New method to run Apple-1 from a switchmode power supply

In post #3, macintosh_nik asked:

"Uncle Bernie may I ask about the fate of your famous tested chip sets? I see that on eBay now you write that this kit is the penultimate one, and the next one will probably be the last? So, is the era of inexpensive and affordable kits coming to an end ?"

Uncle Bernie answers: A bit off topic for this thread, but the answer is short: they indeed are the last ones. I will write more about them once the last one is sold. At the moment I'm too busy.

In post #4,  skate323k137 wrote:

"What are the mA reqirements for each of the 3 voltages ?"

"The PT-65C Provides 5v, +15, and -15. Which variation of the PT-65 are you using ? "

Uncle Bernie answers:

The typical Apple-1 with ACI needs 1.7A at +5V and, depending on the particular type of DRAMs, 100...200 mA on the +/-12V rails. However, this is the "basic" machine. To be able to later add expansions like more DRAM, color graphics, floppy disks, you need a stronger power supply. Such as 3A at +5V, and 1A (or more) on each "+/-12V" rail. Note, these "12V" rails on the 44 pin edge connector come almost directly (after passing through the rectification diodes) from connector J1 and they are NOT regulated. With "direct feed" these diodes are bypassed and so you get +/-12V there . . . at the rated current of the power supply. Actually, this makes "direct feed" ideal to run floppy disk drives. But "direct feed" is not mistake tolerant and there will be mishaps if it were adopted widely - the primary reason why I never published an exact description of "direct feed" and only gave the recipe to hand-selected, experienced builders who asked me for it. "Direct Feed" is NOT "consumer friendly". Hence, the motivation to have another solution. Note that transformers are NOT "consumer friendly" either. And they are a fire hazard unless properly sized fuses are added on each secondary (three fuses needed). And without interlock circuit, one blown fuse might damage the ICs in the Apple-1. I've spent considerable time on sorting all these issues out, and have found solutions, but none of them are elegant.

Modifying an off-the-shelf switchmode power supply appeared to be a more viable solution, especially for vendors who want to provide turnkey Apple-1 systems to those collectors who can't solder. At the moment, Apple-1 owners living in the EU dictatorship have no way to buy an " EU legal" power supply for their Apple-1. In the EU, nobody can cobble together a few transformers and a line cord and sell that as an Apple-1 power supply. You need "CE" certification you can't afford. One such "CE" certified power supply would costs thousands of Euros - just due to the certification and the small production run. So the idea came up to modify a "CE" certified power supply to the required voltages and have the manufacturer make these mods, preserving the "CE" certification. Alas, they are not interested in making a few dozen modified ones.

I've based this work on a PT-65B which makes +5V and +/-12V. These voltages must be increased to a level where they are high enough (after passing through the rectifier diodes on the Apple-1 motherboard) to run the regulators. The mod is trivial and safe EXCEPT that some of the electrolytic capacitors on the low voltage side of the PT-65B are rated for 16V and not for the ~20V we end up with. Electrolytics have a "surge voltage" rating which typically is only 15% above the "rated voltage", so this is clearly a no-no. So some of these electrolytics must be replaced  with suitable ones. Which would be trivial if there wasn't that "goop" they have squeezed between them. I'm still trying to figure out how I could get these electrolytics out of the PCB despite of the goop glueing everything together to a lump.

The PT-65C would be no solution as the output voltage on the +/-15V rails would increase to even higher voltage levels.  The issue is the pulse transformer which has a fixed winding ratio we can't change. The manufacturer of course could do that but you need to order a few thousand of your "special version".

Again, all these complications exist only because of the EU being a corrupt dictatorship: unelected Soviet Union style "Polit-Commissars" who take bribes from the corporations to pass laws which destroy freedom, liberty and the small businesses. Try find any edible sausage in the EU - after they have banned butchers from running their own slaughterhouse, you won't. Most EU sausages now are gross, factory made, chemistry laden crap even the dogs refuse to eat. The EU parliament is believed to be elected by the people (who counts the votes ?) but has no say in the lawmaking process. My recommendation: get out of there as long as you can ! It can only get worse ! These absolutely evil EU "Polit-Commissars" will not stop until you live in a shoe box, have no car, and eat those ground up mealworms and cockroaches. Yummy !

- Uncle Bernie

In post #6, CVT wrote:

"230V AC is not a joke and I am glad the CE certification requirement exists. I don't want to have to open everything that has a power cord and make sure it wasn't put together by an idiot with a soldering iron in his garage."

Uncle Bernie answers:

True that, but if you are "glad the CE certification exists" you never have been an entrepreneur in the EUSSR, and you don't understand how destructive this EU legislation is. There used to be a better time where any M.S.E.E. or "Master Craftsman" level person was allowed to put together equipment which ran from the line voltage, for sale to end customers. There always have been rules / electrical code to be followed, such as "VDE", and you had to use approved components, and everything was safe.

Now you can't build or sell *NOTHING* in the EU unless you are a big corporation able to self-certify "CE". You can't even make a small, harmless, low voltage electronic gadget to sell it. Heck, you can't even make a simple soccer ball anymore.  Because without the "CE" seal of approval, you can't sell it, so it's pointless to make it. It's a coordinated takedown of all economical activity outside of the large, anonymous, faceless corporations which in the end will own everything.

You will "own nothing and be happy". And you will eat "zee bugs" like all other proles. Steaks from real beef will only be available to the political class and their cronies.

- Uncle Bernie

In post #9, CVT wrote:

"LOL! I think you need help!"

"Perhaps we should all move to Russia and be happy like a certain ex-NSA officer - is that what you are suggesting ?"

Uncle Bernie answers:

Oh, you are mistaken. I will eat 100% grass fed beef steaks and sip Dom Perignon and fly around in a private airplane long after most people in the world walk by foot or ride a bicycle, and eat synthetic crap made from insects, or "lab grown meat" which actually are cancerous tumors they grow in a vat. I will be able to heat my house and run the air conditioning system long after most people can't afford it anymore. So I'd say I'm fine, thank you.

I suggest let's stop these off topic rants. All I wanted to point out how the evil "CE" legislation affects us all when it comes to hindering small entrepreneurs from providing solutions to the Apple-1 power supply problem. But somehow this got off topic too much.

