Tektronix 564 Revisited

I recently picked up a somewhat rare 3A5 auto-ranging plugin for the 560 series, and decided to fire up my 564 to test it out. Since it had been over a year since I powered up my 564, I tried it first in it’s current config, a 3A72 Dual Trace plugin, and a 3B3 Timebase. Two problems were readily apparent:

  1. The varying intensity – I strongly suspect this has to do with AC ripple, since it holds steady when line triggered
  2. The trace is about 2 CM short.

Here’s a table showing measured supply rail voltages & ripple. All voltages are within spec, but the ripple on the -100 V, 125 V & 300 V lines are out of spec.

RailDC VoltageRipple
-100 V-100 V120 mV
125 V125.5 V40 mV
300 V301 V120 mV
-12.2 V-12.21 V2 mV
355 V357 V
420 V435 V1.5 V
475 V493 V280 mV
-3,300 V-3,290 V8 V

Replacements:

  • -100 V: C640, a combo 340uF & 10uF 200 V Electrolytic
  • +125 V: C642, a combo 340uF & 10uF 200 V Electrolytic
  • +300 V: C644, a combo 340uF & 10uF 200 V Electrolytic – this is only a 200V cap since it’s low side is off the 125V supply.
  • +475 V: C646 2x 40uF

So that’s a rebuild of 3 cans. C642 & C644 should take care of the ripple on the +420 V unregulated rail, which feeds the high voltage supply.

Using a fluke 6kV probe into my 7854 scope, I was able to measure about 50 V of high frequency ripple on the 3,300 V supply, however I’m not 100% certain that measurement was accurate. I could pick up about that much just by having the probe within a few inches of the HV supply, so I’m not sure how much of that was inductive pickup through the probe, or even the cable of the probe? I’d love someone to set me straight on that.

Following some recommendations on this old post on the tekscopes group, I gave the ceramic binding posts a thorough cleaning with IPA & cotton swabs. So far, I have not seen the intensity modulation problem return that didn’t seem to eliminate the problem, as it returned the next day, albeit not immediately, but after 30 or so minutes of operation. I did discover that with the cover reattached over the HV transformer & diodes, I was no longer able to detect the HF ripple when measuring high voltage. The AC ripple still disappears as soon as it’s measured.

Replacing C342 & C344 greatly reduced the ripple in the 125 V & 300 V supplies, but did nothing for the ripple on the 420 V unregulated, and hasn’t fixed this intensity modulation issue (I was pretty sure it wouldn’t though). I have confirmed that the intensity modulation is at 60Hz, and isn’t related to any errant signals from the horizontal plug-in by running the vertical plug-in in the horizontal plug-in’s slot, and feeding it a 60Hz ramp. The issue reoccurred.

As for the sweep length on the 3B3 – For now I was able to get the trace to the correct length using the sweep length adjustments pots, however, I had to crank them both (the normal and delayed sweep) fairly far clockwise. Once adjusted to the correct length, I do get sweep waveforms as documented in the manual, so I don’t suspect that I’m compensating for a problem further downstream.

Beam Intensity

There are, I believe 3 distinct different methods to modify the beam’s intensity:

  1. J21-14: Intensifying Pulse – changing the intensity of the beam by altering the ground reference of the control grid supply
  2. J21-13: Unblanking Pulse – deflection blanking
  3. J11-24: Chopped Blanking Pulse –

Continued here, 2 years later…

Akai S900

Taking a quick detour from oscilloscope repair to attempt to fix a RAM error on an Akai S900 sampler that I just scored for cheap. The two issues were a dim backlight, and RAM error “Bad RAM Bit DK09”

the logic board – check out that honkin linear power supply

Thanks to DOSputin for compiling a reference of what codes refer to what chips here.
The best copy of the service manual can be found at this page, along with a wealth of other info on the S900.

DK09 refers to IC62. This thread confirms that there’s no way to test them all at once; you only get one bad ram message at a time, which means you have to take the whole damn thing apart, change the ram, reassemble it, then do the test again to find out where the next problem is.

Someone has been in this unit before. 3 of the 6 screws to attach the voice board were missing. It appears someone had dealt with a RAM issue before, as IC60 is now socketed. I’d planned on doing the same. After replacing IC62, I get another error: DJ12 – IC57. Fortunately, that was it, now the RAM test passes.

The latest firmware is V1.2C. I have V1.2A. I don’t think this matters, as I boot to what’s either called OS2 or OS4 with a floppy.

Tektronix 514 – Back at it

Well, after 4 years and two moves, I’m finally back at this old beast. I’ve chronicled my previous work here, and where I left off was getting a sharp, swept trace, if I disconnected the +225V line from the upper deck where the Vertical circuits are. There are some leaky caps up there, and with them in the circuit, they pull a bunch of other things low, including the HV section (I don’t remember why, but it made sense at the time).

This is now a SEVENTY YEAR OLD machine, and given that every Black Beauty & Bumble Bee capacitor I’ve come across so far has been the cause of trouble, I think the best course of action is simply to replace all of them.

