KryoFlux Public Beta 1.0 - AVAILABLE NOW! - Page 8

Loedown · 11 May 2010, 16:57

No new updates SPS people?

Your last WIP is a bit thin on info.

mr.vince · 12 May 2010, 15:37

Summer... real life... job... vacation... It's just. a bit silent but we're making progress. Please bear with us...

Loedown · 12 May 2010, 17:21

Quote:

Originally Posted by mr.vince

Summer... real life... job... vacation... It's just. a bit silent but we're making progress. Please bear with us...

ETA WIP?

Please

Amiga1992 · 13 May 2010, 16:16

I will ask something that may have been asked before so forgive me in advance if thsi is the case.
Can this device be used with, for example, emulators, to read real media and use it with an emu for native support of disks?

IFW · 13 May 2010, 19:18

Yes, but you risk damaging the disk - normally it's very fragile by now.
You are way better off imaging it first then using the image.

orange · 14 May 2010, 23:50

it works with 3" floppies?
(sorry if already someone asked)

Interceptor · 21 May 2010, 22:05

there is no reason why it shouldn't, and we certainly intend to use it with one!

it will just need an extra adaptor cable (or a custom cable) to convert the 24pin header to the 34 pin one.

i for one will be playing with the 3" drives once i get a production board in my hands.

Loedown · 11 June 2010, 16:59

Updates on Kryoflux, it's been rather quiet lads, production schedule or problems?

Interceptor · 12 June 2010, 00:58

in short, it all works, its done.

there are a couple of minor issues being sorted with the final board design, all boring stuff like power flow and resistors in the right places etc, ensuring that the boards dont go wrong if looked at in a funny way etc.

so really the next thing i expect you will hear will be, come buy one!

mr.vince · 12 June 2010, 09:16

New protoypes coming in next week. These are RCs. If all is fine, we will start making boards shortly after. We will also offer preregistration, too make sure people can enqeue. We expect the first batch to sell out very quickly.

Dic_Ray · 12 June 2010, 10:46

Will we see the final Kryoflux in Essen to 25th birthday of the Amiga?

gilgamesh · 12 June 2010, 11:15

I can't download the beta releases. Are they intentionally revoked, or just temporally unavailable?

fiath · 13 June 2010, 09:18

For some reason, the server that holds the KryoFlux distribution files is dead, unfortunately it will probably won't be restarted by the hosting company until tomorrow.

gilgamesh · 14 June 2010, 13:51

The server is running again. Do you know an easy way to display the schematics without Eagle in, say, Kicad or gEDA?

mr.vince · 14 June 2010, 16:05

These are EAGLE files so you need... errm... Eagle to open them. There's a free version available which can do exactly what you want.

A word of warning: We identified several small errors in layout which were changed before prototype production. But we did not fix them yet in the beta release because there was no beta release after these changes.

So please don't produce any bords based on this information, otherwise you end up with broken boards. If you just want to see how things work, please feel welcomed to take a look.

22 July 2010, 18:31

I'm a noob - just found out about this project - so please excuse if my questions are rudimentary. I have some old TI 99/4A home computer disks with assembly language source code that have become unreadable.

I had been contemplating a project such as Kryoflux that would heavily oversample the data from my unreadable floppies, then post-process them on the PC to see if they can be recovered. If this was unsuccessful, I was even considering modifying a floppy drive so that I could digitize the analog signal from the head to provide more information to post-process. Hopefully, the oversampling provided by Kryoflux will eliminate the need for a floppy modification.

It might be worthwhile to include a hook so that the AT7S256 onboard A/D converter could be used to acquire data. The max sample rate is 533 ks/s at 8 bits with an 8 MHz ADC clock. However, if the ADC clock can be run faster, then higher rates might be usable with reduced number of bits/sample.

Am I correct in assuming that the software - both the PC and the embedded ARM7 code - will be released as open source?

If so, when is the planned initial release date for the source code?

IFW · 22 July 2010, 21:31

Chances are you don't want to do that - on the fairly advanced drives you are likely to use to read a disk these days the glitch flitering etc should be very efficient already.

Your usual reading problems may stem from different sources such as physical surface damage, magnetic damage or bitshifts. Other contributing factors can be various alignment and mechanical issues.

