RUNNING AN APPLE-1 FROM ONE CHEAP SWITCHMODE PSU

So let me address all the issues and questions raised !

In post #3, macintosh_nik wrote:

"The topic of alternative power sources is very interesting to me. Our mains voltage is 220v, I, like many people in Europe, use Hammond 186D10 and 186D28 transformers. "

 "Very cool "pigtail" connecting the board and power supply, I like it."

Uncle Bernie answers:

I know ! It has been a huge problem for me. This is why my lab runs from a huge isolation transformer sourcing  110Vac and 220Vac on its secondary side. So I can use equipment from all over the world, and I can touch line voltage powered items on the lab bench without getting an electrical shock - it's only safe however if the old rule is followed to only use one hand/arm at a time, never two. Every professional learns this rule (and others).  For typical hobbyist types, it is not advised to do these things that way, but for professionals it's not always inevitable to work on "live" high voltage powered items.

As for the transformer conundrum, in my "Tips & Tricks" I recommend TRIAD "World Series" transformers which have a split primary winding with four pins. You can put these primary windings in parallel for 110Vac operation and in series for 220Vac operation. This is a very flexible solution because you can configure the line voltage input with four pins and then plug in the different line cords, which then select the series or parallel configuration.

Modern switchmode power supplies however can be fed with a wide range of line voltages, i.e. from 90Vac to 260Vac (or even DC). Which makes them "universal" they work anywhere in the world and they are ideal for places with a third-world standard power grid (such as the People's Republic of California). They have a big filter capacitor in their input circuit which has the benefit not only being able to kill you if you touch it, but it also allows the switchmode power supply to "ride" through brownouts.

So use of switchmode power supplies has merits !

About the braided cable between the PSU and the Apple-1: not only does it look nice, but it is also robust, still flexible, and it stays put. It is also cheaper in terms of materials (no heat shrink tubing which is expensive). So I recommend braiding such wires !

In post #4, brouhaha wrote:

"Some linear regulators (78xx, etc.) can be damaged by having the output powered but not the input. This varies by manufacturer, so cannot be determined solely by the part number".

Oh, I know. I worked as an IC designer for that famous semiconductor company which is known as "the switchmode power supply IC company with an attached luxury shop for precision analog ICs at boutique prices".  I know regulator ICs inside out, literally, at the transistor and layout level. Guess what the two weeks of daily work in my lab were for ? I characterized every aspect of every suitable regulator I found in my lab, but with emphasis on the regulators I furnish with my IC kits. They are all from the same manufacturer and the same production lot. They were found fit for this type of application. Some regulators from other manufacturers failed my criteria. Unless you are willing to repeat my tedious lab work,

DO NOT ATTEMPT TO USE ANY REGULATORS NOT FOUND IN MY RECENT KITS FOR THIS DIRECT FEED METHOD !  BE WARNED !

However, these operating conditions do also exist in any normal application when the power is turned off and there is a sufficiently large bypass capacitance on the output of the regulator. This is what happens: the power is turned off, the voltage on the input of the regulator drops, depending on the size of the input filter capacitor and the load current, and when the input/output voltage difference drops below the "dropout" voltage of the regulator, it enters the dropout region of operation where it can't keep the output voltage constant anymore, and despite the error amplifier tries to turn the series transistor up, it won't source enough current anymore to the load. At this point, the output voltage of the regulator starts to collapse. Where it gets interesting is when the output side has a large enough capacitance such that the input voltage of the regulator collapses faster than its output voltage.  This is reverse mode operation which is abnormal. As it can only happen under specific and rare conditions, which almost never is the case unless the input voltage is suddenly shorted to ground,  and there is a huge capacitor on the output side, early three pin linear regulators could get damaged and destroyed by this "reverse mode". It was simply a design oversight.  This problem was mitigated by adding the "reverse" diode you mentioned in the application notes. It protects the regulator from the "reverse mode" condition. Now, if you draw the circuit of the Apple-1 power supply and add a connection from the J1 pins 1 and 2 to the output side of the LM323K, you can see that the rectifier diodes will turn into these protection diode configuration, and all is good.

