Z8S18020vsg versus Z8S18033vsg

[Nick Brok]

HI,

I have a very strange thing happening:

I bought a Z8S180020 and a Z8S18033. In the first one the clock multiply register doesn't work (0x1e). After reset it reads 0xff

Using the Z8S180033 doesn't have this problem. The clock multiply register (0x1e) reads after reset 0x3f and it work.

Both chips come from ZIlog. Do anyone have a clue on this? Reading the datasheet doesn't tell anything about the difference between the two versions. The only dfference is the maximal clock 20 MHz versus 33 MHz.

The representative of Mouser does also not understand this problem :-(

Greetings,

Nick de pe1goo

[Steve Cousins]

Hi Nick,

I've had similar problems. 

See here: https://groups.google.com/d/msg/retro-comp/acwYaqnlTkI/BzjZNX4uFAAJ

It appears there have been a number of functional revisions to the chip but the part number has not changed.

Steve

[Phillip Stevens]

IIRC the CMR has to be set before setting the CCR, otherwise it doesn’t work correctly. The first thing to check you’re doing.

Specifically setting CCR first leaves CMR unable to be set to doubled clock.

P.

[Steve Cousins]

Another difference I have found in chips that do not seem to have the clock multiplier is the use of DCD0. Those with the multiplier also allow DCD0 to be disabled as a serial port control line and used as a general input. To allow for this configuration feature not always being present I have a jumper on DCD0 so that is can be used as a general purpose input (where supported) or tied to zero to enable serial receive (when not supported).

Steve

On Thursday, 8 August 2019 13:53:19 UTC+1, Nick Brok wrote:

[Nick Brok]

The register reads 0xFF, so the bit seemed to be set already.... but the clock isn't multiplied. My other z8s180 (33MHz) works as expected.

Op donderdag 8 augustus 2019 15:18:19 UTC+2 schreef Phillip Stevens:

[Wayne Warthen]

Hi Nick,

It would appear that the Z8S180 that will not multiply the clock is a Z8S180 Rev. K.  The Rev. K does not have the clock multiplier capability.  The register will always read 0xFF no matter what order it is written to.  Your other Z8S180 that does achieve 33MHz must be a Rev. N which does support the clock multiply bit.

Sadly, the revision of the Z8S180 is basically not marked on the chip at all.  Additionally, suppliers do not understand this nor show it in the part specs.

If you are using RomWBW, you will see either Z8S180-K or Z8S180-N listed in the initial boot message which indicates which chip variant you have.

I am attaching a tech note from Zilog that specifies the differences in the revisions.

-Wayne

[Nick Brok]

Wayne,

Thanks for this info. So the chips are not defective, but  have not the multiplier function... Now I can tell this to the supplier. I have no romwbw, can you point me to the code that checks this?

So my 33 MHz versions are the newer ones and those 20MHz are older, am I right?

Op vrijdag 9 augustus 2019 17:46:44 UTC+2 schreef Wayne Warthen:

[jopil]

A rather more complete answer from the past (2004):

https://www.embeddedrelated.com/showthread/comp.arch.embedded/5940-1.php

[Steve Cousins]

Unfortunately "The only way to tell which part is which is that the Rev K version has "SL1960" printed on the package whereas Rev N does not;" is not correct.

I have chips here which RomWBW identifies as revision K and that do not have features like the clock multiplier, but are not labelled "SL1960". There appears to be no way to identify these as revision K short of trying them. And that assumes there are no other revisions in the wild. Perhaps the chips I have are actually revision J or L. Who knows because they are no identifiers I can see. 

I think RomWBW classifies the chips as either K or N so other slight revision changes may be incorrectly described.

Steve

[Wayne Warthen]

On Friday, August 9, 2019 at 9:34:41 AM UTC-7, Nick Brok wrote:

Wayne,

Thanks for this info. So the chips are not defective, but  have not the multiplier function... Now I can tell this to the supplier. I have no romwbw, can you point me to the code that checks this?

So my 33 MHz versions are the newer ones and those 20MHz are older, am I right?

 

RomWBW tests for Rev. N by writing 0x00 to ASCI1 Time Constant Low Register (I/O Address: 0x1C), then reading it back.  If the value read back is 0xFF, then the chip is designated as a Rev. K.  If the chip had been a Rev. N, the counter would read back as 0x00.  The I/O address 0x1C will actually be the offset from the Z180 internal register base address.