2. Tek 7912AD 500MHz 100GSs Transient Waveform Digitizer
(a) This was sold on eBay UK for $100 with the words 'Pot knobs at front slightly bent easy fix'. The cheapest I've seen on eBay is $999 eBay USA so I was already prepared for the worst. In reality, all three plastic knobs were shattered. The graticule intensity control is a Tek 311-1763-00 aka Clarostat 19-8030 10k pot & DPDT, but the hollow spindle to the pot had snapped off completely. The front panel was easy to remove but due to the DPDT switch the Clarostat is a rare beast indeed. The switch still worked so I left it there and instead drilled out a new hole in the front panel and added a quality
AB 10k pot and moved the pot wiring on the Clarostat across. Straightened the remaining spindles.
The broken knobs should be alright once I get some more Sugru to rebuild them.
(b) Examining the inside, I found the PSU fed into power distribution and fuse board A62. Checking the fuses, F438 rated 32V 7.5A was open circuit solely from mechanical misalignment. Re-aligned.
(c) Checked all capacitor ESRs on power distribution and fuse board A62. All ok.
(d) Quick power check.
I hooked up a video monitor and slowly applied power via the lab mains variac to see if the 7912AD had any chance of firing up and producing a video output. The fan very slowly began to turn and a little later a small puff of smoke appeared from the right hand side. Shutting power off immediately I took a look. I found a blackened 10Ω 1/4W R512 on the Scan Amp & Read Gun Supply pcb A38. Examining Service Manual Vol. 2 (hereafter SM2) P.315 (parts list P.222), R512 is a protection resistor for a familiar blue tantalum C414 10µF on the +15V supply. Unsurprisingly its ESR was 8.2Ω.
I found 3 more 10µF tants in similar circumstances and decided to replace them all regardless. There is a potential issue of overloading the PSU by replacing old high ESR caps with modern low ESR ones but I decided 4 caps would not make a lot of difference. A bigger cap should also not make a lot of difference unless the 10µF was specifically chosen to remove a particular noise frequency, but I think on the +15V supply it's more likely a cost cutting exercise.
Replaced all 4 10µF tant caps with 4 x Panasonic EEUFM1E101 100µF 25V 0.13Ω 105°C gold electrolytics (an old rule of thumb used to be to replace a tantalum with an electrolytic 3x its value to get the equivalent performance, but that was before modern low ESR electrolytics). Also replaced R512.
(d) The fan didn't sound too healthy, so I ordered a new replacement off eBay - the cheapest I could find cost more than than half the price of the entire 7912AD!
(e) Next check every cap on every pcb and replace where appropriate:
A16 Memory Control - C043, C145 100µF 10V 85°C replaced with 100µF 25V 105°C, ESR 0.08Ω.
A18 RAM Memory - C010, C020, C030 100µF 10V 85°C replaced with 100µF 25V 105°C, ESR 0.08Ω.
A20 Buffer Control - C034, C520, C540 +5.1V/-5.2V/-2V supplies electrolytic 1000µF 10V 85°C. Still within spec (ESR measured ~0.08Ω) but leads corroded and end seals pushed outward. Bought and Replaced with eBay 760mA ripple Vishay MAL211816102E3 1000µF 25V 125°C, measured ESR 0.08Ω.
A26 X + Y Ramp Gen - C340, C642 +50V & -50V 3µF 150V, Still within spec (measured ESR 1.04Ω & 0.81Ω) but leads badly corroded, C642 corrosion has also damaged pcb track. These caps cost $40 each new on eBay! Replaced both with 3 x 1µF 150V new vintage tantalums costing $1 each. Replacements are larger, so had to mount 2 on top surface and one on rear surface of pcb. Repaired pcb track with copper wire direct to C642. Measured ESR for 3 in parallel 0.29Ω & 0.24Ω.
PSU - removed, cleaned, and inspected caps - only suspect cap is on A86 Rectifier pcb - C304 1µF with ESR 8.2Ω, but it's a Sprague 34D which is not very high spec. Could not find a datasheet but a modern 34D 64µF spec = 3.3Ω, so it may be oK. Left it for now as the PSU is a pain to disassemble.
(f) In accordance with Service Manual, check all jumpers on all boards. All Ok except for A26 X + Y Ramp Gen: Fast Scan not set on all jumpers, and A46 Plugin Interface: Fast Scan not selected, but should not cause major issues (it was obviously working like this before, so it is possible the SM is wrong).