We can't change anything - all we can do is to watch how their evil plans unfold. "Carbon Credits" are the centerpiece of their scam to enslave you. Look for the "Verra" scandal - their CEO, David Antonioli, just stepped down after selling fraudulent "Carbon Credits" to the likes of Gucci and others. A scam within a scam ! It's all about parasitically leeching on all economic activity - since all economic activity is based on use of hydrocarbons.

Alas, most people are not aware how these parasites get their suckers into all economic activity, step by step. And the brainwash runs so deep that most people - despite being victims of the scams - praise the scams and the scammers to "save the world". Fools. All I can do is to mock them. As far as I'm concerned they just can follow the herd down the chute to the slaughterhouse. I don't care. I'm not part of the herd. Never was. Always have been an independent, thinking, sovereign person, with no allegiance to any nation, country, or ideology.

- Uncle Bernie

Hi -

it turned out the white goop is some sort of RTV silicone and it can be cut with a sharp knife, so that the electrolytic capacitor becomes "moveable" and can be desoldered. Here is a photo which one needs to be replaced (red marks):

PDRM3941.JPG

Note that there seem to be at least two different versions of the PT-65B, as seen in the photo. Both versions need more examination - although the PCB has the same part number (PT-65-R19VAI), some components are different / have different values.

As far as my current state of work is, the following steps are needed for the mod:

1. Short resistor R18 which is in series with the trim pot.

2. Cut the GND trace to the 7912 regulator and add a Zener diode there to increase volts.

3. Replace the Zener diode in the overvoltage protection circuit.

4. Replace the electrolytic capacitor of 16V with a 25V type (this one lives on the +12V rail which will increase to +20.5V).

5. Adjust the trim pot to get +7.5V on the 5600uF capacitor on the Apple-1 motherboard.

All these mods are quite straightforward and easy to do.  I'm also looking into modifying the RT-65B:

PDRM3942.JPG

... but I've already found out that it has a different overvoltage protection circuit than the PT-65B. In the RT-65B I have opened, they moved the SCR from the line voltage side to the low voltage side - I consider this to be a bad idea. The PT-65B got it right: SCR on the "hot" side.  So in case of an overvoltage, it will turn off the primary circuit until the line voltage has been removed. The RT-65B won't, it will try to restart, so the overvoltage condition may be entered again, and again, and again. Not good !

PLEASE do not experiment on your own and wait until I have a final bill of materials ready. There is still a lot of lab work to be done so assess whether the mod affects the electrical safety of the power supply in any way. We do not want to build a death trap which explodes into the face of the user or sets the house on fire.

But so far everything looks good and it seems that this mod yields a true "world compatible" switchmode power supply for the Apple-1 with no mods to the Apple-1 motherboard. A worthy and perfected replacement of the "direct feed" method. But of course, if you do the mod, the PT-65B loses its "CE" certification. Too bad ! What a heinous crime ! Maybe they will haul you away into a gulag ? Who knows. (Enough snark against the evil EU dictatorship).

- Uncle Bernie

In post #14, CVT wrote:

"LOL! Funny, but I wouldn't be surprised if you actually believed that. "

Uncle Bernie answers:

Do not confuse my dark humor with what I actually believe. I am well aware that the EU has no "gulags" - yet - but we can see where their legislation is heading. They want to erect a "carbon neutral" Utopia populated by mere slaves who own nothing and pretend to be happy. And eat zee bugs. Let's see if these evil plans get stopped  by revolt / revolution / collapse of the EU before they can get there. Imagine what happens when the car industry in Europe dies from the planned EU-wide ICE ban. 10's of millions of unemployed workers. Pandemonium !

"CE" was the nail in the coffin of many small entrepreneurs or craftsmen. Ayn Rand wrote about this topic in "Atlas Shrugged": when producers need to obtain permit from the State to produce goods, then it's going downhill. The costs, time and energy required to obtain the permit may be prohibitive for the business opportunity at hand. So this product is dead / won't be produced. No wealth is generated.

In the EU, there are lots of other obstacles other than "CE" - such as RoHS - and EU customs have started to crack down on shipments of vintage electronics components with unknown RoHS status. Which soon will make it impossible for EU based Apple-1 builders to import the needed vintage components into the EU. Be aware of that !

- Uncle Bernie

Enough distractions provoked by CVT. It's a sad fact that the EU is heading towards utter economic disaster, based on their insane policies revolving around the "Climate Change / Carbon Certificate" scam. Note that there was a glitch (or fat finger) in editing the text of my post #15 so CVT cites text parts which I deleted ---- I copy/paste most of these political opinions / observations from my ongoing blog / chronicles on the decay of Western Civilisation, lots of text material, and don't write them specifically for applefritter. So I have to edit them down to make my point without going on for pages upon pages. CVT seems to love that such much that he grabs it while I'm still editing a post --- and then he slings it around. It appears to be pathological.

Now, back to a question I often get asked: what are the risks with transformers vs. switchmode power supplies ?

The problem boils down to the very nature of transformers: if you load the secondary winding, the transformer will do its best to provide the power requested by the load, which it takes from the primary. This process has losses which heat up the transformer. When the transformer is heated up enough, the insulation of the windings will start to melt, producing shorts within the windings. These shorts further increase the losses and the heat - until the transformer starts smoking and is done for. Unless so-called "flame retardents" are involved, it may even catch fire, and burn down your house.

To prevent such disaster, fuses are added to the transformer. The original Apple-1 power supply schematic shows  one 0.5A fuse feeding the two primary windings of the two transformers. Do you think this is enough to protect against all cases of transformer overload ?

It does not. See, a fuse does not blow instantly at the rated current. Instead, the fuse datasheet gives a table at which multiple of the rated current and after which time the fuse will blow. For a typical 0.5A slow blow fuse, at factor 1.5 (0.75A current) the fuse is guaranteed to blow after a wait time of "more than" 1 hour.  These 0.75A at 110 Vac mean 82.5 Watts of power, and that exceeds the spec of either transformer. So if there is an overload condition which just causes the 0.75A to flow, the affected transformer will start to smoke before the fuse blows. However, at 2.1 times the rated current, 1.05A, the same fuse will blow after only 2 minutes, maximum. Maybe the overloaded transformer will survive that. I didn't try. To blow the fuse in less than 10 seconds (which the transformer will survive), 2.75 times the rated current is needed (1.375A). This might happen if a transformer secondary is fully short circuited. But this is not the typical failure mode of the Apple-1.