What makes the task difficult is that:
- bit cell tracking can fail due to slight glitches (PLL or similar level)
- a dedicated FDC will try to deal with errors on its own (microcode, firmware or software level)
- on top of that the OS level may reject bad reads as reported by the FDC in the first place (OS level)

Since we are sampling the flux reversal periods none of these obstacles have any effect on the data read and translated.
How the flux reversals are interpreted (if at all) is up to the host software.
It is already very efficient in dealing with hard to read disks, including dealing with the original drive rotation or wobble; finding the correct cell width automatically etc.

It can recover very hard to read data as is and also does allow for saving damaged data - as much as can be recovered.

Doing a quick search on TI it seems to be using a generic FM/MFM controller.
As such, KryoFlux will deal with most of these problems for you automatically already.

It is also possible to take a look at a stream dump made with KryoFlux and see the kind of damage you have.

What is worth exploring though - if the data you want to recover is extremely valuable worth investing hundreds of hours and tons of money - is to read it with a drive that has a much finer TPI for its stepper motor than the normal ones, ie read "between" the tracks and then reconstruct.

23 July 2010, 03:14

If the magnetic field on the media is weakened to the point where flux reversals don't reliably exceed the threshold circuitry, then capturing the weakened, pre-threshold signal might provide a method for recovering more of the data. Of course, the drawback to this approach is that one would have to tap into the read electronics of the drive directly, since the analog read signal isn't available on the interface. My suggestion to read the analog data from the head read circuitry prior to the threshold operation was a worst-case recovery option.

I understand that the the oversampling and capture of the read data is occurring immediately at the drive interface, before the data separator logic. The data/clock separation and decoding of the oversampled data occurs on a computing platform (a PC) with orders of magnitude more processing power and time than what is available in the floppy controller, thus potentially yielding results even when the magnetic pattern on the media is severely degraded.

In regards to my particular media: There were two flavors of disk controllers for the TI 99/4A. The original TI version (by far the most popular) was based on an Western Digital FD1771, the standard at that time for floppy controllers, which only supported FM recording (90Kbytes/side). Later, 3rd party controllers became availalbe that used either the FD1791 or the NEC uPD765 and had MFM support. My media is written in FM, so it should be the most tolerant of the formats to any degradation and hopefully the easiest to recover..

I suspect that I'll be able to recover the data just using the oversampling techniques, as you suggested. My point was that it might be worth giving consideration to at least reserving one of the A/D pins and bringing it out to a connector. This way, the hooks would be in place to allow a separate 'probe' to be added and attached to the requisite spot in the drive read electronics circuitry for those instances when 'extreme forensics' might be needed to extract data from the media. This attachment point would vary from drive to drive and would require some knowledge or reverse engineering of the drive - hence my characterization of this technique as 'extreme'.

BTW, do you have any information on if/when the ARM7 and PC source files will be released?

IFW · 23 July 2010, 11:07

Indeed, both of the last resort solutions mentioned by you and me are a bit extreme

Hopefully there is no real physical damage on the disks...

Recovery can be based on encoding violations; those could give hints about bitshifting in the case of FM.
So do timing changes in the data window.

As for source release: the host will be officially released when it gets out of beta, which is when draft support has been finalized/added.
Some developers do have early access to it already though.
The fw is not getting released for the time being.

IFW · 23 July 2010, 11:14

btw: you can download the beta from here:
http://www.softpres.org/news:2010-03-16

