On 18/06/2009 08:05, Pete Turnbull wrote: Let's dispel a few inaccuracies that have been presented about this. First of all, it *is* possible to have two hard drives in a BA23. However, it was not supported by DEC, partly because the particular drives they supported at the time drew too much power in a fully-populated system, and partly because the chassis would not then accomodate an RX50 or TK50 and would therefore not be field-maintainable (no way to load diagnostics from floppy or tape). Secondly, the RQDXE and RQDX1-E extenders are intended to provide additional connections for external drives, and are not used in a BA23 or BA123 for internal drives. I don't have one of either right now to check but I remember they're both purely passive devices. The RQDXE is M7512 and is used for an RQDX1; the RQDXE is M7513 and used for an RQDX2 or RQDX3. The card in a BA123 for internal distribution is an M9058, and is a buffered device. It's called a Signal Distribution Board, not an extender. The two systems (BA23/BA123) treat the drive selects in somewhat different ways, which is why you can't always simply take a drive configured for a BA123 and move it (successfully) into a BA23. An RQDXn controller provides four drive selects for hard drives. In a BA123 with M9058, each drive select is taken to one (and only one) of the 34-way connectors (J1 to J4) for the drives, but it's connected (unless you change jumpers) to all 4 drive select lines on that connector. However, as is usual, all the other signals are bussed to all connectors in parallel (though some are buffered separately to different connectors to reduce loading effects). This means you can (and normally do) set the drive select on each drive the same way - normally to DS3. This is analogous to the way you would set all drives to the same ID in a PC system, only there you use a twist in the cable to re-route the drive selects. Although all the signals are bussed, many (including Write Gate) are gated /in the drive/ with the drive select. Because each connector is connected to a different drive select, no two drives can be selected at the same time. However, this is not done in a BA23 distribution board, where all the bus signals are simply bussed. Normally in a BA23 you set the first hard drive to DS3, as that's the first drive select the RQDXn tries when it looks for drives. That's why the convention for the BA123 also became to set every drive to DS3. If you use an expander to connect an expansion box, again you set the first drive in the expansion box to DS3 but you set the second drive to DS4. If you add a second internal drive to a BA23, you must do the same thing: set the drive select to DS4. This is actually stated in the microPDP-11 Maintenance Manual, in more than one place. If you set two drives with the same drive select jumpering, they will both be enabled at the same time, regardless of position. Actually, you could set the second drive to any select except DS3, at least for RQDX2 and RQDX3, but you need to be aware that the RQDXn controllers use the same drive select signals for RX drives as for RD drives. In other words, it has only 4 drive select lines for all the drives connected to it, and it tests them in order DS3, DS4, DS1, DS2. You're supposed to reserve DS1 and DS2 for an RX50 if using an RQDX1. I've never tried this in a BA23 with its own distribution board, so I don't know if it actually connects all 4 drive selects. My 11/83 in a BA23 is set up exactly like this -- DU0: uses DS3 and DU1: uses DS4. It's always been this way, and has always worked. Nothing else has been modified. For the record, it has a pair of RD54 drives in it, and has been running 2.11 BSD for years. There is another gotcha, though. The RD console board on some BA23 boxes -- that's the small card just behind the front panel -- only has connections for one drive. That card has two pushbutton switches, for Write Protect and Ready, along with some very simple electronics including pullup resistors and LEDs. When the switches are open, the drive is enabled (able to go Ready) and write enabled (not Write Protected). Usually the same thing happens when the board is absent because the two signals tend to float high (though not always reliably). Each switch is arranged to short the corresponding signal to ground when pushed in. The LED on the Write Protect is connected between +5V (via a 150R resistor) and the WP switch, and lights up when the switch is pushed in (Write Protected). The LED on the Ready is fed via a transistor to invert it, so it lights up when the switch is /not/ pushed in (drive is enabled) but goes out when the signal from the RQDXn pulls the line low (as it does when it accesses the drive), or when the switch /is/ pushed in (not enabled). There are PostScript files showing the schematics of the M9058 and console boards, and the layouts and pinouts of related boards, at http://www.dunnington.u-net.com/public/RQDX/ I've just added PDF copies for the challenged. -- Pete Peter Turnbull Network Manager University of York