I play with quite a lot of these things and you have a combination with special risk:
- early generation ICs with low over-volt tolerance
- unobtanium ICs
- series pass regulator transistors that can fail short
In this situation at a minimum I recommend that you
- remove the boards (careful with ESD if there is any early MOS in there, I cannot recall just now)
- test that power supply with a reasonable dummy load.
Myself, after seeing an early Victor with 10% of its ICs actually exploded and blown apart after a series pass short -> overvoltage, I install simple fuse-zener-power transistor volt clamps on vulnerable supply rails. That transistor might be OK today, but tomorrow.....
Good luck, if you want further info then you could contact me via the dopecc.net site mentioned above. My site, it's nice to see that someone else has read it (!)
How do I load test this power supply? I usually fix old computers and there I just use old hard disks as ballast, but these voltages are nothing I ever worked with. I don't have any load resistors nor would I know their values.
I took the boards out of the backplane thing and am now slowly ramping up the power supply (to give the capacitors a chance, but they'll be replaced if out of spec). The circuit boards look pretty wild compared to what I'm used to seeing.
Additionally, which Zener diodes do I use and where do you place them? Across the supply rail to ground? I guess they need to be stronger than the fuse they're meant to blow?
I'd put those boards aside for now.
It would be helpful if you could get top and bottom shots of the power supply board too.
Given the complexity of the PS board, I would expect to see a design similar to the attached.
Try to reverse-engineer the PS board to identify, number of transformer secondaries (incl. center taps), rectifiers (half or full wave), bulk cap, and pass transistor circuits. There is a chance that the +180V supply may not be a transformer secondary but directly from the mains (diode+cap). You need to confirm this before attaching an oscilloscope to any output!
The object to testing the power supply is to confirm:
(a) the rectifier diodes are okay; not shorted or open.
crowbar is better because zener diode have series resistance/slope(1W 24V ~30ohm) similar for TVS diodes
need some transient bench tests with oscilloscope
what does that core memory do on display module?
See attached for my partial reverse-engineering notes and observations below.
You'll need to open it in a new window to see the text.
- From your earlier shots, the large TO-3 packaged pass transistors appear to be 2SC1030 80V NPNs. I can't read the other smaller pair. Likely at least one is a PNP?
- There are four pass transistors (suggests four regulated outputs) but I could only track three of them to the output terminals on the right side of the PCB.
- The +180V output appears to be unregulated and not coming from the mains-side of the PCB (a good thing). I can follow the brown secondary lead to the fuse to a [likely] diode to a HV cap to a resistor/inductor to the other HV cap then exits just above the 200mA fuse. I couldn't trace the circuit just to the right of the pot (I'm assuming pot case is the wiper).
- I didn't see a bleed resistor for the high-voltage caps (could it be the 100K?). BE CAREFUL!!!
- Each pass transistor has a small driver transistor (represented by circles); likely in a darlington arrangement with the pass transistor.
- At least for the regulated output from the negative bridge leg, it follows the basic design I posted in my previous message (though I would expect the pass transistor to be a PNP). That is, GND to zener to pass transistor base to resistor to pass transistor collector (unregulated side). The other positive bridge leg is a bit more complicated involving two driver transistors. The positive diode pair secondary involves an unknown (to me) 4-legged brass spider-looking device (some exotic zener?).
- I'm assuming the gray lug eye terminal on the right is connected to the GND PCB trace and brown wire to the edge connectors and was attached to the chassis?
- There's a divider network output in the 2nd lowest output on the right side of the PCB. The pot wiper (lower right) seems to adjust the 4-legged brass spider-looking device that controls the positive diode pair secondary regulation.
What does this all mean?
A:
- you probably don't have to do a thorough reverse engineering of all components. You just need to find out what the last pass transistor does (4th regulated output?) and identify where all the regulated outputs exit (wires to the edge connectors).
- then you can load test these outputs (see my previous message) to confirm that the voltages don't sag. Don't bother load testing the un-regulated outputs since they'll sag; just note their voltage for future reference.
