I just managed to get my hands on a top hat flash.
it is an empty plcc32 socket to flip over an soldered in flash chip. attached to the socket is a secondary flash mem and some circuitry to handle the "chip enable" signal, i.e. pull down the primary one and daisy chain the data/addr bus.
now I wonder, whether it is risky to use it on a non Elitegroup mainboard and it might brick a mobo with no matching circuitry. Ideally, I hope it works on a Gigabyte M57. should I give it a go ? --Q
On Sun, Apr 29, 2007 at 08:06:08PM +0200, Quux wrote:
it is an empty plcc32 socket to flip over an soldered in flash chip. attached to the socket is a secondary flash mem and some circuitry to handle the "chip enable" signal, i.e. pull down the primary one and daisy chain the data/addr bus.
I'd love to find out more about exactly how this is done.
now I wonder, whether it is risky to use it on a non Elitegroup mainboard and it might brick a mobo with no matching circuitry.
Depends on if they do something special on their boards to allow it to work safely, or if the top hat itself is safe enough.
Ideally, I hope it works on a Gigabyte M57. should I give it a go ?
Please do, but don't blame me if the board breaks. :)
Myself, I would investigate both the top hat and the mobo it came from thoroughly before trying the hat on another board.
//Peter
On 4/29/07, Peter Stuge stuge-linuxbios@cdy.org wrote:
On Sun, Apr 29, 2007 at 08:06:08PM +0200, Quux wrote:
it is an empty plcc32 socket to flip over an soldered in flash chip. attached to the socket is a secondary flash mem and some circuitry to handle the "chip enable" signal, i.e. pull down the primary one and daisy chain the data/addr bus.
I'd love to find out more about exactly how this is done.
Probably a lot like the TiVo PROM back-to-back socket. Basically two plcc32 sockets soldered back to back with a few pin tweaks. Yet another variation on dead-roaching or piggy-backing a chip.
"tivo prom piggy" should get you there as a search query. I've occasionally wondered how hard it would be to have something like this manufactured, thereby bypassing the whole TopHat / BIOSSavior vendor specific issue.
Regards, -dhbarr.
--- "David H. Barr" dhbarr@gozelle.com wrote:
"tivo prom piggy" should get you there as a search query. I've occasionally wondered how hard it would be to have something like this manufactured, thereby bypassing the whole TopHat / BIOSSavior vendor specific issue.
The OpenGraphics project (http://OpenGraphics.org, http://www.TraversalTech.com) has people with extensive knowledge in hardware manufacturing and a desire to create hardware with completely open specifications. They have been discussing the possibility of making products other than graphics cards in order to generate capital.
It would be great if LinuxBIOS users/developers could make a list of products they would be interested in purchasing (BIOSSaviors and Flash chips come to mind) and send it to the OpenGraphics developers, who can be reached via their mailing list at:
http://lists.duskglow.com/mailman/listinfo/open-graphics
Thanks, Vlad
__________________________________________________ Do You Yahoo!? Tired of spam? Yahoo! Mail has the best spam protection around http://mail.yahoo.com
On Sun, Apr 29, 2007 at 12:05:21PM -0700, Vlad wrote:
The OpenGraphics project (http://OpenGraphics.org, http://www.TraversalTech.com) has people with extensive knowledge in hardware manufacturing and a desire to create hardware with completely open specifications.
While there's people like that also on this list this is a good point and something to keep in mind if and when we want to produce something.
They have been discussing the possibility of making products other than graphics cards in order to generate capital.
The problem with most if not all products that would be used for/with LinuxBIOS is that they either have a very small market or must reach a very low target price, which means there is little to no capital to be gained from the process. Indeed I think the problem with the product ideas we've had so far is that of financing the production rather than the technical knowledge needed. It's just really expensive to make stuff. :(
(And then there's sales and distribution if it's a consumer product rather than developer product.)
//Peter
David H. Barr schrieb:
Probably a lot like the TiVo PROM back-to-back socket. Basically two plcc32 sockets soldered back to back with a few pin tweaks. Yet another variation on dead-roaching or piggy-backing a chip.