The typical failure mode of the Apple-1 is that some IC dies and starts to get hot and hotter. The regulator ICs have internal overcurrent and overtemperature protection circuits, so they will limit their output current to a safe level. The consequence is that the fuse on the transformer primary will never blow, as the required current to blow it is never reached. Only when a rectifier diode or one of the filter capacitors would develop a short circuit, then this short circuit may draw  enough current to blow the fuse. But if that current is not large enough to blow the fuse, the transformer may get fried regardless. Safety not guaranteed !

The conclusion is that the sole fuse in the original Apple-1 power supply schematic won't cut it. It only provides the illusion of safety.

Can this concept be improved ? We could add a second fuse, such that each transformer has its own fuse in the primary circuit. And then we could dial down the rated current of each fuse until it does not blow after hours of regular operation. This is a bit tricky as the surge current when the filter capacitors are first charged up can be several amperes, although for a brief time. This is why "slow blow" fuses must be used.

But is this dual fuse configuration enough to make this power supply absolutely safe ?

The answer is: no. If you look into the datasheet of typical TRIAD transformers, which are UL rated, the datasheet spells out that the UL class 2/3 rating only applies if there is a suitably rated fuse in each and every secondary winding ! The rationale behind this obviously is that for transformers with more than one secondary winding, a single secondary winding could be overloaded and overheat, if the only fuse is on the primary winding. Note that for the less stringent "general purpose" UL rating, this rule does not apply, one fuse in the primary would do. But why should anyone want to settle for a lower safety level ?

All these concerns would go away if the transformers had an internal thermal fuse. Which sits in the primary winding, and interrupts the circuit (forever !) when that winding overheats. Back in the 1980s and 1990s, most of the transformers found with consumer goods had such a thermal fuse.  But most transformers you could buy today don't.

Here is the next problem with fuses: if more than one fuse is present, any fuse which blows turns off that section of the power supply, but those sections with intact fuses continue to provide their output voltages. For the Apple-1, this is bad - because some ICs in it live on more than one regulated voltage, and their datasheets warn against prolonged operation with one of these voltages missing while the other ones are present. The IC might get damaged if the failure condition persists for too long. The only solution for that: add an interlock circuit which, upon loss of one voltage, turns off all of them.  This involves a relay with three sets of switches, and a few other components around it. I have designed such a circuit, but frankly, it's ugly and hard to hide from sight.

So, let's examine how switchmode power supplies solve all these issues.

In a typical switchmode power supply, the line voltage goes through a fuse (aha !) and gets rectified and filtered. This is the 150uF/400V electrolytic capacitor in the PT-65B. It has enough charge in it to kill a horse instantly - so don't even think of touching its leads, ever !

There is an IC to control the whole thing, a MOSFET (or bipolar) switching transistor, and a pulse transformer. The IC contains an oscillator which sets a flipflop which turns on the switching transistor. Current in the pulse transformer ramps up and is compared against a threshold value. Once the threshold is reached, the flipflop is reset and the switching transistor is turned off. A freewheeling diode allows the current in the pulse transformer primary winding to continue to flow (while it is ramping down). The cycle repeats with the next period of the oscillator.

On the secondary side of the pulse transformer, there are several windings whose voltages get rectified and filtered. One or more of the resulting DC voltages is compared to a reference voltage. If the reference voltage is reached, a LED within an opto isolator turns on and tells the "hot" (line voltage) side of the power supply that the current threshold does not need to increase any further. If the LED is extinguished due to output voltage being too low, the current threshold is slowly increased, up to a maximum limit where it can't increase any further. This process has two effects: enough current is provided to make the desired output voltage at any load current, and the current is inherently llimited to safe levels, as it cannot exceed the threshold. Only if something  fails in this control loop, i.e. the switching transistor not turning off anymore, then the current in the primary circuit will exceed limits (regardless of output load conditions on the secondary side !) and blow the fuse. Below the limit, everything is absolutely safe.

There is another protection circuit, which senses an overvoltage condition at one or more outputs, and in case of overvoltage, triggers an SCR via another opto isolator. This SCR stays on until the line voltage is removed, and it shuts down the switching transistor, turning off all the output voltages on the secondary side at once. No interlock circuit needed !

It should be obvious now how a switchmode power supply prevents all of the dubious and unpredictible failure mode conditions of transformer based power supplies. You can't destroy it by overload conditions (as it always has inherent current limit) and if anything goes wrong with any component responsible for that current limit, it will instantly blow its fuse, and the danger is gone.

This is the reason why almost nobody makes AC transformer based power supplies anymore, and why AC transformers are dying out. The only remaining problem for the Apple-1 is to find a switchmode power supply which does not need "direct feed". Such a switchmode power supply is not available off the shelf, but here in this thread we will show how to modify an existing switchmode power supply to do the job.

If you don't feel comfortable with hacking switchmode power supplies, you can buy suitable ones off the shelf, but you will at least need two of them to run the Apple-1, i.e. a 8-9Vdc at 2-3A one, and another one which makes +/-15V at 0.5-1A. This is less economical (and uglier) than modifying a triple output switchmode power supply.

Oh, and finally, you still can use AC transformers to power the Apple-1. But due to the issues with safety under all possible failure modes, you must keep that Apple-1 under constant adult supervision as long as it is powered up. If any component dies (which will manifest by lost video or program crash), immediately turn off the power, and nothing bad will happen (other than the original event). No sophisticated fuse arrangements needed (as discussed above, under certain failure conditions, they may be useless to prevent disaster).

I have used both AC transformers (with and without fuses)  and various switchmode power supply solutions (direct feed with unmodified PT-65B and this new method with modified PT-65B) and there was only one incident where a transformer was fried and started smoking - and in this case I turned the Apple-1 on and went to the bathroom, hereby violating the "constant adult supervision" rule. So the danger is real ! And with a switchmode power supply, this would not have happened.

