• Please review our updated Terms and Rules here

RK05 problems

A

anders_bzn

Experienced Member
Joined
Jan 27, 2016
Messages
239
Location
Sweden
I had my RK05 working, but now it wan't pass the MAINDECS anymore. It fails the "drive control test" at address 3277: http://svn.so-much-stuff.com/svn/trunk/pdp8/src/maindec/08/dhrkb-e-d.pdf

As I understand it all data written to the disk and is verified in the test before. The test that fails is the random read data test. All test has passed before and I have passed them with three different RK8E controller sets. Now the test fails in the same way with all of them, so I suspect the drive.

I haven't done the drive maintenance procedure yet, is that what I should start with?

Any more ideas are appreciated!
 
In my experience this is the test that fails most often with a ‘nearly working’ drive. The drive works overtime throwing the head randomly all over the place and eventually the head ends up on the wrong track. It’s worth checking the difference between the track seeked for and the track actually landed on. If it a long way out it’s likely nothing to do with the servo control so you need to start checking that the drive is actually seeking for the correct track, but if it’s the next track then likely the servo isn’t accurate enough. You might even find that the vibration is disturbing contacts or joints randomly :(

If it’s ending up a long way from the correct track it’s likely a single bit error in the address which you might be able to chase down. If it’s a long way and random it sounds like connections.

If it’s the servo then is it a previously formatted disk or have you just formatted it in this drive?
If it was previously formatted then possibly the alignment is just a little out. You’d expect a few read errors too, but on random seeks there’s a lot more chance that the head will overshoot just a little. If you can afford to, format a disk and see if it performs better on one it formatted itself.

If it’s newly formatted I’d check how clean the graticle is and also the current control to the solenoid to be sure it isn’t being driven too hard as it nears its destination.

Good luck. This can be a real pain.
 
Last edited:
anders_bzn said:
As I understand it all data written to the disk and is verified in the test before. The test that fails is the random read data test. All test has passed before and I have passed them with three different RK8E controller sets. Now the test fails in the same way with all of them, so I suspect the drive.

I haven't done the drive maintenance procedure yet, is that what I should start with?

Any more ideas are appreciated!
Click to expand...
If the diagnostics that pass write a track and read it, while the diagnostic that fails does random seeks, I'd suspect "seek slop" where you're ending up slightly off-center on the track you're seeking to. In the old days I'd first check the heads for any gunk (which would affect their flying height) and then load an alignment pack and do repetitive seeks and look at the pattern with a scope to confirm that this was the cause of the problem. Then I'd dive into why it was happening.

I've never looked at the inside of an RK05, but I believe the Diablo drive that used the same packs (is the RK05 a re-badged Diablo?) used an optical sensor that looked at a strip of alternating clear / black sections to decide where the tracks were. On some types of drives the strip was glass, on others it was plastic. Any slop in there will result in off-center seeks. I'd be VERY careful when investigating this area as those parts are likely unobtainium at this point.
 
Terry Kennedy said:
If the diagnostics that pass write a track and read it, while the diagnostic that fails does random seeks, I'd suspect "seek slop" where you're ending up slightly off-center on the track you're seeking to. In the old days I'd first check the heads for any gunk (which would affect their flying height) and then load an alignment pack and do repetitive seeks and look at the pattern with a scope to confirm that this was the cause of the problem. Then I'd dive into why it was happening.

I've never looked at the inside of an RK05, but I believe the Diablo drive that used the same packs (is the RK05 a re-badged Diablo?) used an optical sensor that looked at a strip of alternating clear / black sections to decide where the tracks were. On some types of drives the strip was glass, on others it was plastic. Any slop in there will result in off-center seeks. I'd be VERY careful when investigating this area as those parts are likely unobtainium at this point.
Click to expand...

DEC RK02 and RK03's were rebadged Diablo's 30's. The Diablo 31's and DEC RK05 are different designs. You could use either drive with most early DEC PDP-11 Controllers and also interchange the packs if the sector size/formating was the same.

The Diablo 30's had a servo positioner and RK05 used a rather heavy but faster voice coil. The RK05 used an optical transducer with an incandescent bulb. Adding to Terry's guidance, check the bulb for darkening due to evaporated tungsten being deposited on the glass. That may compromise its operation. Also make sure the head retracting NiCd's are in good shape. All too often we found that they needed to be replaced only when we had a head crash during a power outage.


The Diablo transducer is described in Ken Shirriff's repair stories for one of these drives - http://www.righto.com/2018/03/a-1970s-disk-drive-that-wouldnt-seek.html. The heads were servo loaded and a spring retracted the heads during power outages. I never lost a pack in one of these drives.

Jerry
 
wa2flq said:
Also make sure the head retracting NiCd's are in good shape. All too often we found that they needed to be replaced only when we had a head crash during a power outage.
Click to expand...
Yup. I was "bitten" by an RK07 (a completely different drive design) once. I had the drive powered down at the PDU in the bottom and had the heads extended to just before the load point (where the drive no longer keeps them mechanically lifted) for cleaning. As I was extending the heads, the small amount of remaining logic decided to do an emergency retract. This was the first time this had ever happened on many drives, all of which had good NiCd batteries. I'm not sure what I did to make it mad. Fortunately, I only ended up with some mild bruising to my finger (the retract is just enough to get the heads off the disk, nothing extreme).

