Tektronix 7834 Mainframe readout repairs

DSC_1001(attempted) Repair log continues below

Picked this up about a year ago, and used it solely with my then slightly misbehaving 7D01, in combination with the DF01 Display Formatter.  Because the DF01 was gen’ing all the characters, I never noticed that the readout feature was misbehaving until I went to use a pair of conventional plugins.

7834 readout

There’s a few things wrong with this:
1) The characters are incorrect
2) The top and bottom of the vertical plugin readout are supposed to be for channels 1 and 2 respectively, but they both seem to respond to channel 1 only.

Regarding the incorrect characters, it appears that from this table, the characters are shifted one to the left of what they should be. ‘0S’ should read ‘1V’, and ‘4m’ should read ‘5u’  Columns and rows are selected by way of a current loop between the plug-in and mainframe, using .1ma steps between 0 & .9ma.

Oh, hello:
rowmatch

“Score!” I thought,  but sadly turning it does nothing.  Also, my columns are the issue, not the rows.  UPDATE – Column Match doesn’t do anything either.

So I started poking around with my 7D20 and found something strange coming out of the U3433, the custom timer chip:7834 readout board

Note the often overlooked screen annotation feature on the earlier 7k scopes 😉

So there should be a negative going pulse on pin 16 that corresponds to the waveform on pin 10.  Also note that negative portion of the waveform on page 10 are shorter where the pin 10 pulse is missing.  I went ahead and ordered a replacement U3433, we’ll see if that does anything.  In the mean time, I’m off to the tek message board…

09/16 UPDATE:
Heard back form a gentleman on the forum:
1) What I’m seeing on the output of U3433 is not uncommon and due to display skip (which is described in the manual, I just have to wrap my head around it)
2)  He recommended swapping U3429 row decoder & U3418 column decoder – did that, no change.
3) He recommended swapping U3232 row data switch and U3263 column data switch.  oh fuck, those are BURIED.  Standby for some fun.

9/16 UPDATE:
The power supply slides right out, making U3232 & U3263 accessible.  Well played Tektronix.   Swapping them made no change.

Here’s the bottom of the scope – note the plastic rails on the bottom of the PS on the left.  One of those was loose and keeping the PS from sliding out.  Had to pop the bottom cover off to free it.
DSC_0989

DSC_0992

resocketing a few of the transistors fixed the missing readouts on the left horizontal bay, and the upper/lower duplication I was seeing.  Now every bay has it’s readout, it’s just wrong.

9/17 UPDATE:
what’s supposed to be +4.4v on pin 9 into column decoder U3418 is only +3.84v
what’s supposed to be +14.5v on pin 10 is 14.8v

9/19 UPDATE:
Well, this has suddenly turned into a larger problem.  Yesterday I’d gotten some of the characters to change by paralleling in a decade substitution box.  Today I tried again but paralleled the wrong resistor (R3418), and I as was trying to pull the lead (live – dumb) I shorted something, heard a snap from the power supply, and the display went crazy, then dark.  I shut it down, and when I started it back up, U3447 started to release it’s magic smoke.
I can only suspect that one of the legs of the power supply has gone awry, but the power supply wont fire up when not under load.  I’m at a complete loss as to what to do next…

OK – moving on from despair…
With the readout board fully disconnected, the scope powers up and behaves normally (sans readout).  IMPORTANT NOTE:  When disconnecting the mini coax connectors, note that some of them have shield potentials of 15v, so take care to make sure they don’t touch the chassis.

9/20 UPDATE:
It was U3401 – zeros logic & memory.
I decided to swap out one chip at a time from the 7834 to the known good 7603, in order to asses the damage.  When I got to U3401, the same problem showed up on the 7603.  Total facepalm for not doing that earlier.  All other chips were OK, with the exception of U3418, the column decoder.  So now I’m down:
U3418 Column decoder –  tek part 155-0014
U3401 Zeros logic & memory – tek part 155-0018
U3477 7402 quad 2 input NOR gate – I only have an 74ALS02, not sure if that’ll cut it.

