CS PPC SCSI advice??

[Boot_WB]

I think I am having some termination problems on my Cyberstorm PPC scsi.  

Here's what I have:

A short SCSI cable with a terminator on the end (how do I tell if it's Active?!?), one connector in the middle, then one at the other end.

So I have:

End of cable Terminator --> CS PPC <-- ACARD 7720UW SCSI-to-IDE and then my IDE drive.

This is your problem. I have yet to find any UW scsi device which has built-in active termination - they are all passive in my experience.

For the CSPPC UW-SCSI with one device you need a minimum of four connectors on your 68-way SCSI cable:

1 - Active terminator
2 - CSPPC SCSI header
3 - Device
4 - Active terminator

The ACard UW SCSI-IDE adapter (AEC 7720UW) is a 16-bit device - this is what your cyberstorm is communicating with (The IDE device attached is irrelevant as far as the cyberstorm is concerned).

The only jumpers on the ACard are Terminator Enable and Terminator Power.  Both are enabled.

The ACard manual does not specify if the built-in-termination is active or passive, and the lack of components on the board (compared to an active terminator) suggests passive. This is the most likely cause of your problems.

I would strongly recommend that you get yourself two active terminators. I've used ones similar to these with no problems (although they looked slightly different) - the key is that it is 68-pins wide, and supports SE (single ended) modes, as the CSPPC operates SE only, not L/HVD (low/high voltage differential). Consequently all attached devices must support SE modes.

Once these are attached to each end of the chain, disable any termination on other devices (eg the ACard adapter).
Activating termination Power on multiple devices will not do any harm, but you shouldn't need to activate it on more than one device in the chain.

-I *HAVE* to set the bus-width to 8-bits.  If I set to 16-bits I get a parity error message from the cybppc.device (under os 3.9).
-The drive works fine in OS 3.9, set to 8-bit synchronous.
-I *have* had a lockup under OS 3.9 from which I think was a SCSI lockup.

All of which are probably due to poor termination.

-The drive is NOT detected under OS 4.0 after the OS4 kernel loads and the machine reboots.  The OS4 cybppc.device scans the SCSI bus but does not detect any devices.

Perhaps the OS4 driver is more picky - I have no insight here.

Regards



Rich

PS - I've set up SCSI chains several times on CSPPC, and currently have (in storage) a CSmk3 with SCSI chain (set up as described above) and ACard 7720UW adapter working perfectly.

PPS - You can mix 8-bit and 16-bit devices on the same bus, but both ends of the chain must still be actively terminated on all 16-bits. This is done by essentially extending the lower 8-bits into another chain (but you must actively terminate them also). Eg:

1 - Active terminator
2 - CSPPC SCSI header
3 - Device
4 - Upper 8-bits Active terminator (leaving lower 8-bits unterminated)
(8-bit chain continues)
5 - 8-bit device
6 - Lower 8-bits Active terminator

This is a more complex thing to try AFTER you have a simple (ie all 16-bit) chain up, running and stable. It's not a configuration I would recommend for early troubleshooting!

[Boot_WB]

HammerD,

Your bus is set to 8-bit.

Your terminators (both ends) are terminating upper 8 bytes only. This is what they are designed to do and this is why you ran into, with this issue.

Sorry, but this is completely incorrect. An UW active terminator terminates all 16 lines independently - look at the pcb if you do not believe this to be the case.

You need terminators that terminate lower 8 bytes.

This is true for the bus in 8-bit mode only. However, HammerD has a AEC 7720UW 68-pin SCSI > 40-pin ATA adapter. This is a 16-bit device which is communicating with the SCSI host (albeit the host is currently forcing it to operate in 8-bit mode). Ideally it wants to be working in 16-bit (ie wide) mode, hence he needs 16-bit activer termination at each end.

Your chain must look something like this:

active term (low 8 b) --- CSPPC --- Hard Drive + acard with term power enabled and termination disabled --- active term (low 8 b).

This will work, but only in 8-bit mode. A 16-bit terminator at each end will work just as well, and will also allow 16-bit operation.
However, this is a overcomplicating the issue I fear.

You might need to aquire a longer cable for your particular setup / hardware, due to the acard thingy being at one end of the chain.

You also might need to clarify with the manufacturer if the UW acard can successfully fallback to SCSI-II levels.

Interesting question. Testing under OS3.9 would seem to indicate "Yes", but it could be causing some problems I suppose.
Another source of issues could be a 'dodgy cable' (spacing beteween connectors too far/too close), but I still strongly suspect termination to be the issue.

[Boot_WB]

I found this on Seagate's site.  Indicates that the Seagate drives after 1994 have active termination via the TE jumper:

There's some good information in there, given that you have a ST39173W, which would indicate that this is equipped with activer termination. If so, great.

Re your LVD/320 terminator, LVD is a totally different SCSI mode which is incompatible with SE (single ended). The termination specifications are different (see wikipedia page on SCSI). The Cyberstorm PPC is SE - not LVD. The terminators I have bought in the past have specifically supported both LVD and SE buses. Your LVD/320 terminator may not support SE buses - I cannot say. However, if you have two of the Seagate drives with active termination you could perhaps try one of these at each end of the chain.

I guess in the case of my A4000T with SCSI hard drive I would have:

ACTIVE Terminator --> CS PPC <-- Seagate ST39173W.  

On the Seagate hard drive I would have TE (terminator enable) jumpered and also Terminator Power From Drive.

Looks good, assuming that the LVD/320 terminator also supports SE buses (see above).

I'm not sure about Parity - do I need that enabled?  There is a jumper on the drive for "Enable Parity check of SCSI Bus".  The Default is "enabled".

You can try with/without on this one - it will do no harm.

If there is a 'Force SE' jumper, you may as well activate it (CSPPC SCSI is a SE bus)

For my  A4000 desktop system with the ACARD:

ACTIVE Terminator --> CS PPC <-- ACARD with Terminator Enabled and Terminator power Enabled, then my IDE drive attached.

I'm still not 100% sure if I need an extra active terminator after the ACARD....some people don't seem to have them.

The ACard bridges may - or may not - have active termination. I do not remember if I have  tried the termination on them.
However, I've tried with various hard drives and CD/DVD drives with built-in termination over the years, and nothing has ever worked as well as proper dedicated terminators for me. also, nothing comes close in terms of reliability (and simplicity of troubleshooting if things DO go wrong) as having the correct setup in the first place.
Also, SCSI buses which are not terminated can still work - albeit less reliably, with more data corruption, more chance of bus lockups, etc, etc, etc. It's so particular to each setup: cable length (wrt bus frequency and phase of reflected pulses), spacing between devices, etc.  

The CyberstormPPC scsi bus is a tricky beast even when set up correctly.