I agree with your statement that the inner circuit of the regulator IC may vary from manufacturer to manufacturer, even for the same type number. But once an IC is put on the "Released Product List", no manufacturer will ever change the inner circuit without notifying the customer(s) and without mentioning the change in the datasheet. These changes are extremly nasty and have huge repercussions as the ICs typically will need to be re-qualified in the customer's application(s). This process involves a lot of expensive lab work at the customer. To avoid, customers would prefer to just blacklist this IC from this manufacturer and use second source parts instead. So IC manufacturers avoid such running changes of a RPL'd product like the plague. They rather make an "improved" version with a slightly different type number instead, to address issues, but they won't change the original part unless absolutely necessary.

So I'm confident "my" regulators will not cause nasty surprises. I have overstock and would sell them as sets (no other ICs/parts) if need be.

In post #5, Ronin47 wrote:

"That looks suspiciously like the Meanwell PT-65-B"

Uncle Bernie answers:

It is a "Mean Well" PT-65B which is available from distributors all over the world. Thanks for providing the link to your 3D printable case !

In post #6, Corey986 wrote:

"I think running that switching supply would make the ACI not work very well if you had -12V. "

Uncle Bernie answers:

I tested all my ACIs with this configuration (the LM311 based original circuit and the 741 based circuit copied from the Apple-II) and all worked well, despite running from -12V coming from the switcher. No increase of volume was necessary. The performance / reliability of the ACI was as before, no noticable difference, perfect with AIFF files from a sound card / notebook (both recording, playback, and synthetic AIFF files were tried), and the usual mediocre performance with a real cassette recorder. You mileage may vary, though.

Despite I have spent a lot of time in my lab testing and investigating the ACI, I never really got to the bottom of its lousy performance and its sensitivity to the negative supply voltage, which definitely is there and can be observed both in the Apple-1 and on the lab bench. The irony is that neither the LM311 nor the 741 should have any influence of the negative supply voltage per se, as long as it within the bounds where the comparator / opamp can do its job. If you don't believe me, just study the transistor level circuit of at least the LM311. There may be, however, influences of the power supply once the input differential pair is unbalanced. I have some ideas on what could be going on, especially when considering that the unregulated V- supply on the 44-pin bus has a lot of ripple coming from the rectified AC. If you reduce the AC voltage using a variac, the ratio of the DC content and the ripple changes. This may be the reason why we see an influence of the magnitude of the negative supply voltage and why some people believe  the more voltage, the better. But this is speculation. I never had the time to really get to the root cause. And since my ACIs work fine with a sound card, and my cassette recorder has poor quality replacement drive belts made out of Chinesium, so I hate to use it anyways, any further investigation of the ACI topic has no priority for me. I have more important things to do. But if I ever get bored (maybe this Winter ?) I might look into the ACI again.

Hope now all the concerns have been properly addressed !

Nobody so fas has commented on the electrical hazards less experienced hobbyists might be bitten by if they try to do this. One 3D printed isolation case was contributed by Ronin47. But would them fools - which certainly exist even among the elite group of Apple-1 builders - use such a insulation case all the time or would they fool around and get killed (or fry their Apple-1) ?  This is my biggest concern. Technically, my proposed "direct feed" method is sound and it works. I now run two Apple-1 with it 24/7.

Comments invited !

Hi !

Of course, once "Uncle Bernie" has learned a new trick, his mind starts spinning about what else could be accomplished with the new trick. Here is an example:

Many "purists" know that the original Apple-1 came with LM323K regulators that had a "golden" base. For correct looks, they would be much desired, even in clones. Alas, National Semiconductors soon switched the LM323K to a steel base  and stamped the nasty word "Steel" under the type number. Consequently, "golden" base LM323K are exceedingly rare nowadays, I consider them to be "unobtanium". But after a long search I was able to find a dozen or so National Semiconductor LM309K with the elusive "golden" base. Alas,  a LM309K is specified for 1.5A output current which is too weak for an typical Apple-1 using TTL ICs and an ACI plugged in. I measured the current limit of the LM309K specimen I have to kick in at 1.7A load current, which seems as if it might work in an Apple1, but it didn't (so far I had no time to get to the root of it).