Replacements for U3418 & U3401 en route

9/20 UPDATE:
74ALS02 is fine for U3477, tried it in the 7603 (thanks David)

In retrospect, I should have suspected U3401 given that ‘>’ worked but ‘IDENTIFY’ didn’t, since it’s U3401 that signals the column & row data switches to run through the sequence of characters to spell ‘IDENTIFY’.

9/25 UPDATE:
After an in situ check of components around the affected areas, I pulled U3418 & U3401 from the 7603 and installed them in the 7834, held my breath, and power’d it up.  All good.  The soon-to-arrive replacements will make the 7603 whole again, though for now it doesn’t really matter, since it’s really only host to my 7D20 & 7D01.

 

7834 Readout repairs, continued

 

After my previous antics with the 7834 readout system, the outcome was:

  • U3401 Zeros logic & memory – tek part 155-0018 – the original culprit.  Replaced with one from ebay.
  • U3418 Column decoder –  tek part 155-0014, which I think I killed during repairs. Replaced with one from ebay.
  • U3477 7402 quad 2 input NOR gate – smoked during repairs.  Replaced with a 74ALS02 from stock.
  • It turned out U3232, the row data switch that cycles through the 8 plugin display channels was bad as well,  botching the readout in horizontal A bottom, and horizontal B top & bottom.   Replacement sourced from ebay.

So far, I’ve spent $50 on the scope, and another $50 or so on replacement chips.  Still not bad at all.

Here’s a few shots using the 7B85’s delta time feature, which lets you do fun things like measure frequency, rise-time, and pulse-width directly using the on-screen delay time read-out.  Thanks to Peter on the TekWiki who’s entry on the 7B85 pointed out these features.

These photos are of a rudimentary clock circuit I made using parts from the DF01

FullSizeRender FullSizeRender-1

Colored sql

Status

Thanks to Alexey Kalinin for his grc profile which colorizes mysql.
You also need grc, which is super easy to install by hand (directions in INSTALL worked like charm)

Good stuff!  Below screenshot is from cathode, an OSX terminal program that emulates old terminals (note the scan lines and slight curvature).  It may be trite, and a little hipster, but I get a kick out of it.

screen-shot-2016-12-04-at-8-13-56-pm

Simpson 476 Mirrorscope: Success!

Finally got a trace!  Turns out that the open cathode resistor I replaced that I thought was supposed to be a 2.7k was actually supposed to be a 39k, so after rectifying that (with a 33k), and replacing the incorrect bias resistor, it lit up like a charm.

img_7070

After that, I put a sine in the X & Y, and fiddled with the bias controls until I got maximum linearity.  The sweep & trigger is a little wonky, so I think I’ll run some tests with an external sweep generator to dial the amp sections in first, before I tackle the sweep.  Besides, Benton (who’s scope this is) mostly only uses X/Y mode, so that’s more important.

There is some burn-in visible on the CRT, I may poke around eBay to see if there’s a replacement that’s available and inexpensive.  The filter caps in the power supply are shockingly good, but I may replace them anyway, at least the paper ones.

After a congratulatory sip of Bourbon (4 Roses, one of my faves), I realized that all the troubles in this scope stemmed from bad 39kΩ resistors. I wonder if there was a bad batch, or they were all just under spec’d?  In any event, I’m going to replace them all, and probably up the rating to 2W, just to be sure.  I may also replace the 68k resistors that have drifted as well.

Here it is triggering on a 100kHz square wave, not bad!

img_7073

 

Simpson 476 Mirrorscope – continued adventures

So I took Benton up on his offer of giving me the 2nd working scope, and it’s been super helpful in having a unit to compare to.  It’s hard to diagnose this thing without real schematics – the ones from the 480 are different enough to be more of a source of confusion.

Before I got the second unit, I took the face place off in order to get better access to the components.  Unfortunately the face plate is what holds a lot of the unit together, so it’s in a very fragile state right now.  This thing was not made for easy service.  I have it on a plywood square that lets me move it around and spin it more easily.img_7016

There’s some alligator clips to keep parts of the circuit grounded that were otherwise grounded through the face plate.

