with this.
" When and where have you seen this become a problem? examples? Again I've never heard or seen anything along these lines."
Examples are easy to find, as are spec sheets which detail the drain of a powered off device and IDE fanout capabilities, also there are plenty of writeups on IC fan-outs, "ringing", and other problems this will create.
It's pretty universal, which is why you never see vendors rigging up anything like this. The IDE bus was designed for two and only two devices on it.
Normally you can find more about this subject on sites devoted to problems and info concerning IDE, especially those which show signal characteristics when the cable is too long etc.
Post by: Palmore on June 09, 2004, 02:18:00 AM
3rd, the reason hard drives fail with longer cables is signal loss, tell me you know that, not to many devices on the cable.
4th http://www.llamma.co.../4connector.jpg
doh what's that?? and IDE cable for 2 HDD's 1 DVD and a mobo connection. I think I know the difference between a SCSI cable and an IDE cable..
At first I was interested in what you had to say, now I can see not only can you not back it up, but you take my lack of interest in your rants as a lack of knowledge.
Now, show us some where this has been a problem aside from the Xdeck flops, and maybe what you say will hold some water, seeing as how you didn't know that llamma.com sold moddified IDE cables, I'm assuming you are just going off assumption on this too, after reading something a year ago from a half assed group of kids that had a dream of ripping of the scene.
So for now, until otherwise proven, you will have NO problems with it. hook up the drives, you'll be cool.
How bout you get me some info from someone that's actually done it... oh wait... I did
Post by: Palmore on June 09, 2004, 02:37:00 AM
Anyway,
Aside from being a moderator on this and the case modding forums, spending a year of my life dedicated to multi drive xbox's and reading every tut out their, building and running 2 4 drive box's, having one of the largest and most read threads on XS that's been on going for almost a year now, and talking to almost everyone that's ever done/tried or wanted to try multiple hard drives, this has never came up.
It's either a halfcocked idea from Xdeck that you bought into, or something that's never effected anyone who's ACTUALLY done it, or something I've missed in over a year of reading about and working on multidrive xbox's.
Again, in closing, give us documentation, proof, some good reading, share the knowledge if you have it, but it's gotta be credable.
Better start google'n it, cause I can promise you that's what I'll be doing every breeak and chance I get today at and after work
Post by: opjose on June 09, 2004, 06:32:00 AM
The makeshift cables from LLAMA violate the IDE spec altogether. It's from a small group that caters to the Xbox set and is not a production example.
Show me one production computer that has a three or four drives hooked up to the IDE bus. Show me one manufacturer who produces three or four device cables. You can't.
Show me one technical document that says that you CAN hook up multiple devices to the bus and simply disable an unused one.
If you look the only thing you will find is one company who has gone out of business that made a hard drive switcher that actually also recognized this problem and physcially swapped out the drives with an interlock. Their approach was quite novel, and actually applicable to the Xbox. They did it the way it SHOULD be done.
See: http://www.hothardwa...m?articleid=281
I have one of these switches and tore it apart just to see if it isolated the devices. It does.
The Llama cable was made by and for modders much like yourself, who do not always seem to have any deep EE background or knowledge of basic switching circuit design theory.
I'm not talking off the cuff on this, the IDE interface was specifically designed to be a two device bus and it's electrical characteristics reflect this.
I'm not going to google all the info for you because it is SOOO readily available that if you are interested you can do your homework instead of calling me crazy.
However you can start here:
http://www.interface..._threshold.html
http://www.pcguide.c...de/conf_Sig.htm
http://www.interface...nector_IDE.html
http://www.storagere...if/ide/std.html
http://www.pjrc.com/.../ata_atapi.html
Compare the logic levels against the drive engineering circuit charts show power off loads on the bus as a case in point.
Google for: ANSI standard X3.221-1994
http://www.amazon.co...2046573-5847067
"The ATA-1 standard stipulated that a ATA bus only accommodates two drives. These are addressed using 0 (master) and 1 (slave). Since both drives have an onboard controller, each "hears" all of the IDE commands sent on the IDE bus. Thus, the need is resolved to determine for whom a command is sent. The ATA-1 specification also details two transfer modes in which commands and data may be sent to a disk. These are programmed I/O (PIO) and direct memory access (DMA) [3]. "
Heck I found three spec sites in under a minute which deal with the signaling and line level issues involved.
Hint: the controller OEM chipset specs also list the electrical characteristics and requirements. Look for timing and level specs as well as things dealing with changes to the waveforms.
Hint: Scope your setup out sometime and watch the effect of the unpowered device on the bus. Then come back and tell me I'm wrong.
In searching you may discover a bit more about the problem which does NOT come from me, as you seem determined to discredit what I'm telling you simply because I'm citing a problem in what you are trying to do.