- you can check voltage across the devices that I've identified as zeners. And if you can get a number off of them, we can look them up. If I'm correct, then the regulated output should be about two diode-drops lower (~1.2V due to the darlington arrangement). One of these regulated outputs should be -24V (for the NEC logic ICs). However, if the +180V output is truely unregulated, then there are potentially three other regulated outputs that we can only compare the zener voltage with their output voltage; that is, output=~1.2V below zener and the pass transistor collector >2V (likely more) than the emitter (regulated output).
- since there doesn't appear to be any minimal load resistors (to GND) on those output terminals - those from pass transistor emitters (ie. regulated outputs leading to the edge connectors) - before measuring any output voltage, you'll need to add high value resistors to each of the regulated outputs (100K or 1M (if the +180V is regulated)). The pass transistor emitters can't be left open (when logic and display PCBs are removed) otherwise they won't regulate.
- for future reference, I would document:
= the VAC on all transformer secondaries relative to the GND (black) lead.
= the VDC on positive and negative diode bridge legs, and positive diode pair leg; ie. the unregulated side.
= the voltage across the two zeners that I've identified; I don't know what to measure on the spider-looking thing.
= all regulated and unregulated outputs (right side of PCB) that go to the edge connectors.
EDIT: Regarding your schematic...
The items marked ? on the mains-side are inductors (302M=3mH +/-20%, 200K=20uH +/-10% maybe).
The one item marked ? after the diode on the secondary-side may also be an inductor (though, no matches found for "mb2k0"). Is this the device barely visible below the pot and between the two HV caps positive leads? Or do you mean the black rectangle to the right of the pot? If it's the latter, then it may be a zener because it's attached to a resistor to GND and may be part of the +180V regulation (I can't see much in this area).
- I marked the transistors in the attached picture
- One transistor's emitter(?) is grounded and the grey wire with the ground lug also goes to the PCBs, otoh the brown wire is connected to the ground plane where all the caps' negatives are. Are there two grounds (chassis/signal ground?)
- The voltage bleeds from the caps, anyway at 230VAC input there was around 270V on them, but to be verified under load
- The brass device doesn't read as diodes, still no idea what it is
Thanks, I didn't know about inductor numbering. mb2k0 refers to the pot, but I'm confused how it's wired up and I assumed it's for trimming the regulated voltages. The black rectangle is a regular disk shaped diode.
Don't know what to make of the zener diodes, they're labelled ZR212 Stanley 084 and ZR210 Stanley 083.
I'm going to put the inner chassis back together and try to find some load resistors. What do you mean with the high value resistors? Is this in addition to the load you mentioned before?
Voltages relative to GND / with no load:
- for future reference, I would document:
= the VAC on all transformer secondaries relative to the GND (black) lead.
= the VDC on positive and negative diode bridge legs, and positive diode pair leg; ie. the unregulated side.
= the voltage across the two zeners that I've identified; I don't know what to measure on the spider-looking thing.
= all regulated and unregulated outputs (right side of PCB) that go to the edge connectors.
I tried loading the supply rails down (one by one, not all at the same time), but the results are not what is expected. Possibly a fault?
Red: about 210 V unreg, 206 V with a 5K load
White: 0 V with 10K attached (something wrong with the "opto coupler" / mystery device?)
Grey: 13.5 V (unloaded, can I load this the same way as the other non-HV rails?)
Orange (2): 8.1 V (unloaded, I wasn't sure if I'm supposed to load test it when it's a reference?)
Orange (6): -9.7 with a 60 Ohm load
I went through the schematic, the RV2 5K pot is a 500R I believe. I also measured the resistors (in-circuit) and they're all either barely within spec.. worst offenders R8 at 580R and R3 at 2.3K
Any tips? Should I try the pots to see if it gets better? Measure the transistors too?
EDIT: closest I could find is this TIL108: https://www.radiomuseum.org/tubes/tube_til108.html (https://www.radiomuseum.org/tubes/tube_til108.html)
EDIT2: It seems the internal power supply design is same as the Sharp Compet 361, there is a Dutch thread about it and it seems the capacitos were the problem and an IC popped: https://www.circuitsonline.net/forum/view/116228 (https://www.circuitsonline.net/forum/view/116228)
I probably want to make sure this PSU is in order before I connect any boards
| Multimeter red lead on: | Multimeter black lead on: | Reading under no load: | Reading across 10K load: | Reading across 100K load: |
| RED (1) | BROWN (4) | not applicable | not applicable | fill in |
| ORANGE (2) | BROWN (4) | fill in | not applicable | not applicable |
| GREY (3) | BROWN (4) | not applicable | fill in | not applicable |
| WHITE (5) | BROWN (4) | not applicable | fill in | not applicable |
| ORANGE (6) | BROWN (4) | not applicable | fill in | not applicable |
I can't thank you enough for all this work! It's really interesting learning about this and well explained too :-+Voltages relative to GND / with no load:
- for future reference, I would document:
= the VAC on all transformer secondaries relative to the GND (black) lead.
