The BA11-K supply is benign in this respect. It uses a convenitonal iron-cored mains transformer to drop the incoming mains voltage to 20-30Vac. This is then fed to the regulator 'bricks'. While these contain some large capactiors [1] storing enough energy to blow transistors apart, the voltage is not high enouhg to give you a lethal shock. [1] A friend of mine (not on the list) refers to them as a 'CFE', standing for 'Capacitor, F***ing Enormous'. -tony

You should have done! I can assure you a groudn loop -- a small omne like you'll have here -- will not cause problems withterse sorts of signals. Try again with the 'scope ground led connected. -tony

Agreed on the scope connection: hook up the ground. :-) And check the filter caps - unless they've been replaced in the last fifteen or twenty years (check the date codes) they are Prime Suspects in the sort of noise you're seeing (unless it's from your scope setup). It's worthwhile to pull the big ones and perform a leakdown test on them. Another question: are you certain that all the NPG lines are feeding through? I don't know whether a break in the NPG chain will cause BUS ERR on the display - I've never tempted that fate - but if the machine had a DMA device like a disk controller in it at some point and it was removed, the small G727 is not sufficient in that slot. You need a G7272, that also jumpers CA1 and CB1 - or if you find a SPC slot that doesn't have those two connected, you can just replace the jumper on the backplane. (I'd check it for you on my 11/34, but I have too much stuff stacked in front of it right now - sorry.) That's a guess, as I'm not sure if an NPG failure gives you a front panel error or just hangs the machine. -- Ian ________________________________________ From: cctalk-bounces at classiccmp.org [cctalk-bounces at classiccmp.org] On Behalf Of Pete Turnbull [pete at dunnington.plus.com] Sent: Wednesday, July 27, 2011 1:55 PM To: On-Topic and Off-Topic Posts Subject: Re: Power Ripple in DEC BA11-K On 27/07/2011 12:14, Jochen Kunz wrote: Not necessarily, I do it all the time. And the lack of a local ground is almost certainly where all that hum and noise is from. Classic sign of something wrong with the bus grants. If the grant cards are all the small type, are they the right way round? Is there one in every slot, including the last (slot 9)? -- Pete Peter Turnbull Network Manager University of York

Pete Turnbull <pete at dunnington.plus.com> OK, I'll recheck with the ground clip connected when I am back home on sunday. All emty slots have grand cards. The NPR wires on the backplane are in place, except slot 8. There is a non DEC, double wide grand card in this slot. I checked that it connects the grand pins and NPR. I'll recheck with an ohm meter in case there are any loose contacts. I have to reconfigure the serial console DL11-W also. Currently it is set to 110 BAUD but my VT520 doesn't support this. I need to build a new serial cable also. Thats the reason why I didn't check serial console output. I was in the hope to perform basic checks with the programmers console and care about the serial console problems later.

On Tue, 2 Aug 2011 15:21:46 +0200 Jochen Kunz <jkunz at unixag-kl.fh-kl.de> wrote: Meanwhile I found: http://www.retrocmp.com/tools/pdp-1134-programmers-console which describes how to use the programmers console for basic bus testing. I removed all cards but M9301-YF (bus terminator + ROM), M7859 (programmers console interface) and M930 (bus terminator). I tried to read back the Switch Register as described in the URL above and it doesn't work. I don't even get a bus error, when I remove the terminators. The "Run" LED is always lit. The manual states that examine and deposit commands are only possible when the processor is in halt state. But there is no processor and even if there is one, I get a "Bus Error" when I try to halt the CPU by pressing CNTRL + HLT. So I suspect that the console does not execute examine and deposit because it thinks the CPU is still running. -- \end{Jochen} \ref{http://www.unixag-kl.fh-kl.de/~jkunz/}

On Wed, 3 Aug 2011 21:14:19 +0100 (BST) ard at p850ug1.demon.co.uk (Tony Duell) wrote: This is about where I am at the moment: I try to figure out how the Run LED is controled. Unfortunately the scans of the prints are close to unreadable. The other question is: Does the 8008 microprocessor in the console interface honor the HLT signal at all? I surely don't want to go as far as reverse engineer the 8008 firmware. I am also reading the "pdp11 bus handbook" to educate myself about the Unibus protocoll. Intention is to hook up a scope or logic analyzer to the bus and try to capture bus transactions. Unfortunately the book doesn't talk abut SPC slots and the HALT signals are a SPC only. -- \end{Jochen} \ref{http://www.unixag-kl.fh-kl.de/~jkunz/}

On Wed, 3 Aug 2011 19:13:04 -0500 Paul Anderson <useddec at gmail.com> wrote: Gues what PDF, amog many others, I am currently reading... :-( Meanwhile I deciphered (halfway) the halt / bus busy section of the programmers console. As far as I can make a sense out of this mess (the printset is erroneous, confusing and barely readable), the Run LED should go off as soon as I put the console in bus master mode via the maintenance mode "5". But this doesn't happen. Either because it never enters maintenance bus master mode or some other condition clears this mode immediately... -- \end{Jochen} \ref{http://www.unixag-kl.fh-kl.de/~jkunz/}