I mentioned Xdeck because when they announced their project they were going with opti-isolators to overcome this problem. In a couple of exchanges they seemed to be well aware of the electrical issues involved well before I mentioned them.
That they've gone off half-cocked, or into oblivion this has no bearing on this, other than they failed to live up to their promises.
I'm conversant with this problem because years ago I was looking for a similiar solution to switching multiple drives. I wanted to do it "your way" and I was quickly chastised by some other design people who I worked with at the time. They made me see the light, so to speak.
It was then that I started reading up about the hardware involved, long before the hardware hacks on the Xbox.
BTW: Signal loss (attenuation of signal) is also introduced by a powered off device as the device itself is not tri-stated or otherwise taken off the bus.
There are also other problems which arise.
You can't take those filtering caps and resistors "off" the bus either, afterall, can you???
---
Post by: Palmore on June 09, 2004, 08:34:00 AM
Ok, you found the specs on the IDE cable pinouts, congrats, and it only took under a minute.
"Fortunately, the standard data connector has no live power signals, so damage is not typical if the cable is inserted upside-down (though the drive won't work that way, of course!)"
"Power (+5 / +12volts) is supplied over a four pin connector (unless the 50-pin connector is used). There are a number of power (down) modes defined: Active, Idle, Standby, and Sleep.
Active: normal drive operation.
Idle: the electronics power down but still receives commands.
Standby: the drive spins down and the electronics power down.
Sleep: every thing is powered down, the electronics will not respond except for a power reset."
Thus in standby and sleep modes the drive is essentially OFF, as I stated before.
Those links had alot of information, but nothing regarding this issue directly.
You may be right, or it may never cause a problem. From what I know this second after readin every page you sent, and reading everything else I've read, it's not an issue.
When it all comes down to it, in a real world application, as stated before, I've done it, and have YET to have a problem in a year, others have done it, and have never reported a problem. Even if you feel it CAN happen, does not mean it WILL happen.
I think you've read my attempt to get you to dissprove myself all wrong, it's not to be taken in a hostile way, but in a manor to assist the rest of the scene as well as myself. With all the lip flapping we've both done, the only hard concrete evidence that's been provided is myself as well as many others have done this, and not had an issue, you stand alone right now, that's why I gotta give you a hard time so you work hard to "shut me up" and prove your point
In closing.. to answer you initial questions.
We're not talking computers, or production PC's that do this, or anything along those lines, so why would I care to find something that we all know doesn't exist? Stay on topic.
Why must I show you a technical doc saying you can do it, when it's been done and it works and after 365+ days of running it has yet to fail?
Show me a technical doc that says it WONT, and not just IDE specs and standards, we are far from standard on anything here.
Now, you've amde all these points saying it shouldn't work, or it should do this or this could happen, but it doesn't and hasn't and chances are wont. That's how it is no if's and's or but's.
FYI I didn't call you crazy, I'm trying to pick your brian for the information you say you have, and the actuall proof regarding this exact issue, so don't get cranky it's all in the fun of modding and for everyones benifit.
"The ATA-1 standard stipulated that a ATA bus only accommodates two drives. These are addressed using 0 (master) and 1 (slave). Since both drives have an onboard controller, each "hears" all of the IDE commands sent on the IDE bus. Thus, the need is resolved to determine for whom a command is sent. The ATA-1 specification also details two transfer modes in which commands and data may be sent to a disk. These are programmed I/O (PIO) and direct memory access (DMA) [3]. "
That has nothing to do with what we are discussing, cause the powerd off drive do not interfere in anyway with what drive is seen as master and slave.
so in everything you posted and wrote about in your rebutle, what have we learned?
Fortunately, the standard data connector has no live power signals.
Idle: the electronics power down but still receives commands.
Standby: the drive spins down and the electronics power down.
Sleep: every thing is powered down, the electronics will not respond except for a power reset.
No voltage through the IDE cable, seems like we are running in circles, just to chits and giggles I've been asking around at work, the typical answer I get is...
No it wont destroy the IDE Bus, there's nothing that would cause it to fail, no voltage goes through the IDE cable. No technical documentation is needed, cause it wont fail.
Anyway, lunch is over, back to work =(
Ping... come back with a pong
Don't you love a fun debate =)
Post by: opjose on June 09, 2004, 08:39:00 AM
Hook up a IDE interface cable to a FAST scope so you can view the clock line signals, on a PC or Xbox with two devices connected and powered on. Tease out a line...
The clock lines are the easiest...
Lock the scope onto the square wave form using the rising edge as a trigger.
Set the scope so you can read the voltage levels on the graph.
Ok now take a look at the waveform.
It looks like a lower case "n"
Note the level of the hump of the "n". This is the wave voltage level.
It should be about 4.8volts. The threshold for the IDE interface is about 3.8volts and the extra is overhead for safety.