= the VDC on positive and negative diode bridge legs, and positive diode pair leg; ie. the unregulated side.
= the voltage across the two zeners that I've identified; I don't know what to measure on the spider-looking thing.
= all regulated and unregulated outputs (right side of PCB) that go to the edge connectors.
Brown: 200 VAC
Red: 10 VAC
Blue: 16 VAC
D4: 43.6 VDC
D2/D3: 13.8 VDC
D5 ZR212: 14 VDC
D6 ZR 10.4 VDC
The brass device, I measured the voltage across the left legs is 20 VDC, but it was not conducting at the time of measurement. There is a white dot next to the upper left leg, supporting the theory it is an opto coupler, but 20 volts for an LED is a bit much no...
I found resistors in my stash of radio parts and some of them look beefy enough for load testing, tomorrow I'll go find some suitable ones to test the the regulation under load
EDIT: The two lead device 2SC641 has a third leg in the middle but it's just a stub
Blue: 16VAC (RMS) becomes ~22VDC after the bridge (16V*1.414-0.7V). This is what is expected between GND and one pole. However, you have D4 reading 43.6V which is what I would expect between + and - of the diode bridge (there are two halfs, so double one pole to GND).
D5 ZR212: 14VDC. Is this across the zener leads directly? Or, one lead and GND? Because there are components on both legs (R3 or Q1) and that may be affecting your reading. If this is the reading across the zener leads directly, then this zener cannot be replaced with the 12V zener that lists it as replacing (see link in the FYI section of my earlier message). If so, we should expect the regulated output GREY (3) be +12.8V instead.
Agreed. 20VDC is too much for a LED. What's the VDC across the right pair? Maybe the LED is there.
I suspected that the transistor is being [ab]used as a zener! ie. emitter to +HV (unreg) and base to 150K R5 to GND.
C641 and A618 form a constant current source that regulates a constant current through the 100K R7. And therefore,
a constant voltage at the base of Q8 driver transistor. The regulated RED (1) output is then 1.2V below that.
See https://sound-au.com/ism.htm#p5 (https://sound-au.com/ism.htm#p5) for an explaination on current sources/sinks.
| Multimeter red lead on: | Multimeter black lead on: | Reading under no load: | Reading across 10K load: | Reading across 100K load: |
| RED (1) | BROWN (4) | not applicable | not applicable | 209 VDC |
| ORANGE (2) | BROWN (4) | 8.14 VDC | not applicable | not applicable |
| GREY (3) | BROWN (4) | not applicable | 13.9 VDC | not applicable |
| WHITE (5) | BROWN (4) | not applicable | 0 VDC | not applicable |
| ORANGE (6) | BROWN (4) | not applicable | -9.8 VDC | not applicable |
What is the voltage WRT GND at the base of the driver transistor Q6? (ie. upper-right pin of mystery device WRT GND) If it is also 0V then that darlington pair isn't being turned on at all and you may have a mystery device problem. If there is a non-zero voltage on the base of Q6 then you'll need to check/test Q6 and Q7 out-of-circuit.
I don't think the zeners would change much as they age so ORANGE (6) and GREY (3) regulated outputs should not have changed. Notice ORANGE (6) is about -24V WRT chassis as expected/needed for the NEC ICs. This is a good sign.
If all the carbon resistors increased in value by, say, 15% over time, I don't think that would affect the zeners Vz nor the divider network.
Given the voltages you're seeing on the mystery device left & right pairs, I don't really understand it. Is it behaving as it should? It looks like a hermetically sealed semiconductor so it's probably okay given that it's only managing micro power levels. I agree with your Q1 orientation assessment.