(g) In accordance with Service Manual, check load resistances for PSU. Found short to ground on -15V supply. I had stripped the entire machine and was tracing the motherboard when I discovered A32 Video Preamp board hidden on top under a black metal cover with lots of nasty blue tantalums!! C1998 10µF 25V on -15V dead short. Replaced 3 x 10µF 25V supply caps with 100µF 25V 105°C, ESR 0.08Ω.
Rest of tantalums look ok and I have no replacements to hand anyway.
The resistances I got are close but not in spec. No explanation in the service manual.
(h) Retest.
No programmable plugins available (to avoid high intensity damaging the scan tube), so used known working 7A26 and 7B10 having first set it up on 7834. Applied power whilst monitoring 5V logic supply for reference. Glad I replaced the fan - there is no hesitancy now. The correct sequence of lights appeared on the front panel but after about 15 seconds all lights went out and the 5V supply went down, although the fan continued to run, and I knew from the PSU schematic that it was driven from the PSU rather than the mains, suggesting the PSU was not completely dead.
(i) Wary of applying power and doing more damage, I decided to remove the PSU and test it out of the main unit. I knew switchmode PSUs usually don't like being without a load so I built a multiple resistive load (values based on the SM power resistances table 4.1) to test all PSU outputs to the power distribution and fuse board A62, and determine the source of failure. Removed PSU from 7912AD, and A62. Plugged PSU into A62 board and plugged A62 into pcb connector test lead leading to load bank which also feeds sense connections back to PSU. SM table 4.1 values are to be measured on Scan Amp & Read Gun Supply pcb A38. The load was built with whatever power resistors I had to hand. Tests were to TP538 (0V COM):
Test load other pin connected to all GND SENS & GNDs except where stated otherwise:
Supply A62 to Source +/- -/+ Meas Watts Implementation Load Total
Volts pin pin Vtest Ohms Ohms Ohms Load Need in series Ohms Rating
+365V B39 to GND P032-1 26K 45K 69K 4.93W 10 x 3k3 @ 0.5W 33k 5W
+130V B37 to GND P032-8 8K 10K 14K 1.69W 1 x 10k @ 7.0W 10k 7W
+50V B14 to GND P032-6 440R 440Ω 544Ω 5.68W 2 x 220R @ 7.0W 440Ω 28W
+15V B13 to GND P032-5 95R 95Ω 86Ω 2.75W 5 x 20R @ 25.0W 100Ω 125W
+5V B11 to GND P032-4 55R 55Ω 45Ω 0.42W 3 x 20R @ 25.0W 60Ω 75W
-15V B9 to GND P032-2 100R 100Ω 88Ω 2.25W 5 x 20R @ 25.0W 100Ω 125W
-50V B8 to GND P032-1 710R 1K 2300Ω 2.50W 5 x 200R @ 25.0W 1k 250W
Scan Amp & Read Gun Supply pcb A38 also has the following voltages, so added loads for these too, based on the values of resistance I measured for their loads in the 7912AD:
+5.1V A24 to GND - - 25Ω 1.04W 1 x 20Ω 1.3W
-5.2V A44 to GND - - 369Ω 0.07W 1 x 390Ω 69mW
-2.0V A46 to GND - - 480Ω 8.3mW 1 x 470Ω 9mW
-19V A42 to A43 - - 910Ω 397mW 1 x 910Ω 400mW
+19V B42 to B43 _ - 866Ω 476mW 1 x 910Ω 400mW
The POWER ENABLE signal must also be supplied using a switch, see photos below of test load and schematic from my logbook:
Clockwise:
Schematic for DIY PSU test load based on fuse board A62 connector pinout,
Test load component layout,
Fuse board A62 plugged into it,
PSU plugged into A62 on the test load.
Click on images for larger view.
It was reassuring when I applied power, pressed the POWER ON push button on the test load, and the PSU fan fired up. However when I looked at the supplies, they were all wrong except the +5V on the fan board, which was at +4.83V:
+5.1V = +1.12V, -50V = -1.41V, +15V = 0V, +5V = 0V, +15V = +0.05V, +50V = +0.26V,
+130V = +10V, +365V = +4.6V, -5.2V = 0.35V, -2V = -0.65V, +19V = +3.34V, -19V = -4-8V
Clearly something bad had happened, but as they were not all 0V I thought perhaps it could be repaired. The next step was to take the PSU out of its box and check it out with the test load as before.
