Here’s another blast from the past – remember when network adapters weren’t integrated into motherboards? Remember when Ethernet ran at 10Mbit/s? Remember when Ethernet came over coaxial cable? Or when Ethernet practically won over Token Ring?
I guess that sets the mood for what I have in my hands today – a DaynaPORT SCSI/Link-T, graciously donated to me by a colleague (Robert) at UNSW.
Believe it or not, this is a SCSI Network Adapter, running Ethernet at 10Mbit/s over Twisted Pair (hence the -T suffix). The adapter was not provided with a power adapter, but a 12v 300-800mA power supply with positive centre tip will suffice.
The rear of the device shows two DB-25 SCSI ports for loop-through daisy chaining of the SCSI bus. There is a label in the corner indicating the hardware revision – K.
The front of the device offers a twisted-pair RJ-45 8p8c connector, with a 2.1mm DC barrel jack for power. There is a selector wheel to select SCSI ID and a switch marked 1 and 0 for termination, where 1 means termination enabled.
The bottom of the device gives more information about the device – made in the USA with model number DP0801.
The company itself doesn’t really exist anymore – searches for DaynaPORT give very few results, although it does confirm the highly sought after status for this item as it allows owners of vintage Ataris and Macintoshes to access the network and internet. However, a trip to the Wayback Machine allows us to peek into what the company was like at around that time.
The earliest archive is in 1996, and it appears it wasn’t early enough. The closest related product is the SCSI/Link-3 which was a multi-mode (AUI, BNC, TP) external SCSI Ethernet Adapter with a product code of DP0802. It appears that, by 1998, Dayna itself discontinued their products after being acquired by Intel. I suppose the Intel PRO Networking cards may owe their heritage to Dayna, or maybe not.
It also came with a very peculiar SCSI cable – a DB-25 to HDI-30. Having been mostly a PC user, seeing such elaborate, heavyweight connectors is a treat.
Cracking it Open
So, lets take a peek inside – to open it up, it’s not hard. The construction was very much ahead of its time – being entirely screwless! Removing and unclipping the dark grey clips allow the top and bottom halves of the casing to separate –
This allows us to see the top of the two-PCB stack. The printing of the PCB states it is made in USA, and is a Dayna SCSI/Link Transmit/Power board. Note the vast number of footprints that lay unpopulated – these appear to be for the BNC 10Base2 support. J10 would have been the BNC plug, T2 would have been the magnetics and U16 would have been for the required DC to DC converter module along with a few other components. So it already seems that Dayna had many “related” products based on this platform.
The underside of the top PCB show us that this is at least a triple layer PCB, likely a quad layer PCB. The number under the logo implies it was made in Week 39 of 1993. The white stamp implies that this has been QC tested. The inter-PCB connectors are a bit strange – they have a very narrow pitch and are very tall and thin.
They are also gold-plated where the contact is made, but appear to be tin plated everywhere else. Hmm. Keeping costs down already?
The bottom PCB shows an unpopulated position where the 15-pin AUI connector would have been. It seems that the DaynaPORT SCSI/Link-3 is virtually identical to this one! We can spot some Samsung RAM, an AMD flash chip with the firmware on it, and most strikingly, the Zilog Z180 MPU (a more modern Z80). It’s like having a computer in itself!
The underside of the bottom PCB shows something interesting – near the corner is the word TRUST which I find may be an easter-egg or a code-name for the project. It appears to be dated Week 43 of 1993. We can see a double-row of pins on the bottom that appears to be used for bed-of-nails testing and diagnostics – a very smart design decision. There are a long row of resistors and what appears to be transistors – this is basically SCSI termination built discretely rather than using resistor packs.
Testing it Out
It’s a bit hard to test it out – just because I have a SCSI controller that works doesn’t mean I can make use of it. I decided to hook it up with power just to test the unit and it was recognized by Lubuntu.
Unfortunately the device is not usable because there are appears to be no driver support for it in the Lubuntu kernel. Hmm. Well I suppose it was worth a try :).
And just in case anyone visiting wants to ask – no, it’s not for sale.