22 July 2010, 21:31	#157
IFW Moderator Join Date: Jan 2003 Location: ... Age: 52 Posts: 1,838	Chances are you don't want to do that - on the fairly advanced drives you are likely to use to read a disk these days the glitch flitering etc should be very efficient already. Your usual reading problems may stem from different sources such as physical surface damage, magnetic damage or bitshifts. Other contributing factors can be various alignment and mechanical issues. What makes the task difficult is that: - bit cell tracking can fail due to slight glitches (PLL or similar level) - a dedicated FDC will try to deal with errors on its own (microcode, firmware or software level) - on top of that the OS level may reject bad reads as reported by the FDC in the first place (OS level) Since we are sampling the flux reversal periods none of these obstacles have any effect on the data read and translated. How the flux reversals are interpreted (if at all) is up to the host software. It is already very efficient in dealing with hard to read disks, including dealing with the original drive rotation or wobble; finding the correct cell width automatically etc. It can recover very hard to read data as is and also does allow for saving damaged data - as much as can be recovered. Doing a quick search on TI it seems to be using a generic FM/MFM controller. As such, KryoFlux will deal with most of these problems for you automatically already. It is also possible to take a look at a stream dump made with KryoFlux and see the kind of damage you have. What is worth exploring though - if the data you want to recover is extremely valuable worth investing hundreds of hours and tons of money - is to read it with a drive that has a much finer TPI for its stepper motor than the normal ones, ie read "between" the tracks and then reconstruct. Last edited by IFW; 22 July 2010 at 21:38.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
WinUAE 1.4.6 public beta series	Toni Wilen	support.WinUAE	84	02 February 2008 13:29
WinUAE public beta #5 + #6 + #7 + #8 +#9	Toni Wilen	support.WinUAE	66	30 December 2006 22:07
WinUAE public beta #3 and #4	Toni Wilen	support.WinUAE	5	21 December 2006 19:46
WinUAE public beta #2	Toni Wilen	support.WinUAE	15	06 December 2006 10:39
WinUAE public beta #1	Toni Wilen	support.WinUAE	24	02 December 2006 16:07

11 May 2010, 16:57	#141
Loedown Precious & fragile things Join Date: Feb 2009 Location: Victoria, Australia Posts: 1,946	No new updates SPS people? Your last WIP is a bit thin on info.

12 May 2010, 15:37	#142
mr.vince Cheesy crust Join Date: Nov 2008 Location: Hawk's Creek Age: 48 Posts: 1,383	Summer... real life... job... vacation... It's just. a bit silent but we're making progress. Please bear with us...

13 May 2010, 16:16	#144
Amiga1992 Registered User Join Date: May 2001 Location: ? Posts: 19,645	I will ask something that may have been asked before so forgive me in advance if thsi is the case. Can this device be used with, for example, emulators, to read real media and use it with an emu for native support of disks?

13 May 2010, 19:18	#145
IFW Moderator Join Date: Jan 2003 Location: ... Age: 52 Posts: 1,838	Yes, but you risk damaging the disk - normally it's very fragile by now. You are way better off imaging it first then using the image.

14 May 2010, 23:50	#146
orange Registered User Join Date: Apr 2007 Location: Belgrade Posts: 567	it works with 3" floppies? (sorry if already someone asked)

21 May 2010, 22:05	#147
Interceptor Registered User Join Date: May 2002 Location: Essex, UK Posts: 414	there is no reason why it shouldn't, and we certainly intend to use it with one! it will just need an extra adaptor cable (or a custom cable) to convert the 24pin header to the 34 pin one. i for one will be playing with the 3" drives once i get a production board in my hands.

11 June 2010, 16:59	#148
Loedown Precious & fragile things Join Date: Feb 2009 Location: Victoria, Australia Posts: 1,946	Updates on Kryoflux, it's been rather quiet lads, production schedule or problems?

12 June 2010, 00:58	#149
Interceptor Registered User Join Date: May 2002 Location: Essex, UK Posts: 414	in short, it all works, its done. there are a couple of minor issues being sorted with the final board design, all boring stuff like power flow and resistors in the right places etc, ensuring that the boards dont go wrong if looked at in a funny way etc. so really the next thing i expect you will hear will be, come buy one!

12 June 2010, 09:16	#150
mr.vince Cheesy crust Join Date: Nov 2008 Location: Hawk's Creek Age: 48 Posts: 1,383	New protoypes coming in next week. These are RCs. If all is fine, we will start making boards shortly after. We will also offer preregistration, too make sure people can enqeue. We expect the first batch to sell out very quickly.

12 June 2010, 10:46	#151
Dic_Ray Registered User Join Date: Jul 2004 Location: Germany Posts: 105	Will we see the final Kryoflux in Essen to 25th birthday of the Amiga?

12 June 2010, 11:15	#152
gilgamesh Linux snob Join Date: Sep 2008 Location: Monkey Island Posts: 997	I can't download the beta releases. Are they intentionally revoked, or just temporally unavailable?

13 June 2010, 09:18	#153
fiath Moderator Join Date: Jan 2002 Location: South East / UK Age: 46 Posts: 1,930	For some reason, the server that holds the KryoFlux distribution files is dead, unfortunately it will probably won't be restarted by the hosting company until tomorrow.