Now, I also qualified these LM309K for the "direct feed" method using the switchmode power supply. And so I have an Apple-1 clone with a "golden base" regulator that really looks nice. Here it is:

PDRM1863.JPG

Don't get upset about the low profile heatsink, I have full height heatsinks, too, but this particular build is one of "Uncle Bernie's Specials" which are super low profile and deliberatly deviate from the originals. These are meant to go into super slim acrylic cases which are only about 1" high and have a nice looking handsome tropical wood frame. These can be used as wall hangers (as they are low profile) and then taken off the hooks, put on a table, connected to PSU, keyboard and monitor, and actually be used to demonstrate.

This was an idea of a customer of mine and it's really cool (literally, if a switcher is used) unless you are one of these "purists".

If you are such a purist, you would of course use the correct height heatsink and the full size filter capacitors, and the "golden" base LM309K, but you would carefully remove the "LM309K" type number using a solvent like Acetone and Q-tips and then rubber stamp the correct "LM323K" type number on it using the correct font.

I just wanted to show you how a "golden base" regulator could be obtained and used in a fully functional Apple-1 (with some invisible trickery such as the "direct feed" from a switcher involved).

In post #10, Corey986 wrote:

"As for the ACI reliability with the switching supply.  I did extensive testing for Vince Briel when he was planning on creating an ACI for the replica-1 using both his prototype card and a Mimeo ACI.  What I found was that there was a sweet spot of about negative 16.5 to 18v where the ACI is rock solid with a digital audio source from an iPod."

Uncle Bernie answers:

Alas, I never did search for "sweet spots" in the ACI negative power supply. I am a lazy guy and just do the minimum work necessary to draw the conclusions needed for the task at hand. So, when I did the experiment to find out which transformer secondary voltages would work, before I bought the transformers, I made a quick rig with a variac and noted down which secondary voltages did work. In the range I was interested in I found none which would not work. And I saw no "sweet spots" either.  The ACI was as quirky and unreliable as usual with my cassette recorder which is known to be a POS, and worked well with the soundcard. Alas, I am too cheap to buy an iPod. I can't tolerate any in-ear devices anyways ... the aftermath of all the nasty ear infections from scuba diving (last dive 18 years ago, never again, unless antigravity generators are invented to make the tanks lighter). So having neither an iPod nor a good cassette recorder I don't qualify to discuss "sweet spots" and other magic places as far as the ACI is concerned. I still remember the location of those on women ... 

But I can read IC datasheets and the transistor level circuit diagram of the LM311 comparator and so I would suggest the following experiment:

Take an ACI that works (it should have Mike Willegals 100nF input cap mod). Remove LM311 from socket. Bend pin #4 up about 45 degrees such that it won't touch the contact the in the socket anymore. Same trick as with activating the diagnostics page in my PROMs. Then solder a small jumper from pin #4 to pin #1 without getting solder on the thin portion of the pins (which would go into the socket contacts, no solder wanted there).  Plug back the LM311 into the socket. Take instrument and make sure pin #1 and #4 now are connected to circuit ground but not to bus connector "Z". Now the comparator is configured for the +5V only operation that is advertised in the LM311 datasheet. Try the modified ACI out. Does it still work ?

I did that and the mine works. I also did some actual measurements involving signal generator and an oscilloscope and the performance / sensitivity of my ACI did not change over the whole investigated pin #4 voltage range from 0V (grounded)  to -24Vdc in any significant way.

This begs the question where these different reported behaviours of the various ACI's could come from.  My money is on the possibility that the LM311 you guys are using which apparently need negative supply voltages ain't no genuine LM311 but Chinese counterfeits which are re-stamped older, obsolete comparators that needed negative supply voltages. But this is bizarre. I have no other explanation, though. The opposite could also be true - the "LM311" which I use could be counterfeits,  but based on a much more modern die, which allows them to work with no negative supply voltage whatsoever. All the "LM311" I ever had pretend to be TI and they also had the signature tarnish that plagued many TI parts (silver content too high). I'd think the Chinese counterfeiters did not perfect the reproduction of the genuine "TI" tarnish yet.  It is characteristic and involves a lot of "secret sauce".

In this crazy world, everything is possible when it comes to surprises from obsolete ICs coming from dubious sources ! (Remember the counterfeit Signetics 2504V which were re-stamped 555 timers ?)

And then we speculate why our "genuine" ICs can do things which they should not be able to do. Or vice versa.

