A while ago i felt a desire to connect two computers with token ring NICs with each other, without having to hunt eBay for a MAU.
Hence, i beeped out the PCB of an IBM 8226, two ports only, tossed the result in Kicad and got some PCBs made.
I won’t claim to fully understand what the schematic of the ciruit does, but i figured that shouldn’t be necessary. Famous last words i guess.
Long story short, i can’t get the thing to work.
There are no obvious layout mistakes that i would have noticed after receiving and assembling the PCB. From a distance it appears entirely identical to the original.
At this point I have exhausted my ability to debug the problem myself.
Hence my request: Is anybody here willing to have a look at it?
I am happy to supply a ‘debugging kit’ consisting of one fully assembled PCB, an original IBM 8226 and a Soekris net5501 with a TR NIC and ancient Linux that still has the TR drivers.
Eventually I’ll want the IBM 8226 and the Soekris back. The assembled PCB can be kept as ‘compensation’.
Here are the project files: 
Please let me know if you are inclined to take a look.
 GitHub - blackbit42/MAUreene_8226: Token Ring MAU
super interesting project and I wish I was able to help - but sadly I’m overloaded and backlogged with many existing projects, so I really cannot put another on on my table at this point
ok, so I finally found a bit of time to look at this.
Disclaimer: I know close to nothing about Token Ring at the physical layer. My observations are just based on general EE knowledge.
As per information received in-person from @blackbit, the functionality of the MAU is as follows:
- in idle state, the NC (normally closed) relay presents a loop-back towards the Token Ring adapter
- the Token Ring adapter, when brought up,
- verifies the presence of that loopback
- applies a DC bias voltage
- that DC bias voltage is used to activate the relay to insert the interface into the loop
The protoype functionality has been described as:
- the relay doesn’t switch at all when bringing the token ring interface up
- the opto-coupler triggers (and hence the relay switches) when placing a finger on the PCB beneath it
The second topic is probably explained by the fact that base of the opto-transistor is exposed on pin 6 while the collector with +5v is at pain 5. So your finger presents a resistor between +5V and the base, which will make the collector-emitter path conduct, which in turn pulls up the base of Q2, which drives the relay.
Some general hints/ideas:
- manually apply a DC voltage between J1A/6 and J1A/5 (simulating the DC phantom voltage) and see if this makes the relay switch. Use a current limited power supply to limit to something like 60…100 mA
- if that works, then it means the TR adapter doesn’t send that DC voltage, or it sends it at a voltage lower than your test power supply, i.e. the resistor R10 might need to be reduced
- if that doesn’t work, there’s a bug somewhere in the circuit. Check the LED voltage on the optocoupler U1, as well as the base voltage of Q2 as you enable/disable your lab power supply simulating the phantom DC voltage.
- measure if the TR adapter applies a DC voltage betwen pins J1A/5 and J1A/6 when you bring the interface up
- if that doesn’t work, it likely means that the initial loop-back is not detected by the TR adapter
- if that works, it meams somehow the U1/Q2 circuit to drive the relay doesn’t work as expected