On Thu, 4 Aug 2011 19:37:35 +0100 (BST) ard at p850ug1.demon.co.uk (Tony Duell) wrote: I know. And this is the problem. These signals are available in a SPC slot and used by the console but not covered by the Unibus documentation. I still have to google for a description of those signals. I have 40 channels at 50 MHz or 16 at 300 MHz. That should be enough for some basic testing. A friend gave me an isolating transformer. So I can exclude ground loop problems. He will give me a logic probe also next week. This seems to be a simple but helpfull and effective test tool. I checked the +5 V supply again using the isolating transformer. Voltages look much cleaner now. But I get a dampened oszilation with ca. 0.5 Vpp. This occures at about 17.8 kHz. Seems to be "ring" from the SMPSU as I have only a very low load. http://www.unixag-kl.fh-kl.de/~jkunz/tmp/5_Volt_Ripple__y=0.1V__x=20us.jpg http://www.unixag-kl.fh-kl.de/~jkunz/tmp/5_Volt_Ripple__y=0.1V__x=0.1us.jpg -- \end{Jochen} \ref{http://www.unixag-kl.fh-kl.de/~jkunz/}

My point (and I am being pedantic I suppose) is that this is not an SPC slot. It's a special slot for the console controller. Good luck!. There may be something in an 11/34 techncial manaul, but much of the time on thigns like this I find I have to work them out from the schemcatics. Plenty :-) A logic probe is very useful,but bve warned it doesn't give the full story. The prove will tell you if a clock is running, or if a data line is stuck high or low, but if it shows a signal as changing, it tells you nothing abotu the timing of that signal. That;s what you use the 'scope of logic analyser for (and yes, I have been cuaght by this...) My favourite tool for going over a machine looking for silly faults is a LogicDart, which combines a logic rpboe with a simple3-channel logic analyser. Problem it,s it's not an easy thing to find (and AFAIK there's no modern replacemetn that does quite the same thing). That sounds like noise at the switching frequency. My guess is that it means the output capacitor (IIRC the second-largest one in the module) oin the 5V regualor brick is developing a high ESR. Probably worth testing/replacing it. -tony

It is a very useful tool, sure (I use one all the time), provided that (as with all test gear) you realise what its limitations are. So was I, didn't stop me getting caught though... Let me expalin. The machine in question was an HP9820 (as mentioned in my otehr message). Now, with thse machines, I start by checkign the power supplies (all rails fine), then the clock board (both the bit clock and the microcode clock were running), and then I put enough of the machine togethr for it to run (that is all 4 CPU board,s the memory unit and the display board) and see what it's doing. Normally I strt with quick tests to look for 'sillies', if that doesn't find the fualt I usea full-blown logic analyser to trace the CPU microcode and see jsut what it is doing. Anyway, as I mentioned in my previous message, it's bit-serial. There's a 16 bit shift register claled the 'M register' which does a conversion are shifted into the MSB end of this register (as you'd expect), and after 16 clocks, there's a new address o nthe address bus. The outputs of the M regiser are brouight out to pins on a test conenctor on top of the memory box backplane, and I ran my logic probe over them. This showed that the top 4 bits were all chaning, the bottl 12 bits were all stuck low. Now, the M register is made up of 4 cascaded 4 bit shift register chips. I decided based on taht that the chip handling bits 11..8 had failed so that its ouptus were all stuck low . Therefroe the lower-order chips were being laded with 0's all the time. So I desoldered the appropriate chip on the memroy address PCB and replaced it. No change. Not just 'still nothing on the display', but the M register outputs on the test connector were still doing the same things -- top 4 changing, bottom 12 all 0's Then I useda logic analyser. I found that yes, the top 4 bits were changing, but the timing was craxzy, they were not just chaning on the correct clock edge. It appeared they were not being latched at all. So in fact the next shift register in the chain -- the one I'd replaced, always had a 0 at its serial input when it was shifting. Hence its ouptus were always 0 Out with the highest shift register chip in the M rwegister, and I replaced it with the one I'd taken out from the next position and which I guessed was good. It was. This time I got the correct display. If I'd used a logic anaylser at the start I'd have found the fault first time. -tony

ard at p850ug1.demon.co.uk schrieb: That's what I expected, as the machine came with this terminator instaled in the tape controller. The tape controller is in a separate box, so I moved the terminator into the CPU box, when I pulled the UNIBUS cable from the CPU to the tape controller box. I suspected that I cold get away without terminator when the bus is limited to the 9 slot CPU system unit. So I tried without terminator, but this made no difference. No, as I wrote. I'll recheck with the ground clip connected. Any electrolytic capacitor potentionally suffers from this problem, not only those old ons. But sometimes even very old capacitors work well. Mains-frequency Thanks for the hints. I'll have a closer look when I return to the machine. I don't have an ESR meter. I'll try to find an other way of testing the capacitors.