14 June 2010, 13:51	#154
gilgamesh Linux snob Join Date: Sep 2008 Location: Monkey Island Posts: 997	The server is running again. Do you know an easy way to display the schematics without Eagle in, say, Kicad or gEDA?

14 June 2010, 16:05	#155
mr.vince Cheesy crust Join Date: Nov 2008 Location: Hawk's Creek Age: 48 Posts: 1,383	These are EAGLE files so you need... errm... Eagle to open them. There's a free version available which can do exactly what you want. A word of warning: We identified several small errors in layout which were changed before prototype production. But we did not fix them yet in the beta release because there was no beta release after these changes. So please don't produce any bords based on this information, otherwise you end up with broken boards. If you just want to see how things work, please feel welcomed to take a look.

22 July 2010, 18:31	#156
urbite Posts: n/a	I'm a noob - just found out about this project - so please excuse if my questions are rudimentary. I have some old TI 99/4A home computer disks with assembly language source code that have become unreadable. I had been contemplating a project such as Kryoflux that would heavily oversample the data from my unreadable floppies, then post-process them on the PC to see if they can be recovered. If this was unsuccessful, I was even considering modifying a floppy drive so that I could digitize the analog signal from the head to provide more information to post-process. Hopefully, the oversampling provided by Kryoflux will eliminate the need for a floppy modification. It might be worthwhile to include a hook so that the AT7S256 onboard A/D converter could be used to acquire data. The max sample rate is 533 ks/s at 8 bits with an 8 MHz ADC clock. However, if the ADC clock can be run faster, then higher rates might be usable with reduced number of bits/sample. Am I correct in assuming that the software - both the PC and the embedded ARM7 code - will be released as open source? If so, when is the planned initial release date for the source code?

23 July 2010, 03:14	#158
urbite Posts: n/a	If the magnetic field on the media is weakened to the point where flux reversals don't reliably exceed the threshold circuitry, then capturing the weakened, pre-threshold signal might provide a method for recovering more of the data. Of course, the drawback to this approach is that one would have to tap into the read electronics of the drive directly, since the analog read signal isn't available on the interface. My suggestion to read the analog data from the head read circuitry prior to the threshold operation was a worst-case recovery option. I understand that the the oversampling and capture of the read data is occurring immediately at the drive interface, before the data separator logic. The data/clock separation and decoding of the oversampled data occurs on a computing platform (a PC) with orders of magnitude more processing power and time than what is available in the floppy controller, thus potentially yielding results even when the magnetic pattern on the media is severely degraded. In regards to my particular media: There were two flavors of disk controllers for the TI 99/4A. The original TI version (by far the most popular) was based on an Western Digital FD1771, the standard at that time for floppy controllers, which only supported FM recording (90Kbytes/side). Later, 3rd party controllers became availalbe that used either the FD1791 or the NEC uPD765 and had MFM support. My media is written in FM, so it should be the most tolerant of the formats to any degradation and hopefully the easiest to recover.. I suspect that I'll be able to recover the data just using the oversampling techniques, as you suggested. My point was that it might be worth giving consideration to at least reserving one of the A/D pins and bringing it out to a connector. This way, the hooks would be in place to allow a separate 'probe' to be added and attached to the requisite spot in the drive read electronics circuitry for those instances when 'extreme forensics' might be needed to extract data from the media. This attachment point would vary from drive to drive and would require some knowledge or reverse engineering of the drive - hence my characterization of this technique as 'extreme'. BTW, do you have any information on if/when the ARM7 and PC source files will be released?

23 July 2010, 11:07	#159
IFW Moderator Join Date: Jan 2003 Location: ... Age: 52 Posts: 1,838	Indeed, both of the last resort solutions mentioned by you and me are a bit extreme Hopefully there is no real physical damage on the disks... Recovery can be based on encoding violations; those could give hints about bitshifting in the case of FM. So do timing changes in the data window. As for source release: the host will be officially released when it gets out of beta, which is when draft support has been finalized/added. Some developers do have early access to it already though. The fw is not getting released for the time being.

23 July 2010, 11:14	#160
IFW Moderator Join Date: Jan 2003 Location: ... Age: 52 Posts: 1,838	btw: you can download the beta from here: http://www.softpres.org/news:2010-03-16

Currently Active Users Viewing This Thread: 1 (0 members and 1 guests)