Corey986 wrote:

" My LM311 are not from China since I scavanged most of them myself from vintage cards and all are pre-1980, so it could be that there are differences in newer LM311. "

Uncle Bernie answers:

The mine are TI LM311 and were bought from www.mouser.com or Digikey, I don't remember. But I have heard of enough stories about Chinese counterfeits of the products of my last employer before I retired as an IC designer, that I think anything is possible, not only with semiconductors. They found counterfeit brake rotors in the landing gear of 747 jets, for instance. Nothing is safe from these counterfeiters !

Contrary to your suspicion that "newer LM311" may have "differences" to older ones, in the special case of  the industry standard LM311 comparator I'm quite certain they all have the same inner circuitry, 1:1 copycats from the original,  if they are genuine and not counterfeits. There may be some differences in the bipolar process technology being used, so maybe some small layout differences, too, but any manufacturer who wants a slice of that pie - which is huge - will make damn sure their "LM311" are drop-in compatible. IMHO, they will continue to make the LM311 until the end of this technical civilization (which may be in my lifetime = soon).

The LM311 is THE universal bipolar comparator which can be found almost anywhere. It's one of these great IC designs which are timeless, like the LM324, the 555 and the 741. They even make them in tiny, modern BGA packages.  They just give you the best "bang for the buck". An example of another "sweet spot" in terms of performance / price tradeoff. Its only distractor is that is does not work with a 3.3V only power supply. But it will work fine beginning at 3.5V !

This praise however should not distract from the fact that any comparator, if incompetently designed into an application, may act up in various modes of misbehaviour. Don't have proper power supply bypassing, don't balance the differential input signals, have no good hysteresis, and add a lousy layout, and you got youself a problem, guaranteed. Slow opamps like the 741 are much, much more benign. Maybe Woz could not see the mistakes in the ACI design and so he chose the 741 for the cassette interface in the Apple-II.

I have put the same 741 based circuit copied from the Apple-II schematics onto an ACI card and got the impression that it works more reliably than the LM311 based circuit, but it is still temperamental, and due to slower output slopes it introduces yet another quirk which the LM311 based circuit does not have, or at least does not have all too often.  This work can be seen in this thread:

www.applefritter.com/content/which-aci-improvements-do-exist-and-work

I never had the time to really get to the bottom of the ACI's problems and their causation. No magnetic tape storage system is trivial to characterize and measure properly, and the lab instrumentation needed is demanding, if you want real science and not just cargo cult and wishful thinking. The whole issue is exacerbated by the cassette players we can get today being low quality Chinese made new out-of-the-box junk or decades old high quality ones in various stages of disrepair and decomposition / rot. It's just too many unknown parameters in the parameter space of this riddle.

THE PROPOSED EXPERIMENT (+5V only ACI mod):

So what I propose is a phenomenological approach, like what pill doctors do: just take your ACI that - hopefully - works, at least somewhat, and then put the +5V only supply voltage mod I proposed in post #11 above in. It is easy to undo without leaving a trace, and there is no risk of damage if done right. Check it out if your ACI still works or not. If it does, you can be sure that the "direct feed" from the switchmode power supply will also work and have no influence on the ACI performance.

IF THE EXPERIMENT FAILS (ACI does not work from +5V only supply):

If your ACI stops working due to the +5V only mod, we now know it is possessed by a demon which makes it dependent on obscure (unholy ?) features of the negative power supply rail named V- on the Apple-1 44-pin bus. Remember, the Apple-1 was sold at $666.66 which invites demons, because it's a bad, bad number from the holy book !

If the demon is in there, you are out of luck for now. You must first find a witch doctor who can work his magic, using holy smoke, dances and chants, and burns sacrifices to exorcise the demon. I suggest you buy and bring a live pig so the demon can find a new host and is more inclined to leave your ACI. The trick with the pigs is also in the holy book !

Oh yeah, "Uncle Bernie" can be snarky.