After changing / checking / reforming every cap on the upper deck, I still was having problems with the +225V supply, so I turned my attention back to the regulator. It turned out that the series pass tube wasn’t working: All of the current was flowing through the bypass resistors, so the voltage level was based only on the current. I confirmed this by pulling the Series pass tube (6AS7), and noted no change in the behavior. Working back from the voltage trim-pot, I discovered that the 12AX7 comparator wasn’t working properly. The socket needed some cleaning, and then the whole thing locked right up at 225V.

So, it “works”!
Here it is triggering at 10MHz


Issues:

  • Sweep linearity: going to change out the bumblebee caps in the timing circuit to see if that helps.
  • Sweep magnifier all sorts of weird.
  • Interplay between vertical position & vertical attenuation controls: need to look into this more – The ‘Vertical Attenuation’ control affects the vertical position of the trace. Not sure if this is supposed to happen or not, but I’d think not.
  • Change in intensity yields change in trace width: It seems to affect the horizontal trace width only, and happens in a counter-intuitive way; the trace gets wider when the image gets brighter.
  • Calibrator has an overshooting leading edge.

Sweep Linearity:

note the difference in slope between the beginning of the sweep waveform, and the end of it. Notice the difference in the slope between the start and the end of the sweep in the traces below. The worst of it was on the 100uS per division setting, and a new cap helped.

Sweep Magnifier:

the sweep wave form when sweep magnifier is turned on and swept through

C124, a .1uF bumblebee was the culprit. Replacing that, and restuffing C126, an electrolytic, solved the problem.

Calibrator
The problem turned out to be, what pretty much every other problem in this scope was – a bad bumblebee capacitor. I’m just replacing all of them as I work across the scope; I replaced this one in passing, and the problem was gone on the next power up.

AC Socket
I’ve never seen one quite like this before. It’s recessed, as are many of the later tektronix sockets, but there’s no ground pins, so the blades are centered.
– Screw spacing = 1 1/2″
– Chassis hole diameter = 1 1/8″
– Socket inner diameter = 1″

According to some folks on the tekscopes group, these were not grounded, but rather accepted common extension cords of the time – searching used General Electric extension cords on eBay yields some good hits.

This picture from the 1952 catalog confirms their shape, and that the socket was not grounded. A member on the Tekscopes group says the manual instructs the user to ground the instrument via the front panel banana jack, but I couldn’t find reference to that.

If I wanted to replace it with a grounded alternative, I’ve got a few options:

  • an IEC cord – Would require some filing of the frame hole to fit, and may require some bodging of the steel case as well.
  • a powercon connector. may fit unaltered, at an angle, but might interfere with the back case, or require it to be modified.
  • midget twist-lock – 2 prong will fit, mostly unaltered, but the 3 pin would require opening the hole in the chassis, and re-drilling the mounting holes.

Here’s the drawing of the 3 pole midget twist-lock. I think this is the best compromise, as it doesn’t involve any adapter plates or square holes – just enlarging some already round holes by 1/8″. Incidentally, note the original date on that drawing. I appreciate that there’s a 55 year old drawing on a manufacturer’s site that’s still a current reference.

HP 16500B Logic Analyzer

A collection of notes & research on the HP 16500 series logic analyzers.  I’m buying one in a few days, and trying to understand the differences between versions, and the myriad of features & options.

I believe I’m getting the following configuration:

  • 16500B Mainframe
  • Hard-disk (the owner says he added, and thought that made this essentially a C model, but that’s not quite true)
  • Network option (not sure which one)
  • (1) 16530 400 Megasample per second (MS/s) scope timebase
  • (3) 16531 400 MS/s dual channel scope module
  • (1) 16555A 100 MHz state / 500 MHz 68 Channel logic analyzer 1MS depth

There’s a lot of documentation, and the mainframe and cards have their own manuals.  Generally, everything has a User Guide, a Service Guide and a Programmers Guide.  The mainframe also has a Setup Guide & a User’s Reference.  The User’s Reference covers the basic operation of the mainframe itself. Unfortunately, the only available PDF it has a font issue, making it very difficult to read.
To understand how to use the logic analysis & oscilloscope functions, refer to the manuals for the cards themselves.

Other references online:

Networking

The 16500L card gives you a 15 pin AUI jack, for which an adapter (or Media Access Unit) is required for 10Base-T Ethernet connectivity.
The 16500H card has an RJ-45 jack built in, as well as the expansion card connector (also on the 16500L?) and a “High Speed Port” for connection to the 16505A Prototype Analyzer (is this SCSI?)

The 16500C has an RJ-45 jack built in. X-windows connectivity is supported on the B & C models, but only later cards are fully supported.  For example, earlier scope cards will have a visible interface, but won’t show their trace; only the 16532A Oscilloscope Module will.

Programming via LAN can either happen via sockets at port 5025, echoing commands to \system\program over NFS, or copying a file containing a list of commands to \system\program.

RAM

16500B ships with 8MB, and has been found to support up to 64MB of RAM, though earlier reports from HP say it tops out at 32MB.  It’s unclear if 64MB makes a difference vs 32MB. From Phil’s findings, it seems the mainframe is tolerant of EDO or FPM RAM, and though the spec says parity, you can get away without it.

