|
IBM TDB - Information Panel (block diagram of the 95 XP Op Panel)
General Description General Description Planar provides a 34-pin (2-by-17-pin) Berg connector to the operator panel. A matching connector can be found on the panel itself. The panel consists of: Model 95XP:
Model 95 A:
Model 85:
95 XP Op Panel Schematic (ASY FRU 92F1319, PCB FRU P/N 33F8434, PCB P/N 33F5412)
 (Click on the picture for a hi-res version) No active components aside from the two LED display units and the Power Good LED. SW1 enables/disables the "Remote Power-ON" feature. 95 XP Op Panel Connector Pinout
(?) - needs confirmation 95A Op Panel Schematic (ASY FRU 61G2393, PCB FRU 92F0259, PCB RAW CARD 61G2395)
 (Click on the picture for a hi-res version) Active components: two LED display units, Power Good LED, Hard File LED, and a bunch of 74xx logic ICs. Together the ICs make up a decoder for the "Serial Data Link" interface. There is also a provision for something called "Multimedia Link". R2 is a zero-ohm SMD resistor "jumper". R3 would be another one, but it's not populated. These jumpers change behavior of the "-Remote Power-ON Request" signal (J1 pin 29) in regard to the power switch (SW1). 95A Op Panel Connector Pinout
85 Indicator Panel Schematic (ASY FRU 61G3739, PCB FRU 61G3736, PCB RAW CARD 61G3740)
 (Click on the picture for a hi-res version) Very simple board. No active components aside from the the Power Good LED and Hard File LED. It uses the same connector and pinout as the much fancier Op Panel in Model 95, but most of the pins are unused here. 85 Indicator Panel Connector Pinout
* - signal listed as reserved in the Model 85 Technical Reference Most of the "Reserved" lines are actually connected and wired to support the 9595 Op Panel. Some logic is missing from the planar, but it's possible to bring the feature back. More info HERE. Power Switch
If you can, unsolder the other switch on op panels with two switches. Same type of switch. I used a GC 35-491 (Push On DPDT Power Pushbutton Switch) and it fit perfectly. Pull off the round button and pop the original rectangular button on. Fits perfectly. I had a 9585 that would not turn off reliably. You could push the switch up to five (or more!) times before it would turn off. I tried slowly wiggling the switch and the 85 would come up... LED Display Module It's HDLG-2416LED (datasheet) Thanks to Jelte Roelfsema Smart Alphanumeric Display Built-in RAM, ASCII Decoder, and LED Drive Circuitry Software Controlled Dimming Levels and Blank 128 ASCII Character Set. Each module can display 4 characters with 5-by-7 dot pattern. Now this is where I went to work. Imagine what if you took the codes that are output to the parallel port and hooked them into the HDLG-2416? We would need a hex to ASCII converter to feed the HDLG, but with the capabilities and features afforded by the 2416, why not take the time? The information panel is accessed through an 8-bit data bus, which is controlled through a set of I/O ports. Op Panel Speaker 95 XP and 95A use the same type of internal speaker: 8 ohms, 0.3 W, 66 mm outer diameter. The speaker is marked as: AAP, 66R, 8ohm, 0.3W, PROC Speaker connector pinout:
* - Model 85 Op Panel has another Speaker Ground on pin 2, but this is not used by the speaker itself. The speaker is driven by a simple amplifier that is located directly on the planar board. More info can be found HERE (Describes Model 95A, but audio stage in Models 95 XP and 85 is similar). Serial Data Link (95A Op Panel Only) Serial Data Link aka Serial Diagnostic Link or SDL is a direct communication channel between the Processor Complex (Type 4 only) and the Op Panel. It's explained in greater detail HERE. Serial Data Link Connector (J2) Pinout:
The SDL port is used by the POST routine to output early diagnostic information. Go HERE for more info. The link is mapped to port E1h. If you want to learn how to use it, visit THIS page. The primary benefit of the SDL port is that it can be used to output info to the Op Panel before the system channel is initialized, and even if the planar isn't fully operational. Aside from the +5 V power rail, no other connections between the planar and Op Panel are required for the serial link to operate. As can be seen here:
The two wires connected to the planar connector supply the panel with power, pin 25: +5 V (red wire), and pin 28: ground (black wire). The system was jump-started by shorting pin 1: "-Power-ON Request" to pin 24: ground (this was done directly on the planar). Multimedia Link (95A Op Panel Only) The 95A Op Panel has a provision for something called "Multimedia Link" but the related components are not populated. I'm not sure what exactly was this intended for, but it would likely be some kind of Ultimedia-like option. Possibly a separate board that would fit upside-down into the second "rail" right under the main Op Panel board. The second board would likely provide front panel audio connectors, volume pot (judging by the matching cutout in the DASD cage), audio amp etc. Multimedia Link Connector (J3, unpopulated) Pinout:
Op Panel Presence Detect (by Tomas Slavotinek) Before I forget (again...), here is some more info about the Op Panel presence detect logic of the 9595 2S2P planar. It's actually even more primitive than I originally thought. There are no pullups and inverters on the planar. The pins are simply "floating low" (logic 0), and pullup(s) on the Op Panel and Multimedia Panel are used to force the bits high (logic 1). This unfortunately means that the 9595 2S2P planar can't differentiate between the older 9595 Op Panel (the one w/o HD LED) and the 9585 Indicator Panel. (Pins that are floating low, won't change their state when left floating or when grounded.) It can only detect the newer 9595 Op Panel and the unreleased Multimedia board that would connect to the unpopulated Multimedia Link connector. The Presence Detect bits 0, 1, and 3 are occupying pins 14, 12, 10, and 8 in that order. As can be seen HERE. The four PD bits can be obtained via the Planar POS register cluster in 4
simple steps:
The mapping is straightforward:
There are no inverters, so unconnected or grounded pins will show as 0. Pins tied to +5 V via pullups will show as 1. Base Op Panel Presence Detect Codes
"0000" - the older 8595 Op Panel Multimedia Panel Presence Detect Codes The newer 95A Op Panel with the Multimedia board will come up as "1xy0" where "x" depends on the state of the Multimedia Link pin 1, and "y" on the state of the Multimedia Link pin 3. This gives us 3 other possible values:
"1100" - 95A Op Panel with MML pin 1 tied to 5 V The T4 SurePath BIOS only uses this when Int 15 function CBh ("LED display - print character at position") is called. It will return 86h - "Not supported" if the function is called on a 2S2P 9595 that doesn't have the newer Op Panel connected to it. The older 1S1P planar doesn't have the Presence Detect logic at all, so the BIOS assumes that the panel is connected and blindly outputs the data to ports 108 - 10Fh. The POST routine doesn't actually check the Presence Detect bits it seems. Note: One could use the MML pins 1 and 3 as a cheap GPIO ports... You would have to poll it, and it's just GPI without the O part. Debugging comes to mind as one potential use. It may be useful for our BIOS experiments - to switch between two different code paths or whatever. Devkits/devboards often have a set of GPIO ports (sometimes hardwired to switches and LEDs) for this very purpose. 95 XP Op Panel - Hidden Remote IPL Switch (SW1) What's the switch SW1 for on the op panel PCB? Tom says: Ed.: Now we know its real name - Remote Initial Program Load (IPL) Switch (source). Supposed Origin of Op Panel Board From Peter:
The IBM instructor told us, that the panel card has been taken from some 3270 control unit and the second switch was used to set the IPL mode of that box or to clear audible alarms or such. "It is not used with the PS/2 - therefore it is hidden behind the bezel." If it *were* used we have had two switches in the front. I used to use the second switch a) as a replacement for the power switch, which tends to wear out after some time or b) modified the card and bezel to have the switches in parallel, so that you cannot accidentally power down the server. One needed to pry in a tiny hole to operate the second switch too to get it shut down... Tom reacts: Based on these observations and my experience, I would say that it was designed specifically for the Model 95 line, and the switch was either intended to be hidden - for "operator's eyes and hands only" (you have to unlock and remove the front bezel to access it, but you don't have to poke inside the machine itself to do so), and/or it was exposed only on some special bid 95s... After all the *later* 8595 bezel has a sub-assembly with a provision for the second switch! It would make no sense to design and add that part for the revised bezel, if the switch served no purpose, you would just stop populating it... Ed: Some IBM patents show the DASD cage with an early 8595 dual-button Op Panel bezel installed (i.e. EP0426330B1), supporting the above explanation:  Ed: As I have discovered the switch actually *does* serve a purpose in the Model 95 - it enables the "Remote Power-ON" feature. Cheap Op Panel Lens Redo From Tony Ingenoso:
Thoughts on Adding a HD LED I am thinking of using 2 colored LED, and replace the power on LED. From Peter: From Jim Shorney: It should certainly be possible to wire in a simple switching circuit with an NPN transistor on the HD LED to take care of the difference and allow use of a common-anode bicolor LED, providing the +5 can source a few extra milliamps. I haven't figured out (yet) if this comes from hard power or is a signal output from the PS/planar. The other possibility is to use a common-cathode LED and share the resistor for the power LED, but I'm not sure how well this would work. LED's are diodes, after all, so the two sources would be isolated from each other (in theory), but I'm not sure what this would do to the brightness of the LED's. I may play around with these ideas in the future, but I wanted to keep the initial mod as simple as possible for those who might be comfortable with a soldering iron but don't possess sufficient knowledge to handle wiring up transistors and such. From Peter: *But* the visibility is still bad. What you need were a "NAND" function that switches off the green power LED when the red HD LED comes on... that would make the color either green *or* red but not a diffuse orange when both come on. To simplify the circuit: you only need to check
the HD signal. As long as it is low the green LED part is on (if there
were no power it were dark anyway) and once the HD signal pin comes "high"
the red LED comes on and the green is switched off... In this case you
could even use a "common cathode" bi-color LED, which are more familiar
than "common anode" types IIRC and keep the existing current limiting resistor
on the PCB. From Peter: > An LED behind the display window seems to be the best
overall solution. > One of the triangular ones used in tuning indicators
and tape player directional indicators might look cool. > I think I would go with green, myself. And I'm
still thinking about replacing the power LED with a blue one... Peter's Two Color HD LED Peter's minimalistic "Two-Color LED" Solution.  Principle of Operation You need to unsolder the existing green LED. The +5 V DC contact needs to be fed to the above circuit. Only the common cathode of the bi-color LED gets soldered back in the place where the original LED cathode was. The two LED-parts use the existing 330 Ohms resistor in common. The PNP transistor gets minus-potential over the 47K resistor and the green LED lights up. Even when the HD-pin does not feed minus to the resistor (tri-states) the GND connection is given over the red LED part and the current-limiting resistor, so that the green LED will light under any circumstance. Once the HD-pin gets positive (on HD operation) the positive +5 V pulse will turn Base of the transistor positive and the green LED goes blank - the red LED is then on and indicates HD activity. The type of the PNP-transistor is uncritical. Any type will do. I used this 2SA608 while I had it... In case you want to return to original condition you only need to remove the circuitry and resolder the single green LED. Pretty easy, eh? Note: I found out that my original concept works only with *one* panel card, which sits in a very early Mod. 8595-AH9 (without shutter). All my machines have the same panel card that Jim described and to which he published a "revised version" of my PNP-transistor solution. The two cards only differ through the position of the current-limiting resistor for the "power LED" - but that makes a bit difference. Sigh. Shorney's Shortcut Parts required
Tools required
Instructions 1. Trim the leads of the resistor and the anode lead (the long leg) of the LED short, and solder one lead of the resistor to the anode of the LED. 2. Solder the two lengths of hookup wire to the other resistor lead and the cathode (short leg) of the LED. 3. Cover the soldered connections with the heat-shrink tubing and, well, "heat shrink" it. 4. Pull your 95 apart, and unplug and remove the operator panel display PCB (you do know how to do this, right?). 5. Secure the LED in your desired mounting location (more on this in a bit) and cut and strip the two free wire ends to length; the anode lead (the one coming from the resistor) will be soldered to pin 4 of the ribbon cable connector, and the cathode lead will be soldered to the frontmost of the two speaker connector pins (ground). Make your connections on the bottom of the PCB. 6. That's it! Install the PCB back in your Model 95, taking care not to snag your wires as you slide the PCB back in. Enjoy your hard disk light! Now, as to mounting location for the LED... Being into
vintage ham radio gear, I subscribe to the "drill no holes" school of
mods. Finding a suitable location for the LED, without altering or
detracting from the aesthetics of the machine, is the hardest part of this
