It also depends on what you want to do. To me, a classic computer system is more than just the CPU. The peripherals are part of it as well. Running a PDP11 with flash memory mass storage is not the same as running one with a demountable hard disk. DEC managed with a handful of TTL. You don't _need_ an FPGA. You'd want to do DMA, which means that your device has to source the bus address, but that's not hard (a counter + buffers, basically). You also need the word count register (another counter), and some glue logic. Not that hard to build. One big difference is that the RK07 interface speed is determined by the conotroller, and your device has to keep up, keep on squirting out the bitstream at the right rate. The Unibus can be run as slow as you like (within reason). You don't have to keep up with anything. -tony

Ethan> On Thu, May 13, 2004 at 10:34:22PM +0100, Tony Duell wrote: Ethan> While in principal I agree, in years gone by, I was unable to Ethan> do more than play-load 2BSD on a real 11/24 because my largest Ethan> Unibus disk (that wasn't on a UDA-50) was and still is RL02s Ethan> on an RL11. These days, I have bigger Unibus CPUs (11/44, Ethan> 11/70), but in terms of pre-MSCP disks, 10MB is still a limit Ethan> for me (fortunately, I can get around this with 2.11BSD on the Ethan> 11/44 or the 11/70, but 2.9BSD was it for non Split I&D Ethan> machines) Ethan> Even though I appreciate and admire the older disks, I'm Ethan> pretty much stuck with RK05 and RL01/2. I'd love a solution Ethan> that emulates _some_ DEC disk controller (to avoid burning Ethan> boot PROMs, writing drivers for every OS I'd want to play Ethan> with, etc.). In order, a good solution for me would work with Ethan> 2BSD, RT-11, RSX and RSTS... there's plenty of interfaces that Ethan> fit the bill. If you can find adequate documentation for MSCP -- which should be doable -- then an MSCP controller would make a lot of sense. It's more complex in a sense, but it's command packet oriented, so the CSR piece is trivial. That one you can definitely do with just a register file to provide a register window (plus a DMA, but that can actually just be controller software reaching through a memory window over the Unibus). One clear advantage is that you can pretty much pick whatever device size you want, because drivers query the controller for the disk size and believe what they are told. With older devices, you're stuck with the particular device sizes that existed then. If not MSCP, the next obvious choice is RH11, because that gives you the widest choice of the biggest drives, and it doesn't seem significantly harder than any of the others. paul

On Thursday 13 May 2004 21:09, Tom Uban wrote: <sarchasm>But who'd ever want to run an OLD os on classic hardware??</sarchasm> Seriously, you've got a point, and I hadn't thought about that fact before. Part of the issue is that I don't have any of that "old" versions of software/OS that won't talk MSCP. So, what devices would give us the largest range of possibilties for software compatibilty at the larges size? If we're to use a microcontroller to run this board, then it'd be possible to have "changable firmware", and either change it to emulate the type of device you want, or to emulate multiple devices. On second thought, you probably don't need a microcontroller for that, but it seems easier than going the FPGA/PLD route to do it. <reaches for pdp11 peripheral handbooks> In 1970 pdp11 handbook: - no disks, just a teletype and papertape reader are listed. Added in 1973/4 pdp11 peripherals handbook: - RK11/RK05 - 2.4MB disk pack - RP11/RP03 - 40MB disk pack Added in 1975 pdp11 peripherals handbook: - RP04 - 100MB (presumably same RP11 interface as RP03) Added in 1975 pdp11 peripherals handbook: - RL11/RL01 - 5MB disk pack Added in 1978/9 pdp11 peripherals handbook: - RK06 - RK11, 14MB disk pack - RK07 - RK11, 28MB disk pack - RM02/03 - RM11 MASSBUS interface, 67MB disk pack Added in 1981/2 pdp11 peripherals handbook: - RL02 - 10MB disk pack - RP06 - 176MB disk pack - RP07 - 516MB winchester - RM05 - 256MB winchester - RM80 - 124MB winchester So, it looks like to me the best choices would be an RK11/RK05(06,07), an RP11/RP04(05,06,07), an RX11/RX01 and and RL11/RL01(02) emulator. If I were to pick a single choice right now, I'd probably pick the RP11 controller, for the breadth of possible sizes (from 40MB in 1973 to 516MB in 1981), and put RL11 as a second choice, simply because I've got one. :) RK11/RK05 seems too small to be worthwhile, but the RK06 and RK07 aren't bad sizes. Now, I haven't looked much into complexity yet, but at first glance through register descriptions, RL11 takes the cake on simplicity on just 4 registers, RK11 is middle-ground with 7, and RM11&RP11 are the most complex, with at least 20 registers. So, as a first try, I'd probably go for the RL11 for simplicity and then once that was working, try for the RK11 or RM11/RP11 interfaces. These all seem a lot simpler than MSCP (UDA50/KDA50) interface emulation, and support media that should be "large enough" for most people (1/2GB). In fact, the drive in my VAXstation 3200 is only 650MB, and it feels big with VMS 5.3 on it. For the actually data storage, I think you'll want to use flash or a writeable hard drive (perhaps even a 2.5" laptop IDE drive mounted on the UNIBUS controller card). It seems novel to have a 8 (emulated drive) x 516MB (4GB total) UNIBUS storage card that takes up a single SPC slot. Hey Tom, have you started at all on your RK11 emulator? Want to compare notes on this? Pat -- Purdue University ITAP/RCS --- http://www.itap.purdue.edu/rcs/ The Computer Refuge --- http://computer-refuge.org