CPU

16500B & 16500C both run a Motorola 68EC030, per their Service Guides.  One of the links above claims that the ‘B’ runs a 68020

Hard Drive

Turns out the hard drive on this unit is no good (got a few bucks off because of it). These machines are of the IDE era, and the list of supported drives is small, and mostly Quantum Fireballs.  Ko4bb’s page seems to have the most comprehensive write-up, and I already have a Syba SD-CF-IDE-A adapter on the way.  I’m hoping a 128MB CF card that I already have lying around will be supported, but I’ve also ordered a 256MB card that was explicitly on Ko4bb’s list.

Other things to look at

  • 16505A Prototype Analyzer.
  • 16534A 2GS/s, 500MHz dual channel scope (no extra timebase card needed)
  • 16521A 48 channel pattern generator (I don’t know if this works on it’s own, or is only an expander to the 12 channel 16520A).

First Light

I opened it up, gave it a good cleaning, and reseated all of the connectors.  It was mentioned that the hard drive may be problematic, and so is the touch-screen.

  • With the mouse connected, I get an ‘Impaired’ error about the touchscreen, but it seems to work.
  • The hard drive does not work
  • I can boot off either the original Composite System Disk or the hand-made ‘boot disk’, but in either case, none of the cards are recognized, and it doesn’t even see the network card.

I believe the mainframe need to see the modules for the installed cards on the boot disk, so lets see about making our own.

  • If I make a copy of the boot disk by copying files in windows I get an error on boot.
  • Windows complains about formatting a floppy
  • The mainframe can format floppies.  I’ve had mixed success – it formatted one floppy fine, and whined about another one after having just formatted it.
  • I was able to boot with a disk I made by formatting on the mainframe, and copying files over from Windows, however I’m still seeing unrecognized cards.
  • I removed all but one timebase & one scope card, and still not recognizing the cards.

Hard Drive, revisited

The original hard drive finally spun up to life after a number of reboots and I got a few days of playing around with it until the parts came to replace the drive.  I largely followed the instructions on Ko4bb’s page, but had a few difficulties:

  • When copying files (individually, unfortunately) from the floppy to the drive, you have to go into the name field, and hit clear so that it automatically picks up the name of file. If you don’t do this, it still has the name of the last filed you copied, and you’ll end up writing over it.
  • I had a hell of a time getting file writes to stick at first.  My first few attempts gave me an empty hard drive upon reboot.  I wasn’t sure if it was a bad card, or my process (I followed the instructions), but what finally worked was a combination of the last CF card in my inventory, and rebooting after formatting, and then rebooting again after making the /SYSTEM folder, just in case.  That did the trick and I am now able to boot off the CF card to the latest OS.

Touchscreen

The ‘impaired’ status was ultimately fixed by removing the face panel and cleaning off the bezel on both sides, and brushing off the IR emitter / receivers.  One of the links above incorrectly states that this is an acoustic touchscreen – it’s not: It’s an IR grid, similar to some Tek scopes of the same era.  Works great now.

Pod interface

I’ve got this connected up to the RC2014 bus with the flying leads into a strip of header pins, but it’s not something I’d like to repeat.  Instead, I should build a breakout board.  This guy made some inline adapters for various 8 bit platforms, but hasn’t updated his site with any detailed info.

Here’s the pod pinout, from this post that also has some cool X-rays of the pod and leads.

There are no electronics in the pods, but there is an RC network in the probes.  I’m clearly not the first person who wanted to connect their logic analyzer to a board without using a bunch of leads, and HP has a whole document detailing their series of connectors and accessories dedicated to just that.  This document describes an isolation network that should be built into the connector interface, and is I believe pretty much what’s in the probe body of the flying leads.

This is also the circuit in the isolation adapters, which reduce down to a 20 pin connector.  If these are available cheap enough, it may be better than soldering a bunch of SMD resistors & caps.

The plus side is there only 12 – 15 bucks each, the downside it’s it’s more dangley shit hanging off the board, with a fragile flex PCB lead, and more connection points.   Still, it saves a hundred SMD components.

Update – new cards arrived

the owner contacted me and said he found some other cards:

  • (2) 16510B: 35 MHz State / 100 MHz Timing, 80 Channel, 1K Sample buffer Logic Analyzer
  • (1) 16550A: 100 Mhz State / 500 MHz Timing, 102 Channel, 4K Sample buffer Logic Analyzer
  • (1) 16520A 50 Mbps 12 Channel Pattern Generator

Unfortunately, none of the new logic analysis cards work with the existing one, in so much that they couldn’t stack to add channels, rather they’d function as their own separate logic analysis engine.  I could stack the two 16510B cards for a total of 160 channels, albeit at a rather anemic sample depth & speed, or I could use the 16550 for 102 channels at 500MHz, but still shallower depth than what’s on the 16555.  Ideally, I’d get another 16555 card, which I might if the price was right.

Definitely looking forward to messing with the pattern generator though!

