Tom Jennings, 10 February 2005
I'm bringing up this machine as fast as reasonably possible -- or unreasonaly, in the instance of the 6070 disk system.
I really screwed up, but recovered from my folly. I underestimated the difficulty of cleaning the heads -- it took actual physical force to get the crud off. At the same time it's sensitive to cleanliness, and I don't know about you, but I don't have a clean-room here. You can read about (and learn from) my stupid mistakes.
From email@example.com Wed Jan 26 11:33:36 2005 Date: Wed, 26 Jan 2005 11:33:35 -0800 (PST) From: Tom Jennings
To: "General Discussion: On-Topic and Off-Topic Posts" Cc: "General Discussion: On-Topic Posts Only" Subject: Data General NOVA4 status Well things are moving along very nicely, I'm going to power up in the next few days. I've been picking on it each night (30 min - 2 hrs) and it's pretty much ready to begin power-up. I took another look at the DG LP2 that "faulted". With a dry paintbrush and vacuum cleaned out the dustbunnies and a few spiders, esp. attention to the photo-interruptor. Loaded paper into it, powered up. This time carriage zeroing worked, everything behaved including self-test printing out the charset onto greenbar. The "new" ribbons seem dried out and the print is yellow-gray (likely from Bakersfield heat) but I'll try another or WD40 the ink ribbon. Washed it all down, other than a couple of scuffs it looks like new. Completed all the rack wiring, made a terminal (console) cable up from the stub that got cut off for transport. Fan filter, rack panel fillers all in. Took a look at the other chassis; one now empty (DG salesmammals talked customer into too many (profitable) racks, as Bruce suspected :-) one's got the big analog mux thing (more, next message), the remaining has the second tape drive (6023) and the 4300 expansion chassis. Which I might keep -- ask me in a month -- it contains a 16 in/16 out digital interface and I think a one-channel A/D and D/A. Nice. Big though, we're talking 14" rack space but it's mostly air (lots of slots). Laid out all the docs, culled more dupes, which Bruce will get (hi Bruce :-) Bought binders on the way into work today for never-used docs. My lab, quickly outfit last fall (moved out of my nice, purpose-built lab, which we turned into a rent-generating apartment -- when I get sad, I think of the income :-) has crappy power. I'm not sure it's got enough amps to run the CPU, tape and disk. Umm, we'll see, the obvious way. From firstname.lastname@example.org Mon Jan 31 12:03:39 2005 Date: Mon, 31 Jan 2005 12:03:39 -0800 (PST) From: Tom Jennings To: "General Discussion: On-Topic and Off-Topic Posts" Subject: DG NOVA4/X runs Well I just put a cord on it, plugged it in and powered on. No smoke escaped. The CPU passes self-test (no idea what it really checks, but it alleges to be thorough) after I seated the memory board (I forgot just how tough to seat the 15 in square DG boards are). I can poke memory, registers, etc just fine. Took a bit to get the console wired right (to minicom on a laptop, 9600/8/n/1) as the sample cable (from the orig. install) was misleading. The tape drive probably has a bad vacuum sensor; I'll look at it Tuesday again. Might be cable fell off. It otherwise loads the tape and one vac col does the right thing. Other functions seem to work. I have a 2nd tape drive, not yet tested. The disk (10 fixed, 10 removable) is not behaving right. Problem with load/ready. Looks mechanically OK but I will do one more test before the covers come off and the scope and DVM come out. Tom Jennings From email@example.com Wed Feb 2 10:36:44 2005 Date: Wed, 2 Feb 2005 10:36:43 -0800 (PST) From: Tom Jennings To: "General Discussion: On-Topic Posts Only" Subject: NOVA4 6070 disk booboo Decided to try to get the Data General 6070 disk (10 fixed 10 removable) going last night. There's reasonably complex LOAD/READY logic, but from the indicators it wouldn't go ito either state. Got out the schematics and docs, found a nice startup flow chart and checked each signal. (DG does a great job on docs; there was a flowchart and table of signals, each with a drawing/page/zone callout!) so I went down the list and looked at gate inputs; power supply checks (6 supplies all threshold checked), cover in place, a bunch of random logic for zero RPM, temp difference (fixed to removable), heads HOME, blah blah -- turns out it was the fromt panel LOAD/READY switch! It drives an RS flop but the grease in the rocker switch hardened/corroded so it was a two position OFF/OFF switch. Disassembled, cleaned, LOAD/READY worked. Got proper LOAD state, turned to READY, spindle spun up, brush cycled, ... no head load. ALl the logic was OK and saw the RNENABLE or whateveritwas command, just didn't move. The linear motor had acquired stiction from sitting. I took the dust cover off, EXTREMELY CAREFULLY extracted the head assembly about 1/8" 'til it clicked the HOME microswitch, a few times, it loosened up. Reassembled, READY, spinup, brush, head load.... CRASH. Top surface of the fixed disk unmistakably crashed. While I had the dust cover off the first time I did look at the heads and the platter surfaces. I didn't clean the heads (I do have the right stuff for the job; hydroentangled sterile wipes and 99.5% anhydrous isopropyl) first, and maybe I should have looked closer, but I saw the typical "I" shaped pole in the white ceramic substrate, assumed it was OK. Post-crash, there was definitely a lot of FeO on the head. Cleaned that. Ovbious grooving on the platter; 'far as I can tell all other (3) heads/surfaces are fine. Can't see the bottom of the fixed but thehead remained clean. Upon closer examination of the offending head there is definitely something caked on it on the "far" side of the head (from where I can reach) that won't come off with the cleaning pads. It may have been there, but if so, it would indicate a previous crash. Knowing that the platter was trashed, I took off the dust cover, cleaned the platter with a new pad, and felt for large bumps, and ran the READY sequence again. Crashed two more times then I stopped. I'm 90% sure (...) that there was a preexisting crash, but it's entirely possible it happened on my first power-up. Since it's sat for so long all sorts of subtle things couldhave happened which makes that seem likely, though I did examine it before power-up, though probably not closely enough. It suuuuure looks like the platters on the removable cartridge are physically the same as the fixed platter; and since I have two spare (three total) carts I may attempt surgery. From firstname.lastname@example.org Wed Feb 2 15:02:30 2005 Date: Wed, 2 Feb 2005 15:02:29 -0800 (PST) From: Tom Jennings To: "General Discussion: On-Topic and Off-Topic Posts" Subject: Re: NOVA4 6070 disk booboo On Wed, 2 Feb 2005, Jay West wrote: > Almost always the problem is failure to completely clean the drive; heads, > airfilter and ducts, all platters (both surfaces)... you've got to > disassemble the platters, clean and buff them, clean the head armature, ALL > internal cavities, etc. Ye Not gonna do that for this machine. Not all machines deserve that sort of attention; this one doesn't. There's only so many mountains I can move with a spoon in one lifetime; currently, that will be the LGP-21. The NOVA4 got it's airducts cleaned, PS's bench checked, disk surfaces inspected as far as I could w/o disassembly. I admit I did not deeply examine the fixed head set (esp. the top-surface, hardest to see). Complete disassembly introduces it's own set of problems itself, but for the full tilt boogie job, necessary. > s, it could be a pre-existing crash. The above > would have caught that ;) Yes, but I'd never get to it, #1, and #2 an in-place cleaning would have got this one to the 99% point. I didn't do that. >> Knowing that the platter was trashed, I took off the dust cover, >> cleaned the platter with a new pad, and felt for large bumps, and >> ran the READY sequence again. Crashed two more times then I >> stopped. > Yeah, I bet it did ;) I realized it would never work again! This wasn't a subtle crash, or the >tink< of a skip. I wanted to characterize it (by ear) after the head and surface debridement. > job and buff it. Remove the platter. Also break out a dial test indicator and > measure the runout/flatness (which you have to do when you replace it > anyways). No runoff on the outer edge visible to careful scrutiny; certainly less than .001" > The