April 14, 2023

Entrex - Interface Board Circuit

This is my "discussion" of the circuit for the above board. I am trying to put into words what I am able to learn by studying the schematic and assigning helpful names to signals.

Overview

There is one edge connector (P15) albeit a big one. There are also two good sized on board connectors CG1 and CG2. Why two? We will get to that.

AJ did a nice job on the schematic with signals flowing from left to right. The right side ends in two 8212 device (8 bit latches) driving 16 open collector outputs (7406 devices). These all drive signals on the P15 edge connector that AJ has designated "printer".

So we may be able to call this a dual function IO board. It drives the printer (and we will sort of ignore that) and it provides data from the pair of interface connectors (maybe).

What are the 8274 and 8273 chips?

These are not in common use these days. They look like Signetics parts.

The 8273 is a 10 bit serial in, parallel out shift register.
The 8274 is a 10 bit parallel in, serial out shift register.

We get a pair of 8274 on the board, but only one 8273. This seems a bit surprising, but will no doubt make sense once we unravel the design (or else I just missed a second 8273).

These chips promise to be the heart of the matter given that the outside world interface for the Entrex is some kind of unique serial protocol.

Once we label the data bus signals, we discover that both 8274 are connected (albeit with some strange issues) to the DO bus on the parallel side, which sort of makes sense. What isn't clear is what causes the 8274 to latch (capture) data from the DO bus. Each chip has 2 control signals S0 and S1 with the following 4 states: 

S0 = low, S1 = low -- hold
S0 = low, S1 = high -- clear
S0 = high, S1 = low -- load
S0 = high, S1 = high -- shift
S0 is pulled high (to Vcc) on both, so we just get the last 2 cases, load or shift. So, what causes S1 to go low and data to be loaded? Both chips are controlled by the same signal u8_11, which I am going to rename LOAD* as an active low signal.

IO ports

I have already worked out the decoders for these on the processor board. Many of the port select signals go to the processor board card edge. I can look at the backplane schematic and figure out where these go and which do or do not go to the interface board. Here is what we see: 
IN 0x78 -- enables data read from UE4 (8T97)
	    This looks like status information (5 bits, 3 high, 2 low)
IN 0x68 -- enables data read from UB3 and UC3 (both 74173)
	    This looks like actual data

OUT 0x78 -- to UC2 pin1 (inverter), clocks data all over the place
OUT 0x70 -- no data, goes to inverter UB7 pin 1
	    (I think this does get 8 bits of data, but don't see how yet).
OUT 0x68 -- no data, clocks a "0" into flip flop UF4A (and more)

IN 0x70 -- enables data read from UC1 (8T98) printer
OUT 0x60 -- upper 8212 for printer
OUT 0x58 -- lower 8212 for printer
The port for "OUT 0x78" shows writes with every sort of data in the ROM

Data bus

Since we write to the printer ports and read from the 8T98 at port address 0x70, we must have access to the DI and DO buses, so let's label those. Again this involves finding the edge connector pins on the processor board, then going to the backplane to find out where they go on the interface board, then we can apply labels.

Terminal plugs CG1 and CG2

These plugs are not actually physically present on the interface board. The interface board only has off-board connections via the edge connector. AJ was just being kind to draw these on the board so I wouldn't have to be chasing signals onto the backplane schematic

To start with, I will focus on CG1. Both connectors have 14 pins, labeled with the letters A through R (skipping G). Pins K, L, and M are unused in both plugs.

The signals from these plugs go to a crew of 75113 and 75115 chips. These are differential driver and receiver chip respectively and each chip handles two signals. There are two 75113 and three 75113, so a person can deduce that each plug has two differential signals outbound and three differential signals inbound. Here is how this looks, skipping unused and non-existant pins: 

A out A minus
B out B plus
C out A plus
D in C plus
E - duplex switch
F out B minus
H in C minus
J in E plus
N in D plus
P in E minus
R in D minus
We can neglect naming the signals between the plug and the 75113/75115 chips, but it would be nice to work up some names for the single ended signals on the other side. How about: 
CG1_outA
CG1_outB
CG1_inC
CG1_inD
CG1_inE
The above will do for now. We can rename them again later as we learn more. This leads to an amazing revelation. CG1 and CG2 connect to the same 5 signals internally! So you could plug into either of them. The outgoing signals are duplicated, the incoming signals are "wire-or" due to the "pullup" flavor output on the 75115 being used. How about that!