Currently Installed:

  • A: 16520A Pattern Generator
  • B: 16531A 2 channel 100 MHz Oscilloscope
  • C: 16530A Oscilloscope Timebase
  • D: 16550A 102 Channel Logic Analyzer
  • E: 16555A 68 Channel Logic Analyzer

In storage:

  • (2) 16510B 80 Channel Logic Analyzer cards
  • (2) 16531A 2 channel 100 MHz Oscilloscope cards

Ideal Configuration

  • A: 16555A 68 Channel Logic Analyzer
  • B: 16555A 68 Channel Logic Analyzer (bonded to above)
  • C: 165621A 48 Pattern Generator Expansion (needs 16520A)
  • D: 16520A 12 Channel Pattern Generator
  • E: 16534A 2 channel 500 MHz Oscilloscope (built-in timebase)

Update 1/5/2021:

Thanks to Alex C who pointed out that if you’re using multiple 16555A cards, you need to be mindful of the setting of the Master jumper – as if they’re both set to master, the mainframe gets confused. See the manual for the jumper settings.

Heathkit TT-1 troubles

While checking a 6EJ7 for my IM-21, I noticed a whiff of smoke coming from my trusty TT-1.

Per a suggestion of someone in the Vintage Test Equipment group, I turned to my trusty TU-75 and finally used it’s series bulb feature.  An incandescent bulb in series acts as a current limiter, which both protects the Device Under Test, and gives a visual indication of when the DUT is trying to draw more current than you’d like. Sure enough, when the tube was connected using the ‘normal / disconnect’ switch, the lamp glows, and I see the voltage drop appreciably.

Side note: I finally got around to measuring and labeling the lamps in my TU-75.  According to the wattmeter, the largest lamp is 500W!  I seem to remember an even larger one that unfortunately got broken due to shoddy packing when I bought this 4 – 5 years ago.

 

 

 

 

What led me to the problem was this line adjust rheostat starting to smoulder.

See me. Spin me. Smell me.

It was suggested that this is not the source of the issue, rather an indication of something upstream drawing too much current.  Hopefully not an underlying issue with the transformer. With 40 taps across a half dozen windings, It’s basically unobtanium.

 

 

 

I didn’t recall ever having an issue with this tester in the past, and sure, things age, but what’s different about this test than countless one’s I’ve done before.  Well, for one thing, I don’t ever recall testing a tube with a 600mA before.  I decided to try testing something with a fixed voltage filament instead.  A 12AT6 with it’s 6.3V filament did not yield the same excess current draw, and tested fine under this setup.

This Page has a great document on refurbishing the TT-1.  It also has the last published tube data addendum.

Here’s another in-depth document on the TT-1.  It references Kent’s document as well, and pages 12 – 15 describe the theory behind the ballast capacitor. The reactance is inversely proportional to the capacitance, so as the capacitance goes up, it’s equivalent resistance goes down.   In my case, they’re all measuring a bit high:

Capacitor

stated value

measured value

error

C11

3.6 uF

5 uF

38%

C10

3.6 uF

5 uF

38%

C9

7.1 uF

11 uF

54%

C9 + 10 + 11

14.3 uf

21 uF

47%

The sum is what’s important for this range – looking at the switch, it appears that the capacitors are paralleled sequentially through the last three filament settings (the current settings).  Is a 47% overage enough to account for this error?  That would mean at 600mA, it’s theoretically allowing ~900mA: Not great, but I’m surprised it’s enough to cause this problem.  There’s nothing else in this circuit – does this mean I should suspect a short somewhere in this tap?  It’s the top-most tap of the filament winding, #29. 

I’m going to try a lower current filament and see if I end up with the same issue, as well as some larger, higher current tubes.

Testing a 6L6 with it’s 6.3v filament at 900mA gets me a little bit of a dip in voltage down to 105V when testing with the dim bulb, but nothing like I was getting with the 600mA setting. With the bulb bypassed, it seems to work fine, and nothing is getting warm or stinky.
Testing a 6HZ6 with it’s 450mA gets me a dip down to 80V.  With the bulb bypassed, I get a good test, but started to get just a hint of fresh roasted components, so I killed it before I started to see smoke.
Went back to test a 6AW8 which calls for a 600mA, this time with a current meter on the common line on the ballast capacitor.  With the bulb, the line voltage dipped down to 70V & the filament is only able to draw ~400mA.  With the bulb bypassed, the filament drew ~700mA. I didn’t leave it on this setting long enough to summon the stank.
With the same tube & settings, switching to the 6.3V tap doesn’t cause the same brown-out.

My working theory is that there’s a short somewhere in the part of the winding that feeds the current filament settings, but that does not affect the other taps.  Not being able to use the current settings actually isn’t too big of a problem, as the tubes that call for it also have a published filament voltage.

 

Heathkit IM-21 AC VTVM Repair

This is the first piece in a larger lot that I decided to tackle.  It’s the simple cousin of the Heathkit IT-18, which also measured DC voltages & resistance.

I beat my head against this for far too long until I realized that the calibration pot would sway wildly out of spec depending on the position of the contact – from 40Ω to hundreds of ohms.  I replaced it with a 50Ω pot and it was able to be calibrated to mostly within spec.  I’ve never found these early, cheap VOMs to be particularly accurate, but they’re fine for basic work.