key of course if is there is a designated servo platter, or if the drive > uses a glass reticule with cylinder markings on it and an optical system. I No servo track, reticule. I will remove the fixed platter's heads, clean out the cavity and run the removable cartridge only, at Bruce Ray's suggestion. I'll examine and photo the bad head, see what state it's in. I'll leave the platter in place for now and see if I can find part numbers and if the removable platter's number matches. From email@example.com Wed Feb 2 15:05:33 2005 Date: Wed, 2 Feb 2005 15:05:32 -0800 (PST) From: Tom Jennings To: "General Discussion: On-Topic and Off-Topic Posts" Subject: Re: NOVA4 6070 disk booboo On Wed, 2 Feb 2005, George & Oksana Wiegand wrote: > I had a 6045 disk system (1/2 the density of the 6070) with a head crash on > the top surface of the fixed platter about 3/4" in from the edge. One of the > idlets that are mounted in the white ceramic head came out, so I took this > one head assembly off but first measuring it's alignment with a feeler > gauge, cleaned and super glued the idlet back into the head and replaced it > with the alignment reading taken earlier. It worked, booted Ok, the damage > must have been in the unused portion of the disk platter. What's an idlet? Mine's crashed in the same physical spot, likely it's track0. I imagine it's a common spot. I think the 6070 has double TRACK density, not bitrate density; Bruce Ray thinks it's possible the 6070 heads are the same as 6045 heads. I'll found out later, hopefully tonight. > Is the blower working though the filter to generate a cushion of air for the > head to ride on? I thought this is how it works, not only for dust removal. It is, but that's only to cool & purge. It doesn't lift. Bernoulli & skin effect on the head does the lifting, induced by the high surface speeds. Tom Jennings From firstname.lastname@example.org Wed Feb 2 19:10:40 2005 Date: Wed, 2 Feb 2005 19:10:39 -0800 (PST) From: Tom Jennings To: "General Discussion: On-Topic and Off-Topic Posts" Subject: Re: NOVA4 6070 disk booboo OK, you all get to laugh if this is a spectacular failure, but if it's a success, I expect flowers. So the fixed platter is crashed. It turns out the heads were very dirty, I was an idiot to not to take the heads *out* for cleaning. I tried cleaning in-place, but it appeared that there was deep scoring hard-to-see black streaks). When I got the heads out, the black streaks, on only the upper-surface heads (both platters) looked like grooving. Instead it was very hard gunk/oxide. The heads are ceramic. So I took a clean, new popsicle stick (wood) soaked in alcohol, and scrubbed the crap off. The heads are perfect, with zero scoring or scratches under a halogen lamp and magnifier. Set these aside. The fixed platter is trashed. With my calibrated eyeball (English units) I determined that the removable cartridge has the same, exact platter inside. Seeing how I have three of them, I took the platter out. This involved drilling out (GAK!) four extreme-torque screws [100% tape seal, vacuum, etc]. Extracted spotlessly clean platter. Removable platter fits perfectly as a fixed platter. Zero play at the center ring, zero runout after torquing into place. Heads installed, loosely. There's a head-alignment procedure, but I'll set the fixed heads statically since that platter needs a format anyways. The upper heads I will do the alignment. Haven't spun it up yet. My main concern, besides dirt, is torquing the heads in their floating mount during cleaning. But the way the retracted heads lift, causing a HUGE torque on one side of the head holder seems really awful, and likely swamps and tiny error I introduced. Even if it works, I can't be too proud of fixing a problem I caused in the first place. From email@example.com Wed Feb 2 22:43:30 2005 Date: Wed, 2 Feb 2005 22:43:29 -0800 (PST) From: Tom Jennings To: "General Discussion: On-Topic Posts Only" Subject: NOVA4 6070 disk Well, it worked, but failed anyways. The replacement and cleaning worked for the fixed platter just fine, but I crashed the removable platter. The fixed platter repair went as planned. The platter dropped in, the heads cleaned and installed OK, aligned OK. The upper heads