A tech friend of mine wasn't so lucky. He was doing a head alignment on a Calcomp Trident drive (similar to the DEC RM03) and took a shortcut and didn't drop the security pin into the hole - these drives had a pin that went into a hole on the top of the actuator to keep the actuator parked over the alignment track and mechanically prevent it from retracting when doing a head alignment. He lost a finger to that drive.

Back then I was doing RSTS/E data recovery, subcontracted to a data recovery company. They sent me alignment packs for the RM02/3 and the RM05 as part of the deal, and would send me a pack and a set of heads whenever they had a job for me. They would wash the pack down to remove any crud and send it with a note like "don't go past cylinder 124", so I could recover data from the early cylinders past the crash. I modified some drive logic boards with dip switches to change the upper cylinder limit of the seek a drive would accept. I also built a contraption to keep the heads unloaded until they were past a certain cylinder, to read data on the outside of the disk, past the crashed zone. That was less successful, but worked on a number of packs. Eventually the data recovery company started sending me packs that weren't from RSTS/E and a matching blank pack, and I would copy the low and high readable sectors to the blank pack and send both back, so they could have other people recover RSX, etc. data. I wondered why they didn't just do this themselves, but I think DEC was a small part of their recovery business and it wasn't worth it for them to set up that environment - different companies used different numbers of sectors, etc. on the drives and I think most of their 12-platter work was for IBM 3330 and clones.
 
Thanks, many good answers. I'll look into this during the holidays!

I have taken care of the leaking NiCd battery but not yet replaced it. I'm thinking of replacing it with a supercap, has anyone tested that?
The discpack is formatted by me, so no misalignment problems. I have two others, but those weren't readable so this might indicate an alignment problem. I guess that I wan't get so many packs that has been written in other drives so I don't care about alignment (for the moment).

I will check the bulb and the servo system for a start.
 
If this were my drive, I would build a new set of NiCd's rather than try a super-cap or NiMH. I usually mark the cases of older devices with NiCd or Battery Inside to remind/nag me to remove them before putting them in long term storage. Using NiMH's would require changes to the circuits, as they don't like overcharging. For super-caps, you need to be be cognizant of voltage and ESR requirements.

Note that NiCd help keeps the head is retracted against accidental bumps while during maintenance or when removing the pack, not just emergency power off retract The head slides quite easily when the power is off. If it slides too much with power off, there is a microswitch that will energize the head retract again.

As Terry points out, this "safety" mechanism is very very dangerous. Keep your important bits well clear at all times.
 
I have been running two RK05 drives without the nicads installed. I understand the risks, and have new batteries, but have not installed them yet. They seem to work fine without them.
 
On offline post commented on my substitution of NiMH batteries for NiCd batteries, pointing out the charging rate is probably to high for NiMH batteries. No argument there, but given the higher capacity of the NiMH packs I'm using and the very occasional use of the drives in my hobbyist environment, I think my setup is OK for my situation. For production use (??), one might want to lower the charge rate by half.
 
anders_bzn said:
I'm thinking of replacing it with a supercap, has anyone tested that?
Click to expand...

I wouldn't try to do this. Despite the name, a supercap does not have that much energy storage. And the state of charge is directly proportional to the voltage. To make it work would most likely require a lot larger volume of supercaps and a LOT more cost than just using batteries. You would probably need to go to around 3 or 4 times the voltage of the NiCd and use a high current buck regulator to regulate the voltage down to the voltage of the NiCd battery in order to get the energy needed for the head retraction.

I suggest replacing the NiCd's with the low self discharge NiMh cells. Sanyo Eneloop or one of their competitors. The old NiCd cells really needed to be trickle charged all the time to stay healthy. These NiMh cells can sit for a couple of years and be fine and the trickle charge DEC used to keep the NiCds charged will work just fine. Unless you are keeping the drive energized all the time and you probably arent the modern NiMh cells are a much better option.

I ended up with 2 of these drives and I have not brought any of them back to life yet.

Best wishes!
 
jackrubin said:
On offline post commented on my substitution of NiMH batteries for NiCd batteries, pointing out the charging rate is probably to high for NiMH batteries. No argument there, but given the higher capacity of the NiMH packs I'm using and the very occasional use of the drives in my hobbyist environment, I think my setup is OK for my situation. For production use (??), one might want to lower the charge rate by half.
Click to expand...

What was the capacity of the original cells? NiCd's of the period could be trickled at a C/10 rate without issue.

Old NiMh cells (pre eneloop type) could be trickled just like NiCd's. The Eneloop types will tolerate some level of overcharge but you will know if you have overdone it because they will go into thermal runaway a while after they are fully charged. Put a finger on them after a few hours of being in the drive and if they are hot to the touch you will have a problem at some point.

It is clear that if the charge voltage (open circuit voltage) in the drive is <= 6 volts then you will be safe until one of the cells fails with an internal short. This will cause a large overcharge of the remaining cells and some heating. With a cursory search I don't see schematics for the drive. Anyone have a link to them? Lots of copies of the maint manual show up.

Found the schematics. The battery is charged from the -15 v power supply with nothing more than a blocking diode and a 750 ohm resistor. This gives a charge current of around 13 ma which I suspect is fine for a 2 ah Eneloop battery.
 
Last edited:
Enerloop batteries are possible to buy here, NiCd's are not. Those have to be ordered outside EU.

The lamp in the positioner system is not stable. Sometimes it's not lit when power is applied to the drive. When I tapp on it gently it turns on...
 
Have mounted one of those new bulbs now, works like a charm! Not 100% stable on passing the maindec DHRKB-G every time though.
 
You must log in or register to reply here.
Back
Top