I also grabbed the FLIR One from the office and took some temperature measurements.  Here are the two hottest components:

A resistor in the vertical bias sectionimg_7007

A resistor in the power supply:
img_7003

Both ALARMINGLY hot.  (pardon the mixed units, switched to Celsius 1/2 way through).  The FLIR is a really useful tool – the only issue is that at these close ranges, the visual and thermal layers don’t quite line up due to the physical spacing of the cameras.

I started checking values and taking measurements and noticed a few things:
screen-shot-2016-11-19-at-4-40-37-pm

Unlike in the 480, the 2nd section plate resistors are different between the horizontal and vertical sections.  Horizontal are the same 68k, while the vertical is 39k.  All have drifted high, and in some cases out of tolerance. These are 1W resistors, according to the manual.

R109, 2.7k 1/2W, that connects the cathode to ground, was open.  I replaced it with the closest thing I had, a 2.2k.  After that, I could make tiny adjustments in the vertical position, but the beam shape went to shit.

Shortly after that, Benton dropped off the second unit.  I removed it from it’s case, and started taking measurements.

img_7061

For starters, those resistors don’t get nearly as hot, so I started to get more confident that the problem was in the vertical biasing section (that bit with the pots hanging off the back)

Finally, I noticed the smoking gun:

img_7064-1

That piggybacked resistor pair?  That’s R98, called out as a 39k 1W in the schematic, but that there’s a 270Ω in parallel with a 330Ω, for a total of 150Ω !!.  But wait, It gets weirder.  That same resistor in the ‘working’ unit, is color coded as a 39k, but measures at 260k!  W.T.F!?.  When I place a closer value, 33k in the bad unit, I start to get a little more range, but the beam is still fuzzy.  When I replace it with the 260k resistor, I get tons of range, but still a fuzzy beam.

Meanwhile, on the horizontal side of the bad unit, that 39k resistor (R97) has been replaced with 3 50k in parallel, for 16k.

img_7066

Time to order a bunch of 1W resistors, I guess.

 

Simpson 476 Mirrorscope

Up next on the bench, a Simpson 476 Mirrorscope.

img_6849

This example belongs to Benton Bainbridge, an NYC video artist.  He’s got two, and this one needs the most help, so I figured I’d get it on my bench and get it working again.  That’s the hope, anyway.  As can be seen from the photo, the CRT points upward, and the operator views it through a 45° mirror, that flips up as the top of the chassis.  This yields a benchtop oscilloscope at an unheard of depth of only 8″.
It turns out, the manual for this beast is no where to be found.  I’ve reached out to Dave at Arktek manuals to see if he can dig up something.  I’ve gotten hard-to-find manuals from him before, so hopefully he can work some magic.  I also reached out to Steven Johnson, who’s got a great page and manuals for sale.  The Simpson 480 Genscope is approximately one of these mirrorscopes with a signal generator on either end.  I can’t even begin to imagine how unweildly that is to work on, but at least there’s a manual available.

I met Benton at his place, and we started comparing the “good one” and the “bad one”:

img_6824

Sweet DJ rig, yo

The problem seems to be in the vertical amplifier section, made up of (2) 6K6 twin triodes. One the good unit, the signal to the deflection plates varies between 170 – 240v while on the bad unit, they’re pegged at 350v, which appears to be full anode voltage.  Tubes were tested, and swapped, to no avail.  I brought my Heathkit TT-1, which Benton got a kick out of – it’s such a fun instrument to operate.

There were a few ground connections that appear to have been severed, but reconnecting them didn’t solve the issue.  There are a number of resistors that are running hotter, so much so that they’re discolored; however when measured cold, they appear to still be in tolerance.

