On 2017-Aug-19, at 12:10 PM, Dominique Carlier via cctalk wrote: > Maybe it's better to give us all the useful information these power supplies, > moreover it might be useful to other people with the same computer. > > An overall bloc diagram of the D-116 power supply including G1 and G2. > http://www.zeltrax.com/classiccmp_forum/psu_overall_bloc_diagram.jpg > > The complete schematics of the part of the power supply named G2. > http://www.zeltrax.com/classiccmp_forum/g2_schematics.jpg > > A drawing of the regulation board of the power supply G2 with the physical > locations of the components. > http://www.zeltrax.com/classiccmp_forum/g2_regulator_board.jpg > > The schematics of this regulation board. > http://www.zeltrax.com/classiccmp_forum/g2_regulator_board_schematics.jpg > > And a bit of literature concerning the principle of operation about the > regulation with this PSU (you will understand better why I am a little bit > lost ;-) This principle of regulation with a panoply of verification and > Protection systems everywhere is unusual for me) > http://www.zeltrax.com/classiccmp_forum/psu_regulation_principe.jpg > > I have already tried without the CPU board: same symptoms. Next step, try to > check the capacitors in operation.
>> Results of the observations: >> - This is definitely the regulated +5V of the G2 power supply. More I add >> boards more the + 5v level goes down. +5v, +4.8v, +3.6v, +2.9v. It remains >> stable however with just the CPU and the three core memory boards, it >> becomes difficult for the power supply when I add boards in addition to >> these. >> - This is definitely not a problem at the level of the Power Fail circuit. >> - The big capacitors are not in fault (I rechecked twice). >> - So this maybe a problem at the level of the regulation itself, the +5V >> balancing system ? >> >> Question: a faulty voltage regulator can behave in this way? I always >> thought it worked or it did not work, but not between the two states >> depending on the charge. (In answer to the question, yes, a faulty regulator can produce 'in-between' output voltage.) As is typical for power supplies of this type and era, this power supply includes current limiting circuitry. The current-limiting circuitry will throttle down the output voltage (not shut it off completely) as the output current draw goes above a design limit. This would appear to fit the symptoms you describe. The current-limiting circuitry works by placing a small-value resistor in the current path after the main regulator transistor(s) (aka pass transistors) but prior to the voltage-regulation sense point. A transistor senses the voltage across this R. As the output current increases, the voltage across the current-sense R increases, at some point the transistor starts to turn on, and the transistor is connected in such a way that as it turns on it reduces the drive to the pass transistors, throttling down the output voltage. Arbitrarily using the "A2" heatsink half of the two +5 supplies in this power supply for component references, the current-sense R is comprised of a series-parallel circuit formed by the BE junctions of the 2 pass transistors (A2.Q1, A2.Q2), the two 0.1ohm Rs on the emitters of the pass transistors (A3.R10,A3.R11), the 47ohm Rs between B&E of those transistors, A3.R23, A3A1A1.R8, 376.R9, along with additional influence by 376.R8 and A3A1A1.R10. The current sense transistor is inside the LM376 (see internal schematic), 376.Q16 between pins 1 & 8, controlling the 1st-stage driver transistor 376.Q14. There are various things that could go wrong on this circuitry. Anything that upsets the current-sense resistance network to cause the net R to increase will lower the output current that can be drawn (that is, the current limiting circuitry will start kicking in too 'early'). A likely scenario is one of the pass transistors has failed open. This would take out a parallel leg of the current-sense resistance. All current would be forced through the good pass transistor and it's emitter resistor, raising the current-sense voltage for a given output current. The current-limiting circuitry would kick in at a lower current than the design intention. This is beneficial inasmuch as it would work to save the good pass transistor. The 0.1 ohm 9W emitter resistors are critical, although they're probably wirewound and fairly reliable unless quite overstressed. One of these resistors being open would result in the same operation as an open pass transistor. You'll have difficulty measuring them with accuracy but you could do some sanity checks for continuity around both the pass transistor circuit legs, as well as checking the pass transistor BC/BE junctions. (In principle, the 'proper' thing to do in this area is look at what the current-sense voltage is doing, but that requires knowing what the target V is. Could compare with the other half of the supply.) None of this is to suggest this area is necessarily at fault, or rule out other areas of the supply. Another possible fault that would fit the symptom (decreasing voltage as current draw increases, if that's what's going on) is inadequate (partially failed) drive to the pass transistors. In short, anything reducing the power gain of the regulator error amplifier could produce this symptom.