Apple Power Suppply Question with AR1*

Okay, I finally figured the AR1* transistor. It's a photo-transistor that forms the feedback for the 5V voltage regulation. I noticed a diode with the same annotation on the right side of the schematic connected to the 5V output network. This is the opto-isolated linear feedback on this design. All switch mode power supplies have feedback mechanism for the voltage and load current control. I wasn't finding it until I came across the diode. So, I figured it out --eventually.... And it doesn't help that it is an obsolete way of making this annotation.

-RocketScientists

You might take this question to comp.sys.apple2. There are several hardware wizards that frequent there.

I believe I have started the conversation in our private message thread. Yes, I can fix it. Let's rule out simple and cheap problems before taking the more expensive route.

Maybe this is a good spot to discuss the HOW It works.
The Apple II+ power supply is basically a 45 Watt unit (5V@4Amps, 12V@1Amp, -5V@0.250A, and -12V@0.5Amp) Summed up we get 20W+12W+1.250W+6W=40 Watt converter. I'm quoting from memory so I could be a little off. What's awesome about this design is that it is quite robust on the output side. You can short the outputs, you can "accidentally" leave the outputs unconnected and it won't self destruct like many newer power supply designs today. The only weakness to the design itself are the sensitivity to the voltage input (120V and 240V models) -- Don't interchange them as you'll likely blow most of the semiconductors at the high voltage switching side of the power supply. It is very repairable, albeit a slow process desoldering. Look at that TO-3 power transistor? That'll take some drilliong and high power soldering iron (80 Watt temperature controlled). You get the idea.

The other downside of the design is its heavy usage of electrolytic capacitors of the '70s era. These capacitors had short life and tend to lose its effectiveness as the water in the electrolyte dries up, Or they get internal shorts and overheat. Modern electrolytic capacitors are one or two orders of magnitude more reliable and can handle higher operating temperature. The power supply is enclosed in an aluminium box trapping convected and radiated heat. Like an incubator. It does get hot and is the key reason why the old electrolytic caps die out.

Now the details of circuit operation
The power supply design belongs to a topology class pf "Forward Converter" design. It is similar to the push-pull design with two key differences:

1. Only one switching transistor is used to energize the transformer's primary inductor; the other "half" of the primary has a diode that acts as a freewheeling (energy storage in the form of magnetic energy) and forms the core reset preventing saturation.
2. The power is transferred to the output ONLY when the switching transistor is in the OFF state. This means the magnetic energy is stransferred to the secondary outputs during the reset cycle. EXCEPT for one small case. There's a small secondary in the schematic that generates a reference voltage and a small negative bias voltage for the AR1* transistor that has Capacitor C8 across the base and emitter of the photo transistor.

Transistor Q3 and diode CR3 are the main power switching transistor and energy storage/core resetting diode respectively.

The circuit elements resistors R5 and R6, diode CR11, filter capacitor C7, and Zener voltage reference diode CR5 are use to generate a stable reference voltage to determine the maximum currelt allowed to flow in the power transistor and transformer T1 and provide a reference for output regulation.

The output of the voltage reference circuit above is connected to R8. Resistor R9 is used to sense the current flow throught Q3 by measuring the current sensor resistor R13 (and R12 sort-of) as a voltage V=I*R13 --Ohm's Law. Now, the resistor R14 completes the trio summing network where its is regulating the output voltage with the help of the AR1* phototransistor side of the opto-coupler photo-transistor photo diode pair. The photo diode is on the right side of the schematic measuring voltage at the 12V and 5V outputs --More later.

It is important to recognize the key control knob of regulating the output voltage and over-current protection and reference voltage R8, R9, and R14). The joined connection feeds a "jamb latch" circuit which is designed to rapidly shut off the Q3 power switch transistor. Transistor Q1 converts the summing voltage to a turn on device with the help of PHP transistor Q2. More on this following.

So what turns on and off the power transistor? It is done through three separate mechanisms.
1. When Transformer T1 is energized, the winding on the bottom left side of TR0001 is hooked up to ensure a strong on voltage driving the base of the Q3 transistor through resistors R11, R12, R13, diode CR10, and base-emitter of the Q3. This positive voltage drives forward and hard the transistor Q3 giving enough base current to drive it into a hard ON state (transistor saturation mode). This happens when the Q3 is on that created this hard on voltage.