On Friday 14 May 2004 08:45, Tom Uban wrote: I'd tend to agree with that, and add at least an RH11 (massbus) controller to that, so you have access to large-ish disks (RP07/516MB). As a starting place, I'd probably use an RL11 as the programming interface looks very simple (4 registers) and I've got a real one that I can test against. Right. I'd consider doing this with a couple of PALs and a microcontroller for my first run at it. After working with FPGAs at school, I vehemently hate having to write VHDL for them. As well, PALs use cheaper development tools, and are usually cheaper for small, simple designs, and if you chose the right ones (AMD's PALCE line), they're easily erasable. In fact, I think it'd be simpler, but I haven't looked closely at it yet, so I might be wrong. Ahh, you're about where I am with this. With any luck, I should be able to start working on this really soon, perhaps as soon as this weekend. :) Pat -- Purdue University ITAP/RCS --- http://www.itap.purdue.edu/rcs/ The Computer Refuge --- http://computer-refuge.org

Patrick> So, what devices would give us the largest range of Patrick> possibilties for software compatibilty at the larges size? Patrick> <reaches for pdp11 peripheral handbooks> Patrick> In 1970 pdp11 handbook: - no disks, just a teletype and Patrick> papertape reader are listed. Patrick> Added in 1973/4 pdp11 peripherals handbook: - RK11/RK05 - Patrick> 2.4MB disk pack - RP11/RP03 - 40MB disk pack You left out the RC11 and RF11 :-) Patrick> Added in 1975 pdp11 peripherals handbook: - RP04 - 100MB Patrick> (presumably same RP11 interface as RP03) Nope. RP11 controls the RP03 and RP02 (half of the RP03) drives -- basically IBM 2314 lookalikes. The RP04 is the first PDP11 Massbus disk; it uses the RH11 (or RH70 for 11/70s) controller. It's 88 MB (half of the RP06). Patrick> Added in 1975 pdp11 peripherals handbook: - RL11/RL01 - 5MB Patrick> disk pack Patrick> Added in 1978/9 pdp11 peripherals handbook: - RK06 - RK11, Patrick> 14MB disk pack - RK07 - RK11, 28MB disk pack - RM02/03 - Patrick> RM11 MASSBUS interface, 67MB disk pack RH11... there is no such thing as RM11. The RK06/07 controller is called RK611. (RK11 is the RK05 controller.) Patrick> Added in 1981/2 pdp11 peripherals handbook: - RL02 - 10MB Patrick> disk pack - RP06 - 176MB disk pack - RP07 - 516MB winchester Patrick> - RM05 - 256MB winchester - RM80 - 124MB winchester Patrick> So, it looks like to me the best choices would be an Patrick> RK11/RK05(06,07), an RP11/RP04(05,06,07), an RX11/RX01 and Patrick> and RL11/RL01(02) emulator. If I were to pick a single Patrick> choice right now, I'd probably pick the RP11 controller, for Patrick> the breadth of possible sizes (from 40MB in 1973 to 516MB in Patrick> 1981), and put RL11 as a second choice, simply because I've Patrick> got one. :) RK11/RK05 seems too small to be worthwhile, but Patrick> the RK06 and RK07 aren't bad sizes. So actually that would be the RH11 for max flexibility; that gives you RP04/5/6/7, RM02/3/5, and RM80. You can have up to 8 drives (that's true for most of the others -- but RL11 is limited to 4). Also, two RH11/RH70 controllers per system is a standard config that most DEC operating systems support, while more than one of the other disk controllers isn't. RK11 may be a good choice -- even though the drive is quite small -- for *very* old OS versions. RP11 (RP03 drive) might also serve, but I suspect support for that isn't as common. For example, I know DOS-11 V4 works on RK11, I don't know if there's an RP03 version for it. Speaking of block access devices -- for supporting old OSs, a TC11 emulator might be helpful... :-) paul

Tom> I had considered a TC11, but I'm not sure of the benefits other Tom> than it just being interesting... If you want to install DOS V4, it's about the only choice for distribution medium... paul

