2.8" QD Disk Emulation

[Jeff]

S0 to S8: all have the result 001 001, so I guess it is how it should be.

all of them flipped the WG-assertion shortly after the "disk active" led light went on.

all of them flipped the WG-deassertion at the time when the "disk active" led light went out.

Ok good !

I still adding some debug features in the firmware : The next fw version will save the header bitstream directly from the dma. this will be saved to the end of the QD file. This way i will have the exact pulses timing and i will maybe found some non-mfm timings.

[Jeff]

So attached the fw with the stream capture feature :

EDIT : Removed !

select an image, save something, press gotek buttons to change images and then you can eject the stick to share the image here.
(The stream is saved when you change/eject the currently loaded image)

[hoby]

OK - here is a saved stream

EDIT : blank_S0.QD.zip removed.

[Jeff]

OK - here is a saved stream

Thanks !
In 10 seconds i have spotted the special value right before the sync word !!!

Image2.png (5.98 KiB) Viewed 6174 times

I will check all the sync word values but i think we have found the cause of the issue ! (same for the Roland machines i think...)

[Jeff]

Confirmed ! This controller do an "insane" thing to trigger the sync word detection.

sync.jpg (27.67 KiB) Viewed 6171 times

9.7us, 14,7us, 19,6us are normal values. 4.9us is not normal at all in mfm regarding the bitrate.
The issue with the firmware is that it change this 4.8us duration to 9.7us. Then when the machine read back the track, it doesn't see the 4.8us glitch and doesn't see the sync word right after.

The good news is that this is very easy to fix. A new firmware to try in some minutes.

[hoby]

Aha, interesting. I guess that also explains the sort of timeout when trying to load the files, if it never catches the trigger at the beginning

[Jeff]

The new firmware and a blank image to try with.
Anyway please post the images back here.

EDIT : Removed !

[hoby]

I think you have nailed it.

Both load and verification works !!!

EDIT : blank.QD 3.zip removed !

[Jeff]

I think you have nailed it.

Both load and verification works !!!

Finally !

Thanks for all the tests !




@AMSynths : You should give a try on the Roland S-220 / S-10 !

[hoby]

Man. It was the least I could do. It is you that should be thanked for all the work. I really appreciate it. You have brought life back to this old x7000, and hopefully lots of other people will benefit from this.

[Jeff]

Man. It was the least I could do. It is you that should be thanked for all the work. I really appreciate it. You have brought life back to this old x7000, and hopefully lots of other people will benefit from this.

Spread the word ! (And do a youtube video to show how this works ! )

[hoby]

Good Idea! I'll do that, and also share the link here.

[Jeff]

Official version with the fix :

14 September 2019 : Firmware STM32 USB HxCFloppyEmulator v3.4.7.1a ( https://hxc2001.com/custom_fw/ and WIP_Firmware folder )

- QuickDisk support : Fix MB87013 based QD machines support (Akai S700/x7000, Roland...) -> Now support the special MB87013 sync method.

HXCFEUSB_HFE_beta_firmware.zip

USB_HFE_hxc_floppy_emulator_firmware_release_notes.txt

Custom HxC firmware

[AMSynths]

Hi
Been on holiday, latest Roland test results:

fw 3.4.7.1 loads ok from the latest HxCUSBFE

Could not load the blank QD but OK with Roland Formatted file
SAVE works, but with VERIFY ERROR every time
LOAD works from this saved file

So looks like we just need to sort the VERIFY for the Roland. Any traces/data you want me to do?
I am going to load the Gotek up with all the factory samples and try out my adapter PCB

UPDATE: Success with sample dumping into S-220 the factory library then saving to Gotek USB except for saves that need more than one QD due to the verify errors. Work around is to save each Bank separately.

[Jeff]

Could not load the blank QD but OK with Roland Formatted file

Sure , the blank.qd is blank.

SAVE works, but with VERIFY ERROR every time

blank.qd is a "long" qd image. maybe a "4.6s" image is needed ?

LOAD works from this saved file

ok.

So looks like we just need to sort the VERIFY for the Roland. Any traces/data you want me to do?

Just share the saved image. I will first check if the Roland machine use the same sync "x7000" method.

I will provide you a debug fw version capturing/counting the floppy interface events if necessary.