View Single Post
Old 07 November 2015, 20:09   #11
tnt23
Registered User
 
tnt23's Avatar
 
Join Date: Feb 2008
Location: Saint-Petersburg / Russia
Posts: 324
@SpeedGeek, thanks for your reply.

Quote:
Originally Posted by SpeedGeek View Post
An upgraded Rev. 3.0 A3640 is another serious case of "Your Mileage Will Vary" but you should do the U200 cut and jumper mod as part of the 3.2 upgrade. This is needed to fix the STERM sampling bug.
I made the cut'n'jumper mod but did not feel any difference since.

I have then decided to restore 10ns tap, and this seems to positively impact things. No more errors reported by SCSI tester (http://aminet.net/package/driver/media/stest20).

Quote:
Also, don't forget these A3640's are notorious for electrolytic cap problems
Previous board owner already took care of them by replacing caps with new ones.

Quote:
If modified CLK90 timing didn't solve your SCSI problems then you probably don't have SCSI DMA transfer errors. You most likely have SCSI Target to Host transfer errors.
Well, after I restored original 10ns tap connection, checksum errors became quite rare. Better than it had been before, but they still pop up once in a while.

Quote:
Check your WD SCSI chip Rev. (Stock A3000's shipped with Rev. 4 but most A3000 users have already upgraded to Rev. 8 or the AMD version).
Mine is PROTO WD33C93A-PL 00-04. AMD replacement ordered, expect it to arrive any time soon. I was under impression it could help with removable media, but still worth trying anyway.

Quote:
Also, check your SCSI bus termination, SCSI cabling and set your SCSI Prefs to ASYNCH mode since some SCSI targets are buggy in SYNCH mode.
Resistor packs installed on the mobo, infamous diode checked, TERMPWR measured as +4.5V. Cable itself is 20cm in length, with good connectors. On the disk side (SCSI2SD emulator), termination is also installed.

I am pretty positive I have ASYNC set in SCSI Prefs, but will check.
tnt23 is offline  
 
Page generated in 0.06050 seconds with 11 queries