"tivo prom piggy" should get you there as a search query. I've occasionally wondered how hard it would be to have something like this manufactured, thereby bypassing the whole TopHat / BIOSSavior vendor specific issue.
I 'll keep you guys informed. I even forked out those €8 (incl. shipping) to that greedy vendor at ebay. ;-)
the tivo socket has no circuitry at all. --Q
On Sun, Apr 29, 2007 at 01:38:44PM -0500, David H. Barr wrote:
I'd love to find out more about exactly how this is done.
Probably a lot like the TiVo PROM back-to-back socket. Basically two plcc32 sockets soldered back to back with a few pin tweaks.
Yes, with "a lot like" and "a few tweaks" being the gaps I'd like to fill. :p
Yet another variation on dead-roaching or piggy-backing a chip.
"tivo prom piggy" should get you there as a search query.
Not really with a lot of detail unless I registered at the DDB forum. Oh how I hate forums.
Anyway, at least one of the posts indicated that a trace needs to be cut, and one wire needs to be soldered. This solves the problem I'm interested in, but I'm only interested in ways to accomplish the same thing completely without soldering.
I've occasionally wondered how hard it would be to have something like this manufactured, thereby bypassing the whole TopHat / BIOSSavior vendor specific issue.
The BIOS savior is very generic. Some issues may come from the choice of flash chip inside it, but there would probably be issues on other boards with another chip if that's the actual cause of the reported problems.
TopHat I don't know about. I'm interested in Quux' findings! :)
On Sun, Apr 29, 2007 at 09:44:43PM +0200, Quux wrote:
the tivo socket has no circuitry at all.
Really? Is there a schematic somewhere?
//Peter
On 4/29/07, Peter Stuge stuge-linuxbios@cdy.org wrote:
On Sun, Apr 29, 2007 at 01:38:44PM -0500, David H. Barr wrote:
I'd love to find out more about exactly how this is done.
Probably a lot like the TiVo PROM back-to-back socket. Basically two plcc32 sockets soldered back to back with a few pin tweaks.
Yes, with "a lot like" and "a few tweaks" being the gaps I'd like to fill. :p
I'm requesting permission from a "jmayes" to reproduce his/her documentation under an open license of some kind, preferably in the LinuxBIOS wiki so someone with more skill than me can properly update it (eg. add schematics as mentioned later).
I've occasionally wondered how hard it would be to have something like this manufactured, thereby bypassing the whole TopHat / BIOSSavior vendor specific issue.
The BIOS savior is very generic.
It may be very generic, but it's also apparently covered by patents no. 130648 and 177592 (I find no reference as to which country in which said patents are issued). I imagine similar problems may exist with TopHat.
On Sun, Apr 29, 2007 at 09:44:43PM +0200, Quux wrote:
the tivo socket has no circuitry at all.
Really? Is there a schematic somewhere?
No schematic, but see above comment re: request tendered to one "jmayes".
-dhbarr.
David H. Barr schrieb:
It may be very generic, but it's also apparently covered by patents no. 130648 and 177592 (I find no reference as to which country in which said patents are issued). I imagine similar problems may exist with TopHat.
the tivo socket has no circuitry at all.
Really? Is there a schematic somewhere?
No schematic, but see above comment re: request tendered to one "jmayes".
like they should abandon trivial patents, really. The guy who invented the MS patented double mouseclick - he thinks he is Einstein or what ?
On Sun, Apr 29, 2007 at 04:30:34PM -0500, David H. Barr wrote:
Probably a lot like the TiVo PROM back-to-back socket. Basically two plcc32 sockets soldered back to back with a few pin tweaks.
Yes, with "a lot like" and "a few tweaks" being the gaps I'd like to fill. :p
I'm requesting permission from a "jmayes" to reproduce his/her documentation under an open license of some kind, preferably in the LinuxBIOS wiki so someone with more skill than me can properly update it (eg. add schematics as mentioned later).
Keep us posted.
