Long dead time after write operation
Long dead time after write operation
I found that the HxC2001 is dead after write operation for about 300 - 600 ms.
Some old 8-inch FD I/F doesn’t accept this condition. For example, write operation of the IBM FM format (not MFM) fails with the DSD 4140 for DEC PDP-11 (reading is OK). I think it is not a problem of the format of the disk image file because both reading and writing work with another 8-inch FD I/F.
I checked signals from/to the HxC2001. The HxC2001 doesn’t respond after rising edge of the WRITE GATE signal as shown in the linked file. The DSD 4140 seems to check INDEX pulses and/or READ DATA signals, and causes "no read data or drive not ready" error.
Is it possible to avoid long dead time after write operation?
coredump
https://www.dropbox.com/s/e8sxtlztcolsk ... e.jpg?dl=0
Some old 8-inch FD I/F doesn’t accept this condition. For example, write operation of the IBM FM format (not MFM) fails with the DSD 4140 for DEC PDP-11 (reading is OK). I think it is not a problem of the format of the disk image file because both reading and writing work with another 8-inch FD I/F.
I checked signals from/to the HxC2001. The HxC2001 doesn’t respond after rising edge of the WRITE GATE signal as shown in the linked file. The DSD 4140 seems to check INDEX pulses and/or READ DATA signals, and causes "no read data or drive not ready" error.
Is it possible to avoid long dead time after write operation?
coredump
https://www.dropbox.com/s/e8sxtlztcolsk ... e.jpg?dl=0
Re: Long dead time after write operation
This delay is directly related to the write speed of the sdcard. Try to use class 10 SDCard (Sandisk Extreme Series for example).coredump wrote:I found that the HxC2001 is dead after write operation for about 300 - 600 ms.
Some old 8-inch FD I/F doesn’t accept this condition. For example, write operation of the IBM FM format (not MFM) fails with the DSD 4140 for DEC PDP-11 (reading is OK). I think it is not a problem of the format of the disk image file because both reading and writing work with another 8-inch FD I/F.
I checked signals from/to the HxC2001. The HxC2001 doesn’t respond after rising edge of the WRITE GATE signal as shown in the linked file. The DSD 4140 seems to check INDEX pulses and/or READ DATA signals, and causes "no read data or drive not ready" error.
Is it possible to avoid long dead time after write operation?
coredump
https://www.dropbox.com/s/e8sxtlztcolsk ... e.jpg?dl=0
The linked image seems to show a format process. There is maybe also a problem regarding the index signal state which may generate this error. Which firmware version are you using ?
BTW we are developing a totally different hardware design without this flash slow down effect.
Re: Long dead time after write operation
Thanks, Jeff.Jeff wrote: This delay is directly related to the write speed of the sdcard. Try to use class 10 SDCard (Sandisk Extreme Series for example).
I tried class 10 SDHC card but still have delay problem.
Yes. Previous one is the format operation.Jeff wrote:The linked image seems to show a format process.
Linked image was of the beta firmware, V1_8_6_14a.Jeff wrote:There is maybe also a problem regarding the index signal state which may generate this error. Which firmware version are you using ?
I installed v1.8.2.40 and got signals again as linked image. The RDATA signal doesn't come back in 4 ms.
coredump
https://www.dropbox.com/s/iji8nj43u47w7 ... 2.jpg?dl=0
Re: Long dead time after write operation
Have you tried normal write (not format) ?
Re: Long dead time after write operation
Hi Jeff,
coredump
My second post is not format but normal write.Jeff wrote:Have you tried normal write (not format) ?
coredump
Re: Long dead time after write operation
And you still have error with write sector ? What is the write latency in this case ?
Re: Long dead time after write operation
Yes. I still have problem in write operation.Jeff wrote:And you still have error with write sector ?
I guess DSD 4140 controller requests valid RDATA within small time after rising edge of WGATE signal. This controller disables DRIVE SELECT 4 ms after rising edge of WGATE in my second post and generates "no read data or drive not ready" error.
