Not a module as such. Just a RTC IC

maxtherabbit wrote on 2023-02-20, 03:34:

Not a module as such. Just a RTC IC

Exactly this. Many boards have traces to support the DS12885, an external 32768Hz quartz crystal and a CR2032 battery option, but that option is almost never populated. I half-converted the DS12887 into a DS12885: The modded module uses an external battery, but it still uses the internal potted quartz crystal.

Since we're a bit off topic, I hope no one minds another slight deviation on my part:

I got fed up with my Asus VL/I-486SV2GX4 in my rig and installed the HOT-433 in its stead. I think the SIS VLB implementation on the Asus was premature and buggy (another story for another thread).

So far, everything works fine, exept that the BIOS does not like the way I have my CF Card set up. On the Asus, I had it configured thusly in the BIOS:

TYPE: User
SIZE: 1669
CYLS: 974
HEAD: 255
PRECOMP: 0
LANDZ: 15537
SECTOR: 63
MODE: LBA

Mind you, with a PCI IDE card inserted in the HOT-433, this works fine, but with these settings in the BIOS, I get message on boot that the HDD has failed.

Also, when I input these settings in the Shuttle, it (correctly) reports the size of the CF Card as 8011M (whereas the Asus reports it as 1669MB).

I could probably get this working if I wipe the CF card and start fresh, but I already have several partitions on it and would rather not. And it is detected by the BIOS despite the error.

Any other suggestions?

mockingbird wrote on 2023-02-21, 00:58:

Since we're a bit off topic, I hope no one minds another slight deviation on my part: ... Mind you, with a PCI IDE card inserted […]
Show full quote

Since we're a bit off topic, I hope no one minds another slight deviation on my part:
...
Mind you, with a PCI IDE card inserted in the HOT-433, this works fine, but with these settings in the BIOS, I get message on boot that the HDD has failed.
...
Any other suggestions?

I'm not sure whether your topic has anything to do with adding a PS/2 mouse port to a UMC based 486 motherboard.

And, yes, we use a HighPoint PCI IDE controller in that mainboard to overcome the PIO limitations.

I suggest you to create an own topic for that issue.

Disruptor wrote on 2023-02-21, 21:29:

<snip>
And, yes, we use a HighPoint PCI IDE controller in that mainboard to overcome the PIO limitations.

I decided to follow a different route. PIO mode will be sufficient for this setup. I'm using a 16GB SSD with a SATA to IDE bridge on the UM8886BF controller. I have it set up as master along with a DVDRW as slave on the primary. Using both primary and secondary had issues.

The best thing I suppose would be an Ultra 133 TX2, which works with RLOEW's TRIM. I tested it with this setup and it worked... But I thought it was a bit excessive (and I think it was responsible for stuff not being loaded high into memory).

mockingbird wrote on 2022-12-26, 02:49:

The PS/2 worked perfectly fine with my Avocent KVM (tested with Microsoft Mouse driver and Monkey Island). All I can suggest is to try a different KBC. I used a Mega-KB revision F. I must have pulled this thing off of something decades ago. I never thought I would end up using it.

Indeed: Using a Mega-KB (labeled "Mega-KB-HW-P"), mouse and keyboard do work without any problems on the Belkin OmniView Pro KVM.
The erratic keypresses I had when using the KVM switch are gone. Was using a AMIKEY-2 before. Thanks for the hint!

I came across this excellent topic while repairing my HOT-433.
The keyboard doesn´t work, all the wirings are o.k. - ground, Vdd, KBD clock to pin 57 and KBD data to pin 56. First I tried an external KBD controller and adapted everything like explained in the 1st posting, but omitting the PS2 mouse part for now.
With an ext. KBC the system hangs during POST. I then found that pin 56 of the UM8886BF is "n/c" unlike pin 57 where I measure the typical "diode values" with my ohmmeter.
I think the system hangs, because the "CPU reset" signal from the ext. KBC won´t arrive at the UM8886BF due to it´s blown(?) I/O circuit.
Would there be any way to work around this without replacing the UM8886BF??

majestyk wrote on 2024-08-16, 17:05:

I then found that pin 56 of the UM8886BF is "n/c" unlike pin 57 where I measure the typical "diode values" with my ohmmeter.
I think the system hangs, because the "CPU reset" signal from the ext. KBC won´t arrive at the UM8886BF due to it´s blown(?) I/O circuit.
Would there be any way to work around this without replacing the UM8886BF??

Did you do both directions of the "diode value" check? Some I/O topologies do not have clamping diodes to both Vcc and GND, but at least one should be visible. Furthermore, your comparison to pin 57 makes a lot of sense, as both pins have the same kind of I/O structure (high-drive open collector in "internal KBC", standed input in "external KBC"), so similar clamping behavious is expected. If you measure pin 56 as open circuit, a broken bond wire is indeed likely the case and matches the symptom of a broken keyboard only in "internal KBC" mode and a non-POSTing system (no post codes at all) in "external KBC" mode.

