Whoops said:
Got the schematics for this circuit,
That helps a bunch.
I Measured all the voltages at the transformer secondary and every value is spot on.
Replaced all the PSU Electrolytic caps with new Panasonic 105º
Replaced BD901 and BD902 Bridge rectifiers
Unsoldered Q901, Q902,Q903 and Q904 and verify them with my Atas DCA Pro, all the transistores measured good. Soldered them back on
Tested all the PSU diodes, all tested well
You tested the zener diodes for their reverse zener voltage?
Odds are, if they act as diodes on a multimeter, they are probably fine. They tend to fail either shorted or open, never seen one actually shift it's reverse voltage breakdown point, the main use of a zener. Easy enough to test, look up the number and determine the voltage, power supply set several volts above that, anode to minus, cathode to plus through a 1K resistor, measure voltage across diode.
Problems:
After the BD901 regulator I'm measuring +-15v instead of +-18.
You meant BD901 bridge, something seems to be loading down that transformer winding and bridge in a big way. A major clue.
Collector of Q902 I'm measuring +1.7v instead of +12v.
Collector of Q903 I'm measuring -0.9v instead of -12v.
All the other circuit boards are disconnected from the PSU board.
That doesn't look like a circuit that needs a load to function, but someone with better chops than I, will need to confirm this. Anyone care to jump in? With no load, I believe you should get +/- 12v out of it. It appears to be cross-coupled, so that the - rail tracks the + rail, and if the - rail drops, Q904 shunts some + rail to ground to keep up, hard to tell without resistor values. A tracking supply would be expected with sensitive instrumentation that is flat to DC and is not expected to drift over time. Honestly, I do not have a full grasp of that circuit. But it is obvious that a single component failure in that circuit will take down both sides.
Any idea of what might be causing this?
the only thing that I didnt had time to do today was measuring the resistors one by one because I have to lift one leg up, visually they look fine.
Resistors below 1M almost never fail lower in value, check them in-circuit in both directions with your test leads, wait for the readings to settle as whatever capacitors charge from the measurement current. Any that read much higher than marked in either direction, are bad. Circuit unpowered, of course. This is not a definitive test, but it will catch most bad or open resistors that aren't in parallel with other resistors, easier than unsoldering a bunch of legs.
theres one strange thing with the DB901 bridgr rectifier. I measure it with the + and - legs in the air so it was not connected to the rest of the circuit. I have 16VAC coming from the transformer, after BD901 I'm measuring + and - 14VDC, thats really strange shouldn't the votage after rectification go up by an 1.414 ratio? I changed the Bridge Rectifier for a new one and the same happens.
That's normal, with the bridge output not connected to anything, you have no filter capacitor at all. Your meter is interpolating a lot of 120 (100?) hz half-wave humps between peak and zero, and coming up with an average of 14V. Connect any small cap across the bridge output, and you will get your expected (16V X 1.414) minus one diode drop in the bridge.
Since we seem to have it narrowed down to the +/- 12V supply, check that trimpot (VR 914) that appears to set both +/- 12V, they can get dusty,old or corroded with the wiper going open. Measure for 24V across the output rails, note which way the pot slot is facing, and work it end-to-end about 20 times to burnish off any corrosion on the wiper that may be preventing a decent connection to the carbon track, and if you are lucky and see real output, set it for 24V.
Have you checked the op amps? Another down-n-dirty trick, if they are expected to operate in a linear fashion as they would be in this circuit (not as digital comparators with output swinging from rail to rail driving relays or something), measure both + and - input pins. They should be within a couple millivolts of each other depending on internal offset, as they should be in a linear circuit that is operating linearly. If not, check to make sure the output has swung hard in the direction expected. IE, if the + input is more + than the - input, the op amp output should be pulling towards +. This output current can generally be confirmed by measuring voltages on whatever resistors are connected to the output.
Or just change out the chips as they are in sockets, are they anything unusual/hard to get?
Gene