- Uncle Bernie

In post #21, CVT wrote:

"And was I the one who started the political bullshit in this topic ? "

Uncle Bernie answers:

Criticizing "CE" legislation is not starting "political bullshit", but an necessary evil to point out why inmates of the EU prison can't legally buy any suitable, turn-key power supply for the Apple-1 . . . and not only that. There are *lots* of interesting products they can't buy due to EU legislation. And as I have pointed out, the EU is hellbound to destroy small businesses by putting legislative obstacles in their way. I know that first hand as the "CE" legislation destroyed my EU based business, and this is why I emigrated to the USA. I have every reason to hate everything "EU". It took me many years working in the USA (despite much lower income taxes) until I recovered from the losses these swine have caused to me with their "CE" laws. There is much more to this story, but this would be even further off topic. I may tell it elsewhere.

The important takeaway from these political aspects for the EU readers of this thread  is as follows:

1. You will never be able to legally buy a turnkey Apple-1 system.

2. You will never be  able to legally buy ANY add-on card unless it's a kit, and even then it's borderline legal or illegal.

3. And, the God of Lightning forbid (Thor ?), you will never be able to legally buy a turnkey power supply for the Apple-1.

4. EU customs will ramp up their confiscation / blocking of vintage electronics component imports into the EU  because these have no known RoHS status. These vintage components can't have that as 40-45 years ago there was no RoHS.

The 4th topic is the worst. On my last trip to the EU I met a businessman who is in the electronics business (large corporation) and he told me how terrible all this RoHS fallout is . . . they can't get RoHS conforming components for legacy systems and the existing stockpile of non-RoHS conforming components can't be imported into the EU anymore as customs would confiscate them (or send them back). The consequences are profound. In example, a failing industrial control module may lead to shutdown of a CNC machine worth 100000's of Euros and the effect ripples downstream in the just-in-time production chain. Deindustrialization imminent. Because without legacy spare parts, the production machinery can't be kept running. Unintended consequences of a stupid law made by morons ? No. Fully intentional sabotage / controlled demolition of the economy. It's a multi-pronged attack from several sides. Energy price escalation and provoked electrical grid instability is another prong of the attack.

Apple-1 builders in the EU still may have the delusion that they could fly under the radar and still import legacy, non-RoHS components into the EU, but believe me, I'm literally the "spider in the web", and I know of a lot of  recent issues with EU customs - they really started to crack down on shipments of vintage electronics components. This includes "dead" ICs sent back for a warranty inspection. Instead of waving the parcel through (one ! lousy ! IC !), German customs sent the parcel back to the UK. Revenge for the Brexit ?

As for CVT, he looks like a troll to me who somehow gets triggered when I rightfully criticize any political situation affecting viability of my Apple-1 mission. Guess what - the fact that my kits are coming to an end ALSO was caused by "CE", among other complications, such as RoHS. And the new way that Ebay extorts tax / customs duty from EU buyers up front. And then the EU customs authorities charge the buyer a second time because something in the system is broken and does not work. It's like living in the movie "Idiocracy". Yes, the EU seems to have the same policy of hiring low IQ retards into government positions. Here in the USA it's called "Affirmative Action".  Only retards with an IQ below room temperature may apply. The EU must have an equivalent policy. Watch the movie to see how this will end.

As for macintosh_nik's comment in his post #20:

I think macintosh_nik indeed confuses CVT with another guy who posts these fantasy projects he claims to have no time for actually doing. Despite CVT trolls my threads, I still wish him success with his Apple-1 projects. Hope he can sell them without suffering from the consequences of not having the "CE" and "RoHS" seal of approval. They may not crash through his door (or ceiling) 3 AM in the morning to sack him and haul him away, as seen in the prophetic movie "Brazil" by Terry Gilliam, but if he gets caught, they will destroy him financially. Good luck, CVT !

- Uncle Bernie

In post #23, CVT wrote:

"Claiming you didn’t start the political bullshit (a  completely  false statement) ... "

Uncle Bernie answers:

Here we have proof that "CVT" has issues with understanding the English language (he wrote he came from some totalitarian Eastern European hellhole, so this is a possible explanation).

I never claimed I didn't start the "political BS" but explained why I had to start mentioning this nasty topic - POLITICS in the EU is the reason why Apple-1 builders based in the EU have to suffer all these complications. WE ALL NEED TO BE AWARE OF THIS EVIL. But of course, their legislation sabotages and demolishes the whole EU economy.  Everybody is negatively affected, not only Apple-1 builders.

- Uncle Bernie

(I expect CVT will retailiate again, but from this point on, I will ignore him. I did my duty towards the EU Apple-1 crowd to warn them about what is going on, all I have to say about the topic of "CE" and "RoHS" has been said. Be aware of these evils and keep flying under the radar.)

In post #28, dorkbert wrote:

"Have you tried IPA? Should be able to peel them off by hand after."

Uncle Bernie answers:

Will try that (but it eludes me how IPA = Isopropyl Alcohol ??? could break down / unstick  RTV Silicone).

At the moment cutting it with a knife seems to be the only way:

PDRM3965.JPG

Yellow rings show the components which must be replaced: one electrolytic capacitor and one Zener diode. Another Zener diode is added on the solder side, to increase the output voltage of the 7912 linear regulator from -12V to -15.5V, and a small SMD resistor gets shorted by adding a wire. Electrically, the mod is quite trivial but the mechanical complications suck. The application of the Silicone RTV goop is not consistent from specimen to specimen and unless it is completely cut through, there is no way to get the capacitor out. Great care with the blade is required to not cut into the nearby inductors.

Work on the "mod" has progressed, I got the "final" BOM parts from Mouser yesterday, and now I'm running temperature / heating tests. Want to find out if the somewhat warmer temperatures I have observed in the low voltage section are consistent with the increased power asked from the PT-65B after the mod.

It's still a lot of lab work to be done. Finding the "mod" was easy and quick, but assessment whether everything is still OK and no component gets overloaded is a lot of work. And like always, nasty events like dying batteries in the thermometer slow the work down. Could have ordered them together with the rest of the parts if I had known that earlier.