On Friday 14 May 2004 08:59, Paul Koning wrote: Intentionally, trust me. I didn't think they'd be terribly useful, at their sizes Ok, the handbooks didn't say either way, so I assumed incorrectly. More things that the handbooks don't seem to say. Argh. How different is an RK611 from an RK11 at a programming perspective? Is the RK611 a superset, such that you could pretend it's an RK11? Or not? Ok, then maybe an RH11 and RP11. Hmm. Things I didn't know about, and couldn't consider. Pat -- Purdue University ITAP/RCS --- http://www.itap.purdue.edu/rcs/ The Computer Refuge --- http://computer-refuge.org

Tom> I disagree about MSCP making more sense. If the controller is to Tom> be used on older pre-MSCP operating systems, then MSCP will be Tom> useless. Ok, that's true if you want to run early releases of the various PDP11 operating systems. All the late releases support MSCP. Ok, with the definite exception of DOS and the possible exception of IAS and DSM-11. :-) So for earlier ones, RH11 still seems like a good choice, unless it's really old (like early DOS) in which case the RP11 (RP02/RP03 controller) might be good. paul

On Thu, 13 May 2004 22:34:22 +0100 (BST) ard(a)p850ug1.demon.co.uk (Tony Duell) wrote: Very true. The sound of a spinning disk... But I have only one RK07 medium. So I can not benefit from the posibility to swap media. And: Disk drives are prone to fail. Lots of mechanical stuff that weres out much faster then electronic. The ideal solution is to have the solid state disk for dad-to-day use and the real drive for the feel of it, data exchange, ... Sure. But a FPGA is run time reconfigurable so I can emulate what ever interface I want. I can even emulate multiple different interfaces at the same time as a FPGA has enough complexity. (Maybe a RK07 and a TS11) Meanwhile I got a private mail from someone who has a QBus board in prototype state. It looks much like my proposed UniBus solution with FPGA and PowerPC CPU. There will be a UniBus version if this QBus board becomes a success. So it looks like we just have to wait for this solution to all our QBus / UniBus storage problems. :-) -- tsch??, Jochen Homepage: http://www.unixag-kl.fh-kl.de/~jkunz/

On Sat, 15 May 2004 00:49:28 +0100 (BST) ard(a)p850ug1.demon.co.uk (Tony Duell) wrote: Don't remind me of the sudden death of my /home disk last week... (I had a backup.) Including crashed heads? ;-) No. But I have the design so I can stick the entire design into a single FPGA in 20 years to rebuild my replacement with then current technology. Programming a FPGA is a lot less expensive then building a new machinical drive. I need a workshop and lots of tools to maintain a mechanical drive. This can be expensive too. Depends, IMHO. True. ;-) -- tsch??, Jochen Homepage: http://www.unixag-kl.fh-kl.de/~jkunz/

Hmmm.. Certainly the heads can be replaced at home, but that assumes you have the right replacements. I think making heads is beyond a home workshop, but I would love to be proved wrong (and probably will be sometime -- _never_ underestimate what a hacker can do if he tries...) Anybody who believes that designs will port to a new FPGA without problems has almost certainly never tried to do this. I speak from bitter experience... That depends on thw tools you have and what you'd need to get. Certainly programming an FPGA for me would be _very_ expensive and time-consuming. I think I could make a floppy drive (not a hard drive...) -- apart from the heads themselves -- in less time, and certainly for less cost. I have bitter memories of compiles that took all night, and which had totally mangled my circuit so I then had to spend a day or so sorting out the mess. And then repeating when I needed to make some small change. No thanks... -tony

On Sun, 16 May 2004, Tony Duell wrote: Tony, wouldn't one need a (somewhat :) clean room for such stuff, and (for heads) optical gear and all that?? I still believe (as I just explained in detail to another list member ;-) that such a QUnibus board design should be based on a micro only- the "classic" design, where, if we really have to (read: if the micro is no longer available..) can quickly redesign it with another micro. I'd base it on an Intel chip, *simply because* of their availability by scavenging old PC's. --fred \

On the demountable drives -- ones where the user is expected to interchange the pack -- the heads can be replaced in a normal workshop. The manuals warn you to keep things as clean as possible (and certainly not to smoke :-)), but you don't need a clean room. To _make_ heads is another matter... -tony

I assume the faild servoid was the thing that was smoking. I've had drives smoke, of course -- like other electronic devices they run on magic smoke (when it leaks out, they stop working :-)). If this had happened to me, I would have escorted that idiot off the site, and told DEC to send somebody who actually had a clue if they expected me to pay the service contract! No, I am not joking. FWIW, I forbid smoking anywhere near any of my machines, both for my health and theirs. I don't blame you! I don't know why you avoid such terms. They are sadly all too appropriate. -tony