The BIOS savior is very generic.
It may be very generic, but it's also apparently covered by patents no. 130648 and 177592 (I find no reference as to which country in which said patents are issued). I imagine similar problems may exist with TopHat.
Oh ouch patents yes, that kind of vendor issues.
On Sun, Apr 29, 2007 at 11:44:54PM +0200, Quux wrote:
like they should abandon trivial patents, really. The guy who invented the MS patented double mouseclick - he thinks he is Einstein or what ?
Too many patents that are granted aren't very "novel" - I agree.
//Petre
Peter Stuge schrieb:
the tivo socket has no circuitry at all.
Really? Is there a schematic somewhere?
I also had to bl**dy register with the forum. Those two pdf's just show photo's, so for plain wiring no one bothered to draw any schematics seemingly. I guess it is pretty much established, that it's all about two plcc in parallel with #CE activated individually. not that difficult after all. using a flip over socket avoids cutting the #CE line on the mobo if there is no chipset output driving those 2 #CE signals. the Gigabyte patent also includes the firmware procedure to verify flash content and to automatically select a valid chip without user intervention.
maybe one day they ship M57's with full dual bios : one legacy and the other LinuxBIOS activated by default YEAH ! :-) --Q
On Sun, Apr 29, 2007 at 11:39:01PM +0200, Quux wrote:
Those two pdf's just show photo's,
Oh. :(
I guess it is pretty much established, that it's all about two plcc in parallel with #CE activated individually. not that difficult after all.
Definately. This goes for any flash. I made http://stuge.se/ceswitch.png when discussing this with Anton Borisov.
(Note only one switch should be closed at a time. I had no better symbol at hand.)
using a flip over socket avoids cutting the #CE line on the mobo if there is no chipset output driving those 2 #CE signals.
True, but there usually is.. Right? This is especially true on buses where more than one device is attached. E.g. LPC, with both the flash and a superio.
the Gigabyte patent also includes the firmware procedure to verify flash content and to automatically select a valid chip without user intervention.
Yeah, it was using a watchdog that had to be disabled by the first BIOS software tried within a certain time. If not, the system would reset and try the other BIOS.
maybe one day they ship M57's with full dual bios : one legacy and the other LinuxBIOS activated by default YEAH ! :-) --Q
The big money win is when LinuxBIOS is the default and the other one is blank, then the board will cost $10-$15 less. (License fee Gigabyte pays for the Phoenix BIOS.)
Having the full dual BIOS circuit populated would be nice, but I'm not sure all their customers would want to pay even $3 extra for it, since they would never use it.
It's already very cool that it's so simple to add the manual switch to the board!
//Peter
Peter Stuge wrote:
On Sun, Apr 29, 2007 at 11:39:01PM +0200, Quux wrote:
Those two pdf's just show photo's,
Oh. :(
I guess it is pretty much established, that it's all about two plcc in parallel with #CE activated individually. not that difficult after all.
Definately. This goes for any flash. I made http://stuge.se/ceswitch.png when discussing this with Anton Borisov.
(Note only one switch should be closed at a time. I had no better symbol at hand.)
using a flip over socket avoids cutting the #CE line on the mobo if there is no chipset output driving those 2 #CE signals.
True, but there usually is.. Right? This is especially true on buses where more than one device is attached. E.g. LPC, with both the flash and a superio.
The #CE signal on the flash device is active when the signal is LOW. That is what the # in front of the CE stands for. The #CE voltage is pulled high via a pull-up resistor to the VCC rail of the flash device. The driving signal from the chipset is an open drain on an output buffer FET. The FET actually just pulls down or sinks the current from the pull-up resistor to alow level when the #CE is active. The chipset doesn't supply a high or low level voltage tothe flash device #CE pin.
I'll post a pdf schematic shortly. I hope this ends all the confusion about this.
-Bari
On Sun, Apr 29, 2007 at 05:31:53PM -0500, Bari Ari wrote:
The #CE voltage is pulled high via a pull-up resistor to the VCC rail of the flash device.