Re: Long dead time after write operation
4 ms ? ouch... i wonder what the controller expect with such small time. The write recovery time of some floppy disk drive are longer than this. For sure the sd / flash write completion can't be done into this period.coredump wrote:Yes. I still have problem in write operation.Jeff wrote:And you still have error with write sector ?
I guess DSD 4140 controller requests valid RDATA within small time after rising edge of WGATE signal. This controller disables DRIVE SELECT 4 ms after rising edge of WGATE in my second post and generates "no read data or drive not ready" error.
Anyway there is maybe a way to fool the DSD 4140 by sending a pattern during the sd write. Have you a schematic of this FDC ? Which part are used on this board ? (a board scan may be interresting).
Re: Long dead time after write operation
I think the DSD 4140 has something strange behaviour. Generally, FDD defines minimum recovery time (typically 590 - 690 us) but does't do maximum.Jeff wrote:4 ms ? ouch... i wonder what the controller expect with such small time. The write recovery time of some floppy disk drive are longer than this. For sure the sd / flash write completion can't be done into this period.
Anyway there is maybe a way to fool the DSD 4140 by sending a pattern during the sd write. Have you a schematic of this FDC ? Which part are used on this board ? (a board scan may be interresting).
I have not yet found schematic of it. I have just user's manual. I downloaded it from the link in the following website:
http://www.filewatcher.com/m/040002-01_ ... 572-0.html
The user's manual shows the READY signal is not used but it is not true. The DSD 4140 requests the READY signal like typical 8-inch FDD. I think this FDC has many traps.
I'll prepare a picture of DSD 4140 on Monday.
I will get more signals. I have another Q-bus controller which works with HxC2001 (both read and write). Also, DSD 4140 works fine with real 8-inch and 3.5-inch FDDs (YD-174, YD-180, TEAC FD-235HG and MITSUMI D353M3D); I will check signals with real FDD.
coredump
Re: Long dead time after write operation
I took pictures of the DSD4140 controller.Jeff wrote:Which part are used on this board ? (a board scan may be interresting).
Parts side
https://www.dropbox.com/s/eqmrvyqrp3v6c ... e.jpg?dl=0
Solder side
https://www.dropbox.com/s/w9e9ifgpuj7z1 ... e.jpg?dl=0
This controller generates the STEP signal 700-770us after the rising edge of the WGATE signal. Does the HxC2001 accept that timing?
Step signal after rising edge of WGATE:
https://www.dropbox.com/s/1bf68vtleh55l ... e.jpg?dl=0
Re: Long dead time after write operation
Yes it does (the track change is recorded). Did you have this step with the HxC ?coredump wrote:
This controller generates the STEP signal 700-770us after the rising edge of the WGATE signal. Does the HxC2001 accept that timing?
Re: Long dead time after write operation
Thanks, Jeff. I'll try to find other timings.Jeff wrote:Yes it does (the track change is recorded).
That was the signals with real 3.5-inch FDD.Jeff wrote:Did you have this step with the HxC ?
Anyway, I have uploaded new images of signals with real 3.5-inch FDD and HxC2001.
The DSD 4140 controller becomes crazy after writing, generating strange step pulses, when it is connected to HxC2001. The DS becomes high after rising edge of the WGATE signal, then the controller generate strange step pulses.
I have not yet found why I cannot use HxC2001 with the DSD 4140. What kind of signals should I get?
coredump
Real 3.5-inch FDD
INI command of RT-11
https://www.dropbox.com/s/jq07z3ptm0jbx ... 2.jpg?dl=0
https://www.dropbox.com/s/79srhsti9cmws ... 3.jpg?dl=0
COPY command of RT-11
https://www.dropbox.com/s/qfpqtv0ynirdp ... 9.jpg?dl=0
HxC2001
INI command of RT-11
https://www.dropbox.com/s/8fhryykrb0k7a ... 1.jpg?dl=0
https://www.dropbox.com/s/7tt2ksiwlzvvg ... 5.jpg?dl=0
COPY command of RT-11
https://www.dropbox.com/s/ymq7o2ch03401 ... 7.jpg?dl=0
https://www.dropbox.com/s/76w614wnnhoqx ... 8.jpg?dl=0