What is the voltage WRT GND at the base of the driver transistor Q6? (ie. upper-right pin of mystery device WRT GND) If it is also 0V then that darlington pair isn't being turned on at all and you may have a mystery device problem. If there is a non-zero voltage on the base of Q6 then you'll need to check/test Q6 and Q7 out-of-circuit.
I don't think the zeners would change much as they age so ORANGE (6) and GREY (3) regulated outputs should not have changed. Notice ORANGE (6) is about -24V WRT chassis as expected/needed for the NEC ICs. This is a good sign.
If all the carbon resistors increased in value by, say, 15% over time, I don't think that would affect the zeners Vz nor the divider network.
Given the voltages you're seeing on the mystery device left & right pairs, I don't really understand it. Is it behaving as it should? It looks like a hermetically sealed semiconductor so it's probably okay given that it's only managing micro power levels. I agree with your Q1 orientation assessment.
There's barely any voltage at the base of Q6 (~.015 V). This is the biggest question, is it behaving at it should...
Q6 measures OK with a multimeter in-circuit.
I could desolder the mystery device and check if there is a marking underneath. Tomorrow I'll try to procure the power resistors to verify the other rails.
For RED (1), apply a 2K 20W load resistance and measure MES14 (100mA).
For WHITE (5), apply a 33ohm 5W load resistance and measure MES15 (~400mA). Although, it'll probably still be 0V WRT GND.
For GREY (3), apply a 33ohm 5W load resistance and measure MES11 (~400mA).
For ORANGE (6), apply a 33ohm 5W load resistance and measure MES18 (~300mA).
If you don't have 20W or 5W resistors, you can use a lower wattage one but you have to be quick in your measurement otherwise you risk burnout. Don't go lower than 2K on the HV output as that risks blowing the 200mA fuse. If there is no significant sag with the 33ohm, you can try 22ohm down to 12ohm for the LV outputs but now you're getting into 600mA and >1A, respectively.
Sorry for the late reply, the weather was nice and I spend the weekend outside (gasp :o)
Before desoldering anything I clipped in the load resistors (I found a 2.2K, 33R, 22R with appropriate wattages).
I tried to do an experiment: Loading down WHITE and GREY at the same time and there appears to be a voltage now on WHITE: around 1-1.5 V WRT GND, around .4 V on the base of Q6, ORANGE(2) dropped down to 7.9V ~7.97V (closer to your calculated voltage)
Here the requested values:
For RED (1), apply a 2K 20W load resistance and measure MES14 (100mA).
For WHITE (5), apply a 33ohm 5W load resistance and measure MES15 (~400mA). Although, it'll probably still be 0V WRT GND.
For GREY (3), apply a 33ohm 5W load resistance and measure MES11 (~400mA).
For ORANGE (6), apply a 33ohm 5W load resistance and measure MES18 (~300mA).
If you don't have 20W or 5W resistors, you can use a lower wattage one but you have to be quick in your measurement otherwise you risk burnout. Don't go lower than 2K on the HV output as that risks blowing the 200mA fuse. If there is no significant sag with the 33ohm, you can try 22ohm down to 12ohm for the LV outputs but now you're getting into 600mA and >1A, respectively.
RED: MES9 = 169 DCV, MES14 155 DCV - could be in the ballpark with the real Nixies attached, no?
WHITE: MES10 = 1 DCV, MES15 -13.5 DCV
GREY: MES11 = 13.2 DCV
ORANGE(6): MES13 = -9.7 DCV, MES18 -23.3 DCV - remains pretty stable, also with the 22R load
I have to get up early tomorrow so no more time for further tests, but let me know if you still want me to desolder the mystery device or we can conclude that it's doing "something" and the voltages look alright (+/- 13.5, ~ -24 WRT chassis).
I can't explain why WHITE (5) VDC > base of Q6 VDC. Bad Q7 possibly? Q7 emitter should always be 1.2V lower than Q6 base.
ORANGE (2) is lower because you've loaded down GREY (3). And it isn't down by much +8.14V to +7.97V.
My calculated number was based on the assumption of the Q1 being oriented differently than currently shown; collector on zener and base to mystery device (I forgot to update the prediction). Showing Q1 as it is now adds an extra 0.7V to the zener voltage causing the output to be predicted higher by 0.7V to +13.5V. Working backwards, seeing +7.97 means there is +13.3V on the GREY (3) output which is good with the 22R/33R (whichever) you loaded it down with (7.97÷(2000÷(2000+330+500+500))=13.3).