- Uncle Bernie

I've now applied the mod to three "Mean Well" switchmode power supplies: PT-65B, RT-65B, and RT-50B.

Here is a modded "naked" RT-65B (metal enclosure removed) powering my Apple-1:

PDRM3984.JPG

I made a curious discovery: the undesired heating of the electrolytic capacitors on the low voltage side which I had observed (and at first suspected to be due to an ESR/ripple current problem) turned out to be caused by load resistors which sit near said capacitors. These provide a minimum load to all output rails of the switchmode power supply. Of course, when we increase the output voltages, then the power consumed in these load resistors goes up with a square law over voltage, which is expected. What I did NOT expect is that these would heat up the whole power supply that much. In case of the PT-65B, on some components, the temperature increased from ~46 deg C to 60 deg. C, and calculations showed that one load resistor rated for 1W now consumes 1.27W, which is a 27% overload.  This is not all too bad, and even might work for a short while, but what's worse is the general situation that these load resistors heat up all the electrolytic capacitors in their vicinity, and electrolytic capacitors running at higher temperatures live much, much shorter. Which is not desirable.

In case of the RT-65B the situation is such that some electrolytics near those resistors get too hot to touch. Not good !

So I removed all of these load resistors and lo and behold, now the whole power supply runs cooler, although there is still a little heating of the electrolytics, and the heat is coming over the ground plane, from the heat sinks. This can't be avoided. ESR was ruled out by measurements of the ripple current. So after removal of the load resistors, the modified power supply is thermally sound. The downside is that it can't be run with no load (no Apple-1) attached. I'm now pondering over a solution for that. Increasing the resistor values could help, but here is a suspicion:

Could it be that these sneaky and crafty Chinese designers of these power supplies have INTENTIONALLY put these load resistors so close to the electrolytic capacitors so as to diminish the life span of the power supply ? Planned / designed in obsolescence like with the incandescent light bulb scam ? In which the light bulb manufacturer cartel agreed that nobody should produce light bulbs which last longer than 1000 hours ?

Maybe they found that heating effect out too late and then tried to cover it up with that RTV silicone goop, who knows. Such goop is normally used only to make the device more robust against vibrations, but it also acts as a heat insulator.

In the RT-50B the goop blob is really bad, just look at this:

PDRM3983.JPG

It took me two hours to get all these components out, one by one. So I could see the voltage ratings and make a judgement if increasing the output voltages would require replacement of some electrolytic capacitors.

This begs the question if anyone would ever want to remove all this blob of RTV silicone and components just to make sure that the RT-50B at hand has enough voltage margins on these electrolytics in the blob. The risk is that the factory could change component values at any time.

This said, it seems that modding the PT-65B is the easiest. The RT-65B requires more disassembly, and the RT-50B further adds that RTV/component blob.

How far would you go ?

Comments invited !

- Uncle Bernie

Today I was able to remove most of the critical components from the RT-50B switchmode power supply, and the schematic reverse engineering is almost done:

RT50_schem_lores.jpg

Note that the schematic still has some imperfections (in pencil) which need to be corrected if wrong.

It seems that the only component which could not stand the mod to higher voltages is electrolytic capacitor C53, 1000uF/16V, which, alas, is the one on the center of the RTV silicon blob visible in the photo of the previous post.

I've investigated if it would be possible to rewind the pulse transformer and desoldered it, but it turned out to be glued together and so there is no way to change it.

So to modify the RT-50B, getting the C53 electrolytic out from the middle of the blob is mandatory.

The question to y'all is whether you would want to use a RT-50B (it's cute, so small !) despite of this complication.

Comments invited !

- Uncle Bernie

P.S.: calculation of the power consumption in the various load resistors has shown that even with the increased voltages it would still be a tad below the spec limit of these resistors. But it is still ugly to have all these nasty "heaters" on the PCB. I'm pondering a circuit mod where I could take out the worst offenders (R100, R101) and add some components to the overvoltage protection circuit which would send the power supply into hiccup mode if the +12V rail(after the mod: > 20V)  is not properly loaded by the Apple-1 itself. Note this should never happen if the Apple-1 is connected to the power supply. This measure would remove the heat generated by R100 and R101 which is significant. --- You can see that modding switchmode power supplies is not as trivial as it seems at the first glance.

Some progress has been made with the RT-50B mod. I've put it all together again, replacing the offending 1000uF/16V electrolytic capacitor against a 1000uF/25V. Note that although I did run the 1st proof-of-concept hack with the original 1000uF/16V capacitor, this is not advisable, as the voltage on the capacitor after the hack increases from 12V to >= 20V. The original capacitor did withstand that, at least for the short while I ran the experiment, but I always was wary that some of those capacitors could explode due to overvoltage. Nowadays it's not as disastrous as it was back in the day, before electrolytic capacitors had these pressure relief kerfs - they would explode in a most spectacular way, spilling their shredded guts all over the place. Nowadays the pressure relief kerf opens and some steam comes out, but the guts are not coming out. Still, a steam burst might come out, with could take an eye out, if you don't wear approved safety goggles (which I always do when working on switchmode power supplies - in the company lab this was mandatory to do).

I have also removed two of the load resistors which would have generated 3 Watts of heat. I'm now observing the temperature rise of all the components involved, as I don't like electrolytic capacitors to run too hot - even if they are heated from the outside, this reduces their life span.

The jury is still out if some of the smaller SMD load resistors also should be removed.

After arriving at a conclusion, I will see if the overvoltage protection circuit (which currently is deactivated) must be augmented to also monitor the +12V rail. This would involve adding another two small diodes.

Here is the modified RT-50B powering my Apple-1:

RT_50B_LORES.jpg

You can see it can be done, the RT-50B can be modified, the nasty RTV silicone blob is gone, and the center capacitor which was in the blob has been replaced with the one rated for the higher voltage.

But the question still remains if anybody out there would be interested to use a RT-50B despite of the complications with the RTV blob (see previous posts). Or whether the cheaper and easier to modify, open-frame PT-65B would be preferred. I don't want to write up modding instructions for switchmode power supplies nobody would ever want to mod for powering the unmodified Apple-1.