Yes, but is the pullup usually internally in the flash chip?
The driving signal from the chipset is an open drain on an output buffer FET.
Thanks for pointing this out. There is no risk of damage on an open drain bus as long as all "inputs" to the bus only try to sink current.
In order to control any signal it needs to be proxied or jammed.
The former is what we've been discussing so far and will require cutting a track and some soldering.
The latter permits solderfree chip select override and is what many game console modchips do but for address/data/read/write instead of chip select.
They simply drive the signal "harder" than the chipset so that even if the chipset is sinking current the flash chip will still see a high input.
When wanting to change what the chipset says should be 0 into a 1 externally the chipset will see a lot of current through the output and if there's too much the output will just break unless it is protected against short circuits.
Maybe every single output on every chip does have clamping? Then I don't have to worry, but I think that's too good to be true..
Is it? :)
//Peter
Peter Stuge wrote:
On Sun, Apr 29, 2007 at 05:31:53PM -0500, Bari Ari wrote:
The #CE voltage is pulled high via a pull-up resistor to the VCC rail of the flash device.
Yes, but is the pullup usually internally in the flash chip?
They usually have them but they aren't relied on. It is more reliable to use an external pullup resistor.
-Bari
On Mon, Apr 30, 2007 at 12:12:41AM +0200, Peter Stuge wrote:
The big money win is when LinuxBIOS is the default and the other one is blank, then the board will cost $10-$15 less. (License fee Gigabyte pays for the Phoenix BIOS.)
? Are you saying the per-board royalty is *that* much? I thought it was about 10% of that - $1 or $2?
Thanks, Ward.
On Sun, Apr 29, 2007 at 07:19:14PM -0400, Ward Vandewege wrote:
On Mon, Apr 30, 2007 at 12:12:41AM +0200, Peter Stuge wrote:
The big money win is when LinuxBIOS is the default and the other one is blank, then the board will cost $10-$15 less. (License fee Gigabyte pays for the Phoenix BIOS.)
? Are you saying the per-board royalty is *that* much? I thought it was about 10% of that - $1 or $2?
I don't know how much Gigabyte pays, maybe they have a good deal with Phoenix, maybe Phoenix charges less, but another BIOS vendor charged $15 regardless of volume. I was shocked too.
//Peter
http://www.dealdatabase.com/forum/showthread.php?p=259752
the tivo piggy pack socket for secondary BIOS chip
David H. Barr schrieb:
Probably a lot like the TiVo PROM back-to-back socket. Basically two plcc32 sockets soldered back to back with a few pin tweaks. Yet another variation on dead-roaching or piggy-backing a chip.
"tivo prom piggy" should get you there as a search query. I've occasionally wondered how hard it would be to have something like this manufactured, thereby bypassing the whole TopHat / BIOSSavior vendor specific issue.
I 'll keep you guys informed.
here are some of my findings:
the "top hat flash" ( = thf ) does not necessarily brick your board if non-Elitegroup. it does, however, not boot either if the chip types (spare and original) do not match. (no display on POST card) some boards have poorly placed bioses where thf is impossible to hot unplug.
I believe, thf will we useable on a GA M57SLI board with soldered on plcc32 bios. Advantage : no need to solder or rewire. --Q
David H. Barr schrieb:
I'd love to find out more about exactly how this is done.
http://de.babelfish.yahoo.com/translate_url?doit=done&tt=url&intl=1&...
http://fab51.com/workshop/bios/dual_bios1.html
in babelfish gives som insight. It looks like it would work with many different flash types. --Q
On Sun, Apr 29, 2007 at 10:33:57PM +0200, Quux wrote:
I'd love to find out more about exactly how this is done.
http://fab51.com/workshop/bios/dual_bios1.html
in babelfish gives som insight. It looks like it would work with many different flash types. --Q
This requires cutting a mainboard trace and wiring the incoming CE signal to a circuit. I want to avoid this.
//Peter
-
Bari Ari -
David H. Barr -
Peter Stuge -
Quux -
Vlad -
Ward Vandewege