Re-read my last message. Report back the VDC across C4; this is the blue 33uF cap next to the mystery device. It could be a bad wiper contact, dry solder joint (area around mystery device, Q1,R1,Q6, pot wiper), bad Q1, bad Q7, (unfortunately) bad mystery device, or combination. When you report back, we can dill-down further and eliminate/confirm Q6/Q7 is the culprit.
...
I'll check Q1, Q6, Q7 during the course of the next days out of circuit.
...
With GREY (33R) and WHITE (60R) loaded down:
C4: 9.8 DCV (always), mystery device right leg pair: 8.6 DCV, or virtually no voltage?
minimal loads:
C4: 10.0 DCV, mystery device right leg pair: OL, 3.98 DCV, flashing bar graph?
Not sure if something is intermittent and I'm getting different results on the mystery device, but it's probably wise to exercise the pot and reflow some joints. If you think it's a good idea I could also try to scope the voltages around that area, maybe the multimeter doesn't like the output.
Btw, R1 = 1k (if it's the "dog bone" style resistor) and measures as such, R3 = 2k and is at 2k3
Q:You said that R1 measures 1K but is it marked 2K?
If true, then that changed the current from 11mA to 22mA (based on 22V unreg. rail) but this should not be enough to kill the LED (if it exists) in the mystery device because many old optocouplers have robust LEDs that can take 40+mA continously.
I guess you can try to test Q1 out-of-circuit.
Q: Are you using one of those component testers that shows transistor type, EBC pinout, hFE, etc?
When you do remove Q1, confirm from the underside (based on the tab) exactly which pin is going to GND, zener, and mystery device upper-left leg. Datasheet: https://www.web-bcs.com/transistor/tc/2sc/2SC857.php (https://www.web-bcs.com/transistor/tc/2sc/2SC857.php)
If Q1 is good, then you can remove the mystery device and test it separately by connecting:
i. upper left to 0V, lower-left to 1K to Vcc,
ii. upper right to 0V, lower-right to 10K to Vcc
iii. with your multimeter on VDC on the right pair.
iv. use 22VDC for Vcc from a lab power supply or lacking that a wall-wart/laptop power supply that can do any fixed value between 12VDC and 22VDC.
If it's a working opto coupler then connecting and disconnecting the 1K should cause the MM to go from 0V to Vcc.
Also, report back the Vdrop on the left pair when the 1K is connected.
EDIT: Also, while you have the mystery device out, confirm with your MM set to ohms, that there is very high resistance between any pin on the left side to any on the right side.
I found a 4n35 (Vce=70mV, still ok?)... quick check before I power it up:
Is it wired in correctly? Or do I need to swap the C / E around?
The 4n35 should be fine.
And you have it the right way assuming the LED is on the left, phototransistor on the right.
But before powering up with the logic and display boards installed, let see what the output voltages are under load.
Connect to GND: 10K*, 33R*, 33R, 22R, for RED(1), WHITE(5), GREY(3), ORANGE(6), respectively. (*=more realistic)
Measure them WRT GND (BROWN(4)) and WRT chassis (GREY(3)).
No 10K available (of that wattage) for the RED supply, but with 2x22k at about 10.8K I get 207 DCV WRT GND, 193.5 DCV WRT GREY respectively.
With the 33R load, WHITE is: 8.8 DCV WRT GND, -4.9 DCV WRT GREY, but I noticed that it creeps up with temperature. Is this something to worry about? Is the 2SC830 failing or is this normal behaviour? It went from about 8.6ish to 8.8205-ish (sometimes creeping up, sometimes going back down, more or less remaining there now). If I blow on it changes a little...?
The optocoupler is slightly warm..
I think it's repaired? Time for the logic boards? Or is it better to implement the protection circuit against shorting transistors first?
Good and bad news:
It powers up and the nixies light up too. There was even a 0 on the rightmost nixie, but after 15 seconds it showed random numbers and went comatose.