Hence, I need your feedback / opinions.

Comments invited !

- Uncle Bernie

In post #33, macintosh_nik wrote:

"As for the PT-50B, if its only advantage over the other two switching power supplies is size and its modification is not the easiest, I think it will be of very limited interest. "

"I'm willing to be a beta tester on this project, and maybe later make a comprehensible YouTube video like I did for the "direct feed method". And as before I am willing to be a beta tester ."

Uncle Bernie answers:

I agree that the RT-50B is the most difficult power supply of the three candidates to modify. But it also is the smallest one.

Yesterday I removed all of the load resistors in it, most are 1210 SMD, and, of course, glued to the PCB, so you need to melt the solder at both leads simultaneously and then pry them off with the tip of a small X-acto knife. All this is not easy. But it was worth the effort: afterwards the whole power supply ran much cooler. See here:

PDRM3993.JPG

Today I will investigate what happens with it in case of no load. And then do the necessary mods to the overvoltage protection circuit, which at the moment is disabled. Most likely, this will lead to the same behaviour seen with the Apple II power supply under no load: the "hiccup" mode - tick tick tick. But this is not a bad thing. All this is rooted in Wean Well's attempt to make an universal switchmode power supply which can also work with no (or feeble) load. And so they added these load resistors which heat up everything in their vicinity.

I intend to use the same technique (removing the load resistors and augmenting the overvoltage protection circuit) with all the three candidates. It just makes sense that in case of powering the Apple-1 we do not need the load resistors, ever, and the switchmode power supply runs cooler and lasts longer.

The question about whether I should publish the recipe for the RT-50B is still open. I have to develop the mod for it anyways because I have bought half a dozen of them when the PT-65B was sold out during the pandemic. But doing the mod myself requires only my notes and no fully written out "recipe". So I could save RQLT by NOT writing down the recipe for the RT-50B. But then, if months later people would want that recipe, it would be infeasable to write it, as I usually run all the various notes on loose sheets through my shredder. They are just a burden after a project is completed. I only keep the "final", clean drawings. Which only are made if I want a "recipe" or "building instructions".

Your idea to make a video for the mods is greatly appreciated. Unlike the "direct feed" the danger of damage to the Apple-1 due to a wiring mistake is greatly reduced (unless some fool really gets everything wrong and backwards). And I don't expect that fools mess with modifying switchmode power supplies - the typical fool knows he's too incompetent to do that and he still wants to stay alive, eat, drink, and enjoy life. Still, I will put disclaimers into my recipes like: "DO THIS AT YOUR OWN RISK ONLY" etc., the usual legalese.

People who saw this thread already have asked me if I would sell them such a ready-made modified power supply. Which I won't do. UL assessment and product liability insurance probably would cost 100000 US$, so if I could sell 3 of them, each one would cost $33350 plus shipping. For the same reason a new Cessna 172 small aircraft costs $425000 (a ridicolous price) while it should only cost half of that. The balance was caused by parasitic lawyers.

So all what is left for us is Do-It-Yourself.

- Uncle Bernie

In post #34, softwarejanitor wrote:

"It may also act as inspiration for someone to do something similar with another brand or a later model power supply some time down the road. "

Uncle Bernie answers:

This looks like a good idea, I could publish my lines of thought about the mod here in this thread, so others could learn from it and apply the same lines of thought towards modification of other switchmode power supplies. After all, applefritter is meant for eternity, or, at least for as long as our technical civilization exists. Would be nice if 100 years from now, people could use the information herein to modify other (vintage, of course) switchmode power supplies to run Apple-1 clones.

But I worry a bit about the technical skill / insight / experience random readers of such material would have. It is quite trivial and quick to modify any switchmode power supply to run at somewhat lower or higher output voltages, within reasonable limits of course, but there are a lot of parameters which must be checked afterwards. And this may require know-how and instrumentation beyond the typical hobbyist. My background is that I once worked for a semiconductor company which was known to be "world's #1 switchmode power supply IC producer with an attached boutique for high priced, high performance analog ICs almost nobody needs". Although I was not in one of the two switchmode IC divisions, I occasionally did help their designers to solve problems they ran into. There are some fine points which are not to be found in textbooks and the irony is that even within a company which has a track record of stellar switchmode power supply ICs, original designers retire, and designs get migrated by less experienced, younger engineers. And then they run into problems ... the same problems the seasoned designers had solved 20-25 years ago. It's the attention to these fine points which makes or breaks a new switchmode power supply IC. Bizarre effects can happen, which are mind boggling and these less experienced IC designers can't understand. Such as the infamous "warble" in polyphase switchers. But I digress.

The takeaway is that modifying switchmode power supplies is not for every electronics hobbyist. So the question is if I should show more than the naked recipes. If I show lines of thought people may start to experiment without fully understanding what they do. Line voltage powered switchmode power supplies belong to the most dangerous electronic building blocks in existence. And some required measurements in those are dangerous even when using an isolation transformer. But even the small ones found in battery powered applications (such as notebooks) can explode fiercly enough to cause injury.

So you may see now why I'm reluctant to publish more than just the proven recipes for specific types of switchmode power supplies. You can see from the above posts that it's a lot of work to make sure nothing gets overloaded, overheated, or runs outside the safe operational envelope. The advantage of the three "Mean Well" power supplies cosen is that all of them use a well known, proven PWM controller IC which is current mode controlled. If the designers of these power supplies did the sizing of the current sense and current limit of the main switching transistor (a MOSFET) properly, then there is NO way to get the pulse transformer into saturation. But I'm not yet there to explore the limits of this part of the circuit.

You can see that although the first "proof of concept" kludge to increase the output voltages was quick (~ 10 minutes per power supply), there is a lot of additional work to do to arrive at a reasonable safe "recipe" for the mod.

- Uncle Bernie

In post #38, robespierre wrote:

"3oo VDC is typical on the primary side of a SMPS designed for global use. When used in 120 VAC countries, the line voltage is rectified and doubled (Cockroft-Walton multiplier); in 240 VAC countries, rectification alone is used. Enabling or disabling the multiplier just requires moving a wire link."

