More information...
Pouring over the available schematics, some likely relationships between LANCE chip revisions, board configurations and firmware revisions - are becoming clear.
First I should say that the drawing set is unusually complex (hence "unclear") because it consists of schematics for version
H1 and J1 generations of the DELQA circuit boards. Moreover, between these issues the computer systems used to generate the designs and artwork were changed, requiring the entire database (schematic and artwork) to be re-entered from scratch.
Consequently, although they are superficially similar, it has taken some time to scrutinize these versions to assure they are indeed the same after rework, except where the need for jumpers are no longer required by the newer artwork.
Furthermore, a disclaimer in the package indicates "
Option Revisions were never updated to reflect these changes".
With this in mind, I am seeing the following relationships emerge.
1) The LANCE Ethernet chip (E13) comes in several varieties.
The DEC part numbers for these are as follows...
21-21672-04 REV D LANCE
21-21672-07 REV E LANCE
21-21672-09 ??
21-21672-10 "C-LANCE"
- Notes seem to indicate -9 and -10 are equivalent.
- Rework instructions indicate -07 (and earlier) should be replaced by -10.
- Apparently -7, -9 and -10 all require the same jumper configurations, while others do not.
A nice
Wikipedia article is revealing...
A later refabricated chip called the C-LANCE Am79C90 is made with 0.8 micrometre
CMOS technology. The original NMOS version Am7990 and the CMOS Am79C90 version are differ in some details which may have an impact on
device driver compatibility.
CHIP BUGS
The old LANCE (Rev. C) chips have a bug which causes garbage to be inserted in front of the received packet. The workaround is to ignore packets with an invalid destination address (garbage will usually not match). Of course, this precludes
multicast support.[SUP]
[3][/SUP] The
Amiga SANA-II network interface API has poor multicast support. And this chip bug might be the reason.
No capability for transmit buffer byte count of zero. Receive lockup may occur if bus latency is large. External loopback on a live network may cause reception of invalid loopback failure indications. Receive descriptor zero byte count buffer interpreted as 4096 available bytes. Will
poll computer memory every 1.6
ms for new packets to transmit.
2) Firmware:
There is some confusion over which firmware is required to implement the "TURBO" behavior of the "YM" boards, since both YM and M versions appear to have been initially built with
334/335 ROMs.
Online documents indicate:
M7516-YM ROMs: (v2.0.0)
23-365E5 ( E28? )
23-366E5 ( E29? )
M7516-M ROMs:
23-334E5 ( E28 )
23-335E5 ( E29 )
- From rework instructions, it appears that if your board already has 334/335 ROMs, then it's LANCE chip should already be -10.
- Conversely, if the firmware is older than 334/335, then you might possibly have LANCE chips before -9, in which case not only the LANCE chip would need to be changed, but other rework may be required.
- The "YM" (TURBO) firmware (334/335 and 365/366) must therefore require at least -10 LANCE (possibly -9)
However, since "
YM" Boards appear to have been initially built with
334/335 ROMs, it therefore follows that any boards with them could, and should, be upgradable to firmware
365/366.
3) The DELQA-PLUS Addendum:
Appendix D - has instructions regarding testing for a board that is TURBO capable, and to see if TURBO mode is enabled. It specifically states that Firmware Version 2.0 or above is required for TURBO mode to operate and for the board to respond correctly to these tests.
I should be able to come up with a simple set of ODT commands to verify some of this.
I see no place where DEC might hide a secret bit to enable this function, so it must be contingent only on having capable LANCE chip and minimum firmware Rev 2.0. I'll try to look at the DEC ROM contents for both versions 335/336 and 365/366 to see if I can determine their "Revision" level. [sometimes we get lucky and they're in ASCII]
That's it for tonight. Please, if you have M7516's, take a few minutes to post their particulars.