2. Now when the voltage error summing network (R8, R9, and R14) reaches a high enough voltage due to beforesaid mechanism, the transistor Q1 turns on, pulling down the resistor R10. This drives the transistor Q2 on -- and hard. This one effect shorts and grounds the voltage on the node between Rll and Q3. Resistor R11 is present to allow the transistor Q2 ro short the node to ground safely. Now the power switch transistor Q3 begins to turn off, current is decreasing. And

3. The TR001 winding at the bottom left senses the declining current and wants to swich negative in response (transformer action or Biot Savart Law) And look what happens... Diode CR6 that is across R11 turns on like a perfect switch thereby bypassing the resistor effect of R11. And diode CR7 switches oof and FAST to protect transistor Q2 from what is about to happen. A violent hard yank negative voltage appears at the base of Q3. No current liming here. Actually there is a little one formed by current limiting resistor R22 which discharges capacitor c22 unril either transistor Q3 is completely off and/or capacitor C22 ia completely discharged. Why the violent action?? When controlling power transistors as switches, we need to sweep out any hole/electron pairs out of the collector-base region and FAST during turn-off or else that silicon area of the transistor will overheat and self destruct. vaporize or a slow breakdown mechanism called "secondary breakdown". A phenomenon of bipolar transistors only. once the transistor Q3 is off, the diode CR3 resets the core to a known hysterisis (SP?) state for the next time the Q3 turns on. Once the current has reset to 0 by the reset diode CR3, the TR001 winding at the bottom left (base driver winding) reverses voltage to drive a positive voltage for the next switching cycle in response to the changing of the magnetic field into the opposite direction (again Biot-Savart Law.)

The observant reader would notice that the above scenario requires that transistor Q3 needs to be in the on state in order to "turn on" itself. so how does it get turned on in the first place. Like when you first plug it in and turn on the switch?

Ahh, that's a good question. There is a tickler circuit formed by resistors R2, R3, diode CR2, capacitors C5 and C9. This circuit samples the AC voltage (120V or 240V depending where you live) and chops off the positive half of the sime wave by the action of R2 and diode CR2. The -120v or -240v sine wave gets divided down to around -2.5 volt or so through the capacitor AC divider action of C5 and C9. This pulls the emitter of the power transistor Q3 to about -1 volt. This tickles the Q3 to start turning on and begins the switching cycles.

Next time I will discuss how the input line filter and filter doubler circuits work and the output stage circuittry.

-RocketScientists

Do you know anyone who I can ship an original Apple II power supply to and who will attempt to fix it (for a fee)? The problem is probably one or two blown components due to using the wrong input voltage etc.

Now that the switching behavior and control circuits were discussed. You should have a fairly good grasp of how the power supply works.

Before I begin on the new section of the power supply, I have some loose ends and clarification. Remember I said that this is a forward converter topology where the power transfer occurs only when the transistor Q3 is off and the diode CR3 is on? I also mentioned the reference voltage circuit and the negative bias were the only section that the transfer of engery occurs when Q3 is on? Actually only the negative bias circuit formed by capacitor C6 and diode CR9 that feeds the negative voltage to the photo transistor AR1*. The referemce voltage circuit containing the Zemer diode does not turn on until the first reset cycle begins (after Q3 is off).

The second item I wanted to highlight are the two sets of "snubber" circuits formed by resistor R4, capacitor C4, and the other set resistor R23 and capacitor C23. These devices are extremely important in saving the diode CR3 and the transistor Q3 from being destroyed by voltage spikes resulting from a phenomenon called leakage inductance. If thesee components weren't there, removed, or damaged then there would exist an infinite voltage spike at the time when Q3 is off, or when diode CR3 stops conducting. The snubber resistors are chosen to limit the voltage spike to a reasonable value. Their respective capacitors are chosen such that their effective impedences matched their corresponding resistors. That means R4=1/wC3 where w=2PI*f where f is the switching frequency.

The last loose end. The forward converter operates in a special mode called DCM (Discontinous Mode) in which a new switching cycle does not begin until the transformer's energy is depleted and started fresh. This is why we have the core reset diode CR3. All the stored energy by the Q3 energizing the core is dumped to the secondary outpus during the reset cycle. HOWEVER, this particular design can tolerate unspent energy from the previous switching cycle still present. That is why we have the R9, R12 and R13. If the secondary output got shorted for some reason, it "bounces" the magnetic energy back to the primary in the next cycle. The Resistor R9, R12 and R13 are designed to sense for ant resisual power remaining in the transformer and forches Q1 and Q2 to start the turn off cycle immediately before Q3 has any time to stay on! This feature is one of the best designs I've seen for that time period. As long as you don't mess with the primary side circuits, it is basically indestructible.

Finally the percentage of the time the power supply spends in the reset cycle is always greater than the energizing cycle by Q3 on average. That's the behavior of DCM. It guarantees that the transformer core T001 hysterisis is always reset to a known state.

Noving on to the filter network. Starting with the line filter, this device is formed by winding two wires at the same turn ratio around a ferite core. And two ceramic capacitors (not physically shown on the schematic). The purpose is to keep high switching current noise from leaving the power supply into the main power mains where that noise can cause interference in your home appliances or even a neighbor's.