Looks good.
Now zoom in to see the rising edge spike.
You can now see the effects of noise and ringing on the line. You'll normally see a sinusoidal wave form starting with the rising edge spike.
Do exactly the same with three devices connected and one powered down.
The upside is that the "ringing" is attenuated. It's NOT as severe.
This would normally mean that the extra drive would improve things, right?
E.G. less ringing and noise!
However NOW look at the "n" wave form.
First it looks now like an inverted "u" instead and the rising edge has been lopped off a bit (which is not in itself totally bad, the extra drive actually deminishes ringing and the leading edge spike.).
But check it's total rise level (voltage).
Opps, the signal has degraded to just about the triggering threshold of the IDE electronics. About 3.8v
Therein lies the problem!
Add a little noise into the mix and anyone can take it from there.
What I HAVE NOT tried yet, which would be interesting, would be to do the same thing and leave the drive powered ON but disabled via the HDD disable line.
I would like to see the difference on a scope to compare it against the baseline.
There might actually be some improvement this way which would indicate that leaving the drive powered on but disabled may be a better alternative...
However I'd have to see or get results from someone who has done this.
Post by: opjose on June 09, 2004, 09:07:00 AM
Yes there is indeed "voltage" on the lines.
Otherwise the drive could not be woken up by a signal commanding it to do so.
How does data communication work otherwise?
It's about 4.8v PtP, that is voltage.
Believe me I do not want your "project" to be problematic. Quite the opposite.
However I've seen far too many people doing very silly things (electronically) and then wondering why they are getting strange or intermittent results. Much like "electronic experts" who overload their home AC lines and wonder why their breakers are always blowing. They have little actual knowledge, just experience plugging things in.
In most cases everyone gets quite defensive and yet provides nothing tangible to illustrate that they even understand that there ARE things which can be problematic. I'd rather be told why it would not be a problem with some credible arguement.
Thus far all you've given me, is that "it's never been a problem for me.".
That indeed may be. You may also be working with a machine(s) that is optimally working otherwise and not bogged down with Cold Cathode tubes, LED mods, switches, cold solder joints, etc. exacerbating things.
I'd like to hear though more specifically why it isn't a problem and how the electronics were designed for this type of additional fanout.
AFAIK and everything I read indicates otherwise though.
I also still wonder about those 29 wire mod chips tapping the data and address lines in such an unconventional way.
There is NO doubt that they work, and I've used them, but I've also seen "freezing" problems develop over time that could be attributed to glitches that cheapmods and 29 wire mods may introduce or affect over time much like your project...
When I mentioned this YEARS ago, this too was outright dismissed by some hackers... then afterwards it was recommended that the connection lines be kept short. heh... it was basically the same/similiar problem, but it was being helped by the shorter lines...
Post by: Palmore on June 09, 2004, 11:45:00 AM
So does this all answer your question?
It falls along the lines of attempt at your own risk if you are choosing to go with a power line swapping method.
Now Opjose, not being an EE myself (dunno what gave that away lol
) could you test the powerd on/disabled via the disable line? The more we have available to us the better, and no one seems to know this better then yourself.And as you asked, my tangible form of proof is "it hasn't been a problem for me"
, but I've always been, if it's not broke don't fix it kinda guy lol.Good points, and thank you for showing us the light.
Post by: opjose on June 09, 2004, 01:45:00 PM
| QUOTE (Palmore @ Jun 9 2004, 07:45 PM) |
| Ahh there is the honey I was looking for, excelent reply So does this all answer your question? It falls along the lines of attempt at your own risk if you are choosing to go with a power line swapping method. Now Opjose, not being an EE myself (dunno what gave that away lol ) could you test the powerd on/disabled via the disable line? The more we have available to us the better, and no one seems to know this better then yourself.And as you asked, my tangible form of proof is "it hasn't been a problem for me" , but I've always been, if it's not broke don't fix it kinda guy lol.Good points, and thank you for showing us the light. |
If I can grab one of those fast TeK scopes over the weekend I'll give it a shot.
The ones I have at home are too slow for this.
I'm curious about the enable/disable line as it does seem to point to some inherent capacity to deal with offlined devices.
Post by: opjose on June 09, 2004, 01:57:00 PM
See: http://forums.xbox-s...0
Post by: opjose on June 09, 2004, 06:17:00 PM
| QUOTE (carmoniousmonk @ Jun 9 2004, 11:40 PM) |
| the idea that i am tossing around is one hard drive and one internal dvd with an output to an external dvd. so even though the connection to the external dvd will be infrequent and left unplugged in the back of the box, when it is plugged in and there are three actual devices on the ide cable, i might experience a similar type of problem. how would the disable line idea affect a 2 dvd - 1 hdd configuration? thanks. |
Exactly the same way.