A brief look at these 5 signals shows the following: 

CG1_outA - from UC6 pin 8 (a nand gate)
CG1_outB - from UA3 (serial data out from the lower 8274), ie Txdata
CG1_inC  - to pin 10 on UD3 (the D input of the 8273), ie RxData
CG1_inD  - to various logic
CG1_inE  - connecton on both CG1 and CG2 but nowhere else ??
Take note that there is another 8274 (the "upper" on the schematic) and it is entirely unclear what is going on with the serial data coming from it. Why are there two 8274? How does the parallel data get loaded into these from the data bus?

Clock?

So, where does the clock come from for the Interface board? My guess is that some "master" elsewhere provided the clock to run all of this, and we will look in vain for a clock source on the interface board. A mystery as yet, but not a schematic errata.

How about that DM9324 chip?

Made by Fairchild way back when. It is a 5 bit comparator. What we see doing on is that on one side we have 5 switches (actually we have 8, but 3 of those don't do anything). On the other side we have 5 bits from the output of the 8273 serial to parallel shift register. What this looks like is that the switches set the identity of the terminal and the whole shebang is set up to allow the terminal to recognize that it is being addressed. Pin 14 goes high when there is a match. This is UC4 we are talking about. I'll call this signal MATCH. It gets buffered by an AND gate for no apparent reason, and that buffered signal is routed to two other chips, one being the D input to a flip flop. I'll call the buffered signal MATCH and the unbuffered rMATCH (for raw match).

And what about UC3 -- 74173

The 74173 is a 4 bit D type register. So it will store 4 bits and deliver them when reauested. Pin 7 is the clock, which tells it to grab the 4 bits on the input side. Pins 9 and 10 are "data input enables". The schematic shows them as no-connect, which would be bad (we want them pulled low). This may be a schematic errata and these are actually connected to ground (we will assume that). Pins 1 and 2 are "output control" -- both must be low to have this device drive its output with the contained data. Pins 1 and 2 are contolled by inp_68*, so a read by the 8080 on port 0x68 reads from this chip! It only gets 4 bits though, on DI7, DI6, DI5, and DI4. Pin 7 is driven by u14_3 which I am renaming to DATA_CLK.

DATA_CLK comes from the output of an AND gate (pin 3 of UB6), which interestingly enough also clocks a "1" into a flip flop (UF2).

Don't forget the 8T97 and 8T98 chips

The chips in question are at UE4 and UC1 respectively.

The chip at UC1 is "read" by the control signal inp_70*. Note also that this is an 8T98 and inverts its inputs. So a read (IN 0x70) will get what is "behind" this chip. Only 6 bits are read (the upper 6: D7 to D2). Pin 1 is shown as a no-connect, but I believe this is an errata and that it is connected to pin 15 and thusly inp_70*. What is read are 6 pins (with pull-up resistors) from the edge connector that are marked as from the printer, so this is really part of the printer interface.

The chip at UE4 may be more interesting. It does not invert. It produces 5 bits of data when read, on D7, D6, D5, D1, and D0. It is controlled by signal inp_78*

Writing data

Looking at the ROM code, a write to port 0x70 seems to write data. The schematic shows the port 0x70 select signal going to several AND gates, but ending up at the clock input to the lower 8274. I suspect an Errata and that signal should really go to the LOAD input (and that would be for both 8274). This makes a lot more sense. Why we have two 8274 being independently clocked is certainly a mystery.

Mysteries

These are probably schematic mistakes (errata).

Who drives u34_13? I can find no source for this signal. I have renamed it Reset1* as it resets a bunch of flip flops. There is also an errata on u24_8, the other input to the AND gate must have something driving it.

Feedback? Questions? Drop me a line!

Tom's Computer Info / tom@mmto.org