Heathkit T-3 Restoration

Intro

This is not my first signal tracer, but it’s the first one I’ve gone to this much trouble on.  These have been getting pricey ever since musicians discovered they make cute little practice amps, so whenever I see one at a good price, I’ll snag it.

When I got it, it ‘worked’.  The indicator indicates, and sound played as expected.  It had a bit of hum and noise – not unreasonable, but definitely room for improvement.  At first, I thought I’d just replace any parts that were out of spec, but given how this is constructed, I decided a ground up re-build would be a better approach.

Here’s the before pic – the component count isn’t so high that it’s difficult to trace out in it’s condition, but it’s not exactly easy on the eyes.

Unbuilt Heathkits are getting fantastically expensive, but I figured if I did a complete tear-down, and ordered all new caps & resistors, I’d essentially get some resemblance of the new kit experience.  So off I went ordering parts, and stripping the chassis.

Parts

Everything showed up as ordered, but I didn’t pay attention to the physical size of the resistors.  Everything that showed up was 1/2 W as ordered, but they’re suspiciously small – even though they’re probably ‘correct’, they look out of place. They look like 1/8 W, they’re absolutely tiny. 

All of these really are 1/2 W resistorsThe one’s in question are:
47Ω – on the output
10kΩ – in the power supply
470Ω – on a cathode
There’s also the 1W 1kΩ in the power supply (the one that was looking toasty in the original build). It’s bigger, but it still just doesn’t feel right.  I’m going to go 2W on all these, somewhat for piece of mind, and somewhat for looks.

Wire

The solid-core for the wire used throughout measures .025  which puts it at 22 AWG. The OD w/ insulation is .056.  I have a bunch of 24 AWG solid core PVC jacketed hookup wire in various colors, and some cloth covered 20AWG from New Old Sounds.  The PVC wire has a slightly smaller OD than the original stuff, whereas the cloth covered stuff is on the thick side; slightly thicker than the transformer leads.  24 AWG is really fine for all of this, and I may use the cloth covered stuff for leads that pass through the deck.  The highest current is probably the heaters.  Each tube is 150 mA, and 24 AWG is good for 3.5 A, so we should be fine.

Color Coding

Most, if not all Heath gear of this era uses the same grey, solid-core wire.  I’d like to impart some color coding, and a quick search unearthed this:

Ground = Black
Filaments = Brown
B+ = Red
Control Grids = Green
Plates = Blue
AC Line = Grey

Teardown

I was able to strip out almost everything, with the exception of the transformers and a few of their connections.  There’s really no slack on the transformer leads, so I didn’t want to risk it.

Here it is as disassembled as it was going to get – note I already started on re-stuffing the electrolytic capacitor.

Speaking of capacitors – this one was a bear to empty out.  I think it tested OK, but I was trying to make it future proof.  In my experience, really dried out, dead electrolytics seem to come apart rather easily.  When I opened this one up, it was still a damp, packed, impervious mess.  Picking, poking, & drilling did nothing, nor did soaking in water.  Boiling did the trick – a poor mans double-boiler with a sauce can kept me from exposing my cookware to the nasty innards.

 

Schematic

Tube compliment:
12C8 –
12SH7 –
12A6 –
1629 –

Rebuild

I generally followed the instructions, making a few modifications to component placement, particularly where smaller replacement components allowed for a neater layout.

Started with the twisted heater wiring

 

Power & fuse wiring

The face-plate was installed after getting a polish.  It’s not perfect, but it’s better.  The screw-heads & jack hardware also got hit with a wire wheel – It’s easy, and I think makes a big difference.

Face-plate installation

B+ Wiring

Coupling capacitors & plate wiring

Shielded cable

Shielded cable – termination detail

Upper Deck

Wiring up the eye tube socket

Final Assembly

I’m pretty proud of how this came out.  There was one guy razzing me on the Vintage Test Equipment group about how some of the runs were too long & unsupported, which would cause variations in capacitance when they vibrate, or there was a possibility of exposed leads shorting if the unit was dropped.  I’m dismissing those as non-issues; I’m not going to sweat variations in picofarads on a device that tops out at a few kilohertz, and any drop hard enough to cause these components to come into contact would have much larger consequences anyway.  It may be difficult to tell from these photos, but leads that look close to touching are either at the same point in the circuit, or are a few cm away from each other.  These construction methods aren’t appropriate for high frequency circuits, but for audio, it’s just fine.

Testing

With the final assembly complete, all that’s left to do is power this up, slowly on a Variac.  It came to life with little fanfare – no creaks, pops, crackles, or smoke.  Once I saw the glow of the eye tube, but didn’t hear anything, I was convinced I’d screwed something up – I expected at least a little noise.  I tried playing back some music through the 1/4″ just for kicks, and was pleasantly surprised to hear clean audio reproduced.  It’s sounds as good as can be expected for this tiny speaker, but it’s quiet and clean. I haven’t tested the wattmeter function – I may at some point, even though I’m confident it’ll never get used.  The paint on the outer case isn’t in great shape, but I don’t have the means to strip it down, so that will be a future project – perhaps doing a whole batch in one go.

Success!