I don't see any way to get internal KBC working if the KBC data connection is broken. For external KBC, you would need to provide a replacement reset circuit (onless you don't require AT-like reset-via-KBC at all, but IIRC this reset style is even used by some BIOSes in their POST code). Furthermore, a dedicated reset circuit would only work if the broken CPU reset input to the UM8886 does not cause any internal resets in that chip. For example, I suggest to avoid any "fast A20 gate" functionality if the keyboard controller reset input failed. As the idea of the keyboard-controller generated reset pulse is to reset the CPU only without resetting the remaining system, there is some hope that the 8886 does not perform unintended side effects even if pin 56 is broken.

The CPU will likely not initialize correctly without a valid power-on reset (POR). It is quite likely that UMC integrates the POR circuit in the 8886, and uses the 8886 to control the CPU reset signal, both for power-on reset and for keyboard-controller initiated reset. According to the schemtatics of a Pentium laptop containing the 8886 south bridge, CPU reset output is at pin 29, and master reset input is at pin 24. Possibly you get a solution by disconnecting pin 29, and providing your own reset signal generated from both pin 24 and pin 56.

I checked again and while pin 57 has a low resistance with the red probe to ground, pin 56 has "infinite" resistance both with red and black probe to ground.
I also traced pin 56 - it´s connected to the KBD pin of the DIN connector and to pins 8 (output of inv. 6) and 2 (output inv. 1) of the 7406 hex-inverter. The input of inverter 1 is connected to pin 24 (P23/MSCO) of the external KBC, the input of inverter 6 is connected to pin 38 (KBDO) of the external KBC.

While the external KBC was present, the system stopped at some very early POST-code (that I don´t remember).

Before proceeding I´ll try to find a donor-board for the southbridge. It´s a UM8886AF (not BF) btw.

It´s probably not the best solution to have pin 56 connected directly (without buffer or DC separation) to the keyboard connector. Any EMC issue or DC voltage at the DIN connector can potentially blow the UM 8886.

majestyk wrote on 2024-08-20, 06:35:

Before proceeding I´ll try to find a donor-board for the southbridge. It´s a UM8886AF (not BF) btw.

It´s probably not the best solution to have pin 56 connected directly (without buffer or DC separation) to the keyboard connector. Any EMC issue or DC voltage at the DIN connector can potentially blow the UM 8886.

I'm sorry if I wrote 8886BF, I didn't mean that new edition. The BF is known to implement in internal PS/2 mouse port (which might not be available omn the 8886F or 8886AF), and while the Biostar MB8433-UUD-A uses the "B" revision, I wasn't even sure if the Shuttle HOT433 uses the plain 8886F, or the "A" revision. I don't know how the A revision differs from the original revision. Don't mind the "F" at the end, all UMC chipset chips have it (and AFAIK, all UMC multi-I/O chips as well). It might be a company-wide suffix indication the TQFP package, which is quite meaningless in this case, as the 486 chipset chips by UMC are only available as TQFP anyway.

The direct connection of pin 56 to the keyboard connector is considered a feature, because it allows for cheaper boards with less components. Adding a buffer on that line is non-trivial, because the keyboard signals are bidirectional. Adding a capacitor for DC separation will also not work reliably, because the keyboard interface protocol does not use a DC-free coding scheme. On the other hand, all boards should have an LC filter (L in the signal, C to ground), to catch low-energy ESD pulses and flatten the edges, so the L/C filter and the ESD protection diodes are supposed to protect the chip "good enough for consumer purposes". It's clearly not an industrial-strength design, though.

This was my fault, I brought the "BF" type up myself and found it´s AF when I had a coser look, so I was just correcting my own error 😉

In the meantime I replaced the UM8886AF by a UM8886BF frome some donor board.

Everything´s working fine and I finally have a keyboard again!
I wonder if the PS2 mouse header would work with this revision without adding an additional KBC?

majestyk wrote on 2024-08-29, 16:22:

I wonder if the PS2 mouse header would work with this revision without adding an additional KBC?

The integrated PS/2 mouse port is a new feature in the 8886BF. At least revisions 1 to 3 of the HOT433 do not contain the traces required to support the internal mouse port of the 8886BF. If your board is "rev. 4", you might want to trace whether you can find connectivity from the PS/2 header to pins 199/200 of the 8886BF (possibly interrupted by a non-populated 0-ohm resistor network). See the chapter "integrated mouse port" in the OP.

I have a v4 mobo. I removed the AT keyboard port and soldered in a PS/2 Keyboard/Mouse port, but (obviously) the mouse doesn't seem to work. Is there anything else I should check other than the integrated mouse port wiring?

UPDATE: JP38-40 were jumpers for switching LA19-17. It seems that the default setting (soldered wires, not jumpers) was to connect to LA17-19 of other ISA slots, not to wires from the chipset. Also, JP41 and 42 (no pin headers or wires soldered) had DATA/CLK from the PS/2 port and wiring from the chipset, so I soldered the pin headers and put in jumpers. I just realized that I've lost my PS/2 mouse, so I'll test it after I find it.