adventure. ... I finally got around to trying Peter's nifty circuit for a two-color HDD LED in a model 95. In the process, I discovered that there are at least two variations of the info panel PCB. Mine differs from Peter's in the location of the LED resistor, so I had to change the circuit as shown below:  Circuit function is essentially the same, the change was required because my board has the 330 ohm on the anode of the power LED instead of the cathode, as shown in Peter's diagram. The green LED does not fully extinguish in this configuration because the HD output only seems to source about 3.6 volts under load, but it is dim enough that the red is clearly visible from an angle or a distance of a couple of feet (meters?). That having been said, I also tried a rectangular red LED behind the display panel on another 95. I must say, I like the way it looks. I fixed it with hot glue to the bottom of the rightmost operator display LED so it appears directly below the rightmost character of the display. Now I need to decide which way I like better... Button Sizes The button size between "left-hand" and "right-hand" panels is different - respectively the *length* of the buttons. Ran into this when I swapped a button from the 8595 onto the 9585. Early 95s without Shutter over Power Switch > Early 8595 have no clear springy cover over the power switch. It's kinda funky looking having a beige "ring" around the white switch. Power switch protective shutters became standard in early 1992. 8595 before then lacked them. Tony Ingenoso replies: 8595, 9585, and 9595A Op Panel Bezels and PCBs 8595 Op Panel Bezel, Early Front 33F5408
This is one of the earliest Op Panel bezels, note the lack of a shutter over the Power Switch button, Power Good LED, and the raised edge around the button. Note that the bezel hooks are on the LEFT and the bezel latches are on the RIGHT. (Thanks to Jelte for pix) 8595 Op Panel Bezel, Rear 33F5408
Note the simplicity of this moulded part. See the EMI shielding? That copper colored coating is replaced on later bezels by the silvery Enshield (which was probably cheaper as well...). Note the small locating posts at the upper corners? They fit into holes in the DASD retainer. 8595 / 9595 Op Panel Bezel, Front
This is not the earliest, as it has the shutter over the Power Switch Button. Note that the bezel hooks are on the LEFT and the bezel latches are on the RIGHT. It has a Power Good LED and an Information Panel window. 8595 / 9595 Op Panel Bezel, Rear
Note the small locating posts at the upper corners? Note the modular assembly, with the white button frame being placed onto the bezel (grey). Peter mentioned that the op panel PCB originally came off an older existing device. Also note the LED window is covered with a darkly shaded piece of plastic. If the plastic window got a little scuffed up, you can push the flexible sheet out, and the tabs will pop out of the frame with no damage. To re-install the window, put one tab in the slot, and bow the sheet a little until the tab re-enters the opposite slot. Note the spring powering the shutter rests on a little hook molded into the shutter. If the shutter haltingly raises, czech to see that the end of the spring is in the hook, not dragging between the front of the hook and the white module. Note the white hook just above and right of the end of the spring. Op Panel Shutter Spring, Richtig (Lorenzo)
In this daring piece of realism, Lorenzo Mollicone transforms the genre... Er, on the other hand, this clearly shows the shutter spring in the correct position. Op Panel Shutter Spring, Falsh (Lorenzo)
Note the shutter spring is in front of the little "wings" used to guide the shutter up and down. The shutter spring rubs against the subframe, and the shutter no longer moves smoothly. In addition, the shutter is pulled back into the subframe causing resistance as well... 8595 / 9595 Op Panel PCB, Top (Jelte)
8595 / 9595 Op Panel PCB, Bottom (Jelte)
8595 Frame (Jelte)
See the rectangular holes on the left for the Op Panel Bezel hooks, while there are similar "T" shaped holes to the right. Note there is only ONE set of small holes for the Op Panel bezel at the top corners. Note the small hole for the alignment pin in the upper left corner is horizontally elongated, while the upper right pin hole next to the latch hole is round. This makes sense, as the hooks on the left are inserted first, and the left pin makes an arc as it enters the hole. The rightmost pin is near perpendicular as it enters the hole near the latch. 9595 Single Serial Frame (Lorenzo)
Same as the -AKD, I would surmise the bezel mounting holes changed with the 9585 (K/N?) and 9595A (Dual Serial/Parallel) models. Note that the hook /latch holes are slightly inward from center. Maybe because it was better to not have the drive retainer hooks in-line with the bezel fasteners? Dunno. 9585 Op Panel Bezel, Front 41G9550