Tektronix 7854: Continued Woes

Status

Well shit – it seems as though something else went wrong with this scope while I was slogging though the ROM problem.  It’s not the display board.  We’re not even getting to the point where that’s being checked.  It seems to quickly go into this vegetative loop pretty much immediately.  I put back the old (bad) ROMs to at least see if I could get the ROM error pattern again, but nothing – just all lights lit.

Did some probing on the chip-select lines on the ROM card, and it’s clear that it’s going through some quick little loop, but it’s not always the same loop on each restart.  I’m hoping to find a way to reset the CPU w/o cycling power.  I think I can short TP1200-1 to ground, but waiting for someone to confirm so I don’t invariably nuke something else.

I confirmed that all power supply rails are spot-on.

onward…

Tektronix 7854 – back on the bench

I bought this unit a few years ago when I happened on a cache of Tektronix gear being sold by the sons of a deceased collector in NYC. I remembered that I powered it up enough to discover that it suffered from the infamous ‘ROM Rot’ issue that plagues these machines.  The machine is seen in it’s vegetative state below – This specific pattern of illuminated mode switches means the machine can’t initialize itself, because it lacks the instructions to do so.

Dude where’s my socket?

Unfortunately, I’d since removed the IEC socket – I vaguely remember it was cracked. Replacing that would be the first order.

Getting to the inside to attach the connector is a bit of an exercise in contortion, but it’s manageable.  The biggest annoyance in working on these supplies is that even after disconnecting all of the connections, it’s still tethered to the mainframe by the cord to the power switch (I’ll bet there’s a way around this, but it’d be at the switch end).

It’s usually hard to plug things in incorrectly due to how they’re keyed, but I take photos just in case.

The other annoyance is reconnecting the line to the HV power supply.  It passes through a hole in the shield of the supply, and once you disconnect it, you can’t plug it back in with the shield in place.  And, there isn’t enough slack to remove the shield, so you have to:

  • disconnect the cable from the other end, up in the high voltage section (careful!)
  • run it through the hole in the shield
  • plug it in to the power supply
  • reattach the shield
  • re-route and reconnect the cable back up into the HV supply.

With the power supply reassembled, we’re back to square 1 – no ROMs.

First attempt with the EP-1 programmer

Finally got this thing working.
Funny story – I actually had one of these in storage, but it was cheaper just to buy a new one than to rent a zip-car and make a special trip just for this one piece.  So now I have two BP Microsystems EP-1 programmers…

Here’s the Manual
UPDATE: Thanks to John Spina on the Vintage Test Equipment group for pointing me to the original software and firmware updates.

UPDATE:  Thanks to Rodger Whitaker for pointing out this modern USB programmer from Batronix that supports older chips. ~200 euro.  Not cheap, but not as expensive as I expected.

I was able to use Serial.app in OSX to connect to my old Radio Shack USB-serial adapter, which I couldn’t get to work on my Win10 box.  The programmer doesn’t need any additional software, it serves up it’s interface over the serial port.  It automatically detects the baud rate, up to 38,400.
To read a chip:

  1. ‘C’ brings up a menu to select the chip by make then model
  2. ‘PROTO’ lets you select the protocol – I had success using XMODEM.
  3. ‘RH’ reads a hex file – it will send the file once you initiate an XMODEM receive via the terminal emulator

Some other useful commands:

  • STAT – lists the current baud rate, firmware, selected chip type, and protocol.  It also has the phone number for BP Microsystems, which is still in business, and still has the same number (contrats guys!)
  • LIST – shows the contents of the chip, with a familiar hex viewer layout
  • BLANK – confirms that the ship is blank (all FFs)

There are also a number of commands for reading files in different formats (Intel, hex, Motorola, and Tektronix (?)), manually programming segments of a chip



Programming a chip:
This is where I’m getting stuck.  I’ve already incorrectly programmed one chip with a single character.  I have to find the correct combination of protocol and upload settings.
OK, I’ve got a bunch of EPROMs I can try programming once before my eraser shows up.

Attempt 1:
Set programmer for XMODEM,
Set transfer for XMODEM, no 1K block size
Appears to work correctly, but then I get the following:
Bytes received; 6 hex file errors;
1 bytes programmed correctly, no errors.
Only the first byte is programmed, and not even the correct one.

Attempt 2:
Same as above, but convert to HEX file first.
Appears to be working correctly, and takes a lot longer.
Appears to work correctly, reports:
00 Bytes received;
2000 bytes programmed correctly; No errors.
2000h = 8kB, so that’s promising.

I used this site to make the bin to hex conversion (would love to find an offline solution) What worked for me was no delimiters, but newlines every 16 bytes.

When we read it back and diff with the original, we get a file that’s 1kB larger, and seems to have carriage returns (0x0D) every 32 characters.  I can’t tell if this is a readback error, or a programming error.
It’s a read-back error.  Using ‘RB’ gets the file back in straight binary, and comparing it to the original file from the wiki I used to burn it from shows no errors.
SUCCESS!

SO: Even though the programmer can send us back binary files, we have to send it hex files.  Lesson learned!