After leaving the line filter block, the resistor R1, Bridge Rectifier CR1, fuse FU1, and the three (four) electrolytic capacitors complete the filter circuit. The resistor R1 is really a thermister element, an NTC (negative Temperature Coefficent) device. When cold it has about 7Ohms resistance. When hot its resistance drops (I'm guessing) around 1Ohm. This resistor is a current inrush limiter. It prevens CR1 and capacitors C1, C2, C3, (and C4) from getting a large current spike when the power supply is turned on and blows those parts. There are other features of this resistor. It has two large solder balls on either side of it connecting two fusing wires to it and to the circuit board. That is not solder. It is a special low melting alloy similar to Wood's Alloy. Designed to melt and break connection thus preventing a fire from occuring. The bridge rectifier CR1 consists of four internal diodes which are used to convert the incoming AC sine wave voltage (120V or 240V) and converts them to " halversines" a bumpy DC voltage that resemble a doubled, folded sine wave with only positive voltage component at 120Hz or 100Hz depending where you live (50Hz or 60Hz). That's why the + and = symbol on the CR1 denoting positive and negative terminals.

The C1, C2, and C3 (C4 exists in newer designs) are used to filter the haversines into a low voltage ripple fairly pure DC voltage. It does not need to be perfect because the power supply regulation will basically eliminate that ripple and you wont see it at the secondary outputs. Now, in more recent designs, there are four filter capacitors and are arranged with CR1 along with a jumper wire which at the factory can un plug the jumper wire and swap it with the incoming power connection for 240V line voltage. By leaving the jumper wire in its default position it would work for the 120V line voltage. This trick worked by the fact the jumper wire re-arranged the C1,C2,C3, C4 and bridge rectifier connection to form a voltage double for the T001, Q3 and CR3. The rectified 340V DC is generated.

In the 240V setting, the jumper setting doesn't allow the doubler function from occuring and the bridge rectifier just generates 320V from the 240 line voltage. One would ask, how does a non-doubler circuit generate 340V DC from a 240 V AC line voltage? AC line voltages are rated at RMS voltage. A 240V actually peaks to 320V (multiply 240 by sqrt of 2). Again, this doubler circuit does not exist in the original version of the power supply (and the schematics do not show this feature either)

I do want to point out the fuse location in the schematic is actually a design flaw. It really needs to be placed before the line filter. If the bridge rectifier or the capacitors got shorted, theres no fuse to prevent high current from blowing parts or burning circuit traces. More recent designs have corrected this problem.

The secondary outputs:
The secondary output circuits are all very similar. they all have capacitor, choke inductor, and capacitor configuration. This applies to all four outputs. Since the power supply switches around 25Khz, the filter capacitors and choke inductors eliminate the AC ripple to within .1 percent. That's a very pure DC output voltage. Rectifiers CR12, CR13, CR14, CR15, and CR16 are DC voltage rectifying devices for the ripple capacitors and chokes. I do want to point out that the rectifier CR12, CR13, and CR14 are special germanium diodes. They have a low forward voltage drop (0.45V in my PS). They also can handle about 10Amps of average current. They are also slow switching devices compared to the modern day schottky diodes counterparts. Note that these diodes are soldered onto sheet metal heatsinks as they run hotter than schottkys. Should these diodes fail on you, you cannot find them anymore. Just use the schottky diodes of similar ratings (I can help make the selection from digikey.com after I make my replacements) The schottky replacements should run coooler and has an extremely fast switch time.

Now, you should notice the totem pole circuit connection between the secondary windings of T001 and the diodes CR12, CR13 and CR14. Diode CR13 and CR14 are parallel connections for the +5V output. the next secondary winding (the one with CR12 connected) generates 7 Volts. Note that +5Volt + +7Volt gives the +12V. That's how it is done. The same trick is done for the -5V and -12V secondaries formed by the diodes CR15 and CR14, and capacitors C13, and C14. Again the totem pole connection to get the -12V (-5V+-7V)

The next set of devices formed by resistors R15, diodes CR17, CR18, Zener diode CR19 and capacitor C18 form another voltage reference for measuring the +12V outwput. The diodes CR17, CR18, transistor Q4, and pointometer r16 are used to tweak the the +5 and +12 volts at the factoty to calibrate the outputs. Remember that the +12V is made from the +5V. So by putting the photodiode AR1, resistor R18 and R19 are bias networks that are tuned by the pointometer. The photodiode is optically coupled to the phototransistor AR1* completing the voltage feedback regulation circuitry. The photodiode sees higher voltage it emits more light to the phototransistor. Biasing the voltage on the photodiode allows us to tweak the output voltage.

Finally theres a criwbar circuit that is used to short out the +12 and +5V output in the event the voltage regulation fails. This crowbar is formed by SCR Q5, Zener reference voltage diode CR20, and resistors R21 and R27. A reference voltage is formed by Zener CR20, R21 and R27. Should the +12V go over 13V or so, this turns on SCR Q5. Once you turn on an SCR, you cannot turn it off. At least not in this configuration. So the SCR shorts the +12 and +5 volts to 0v preventing unregulated voltages from damaging your Apple. When this happens, there's a problem with the power supply that needs repair. Better save the computer at all cost. That's the philosophy of the crowbar cirvuit.

That's it!
-RocketScientists

Thanks for the note.... I just noticed lots of keyboard typoes... Grrr.
-RocketScientists