The PSU is quite complex with several PCBs. The loosely populated Fan Controller board A80 at the top contains the 5V logic power up logic and a supply circuit for the 90Vac fan. The Inverter board A82 is on the right side (viewed from the front of the 7912AD) of the PSU. A82 contains the two huge mains input smoothing capacitors screwed onto its reverse side with their bodies in the centre of the unit. A82 also contains the oscillator and driver logic for the SMPSU, consisting of a two transistor totem driver on the primary of pot core T360. Four separate secondaries from the pot core lead into A84, a small board inaccessible in the centre of the unit that holds four TO-3 transistors arranged as an
H-bridge switch-mode driver, in turn switching the large primary transformer T014 on the Rectifier board A86 at the base of the unit. Multiple secondaries from T014 feed into the Regulator board A88 on the left side of the PSU. The fan blows through the middle, cooling the entire assembly:
The Fan Controller board A80 is at the top:
The Regulator A86 The Inverter A82
is on the left: is on is on the right.
The Rectifier board A86 is on the bottom
Before powering up the now exposed PSU, I first checked the ESR of every accessible capacitor, leaving a label indicating the value and where measurable, its capacitance. I found nothing amiss apart from C304 on A86, a big Sprague 109 1uF measuring 1.12uF with a rather high ESR of 8.5Ω (see red label below), but although I could find no spec for it on the internet, other values in the range suggested I could expect its ESR to be several ohms. Given the capacitance seemed ok, I left it.
Above: The Fan Controller board A80 Bottom of Rectifier board A86
Click on images for a larger view
Below: The Regulator board A88 The Inverter board A82
I applied power. Thoughtfully, Tek included a neon across the main capacitors on Inverter board A82 to indicate a lethal voltage is present (see bright spot just below and to the right of the red square capacitor between the DMM leads in the photo below). I had a protective RCD (Residual Current Detector) on the supply to the PSU should I inadvertently short it to ground through my body, but the neon was a very useful reminder of the dangers present and I confirmed 330Vdc was present across the capacitors:
Next I checked for signs of life in the switcher and measured 40.32kHz on the oscillator on A82. Tracing the signals, I confirmed the pot core was being driven and drive signals were being produced to the big transistors on A84, although with no helpful timing diagrams in the SM I could not work out if they were the expected waveforms. I also found ac signals on secondaries of the big T014 transformer, and voltages on the dc outputs of the Rectifier board A86, but they looked very odd:
+17V = +4.88V, -17V = -5.5V, -7V = -20.6V, +54V = +12.5V pulsing, +7V = +1.6V,
-4.5V = +3.6V to -3.6V spikes, +140V = +106V pulsing, +400V = +130V pulsing.
Below: Tek's schematic for the PSU excluding the Regulator board (click for a larger image):
Below, I confirmed the pot core totem pole driver was generating pulses to its primary, then I verified the pulse width logic using the Logic Analyser facility on the Rigol:
Below - once I saw the pot core being driven, I moved onto the H bridge driver for T014:
Above: the small A84 board holding the four big bridge driver transistors.
Left: the topology of the H bridge transformer driver only became apparent when I redrew Tek's schematic.
It seemed to be partly working, but I could not work what it was doing. After fruitlessly running around all of the boards finding weird voltages, eventually I dismantled the entire PSU, checked the capacitors on the Rectifier board A86 that was now accessible, and finally, having disconnected the rectifiers, measured the inductance of the secondaries on the big transformer T014.
What I found was not good: the ±17V and +7V secondaries had an inductance of ZERO, see photo to right.
The remaining secondaries measured in the mH range, but they were no use without the +7V supply: I concluded the voltages I had seen were somehow bleeding into their outputs from other secondaries, probably through the regulator transistors missing a biasing supply, and because +7V was missing, the +5.1V supply was also missing, which meant the SMPSU comparator circuit didn't have a feedback reference, so the outputs were all over the place.
The T014 transformer is a custom Tek part and of course unobtainable.
However after much Googling I did find someone in the USA with an old 7912AD they wanted to sell, and who was willing to sell me just the transformer and a few extra parts such as replacements for the broken switches and knobs on the front panel. The seller was John Griesson, who I also discovered was selling a low cost aftermarket alternative to Tektronix 067-0616-00 7k series plugin flexible extender. Having already discounted the flexible extenders on ebay due to prohibitive cost, I bought two sets of his pcbs as well:
http://ecosensory.com/tek/tek_7K_flex_sales_blurb.txt
It is now August 2017 and I have the PSU and parts, but being engrossed in other tasks, I haven't yet returned to the 7912AD and tried the new PSU. Watch this space!