This bezel has a shuttered Power Switch Button, a Power Good LED, a HDD Activity LED, but the Information Panel window has a opaque plastic insert. Note the bezel latches are on the LEFT and the bezel hooks are on the RIGHT. 9585 Op Panel Bezel, Rear 41G9550
Note the small locating posts have been moved from the upper corners to just above the lower hook / latch? Note that the Information Panel window insert (57F2085) has the same tab arrangement as those for LED displays. The power button frame has lost the well for a second push button, and a second LED light pipe has been added. Note the spring powering the shutter rests on a little hook molded into the shutter. If the shutter haltingly raises, czech to see that the end of the spring is in the hook, not dragging between the front of the hook and the white module. Note the white hook just above and right of the end of the spring. To remove any spring tension from the shutter, place the end of the spring into the hook. 9585 Op Panel PCB, Top 61F3736 (David Beem)
9585 Op Panel PCB, Bottom 61F3736 (David Beem)
9595A Op Panel Bezel, Front 41G9550
This bezel has a shuttered Power Switch Button, a Power Good LED, a HDD Activity LED, but the Information Panel window has a darkly shaded piece of plastic. Note the bezel latches are on the LEFT and the bezel hooks are on the RIGHT. The 9585 and 9595A use the same bezel. 9595A Op Panel Bezel, Rear 41G9550
Note that the Information Panel window insert has the same tab arrangement as those for LED displays. The power button frame has lost the well for a second push button, and a second LED light pipe has been added. Note the spring powering the shutter rests on a little hook molded into the shutter. If the shutter haltingly raises, czech to see that the end of the spring is in the hook, not dragging between the front of the hook and the white module. Note the white hook just above and right of the end of the spring. To remove any spring tension from the shutter, place the end of the spring into the hook. 9595A Op Panel PCB, Top 92F0259 (Jelte)
9595A Op Panel PCB, Bottom 92F0259 (Jelte)
9585 and 9595A Frame (Jelte)
See that the hook -AND- latch >BOTH< use a "T" hole? Note there are TWO sets of small holes for the Op Panel bezel pins, at the top corners -AND- above the bottom hook / latch holes. I suppose you could use one LED bezels on 8595 >AND< 95A frames... Note the elongated hole for the top posts is on the left, and the elongated hole for the bottom posts is on the right. What is the dual card guide doing here? See the Op Panel board is in the upper card guide, and the narrower bottom guides are empty. Note the notch in the frame in the lower right (below right 4 character LED), and the semi-circular relief in the plastic DASD Support Structure... Almost resembles a relief for a thumb wheel... Power Switch Test / Jump Start Pull Op Panel out front of system. Leave PSU plugged in. Unplug the molex connector(s) from the PSU if you don't want your hard drives to spin up/down during the testing. For the single LED (95 XP) Op panel: Short pin 1 (or 4) to pin 2 (or 5) and the PSU should start.
For the dual LED (95A) Op panel: Short pin 1 to pin 2 and the PSU should start.
Operator Panel Registers See Hex 0108-010F for info about Op Panel registers. Panel Programs
PS2INF.ZIP Small program to access the LED Panel (DOS, OS/2 driver). BASIC Way
A$ = "133 MHz " You can as well use DEBUG's "OUT {portnumber} {value}" command to pipe into the display. Write a text that includes 8 lines of out-commands to the ports 108 - 10F with the hex character values for {value} and use DEBUG<mytext.txt to bring it into the panel. 30 is "0", 39 is "9", 41 = "A" 5A = "Z" ... you will get the scheme. Linuxinfo LED Panel Odd Behavior From Alfred Arnold: From Peter: Alfred Arnold retorts: And finally, the moment of discovery: 8 "Lights" on Op Panel I have been having some trouble with getting a Mod 95 to see the optical SCSI drive being added. I went in and unplugged the SCSI cable, and power cable to reset everything and start over. Now I have 8 positions across the op panel with a pattern of small dots (LEDs) and the machine goes no farther in the boot process. What do the 8 position, all LEDs on suggest. It suggests you pulled one end of the cable going to the LED display loose at one end or the other or that you somehow damaged the cable to it. Check for this small flat cable and make sure its OK. Jack, I had high hopes that it would be that simple. I checked and re-plugged both cables and did not change my symptom. Any other ideas. I have an HMM dated Sept 1993. Where did you find the symptom info.? Did you get it from the HM manual? From Jack Gulley: |