I think it's the fact that he was burning something near a dismantled disk drive that was the problem, not whether he was deliberately inhaling the fumes :-) Well, I suspect you could argue that the person clearly didn't have a clue about the operation of hard disk drives (I assume _you've_ seen that picture in the DEC manuals showing the head flying height in comparison to a smoke particle, human hair, etc), and that therefore DEC were not providing the service they were being paid for. How do you spot the faild servoid who had run out of petrol (gas)? He's the oen swapping out the tyres to find out which one is flat. What does he do if he can't find the flat tyre? He replaces the alternator. Hmm.. I have no arguments with the first 3 groups. But the whole attitude of field service, at least the ones I've come into contact with, seems to prevent logical thought. And that results in random parts being swapped, and the problem still not being cured. I will say it again. You will _NOT_ fix a computer by swapping parts until the problem goes away. It is _ESSENTIAL_ to make measurements first. Actually, that applies to fixing just about anything. -tony

Where are guys like this now? Do any of them participate in "classic computer" activities, or have they moved on to other things, such as fishing or woodworking? The hobbyists sure could use guys like this from time to time, I would think. Ashley ----- Original Message ----- From: "Paul Koning" &lt;pkoning(a)equallogic.com&gt; To: &lt;cctalk(a)classiccmp.org&gt; Sent: Tuesday, May 18, 2004 9:45 AM Subject: Re: DEC RK07 drive interface specs wanted

I always admired and respected the true debuggers. I've recently acquired (and am in the process of acquiring more) 1970s vintage DEC equipment (PDP-11s, RK05 drives, VT52 teminals, etc). As is to be expected, some things are showing their age and may need the level of debugging that you have described in order to determine and correct their problems. I am going to be learning a lot! Ashley ----- Original Message ----- From: "Paul Koning" &lt;pkoning(a)equallogic.com&gt; To: &lt;cctalk(a)classiccmp.org&gt; Sent: Tuesday, May 18, 2004 10:25 AM Subject: Re: DEC RK07 drive interface specs wanted

Tony Duell wrote: Nope; then you find out about dirt and worms. :)

On Tue, 18 May 2004, Paul Koning wrote: And it must also be remembered that these folks did this when it was economically feasible to spend a day or two tracking down and repairing a problem. Business is time, and time is money, so as technology evolved and came down in price, it became more practical to just swap boards (both for the user and the supplier). Sometimes we lose a sense of the more pragmatic aspects of tech work. If your business is halted because a computer system is down, would you rather your tech take a few hours or a day or two to track down and fix a problem, or would you rather they swap a few boards in an hour or so until they find the problem? -- Sellam Ismail Vintage Computer Festival ------------------------------------------------------------------------------ International Man of Intrigue and Danger http://www.vintage.org [ Old computing resources for business || Buy/Sell/Trade Vintage Computers ] [ and academia at www.VintageTech.com || at http://marketplace.vintage.org ]

On Tue, 18 May 2004, William Donzelli wrote: My customers are ready to hang me if their systems drop out for more than, say, half an hour *scheduled* time, or 5-10 mins unscheduled. So, they invest in lots of redundancy, which helps. --f

On Wed, 19 May 2004, Tony Duell wrote: Tony, in some cases, a business being down a whole day means that business goes under completely. It's not as cut and dry in the "real world". -- Sellam Ismail Vintage Computer Festival ------------------------------------------------------------------------------ International Man of Intrigue and Danger http://www.vintage.org [ Old computing resources for business || Buy/Sell/Trade Vintage Computers ] [ and academia at www.VintageTech.com || at http://marketplace.vintage.org ]

On Tue, 18 May 2004 07:37:49 -0700 (PDT) Vintage Computer Festival &lt;vcf(a)siconic.com&gt; wrote: It is also worth noting that often times the 'real' tech people aren't deployed in the field. I've done component level troubleshooting work in the past at the shops where the pulled cards are shipped. It often isn't efficient for the field staff to do that sort of troubleshooting. Repair and testing of the pulled boards DOES happen, in facilities where the equipment, expertise, and documents are more complete. Sadly, this isn't always the case anymore, but generally up until the recent past it has been. Certainly in the case of higher-quality equipment from the likes of DEC, IBM, etc., but there's a lot of throw-away junk getting deployed these days.

On Sun, 16 May 2004 00:20:06 +0100 (BST) ard(a)p850ug1.demon.co.uk (Tony Duell) wrote: [FPGA work] I have to admit that I never worked with FPGAs. I work for a company that is producing an embedded product that is based on a StrongARM and has also a FPGA inside. I can't remember that one of our hardware developers ever complained that hard about the FPGA programming software. But maybe the tools they use are very different from yours. AFAIK they use some shematic capturing to design the FPGA, no VHDL or the like. -- tsch??, Jochen Homepage: http://www.unixag-kl.fh-kl.de/~jkunz/