went in OK, did the basic mechanical alignment and since they're accessible, did a final clean in-place. (While the bottom platter was out I wiped everything down with a kimwipe and 99.5% isopropyl.) I was exceedingly careful and clean, and I don't know how to diagnose what happened. Possibly the head got canted in it's holder during cleaning, removal or installation, no way to tell. I went out of my way to not apply torque to the end of the head, and it wouldn't seem too delicate; when the heads are retracted they are lifted on one side of the carrier, which applies a lot of off-center force. I followed the remove/install procedures to the letter. Iropnically, the lower platter heads seem to have floated OK so the repair worked. From firstname.lastname@example.org Fri Feb 4 00:25:04 2005 Date: Fri, 4 Feb 2005 00:25:03 -0800 (PST) From: Tom Jennings To: "General Discussion: On-Topic Posts Only" Cc: Tom Jennings Subject: NOVA4 6070 disk -- whew! Success! But what an ordeal! I removed the top head, got real zen with an 8X magnifier and the desk halogen. Damned head was dirty; I did a crap job cleaning it. I carved off a thin sliver of wood with which to clean and explore the aero holes, leading edge, etc. Scrubbed off some carbony crud. Got it spotlessly smooth and clean, washed it all out. Re-installed and did static alignment. (There's an electronic alignment procedure, similar to aligning 8" floppy heads with an alignment disk, but (1) I don't have an alignment disk and (2) pack interchangability isn't an issue and (3) I don't have any existing packs to worry about so I'll just re-format to this particular alignment.) So I've got one cartridge left, the brand-new one still wrapped in DG plastic. It's spotless. I mount it, spin up (with the servo DISABLE/ENABLE switch set to DISABLE *), ... a horrible rumbling noise! It's coming from the new pack, definitely something spinning that is roughly dragging on something not. I demount, no crap on the platter? Remount, check the plastic container's alignment (the platter magnetically attaches to the spindle, floating free, but close, to the enclosing plastic housing). Check the housing all around, etc. Spin up -- NOISE! Sigh, removable cart on the bench. Long story short, I hear a rattle, and a broken screw head falls out... half the mag-latch mechanism isn't attached, and falls down with gravity to drag on the inside of the platter's mounting hub. So that's why this brand-new, never-used DG 10MB removable cartridge is sitting there in a plastic bag. It's broke. CRAP! Well I had all the parts from the sacrificial cart that I turned into the fixed platter. However it's of a different latch design, so I had to transfer the removable platter into the old housing and mechanism! HAIRY! The inner hub was filled with turned and melted plastic, luckily the design of the hub fully contained the shavings. It worked. Well, heads load without crashing, a good start. Electronics, if broken, I can fix. POST MORTEM: So what did I learn: One, I did a shit job of inspection. Though I can't know, I have to assume the initial crash was my fault. I underestimated the subtlety of the dirt. The dirt was very old, and very hard. Plus, my eyes are 50 years old now, not 21; I was actually surprised how much more was revealed with the magnifier. Two, I didn't *fully* learn the lesson of the first crash when I finally did remove the heads for a 'proper' cleaning after the second. So I rethought my general approach (non-heirloom: assess condition (overall: excellent) therefore fix only things needing fixing) and still think it is correct for this machine, but I neglected to consider that that disk-drive heads/platters is simply a unique situation. Heads and platters are simply to precious, and problems manifest too disastrously. A counter example is the tape drive(s); while I'll be paying a bit more attention (ahem) the relationships 'tween head and media is vastly simpler, easier to grok and to fix. And I'm testing with a tape marked "bad" :-) tomj (*) The drive sequences airpump on, spinup, brush cycle, head load. I disabled the servo so I could let the air pump run extra time, and manually trigger 3 - 4 extra brush cycles, and generally flush things out.