Supported Chips

If anyone is interested, here’s a complete list that the programmer spits out with the ‘PARTS’ command:

AMD
8753H *1B,C 8751H *1B,C 87C51 *1C Am27128
Am27128A Am2716 Am27256 Am2732
Am2732A Am2732B Am27512 Am2764
Am2764A Am27C128D Am27C128P Am27C256
Am27C512 Am27C64D Am27C64P Am2817A
Am2864A Am2864AE Am2864B Am2864BE
Am28C256 Am9716 Am9761H *1B,C Am9864

Atmel
AT27C128 AT27C256 AT27C256R AT27C512
AT27C512R AT27C513 AT27C515 AT27HC256
AT27HC256L AT27HC64 AT27HC641 AT27HC64L
AT28C04 AT28C04E AT28C04F AT28C16
AT28C16E AT28C16F AT28C17 AT28C17E
AT28C17F AT28C256 AT28C256E AT28C256F
AT28C256E AT28C64 AT28C64E AT28C64F
AT28C64X AT28HC16 AT28HC16L AT28HC16L
AT28HC191 AT28HC191L AT28HC256 AT28HC256E
AT28HC256F AT28HC256L AT28HC256LE AT28HC291
AT28HC291L AT28HC64 AT28HC64E AT28HC64L
AT28HC64LE AT28PC64 AT28PC64E

Bowmar/White
8014 8020 8023

Catalyst
CAT27128A CAT27256 CAT27512 CAT2764A
CAT27HC256 CAT28C16A CAT28C17A CAT28C256
CAT28C64A CAT28C65A

Dallas Semiconductor
Dense-Pac
DPV27C256 DPV27C512

Electronic Arrays
EA2716

EXEL
XLS2804A XLS2816A XLS2817A XLS2864A
XLS2865A XLM46C15 XLM46C16 XLM46P15
XLM46P16 XLS46C15 XLS46C16 XLS46P15
XLS46P16

Fujitsu
MBL8742H *1A MBL8749H *1A MBL8749N *1A MBM27128
MBM27128-X MBM2716 MBM2716H MBM27256
MBM27256-W MBM27256-X MBM2732 MBM2732A
MBM2764 MBM27C128 MBM27C128P MBM27C256
MBM27C256A MBM27C256A-W MBM27C256AP MBM27C256H
MBM27C512 MBM27C512P MBM27C64 MBM27C64-W
MBM27C64-X MBM28C64 MBM28C65 MBM83256
MBM83512

Generic
27011 (12.5V) 27128 (21V) 27128A (12.5V) 2716 (25V)
27256 (12.5V) 2732 (25V) 2732A (21V) 2732B (12.5V)
27512 (12.5V) 2764 (21V) 2764A (12.5V) 27C128 (21V)
27C16 (25V) 27C256 (12.5V) 27C32 (25V) 27C512 (12.5V)
27C64 (21V)

General Instrument
27256 27C128 27C256 27C512
27C513 27C64 27HC64 27HC64L
28C04 28C16 28C17 28C64
28CP256 28CP256A 28CP256B

Greenwich
GR27128 GR27256 GR27512 GR27513
GR2764 GR281 GR881 GR3281

Hitachi
HN27128A HN27128AG HN27128AP HN27256
HN27256G HN27256P HN27512 HN27512G
HN27512P HN27C256 HN27C256FP HN27C256G
HN27C256HG HN27C64 HN462532 HN462716
HN462732 HN4827128 HN482732A HN482764
HN58064 HN58C65 HN58C66P

Hyundai
HY2764 HY27C64

IDT
IDT78C16A IDT78C256A IDT78C64A IDT78M64
IDT78M64S

Intel
27011 27128 27128A 27128B
2716 27256 2732 2732A
27512 27513 2758 2764
2764A 27C128 27C256 27C512
27C64 2816A 2817A 2864
2864A 68C257 8041A *1A 8042 *1A
8048AH *1A 8049AH *1A 8050AH *1A 8741A *1A
8741AH *1A 8742 *1A 8742AH *1A 8744H *1B,C
8748 *1A 8748H *1A 8749H *1A 8751H *1B,C
8755A *1A 87C256 87C257 87C51 *1C
87C64 P27128A P27128B P27256
P27512 P27513 P2764A P27C128
P27C256 P27C64

Microchip Technology
27256 27C128 27C256 27C512
27C513 27C64 27HC256 27HC256L
27HC64 28C04 28C04F 28C16
28C17 28C64 28C64A 28C64AF
28C64AX 28C256 28CP256

Macronix
MX27C256 MX27C64

Mitsubishi
M5L27128K M5L27128K-I M5L27256K M5L27256K-I
M5L2732 M5L27512K M5L2764K M5M27128P
M5M27256P M5M27512P M5M2764P M5M27C128K
M5M27C256K M5M27C256AK M5M27C256P M5M27C512AK
M5M27C512AP M5M28C64AP M5M28C64P