I am in this snarky mood because after a careful analysis of the ACI I just can't see any possible reason why its - albeit mediocre - performance should depend on the negative supply voltage at all. Inside the LM311 there is nothing which could cause any significant effect on its gain or offset voltage from its negative supply rail. It has a PNP input stage in which everything references upwards to the positive supply rail, even if it is imbalanced. The following NPN differential pair has a good PSRR from the V- supply pin (according to my SPICE simulations). BTW, most of the "random" timing jitter seen at the output of the LM311 comes via the 47K hysteresis resistor and the noisy +5V supply. Adding a 100nF power supply bypass capacitor helps to reduce the jitter, but it should be put where noise is generated, and these are the PROMs (location 3 and 4). I started to use DIL 16 sockets with built-in bypass capacitors in these two locations and measurements show the jitter is significantly reduced. But, alas, it does not really help much with the mediocre performance of the ACI.

WHY IS THE ACI SO BAD ?

So what is really wrong with the ACI ? I don't know. Maybe instead of speculating and trying to analyze what is wrong with the ACI circuit design and layout we should just toss it into the trash can and design a new ACI from scratch. I intend to do this if there is enough space on my floppy disk controller card. Not that I'm out of ideas what may be wrong with the ACI circuit, but each of these ideas would take many hours of lab measurements to confirm or debunk. And in a fraction of that time I could design an ACI which really works. It may even use a LM311 again. It's a great comparator that stood the test of time. But there is no way to make a good tape read channel circuit out of it with that minimum number of passive elements Woz did use. Making a good circuit would require more resistors and capacitors. Filtering of certain nodes being the most important task. The ACI in its present form does not have that. And it shows. Such a circuit may work fine on the lab bench when outside of the Apple-1 and powered by a clean lab supply. Plug it into the supply from hell (the Apple-1 with all its noise and ringing) and its performance falls apart. This is how I see it.

CONCLUSION

We don't know why the ACI is so temperamental.   ("These are not the wizards you are looking for !")

But we can do an easy and quick experiment to find out if it would work with the "direct feed" from a switchmode power supply without even having bought that switchmode power supply yet.

So, dear readers, do this little experiment with the +5V only operation of the ACI. And report here what your observations are.

Can you see any of the gain loss issues ? Or not ?

I'm curious !

In post #19, macintosh_nik wrote:

"Novichok one of the easiest and cheapest ways to connect your keyboard to the Apple-1."

Uncle Bernie respectfully disagrees.

The cheapest way to attach a keyboard to an Apple-1 is my keyboard emulator seen here in this post:

https://www.applefritter.com/content/apple-1-keyboard-emulator-cable-plans

All you need is an old PC or notebook running DOS and having a parallel printer port. You can pick these old notebooks from the 1990s up on flea markets for $5 or for a song. I used to pull the mine out of recycling bins for free !

Then you need a flat band cable, a transistor,  a printer connector, and the header that plugs into the keyboard socket on the Apple-1. These headers come with my famous IC kits.

The advantage over any other keyboard solution for the Apple-1 is that you can let the software "auto-type" anything you want into the Apple-1.

You just put the text file what you want to be typed into the machine on the notebook, and use F3 to auto-type it.

I've recently added a "turbo" function to the keyboard emulator software which allows me to load large pieces of software in just a few seconds.

I would also make a Linux version if there was enough interest among the Apple-1 crowd. Just send me a personal message here on Applefritter.

I developed my keyboard emulator as a tool for bringing up Apple-1's which sit naked on the lab bench or for showing them in museums (use autotype to make them do something and not just blink the stupid cursor).  It is not suitable if you want to put the Apple-1 into a nice enclosure. For this, I would use a real Apple keyboard modified for the Apple-1.

Alas, I have so many goodies which are half finished because of lack of interest from the Apple-1 crowd. If you want something from me, you have to motivate me.  Not with money (I have enough green toilet paper, thank you). But a little bit of feedback and interest from y'all would greatly help me to find out which projects are worth to pursue any further and which are not wanted and just a waste of time, precious RQLT, of course. I did not work on my Floppy Disk Controller since my last post on it, months ago, for this very reason. As far as I'm concerned, I have it developed to a point where I know it works, and all the problems have been solved. So the intellectual challenge to continue with this project is gone.

