UMB
in UMB3
probably stands for universal motherboard
.
Apparently its designers have thought of themselves very highly, as the PCB
silk screen reads UMB3 MENSA
in the front left corner next to the
power switch.
The motherboard features a power supply with 12 VDC power input,
an 8-port 10BaseT Ethernet hub, two WAN module slots with an external RJ45-style
jack for each, and console and auxiliary ports.
The WAN modules are mounted internally and normally don't have their own
external connectors on the rear panel.
Instead the WAN 1
and WAN 2
RJ45-style jacks are
actually located on the motherboard, and two separate internal connections
are provided to each WAN module: one is the passive connection to the external
jack and the other is the main internal interface.
The CPU is MC68MH360, which is the classic QUICC from Motorola with the
Multi-HDLC
additional feature.
The board design looks overly complex. There is a discrete PC16550DV UART
(the 4 SCCs plus SMCs in the QUICC weren't enough?), a GAL22V10D PLD, a bunch
of random logic and a bunch of headers serving unknown (debug?) purposes.
There is some RAM on the board (4 MB) as well as two SIMM slots;
I have never seen a unit with the latter populated.
The Ethernet hub is implemented in two LXT915 chips and presumably
connected to one of the SCCs on the MC68MH360 in the Ethernet mode.
It seems that the interface between the motherboard and the wanlets is more
like a synchronous serial DTE-DCE interface rather than a microprocessor bus,
and each WAN port is connected to one SCC on the QUICC.
Note the use of an MH
part rather than the regular MC68360:
I wonder if they have used the Multi-HDLC
feature with their ISDN
wanlets.
The part of real curiosity is the auxiliary port
. It's a DE15F
connector (obviously non-standard) that is claimed to serve two completely
different purposes. On units functioning as routers it supposedly acts as
an EIA-232 asynchronous serial port for connecting an analog modem for dial
backup, yet on the D-series pseudo-DSUs it becomes a
V.35
synchronous serial port instead.
The same port is also claimed to be capable of Apple LocalTalk mode operation.
Where this port actually fits hardware-wise is a mystery — my first
guess would be the last SCC on the QUICC (last after having one taken up
by Ethernet and two more going to the WAN slots).
The transceiver chips seen around it as well as the console port are 26LS30s
and 26LS32s.
There is only one flash chip on the board, and it's a soldered TSOP. The capacity is 1 MB. There is also a socketed PLCC28 which at first seems to be some kind of boot or configuration PROM, but it's actually a GAL22V10D PLD (as revealed when one peels the sticker off).
Hmm, the firmware flash is soldered but the PLD is socketed? Makes one wonder if the PLD might differ between various router and pseudo-DSU products based on this motherboard. The sticker on the PLD reads the same programmed part number (5000040-00-01) on our D7100-C and R7200-T units though, so maybe it's invariant after all.
As always, the greatest flaw in Netopia's products is software, not hardware. Could we possibly hack the UMB3 platform by putting our own software on it? In theory yes, but in practice it may not be the best option: this discussion has been moved to its own page.