Uncle Bernie comments:

That is typically done only on switchmode power supplies of higher output power classes. And it's often done automatically (the switch is a transistor or SCR/TRIAC, in older ones a relay). But for the "Wean Well" RT-50B, RT-65B and PT-65B, which are low power, the line voltage is directly rectified by a bridge rectifier and then fed into a beefy 100uF/400V electrolytic capacitor. When it is charged up, the charge contained therein is lethal - it can kill a horse ! So avoid touching the voltage there ! After unplugging the line cord, wait a few minutes before handling the switchmode power supply. This allows the lethal charge on this capacitor to get dissipated by bleed off resistors the designers have put into the circuit.

An important safety measure when working on line powered switchmode power supplies is to clamp them (gently) in a vise so they can't move around on the lab table but you still can probe circuit nodes (using probes and probe cables of sufficient voltage rating, but most multimeters  have 200V/500V/1000V settings and come with appropriate probes and probe cables).

Since measurements must be taken with the power supply running, the work is inherently dangerous. You must focus and be well prepared to probe the right points, and have steady hands to avoid slipping with the probe tips. I use probes with robust, needle sharp tips, which won't slip that easily when placed on soft solder joints. But the typical cheap multimeter "Made in China" does not come with such probe tips, those are blunt and slip more easily.

Hobbyists / Amateurs must avoid working in the "hot" or "line voltage" section of the circuit, there are lethal voltages inside. Professionals typically use isolation transformers to  mitigate the risk of electrocution, and the rule still is to never work alone in the lab. There always must be an observer watching you working on the circuit, and this observer is ready to hit one of the the red mushroom power cutoff buttons which are on every lab bench and every wall. This will shut down the line voltages in the whole lab. There also is a defibrillator to restart a stopped heart and everyone working in the lab has been trained to use it. Since hobbyists / amateurs typically don't have that safety equipment, DON'T WORK ON LINE VOLTAGES !!!

But you can greatly mitigate your risk when you have a second person observing the work, who is instructed to not touch you under any circumstances before turning off the power strip feeding the experiment.

Still, it is dangerous work.

In the meanwhile, I have made some more progress with the mods, and have taken out all these pesky load resistors from all three power supplies, which needlessly heat everything up. Alas, now I've got slight overvoltage conditions on some of the unregulated outputs under some (pathological) load conditions and need to replace / add a few more components to prevent that from happening. One of them is a 24V/1W Zener diode which costs $0.28 - - - it only conducts when the pathological load condition happens, and otherwise stays cool. This pathological load condition is full specified load on the regulated rail (+5V increased to +8.4V) and no load on the +V2 and -V3 rails. Of course, due to the nature of transformers, then the unregulated voltages +V2 and -V3 will climb up. When running an intact Apple-1, this pathological load condition never happens, and the loads it presents at +V2 and -V3 are large enough to keep the voltages there within the safe operational envelope. But if something would happen which disconnects the +V2 and -V3 loads and then increases the load at the +8.4V rail significantly (the Apple-1 dying and spilling smoke ?) then some electrolytic capacitors in the switchmode power supply would be stressed with overvoltage - according to my measurements, still a tad below their forming voltage, but definitely above their rated voltage, so something has to be done against that.

This of course delays my work by a few days, as I have to order parts at Mouser again.

Stay tuned !

- Uncle Bernie

Although it was more tedious and more time consuming that I thought.

The missing work to complete the mission was to remove the "minimum load" resistors the manufacturer has put into the RT-50B, RT-65B and PT-65B switchmode power supplies. These are unfit for operating at the higher voltages seen after the mod (nothing dramatic, though). The problem is that power dissipated in a resistor is a square law over voltage across the resistor, so we need to take the original ones out. They could be replaced by resistors of higher resistance values. But still, this would be an ugly solution. I just don't like the idea that the nearby electrolytic capacitors get heated by those resistors and punish this ordeal with a much reduced operational life.

The point is that when modding a power supply for the Apple-1, we don't need a "universal" power supply which can operate even with no load on the +V2 and -V3 outputs. The Apple-1 provides the minimum load needed to keep these unregulated voltages at bay. Unless there is some catastrophic failure in the Apple-1 itself !

And since I am the type of engineer who wants to err on the side of caution (unlike the clowns who designed, built and operated that "Titan" deathtrap mini submarine to the Titanic --- seems they had no know-how where to use CFC and where to avoid --- literally "DIEversity" engineering, like that killer pedestrian bridge that collapsed shortly after opening).

So I have carefully evaluated all conceivable failure modes of the Apple-1 where it could draw a high current from the regulated output of the switchmode power supply while stopping to draw the minimum current from the other two outputs. And lo and behold, I found a scenario where the voltages on these two outputs (called "V2" and "V3") would exceed the nominal voltages on some electrolytic capacitors. Including one of the 2200uF filter capacitors on the Apple-1 motherboard.

As a remedy, I designed and developed additional overvoltage protection circuitry to add to the switchmode power supplies.

This work was greatly delayed due to the sad fact that the last electronic components shop in Colorado Springs went out of business 3 years ago. Otherwise, a missing part not stocked in my lab might have stolen just one hour of my time to drive downtown and buy it from that shop. Oh, and the greedy landlord who had increased their rent so much that they had to give up could not find new tenants for  three years. The place stood empty all the time. Nice profit from the increased rent, Mr. Landlord. Now a dubious "church" has moved in. I wonder if they have to pay any rent at all. Maybe they just pay the utilities and the sole benefit for the greedy landlord is that the place now is heated in winter, reducing damage to the structure.

Anyways, each missing component now costs me at least a week of delay as I have to order it online and wait for the arrival. This is why progress with this work was so slow. But I have found a very cheap solution which now is under extensive tests. So please be more patient and stay tuned.

- Uncle Bernie

Dear fans / followers -

it seems that I do now have a complete solution how to mod the Mean Well PT-65B open frame switchmode power supply for powering the regular, unmodified Apple-1 in lieu of the transformers.

The benefit is that when being fed with these DC voltages (+8.4V, +20V and -15V), all the regulators in the Apple-1 run much cooler.  Such an Apple-1 also can run off any line voltage in the world.

"DIRECT FEED" IS DEAD !