Here are the voltages, with the RED pot all the way down (one dot on the nixies still lights up, I guess that's good enough voltages)
WRT GND, chassis in VDC:
RED: 190, 176.8
WHITE: 5.6, -7.7
BROWN: 0, -13.4
GREY: 13.4, 0
ORANGE(2): 2.9, -10.5
ORANGE(6): -9.8, 23.3
WHITE and ORANGE(2) deviate from the prediction.. close, but no cigar
EDIT: Attached are some scope pictures.. not sure what to make of it, WHITE still looks weird compared to the others
WRT GND | WRT chassis | My Comments | |
| RED(1): | +190V | +176.8V | Good. Within acceptable limits. Nixies light-up (typ. >165V) nor excessively bright. |
| WHITE(5): | +5.6V | -7.7V | Bad. Likely a consequence of ORANGE(2) droop. |
| BROWN(4): a.k.a GND | 0V | -13.4V | Good. No drooping. |
| GREY(3): a.k.a chassis | +13.4V | 0V | Good. Confirms reverse of BROWN(4). |
| ORANGE(2): | +2.9V | -10.5V | Bad. See below. |
| ORANGE(6): | -9.8V | +23.3V | Good. No drooping. |
ORANGE(2) should be close to +8.1V WRT GND or -5.4V WRT chassis. Because it is derived from a resistor divider, that implies that whatever it feeds must be high impedence otherwise it will cause it to droop. Which is what is happening. Unfortunately, you are going to have to trace where the ORANGE (2) wire goes; which board(s), which component(s) it immediately touches. It is probably used as a reference (my guess) so it shouldn't be used in lots of places; probably leads to a [bad] buffer amp that's loading it down.
Q1: Immediately upon power-up, is ORANGE (2), WHITE(5) normally at +8.1V, +8.6V, respectively, and then they droop?
Q2: Immediately upon power-up, can you type on the keyboard and does the machine respond normally before it goes comatose?
Q1: Immediately upon power-up, is ORANGE (2), WHITE(5) normally at +8.1V, +8.6V, respectively, and then they droop?
It looks like it stabilises at 5.6 V immediately.
I traced it to yet another alien looking thing (see attachment). The second legs seems to go to a ground plane as the trace then stretches across the board connecting many other components...I haven't found out if it's one of the established grounds (BROWN or GREY). I guess to know what the voltage should be we need to figure out this component too.
Quote
Q2: Immediately upon power-up, can you type on the keyboard and does the machine respond normally before it goes comatose?
See my second post after the scope pictures. It seems to power up correctly without keyboard and there is possibly a fault there.
So, without the keyboard (all other boards installed), the voltages are still those below?
WHITE(5): +5.6V -7.7V WRT GND/chassis
ORANGE(2): +2.9, -10.5 WRT GND/chassis
And, also without keyboard, does it still go comatose after 15 seconds?
If it DOES NOT go comatose without the keyboard then I'd give the keyboard a good cleaning and test each of the switches for continuity (high-resistance when not pressed, low-resistance when pressed). Then reboot with keyboard installed and see if it still goes comatose. A key may be loading down a power rail.
The comatose issue or low ORANGE(2) may not be related.
The voltages remain the same with the keyboard disconnected, the only difference is the 0 doesn't show up (it only appears comatose with the keyboard connected).
Working assumption: the boards regulate ORANGE(2) and WHITE somehow and there are no faulty components that influence it.
Any idea how to disassemble those type of keyboards? I've never handled one like that... the wires are soldered to phenolic boards sandwiched between metal plates. Is it possible to bend those metal pieces back..? I guess that's how you un-sandwich it...
I would like to throw this in an ultrasonic cleaner but I don't own one and the keyboard can't be disconnected from the backplane. Alternatively, I'll try to map out the keyboard. My hope is that maybe just one or two keys are stuck close and hinder the calculator from operating correctly.
PS: There's a really bizarre connector on the back.. what could be the purpose? Remote control... factory testing?
PS: There's a really bizarre connector on the back.. what could be the purpose? Remote control... factory testing?
Maybe you can first separate all key groups from the top of the enclosure. It looks like it can come off as one unit. I think that row of brass phillips screws above & below the blue wires (from your first message) secures them. That should give you access side-on to spray contact cleaner and exercise the keys. I don't think you should separate those plates as they are punched along the left and right edges (4 times) of each key group. It'll probably cause you grief to put back together with multiple springs and alignment points. And never be flat again.