There are a few differences between the 480 and the 476:

  • The 5Z4 rectifier is absent in the power supply.  This appears to power only the oscillators, so no surprise there.  The power supply consists of 2 6X4s, one for the CRT HV, at about 900v, and one for the rest of the circuits, at about 350v.
  • There’s an extra 12AU7 not present in the schematics for the 480.  I suspect this might be a pre-amp for the vertical input.  The 476 has more rages than the 480: 4 steps from .5v to 500v as opposed to ‘low / high’ on the 480.  Alternately, it’s a part of the sweep / trigger circuit?  I know that this is scope is capable of ‘synchronized’ operation, but I can’t for the life of me see how this thing triggers by looking at the schematic.
  • Some other component value differences.

Here’s the relevant part of the 480 schematic.  simpson-480

Benton offered that I should take both, to have  good one to compare with.  I declined, only wanting to bring two additional cubic feet of oscilloscope into my apartment, but I may take him up on it if I can’t get this figured out.  On old gear like this, I immediately suspect old caps in the power supply, but since the horizontal sections seems to be working (ish), that doesn’t screen out as the culprit.  I think I’ll just start checking / changing out caps in the PS for good measure, as I familiarize myself with this unit.

The other thing I’m definitely going to do is grab my boss’s FLIR One to scope out any really hot (or suspiciously cold) components.

But before I do anything else, I gotta order some more caps, my HV Electrolytic stock is almost depleted.  I kind of like running out of parts, it means I’m keeping busy.

Heathkit H89A – Continued boot failures

So I’ve tried just about everything, and the consensus is that I’ve just got bad disks.  A bunch of people on the SEBHC Board has been super helpful, a few even offering to send me boot-disks.  I took one gentleman (Ken) up on his offer, so hopefully that’ll get me up and running.
Lee suggested some lines on the floppy connector to probe, which I’ll get to tomorrow.  May even sick the 7D01 on this…
This manual has me convinced that the BIOS is correct. It’s a newer one to go along with the floppy controller card, which just isn’t referenced in any of the printed documentation I have.
Ken confirmed my understanding that the floppy drive is a fairly dumb device, and it’s the controller that determines sector firmness, track count, & density.  He’s even shared wiring instructions to get 3.5″ drives.  I’m eager to try some newer drives and disks once I get up and running.

Heathkit H89A – floppy config & attempted boot

I’ve tried every disk in every drive with what I thought was the working configuration, to no avail.  Every failure is the same: hitting ‘b’ to boot, getting ~5 seconds of disk activity, and failing with ‘? Boot Error’.  Heads appear to be in impeccable condition, and the disks look clean.  Here’s a closeup of the heads on the hard sector drive:

Floppy drive head closeup

And here’s the collection of disks I’m working with:

img_6437 img_6439

SW501,4 controls which card to boot from:
0 = H88-1 Card at P506/P512 (the right most slot), I/O Port 174
1 = Z89-37 Card at P504/P510 (the left slot of the right bank, I/O Port 170

I’ve tried both; 0 boots to the internal, hard sector disk, 1 boots to the 1st drive in the H-37 enclosure, a 96tpi soft sector drive.

The internal drive is set to Unit Zero (or DS1) for use with the H88-1 card
Screen Shot 2016-08-14 at 3.45.36 PM

But wait, we can also use this drive as a soft-sector drive on the the Z89-37 controller.  Note that the drive settings are different:
Screen Shot 2016-08-14 at 3.45.44 PM

Yes, it’s backwards.  My predecessor was nice enough to not cut the links, but to simply bend the pins of the package out.  unfortunately, they’re super fragile, and a single attempted bend broke the pin, several times. I ended up using a portion of the original package and some wire.

Note that J4-J7 needs to be set correctly on the Z89-37 card.  It’s now set to J4, meaning the single drive ID’s as DS1 plugged into P3 (the top port on the card) is Drive 0.

As I was shuffling connectors and moving the board around, this happened (took me a few minutes of failed boots to spot):
img_6436

I temporarily crammed it back in, and got to the H: prompt.  Still won’t boot to a disk.

I wonder if this is noteworthy:
Every boot attempt seemed to take the same amount of time to fail (3-4 seconds), regardless of which configuration or even if a disk was present.  When I jump J4 on the Z89-37 (what I believe to be the correct config), an attempted boot without a disk tries indefinitely and never fails.