And unlike with the "direct feed" method (which I now consider to be obsolete), there is no danger with making mistakes that might blow up ICs on the Apple-1 motherboard. I always had 2nd thoughts about "direct feed" and this is the reason why I did not publish the instructions before one of my beta testers (macintosh_nik) spilled the beans on his youtube channel. He used wire color codes which could have lead to disaster, as they would allow confusion of wires carrying different voltages. Unlike the color codes I had used.

THE MOD IMPROVES THE PT-65B TO RUN COOLER AND LIVE LONGER !

I also succeeded to make the PT-65B run much cooler by removal of two "minimum load" resistors the manufacturer has put in. I replaced these with an added circuit that reliably prevents overvoltages under pathological load current conditions. My added circuit just sends the PT-65B into "hiccup" mode if there is a large load on the +8.4V rail (driving the LM323K regulator in the Apple-1 motherboard), but no load on the +20V and -15V rails. After removal of the "minimum load" resistors (which cook the electrolytic capacitors near them badly) this pathological load current condition would otherwise lead to excess voltage on two of the electrolytic capacitors in the PT-65B.

ABOUT THE COMPLEXITY OF THE MOD

On the PT-65B, one electrolytic capacitor, one zener diode, and two resistors are removed. All these are leaded components, not SMDs. Then, the empty place for the electrolytic capacitor gets one with higher voltage rating (25V instead 16V) and three Zener diodes are added. Two PCB traces must be cut. The difficulty level is moderate, at least for people who were able to build an Apple-1. The most difficult part probably is to cut into the RTV silicone goop the manufacturer uses to glue some components together.

BETA TESTERS WANTED !

Before I publish the mod, it should be beta tested. Beta testers should be residents of the continental U.S. as shipping costs should stay reasonably low. I can provide kits with all the required components to modify a PT-65B and to build the new low voltage power cable to the Apple-1. I can also throw in an unmodifed PT-65B at cost ($24 ea). I will NOT provide a line voltage cable - you have to find one yourself. And YOU do the mod entirely at your own risk. I shall not be liable for any incidential or consequential damages, including, but not limited to, loss of property, loss of health, or loss of life. The PT-65 is an open frame power supply and the "hot" / line voltage side is exposed to be touched by curious or careless fingers, kids, or pets. If you have kids, pets, or idiots in your household, do not become a beta tester for this mod !

ARE YOU INTERESTED IN BECOMING A BETA TESTER ?

If so, use the "send PM" button to send me a message. State that you want to become a beta tester for the PT-65B mod and that you  will do it at your own risk, and not hold me responsible for any damages. Expect a mod kit without the PT-65B to cost $10 including postage. Adding a PT-65B would increase that to $34 and maybe the next postage bracket - but I don't think it will cost more than $40. (Sorry foreigners - you are better off waiting for the beta test phase to be completed and then you can order all the parts from your local distributor like Mouser or Digikey. Don't even ask me to sell you a kit - too much customs paperwork and heinously high, inflationary postage fees even for uninsured parcels. I just won't ship anything anymore to destinations outside the continental United States, sorry.)

Hope that some Americans volunteer to become beta testers.

I case of doubts, just state your concerns here in this tread. I would be glad to answer them (and learn hereby which concerns people might have about the mod and about running their Apple-1 from that modded PT-65B).

- Uncle Bernie

In post #42. Macintosh_nik wrote (referring to "Direct Feed"):

"I think this method is simple and safe, most Apple-1 fans only have one board, so there is no chance of accidentally connecting the power supply to an unmodified board. But it's a dirty trick."

Uncle Bernie objects:

No, "Direct Feed" is  not a dirty trick, and it's not simple nor safe.

Actually, it is a clean trick if used with the LM323K, 7812 and 7912 I have qualified for this trick, and these were put into most of my kits after "Direct Feed" was developed. But months later, I ran out of these qualified regulators, and the last kits had different ones. Same generic type, but from other manufacturers. There may be subtle differences in the inner circuits or the process technology, despite these are generic, industry standard parts. So there is a risk when using random regulators you may find. You can qualify them for "Direct Feed" all by yourself, but believe me, it's a tedious process requiring careful work in the lab. It involves measuring all the current flowing in and out of all the three pins of each regulator, under various voltage conditions. Note that the voltages on switchmode power supplies do not come up and disappear in a defined sequence. There are some fine points with this "trick". Note that some manufacturers of these regulators recommend a reverse bias prevention diode across its input and output nodes, to protect the regulator from damage which they found out to happen with their regulators under certain conditions where the output voltage is higher than the input voltage. Other manufacturers of the same type of generic regulators do not make such recommendations - their regulators can take the abuse. This may be rooted in subtle differences in the process technology being used.  The process technologies of 50 years ago when these regulators were developed don't exist anymore. So they are now being built on process technologies the original designers of the circuits within did not know.

SO you can see there are fine points and hidden traps in the concept. It looks "simple" but it isn't.

And it's not safe in the sense of "foolproof". If you plug in the header which goes into the J1 connector on the Apple-1 motherboard offset and / or turned by 180 degrees, your Apple-1 will go "poof". Not good ! Stupid mistake !

Now compare to the new method proposed in this thread:

- No modifications to the Apple-1 motherboard

- Rectifiers still being active, protecting the Apple-1 against possible reversed polarity on J1

- Regulators on Apple-1 still active, not bypassed, they will protect the ICs from overvoltages.

So the new method is vastly superior and much, much safer.

Alas, so far no volunteers as beta testers. Chicken and egg problem here - I will not publish the modding instructions before they have been beta tested. And this beta test should be done in the USA where I can ship the parts cheaply and without getting sabotaged by customs goons. I do have one beta tester in the EU who will buy his own components needed for the mod. But this is not enough.

Are there really no volunteers as beta testers for the new way to power the Apple-1 from a switchmode power supply ?

So many people have whined about the issues with the cruder methods outlined in the "Tips & Tricks" that I finally had pity and developed the mod. And now nobody wants to be a beta tester.

Maybe people reading this thread have not yet discovered the "LAST POST" button. Who knows. Should I try to make an extra thread which looks for beta testers ?

- Uncle Bernie