Mostek
ET2716 ETC2716 ETC2732 MK2716
MK2764 MK38XXX

Motorola
MCM2532 MCM2716 MCM68764 MCM68766

National
MM2716 MM2716E MM2758-A MM2758-B
NMC2732 NMC27C128B NMC27C128BQ NMC27C128BN
NMC27C128C NMC27C128CQ NMC27C16 NMC27C16H
NMC27C16HQ NMC27C16Q NMC27C256 NMC27C256Q
NMC27C256B NMC27C256BN NMC27C256BQ NMC27C32
NMC27C32B NMC27C32BQ NMC27C32E NMC27C32EH
NMC27C32H NMC27C512 NMC27C512A NMC27C512AN
NMC27C512AQ NMC27C64 NMC27C64N NMC27C64Q
NMC27C64B NMC27C64BN NMC27C64BQ NMC27CP128
NMC27CP128Q NMC9817 NMC98C64A

NEC
8748HD *1A uPD27128 uPD2716 uPD27256
uPD2732 uPD2732A uPD2764 uPD27C256
uPD27C256A uPD27C512 uPD27C64 uPD28C04
uPD28C05 uPD28C64

OKI
MSM27128A MSM27128AS MSM27128AZB-RS MSM2716
MSM27256 MSM27256AS MSM27256ZB-RS MSM2732
MSM2732A MSM27512 MSM27512AS MSM27512ZB-RS
MSM2764 MSM2764A MSM2764AS MSM2764AZB-RS
MSM2764RS MSM27C128AS MSM27C64AS MSM2816ARS

Quick Pulse
27011 27128A 27256 27512
2764A

Ricoh
RD27C256 RD27C64

Rockwell
R2764 R2764C R27C64 R2816
R5213 R52B13 R52B33 R87C32
R87C64

Samsung
KM2816A KM2816AI KM2817A KM2817AI
KM2864A KM2864AH KM2865A KM2865AH
KM28C64A KM28C65

Seeq
27128 27256 2764 27C256
2804A 2816A 2816AH 2817A
2817AH 2864 2864H 28C256
28C64 28C64A 28C65 36C16
36C32 38C16 38C32 5133
5213 52B13 52B13H 52B33
52B33H 5516A 5516AH 5517A
5517AH 55B33 55B33H 82005
82025 86063 E52B33 E52B33H
M52B33 M52B33H

SGS
ET2716 ETC2716 ETC2732 M27128A
M2716 M2716P M27256 M27C256B
M27C512 M2732A M27512 M2764
M2764A ST27128A ST27256 ST2764A
ST27C256 TS27C256 TS27C256P TS27C256Q
TS27C64 TS27C64A TS27C64P TS27C64Q
TS28C16AC TS28C16AP TS28C64C TS28C64P

SGS/THOMSON
ET2716 ETC2716 ETC2732 M27128A
M2716 M2716P M27256 M27C256B
M27C512 M2732A M27512 M2764
M2764A ST27128A ST27256 ST2764A
ST27C256 TS27C256 TS27C256P TS27C256Q
TS27C64 TS27C64A TS27C64P TS27C64Q
TS28C16AC TS28C16AP TS28C64C TS28C64P

Signetics
27C256 27C512 27C64A 87C256
87C64 SC87C51 *1C

SMOS
SPM27128 SPM27128C SPM27128H SPM27C256
SPM27C256H SPM27C64 SPM27C64C SPM27C64H
SPM2864 SPM2864C

Synertek
SY2716

Thomson
ET2716 ETC2716 ETC2732 MK2716
MK2764 MK38XXX TS27C17AC TS27C17AP
TS27C256 TS27C256P TS27C256Q TS27C64
TS27C64P TS27C64Q TS28C16AC TS28C16AP
TS28C64C TS28C64P

TI
SMJ2516 SMJ2532 SMJ2564 SMJ27C128
SMJ27C512 TMS2516 TMS2532 TMS2564
TMS25L32 TMS27128 TMS2732A TMS2764
TMS27C128 TMS27C256 TMS27C512 TMS27C64
TMS27P32A TMS27P64 TMS27PC128 TMS27PC256
TMS27PC512 TMS27PC64 TMS28C64

Toshiba
TC54256AF TC54256AP TC54512AP TC57256
TC57256AD TC57256ADI TC57256D TC57512AD
TC57H256D TMM23128-H,H TMM23128-H,L TMM23128-L,H
TMM23128-L,L TMM2364-H,H TMM2364-H,L TMM2364-L,H
TMM2364-L,L TMM24128AF TMM24128AP TMM24256AF
TMM24256AP TMM24256BF TMM24256BP TMM24512AF
TMM24512AP TMM24512F TMM24512P TMM2464AF
TMM2464AP TMM27128 TMM27128A TMM27128AD
TMM27128ADI TMM27128D TMM27128DI TMM27256
TMM27256A TMM27256AD TMM27256ADI TMM27256BD
TMM27256BDI TMM27256D TMM27256DI TMM2732
TMM27512 TMM27512AD TMM27512ADI TMM27512D
TMM27512DI TMM2764 TMM2764A TMM2764AD
TMM2764ADI TMM2764D TMM2764DI

VLSI Technology
VT27C256 VT27C512 VT27C64 VTC27C256

Waferscale
WS27C128F WS27C256F WS27C256L WS27C512F
WS27C512L WS57C128F WS57C256F WS57C512F

White Technology
8014 8020 8023

Xicor
X2804A X2816A X2816AI X2816AM
X2816B X2816H X28256 X2864A
X2864AI X2864AM X2864B X2864H
X28C256 X28C64