I have *lots* of such projects were I lost all interest after the "hard nuts" were cracked and the solution was demonstrated. But you have to keep in mind that turning such a "demonstrator", whether it is hardware or software, into a sort of "product" others can build or use, requires a lot of drudge work far below my pay grade (in terms of RQLT). It's not a matter of money. It's a matter of whether the pursuit is worth the RQLT. Young people can't understand this, as their glass of life energy / life time is still almost full. But in my age I need to scrutinize how I spend the little RQLT left over in my glass of life. This is also why I live in a totally messy house - getting order into chaos is just a waste of  time. Sometimes it's cheaper to order some misplaced item online than wasting the time searching for it. This is how I ended up with half a dozen torque wrenches, for instance.

So you get the idea. Comments invited !

In post #21, Crud 3 wrote:

"That "auto-type" functionality sounds interesting... Will that work only for entering machine language through WozMon or can it also be used after you have loaded and started Basic?"

Uncle Bernie answers:

Auto-typing can do that, of course. It just turns files on the notebook/laptop/tower file system into keystrokes it feeds one by one to the Apple-1.

So it works with any Apple-1 software running: Wozmon, BASIC, games, everything.

More on this in the actual keyboard emulator cable thread (it's off topic for this 'direct feed' power supply thread), see post #6 there:

https://www.applefritter.com/content/apple-1-keyboard-emulator-cable-plans

Hi fans -

here is some more recent information about this topic:

1. Four beta testers of this "direct feed" method so far, from all over the world, and all of their "direct feed" Apple-1 work fine !

2. The only little issue seen so far is that the switchmode power supply will start up with a different voltage ramp than the transformer solution does. Which in some cases causes weird behavior of the Apple-1 after the power is turned on. A transformer fed Apple-1 typically shows a static field of more or less random characters (some regular structure in it if the power was off long enough) which goes away with CLR SCREEN and then you have the usual sole blinking cursor in the top leftmost position. But the "direct feed"  may cause the random character field to appear which then scrolls upwards upon power on. Which may spook users. But it's harmless ! After the CLR SCREEN all is OK, as with the transformer feed.

3. I did an experiment what happens if the usual transformers are being plugged into a motherboard configured for "direct feed". I wanted to see if there is an explosion (the electrolytic capacitors !) or fire hazard. Keep in mind the little red placard saying "DC ONLY" at the J1 connector could fall off after a certain time - I see this in airplanes I fly from time to time, these DYMO stickers are quite durable but don't stick forever, they tend to fall off after a few years ... but back to the Apple-1 configured for "direct feed": which dangers are lurking in it when the sticker falls off and transformers are plugged in despite they shouldn't be plugged in any machine configured for "direct feed" ? Will some capacitor explode into the face of the operator or will some IC ignite and set the house on fire ? I chose a really beaten down lab rat build which did not work too well anymore for this test. It originally was one of my first burn-in rigs and its IC sockets were pretty much used up. Then I used it for testing some circuit mods involving cutting traces. It was pretty much on its last legs anyways. But good enough for some real cruel tests. I put a nasty IC set in which could not be sold because of unfathomable problems it had but worked OK enough for this experiment. The outcome was that all regulator ICs just got hot and died. But no smoke. No capacitor exploded. All capacitors stayed cool. Amazingly, none of the other ICs on the motherboard died. The transformer was able to sustain the situation long enough for even the slowest human operator to notice something went wrong, and would turn off the power again. Then I plugged the PT-65B back in (boldly risking the immense sum of US$20 it's worth), and I saw the Apple-1 did not work anymore because of the dead regulators essentially being short circuits (which may have saved the other ICs). After all the four regulators were replaced, and the PT-65B turned on again, the Apple-1 came back to life as if nothing had ever happened. I'm currently running diagnostics 24/7 to see if some IC was crippled and would later die from its "wounds". After that I will do another destructive test (with a longer power up time) and then throw the whole thing into the trash, motherboard, ICs and all. But I will salvage the $6 heat sink ! You might ask why I do destructive testing - it has to do with "due diligence". You want to see, under controlled conditions, what could happen, if the product is being abused in the most likely ways, by operator error. In this case, the "product" is the "direct feed" method. If something could explode or catch fire under the most common mistakes that are possible, in this case plugging in the wrong power supply, it would be unwise to release the full description of "direct feed" to the public, because parasitic lawyers could construe this act to be "reckless" or "negligent". So you have to do your homework / destructive testing and document it well. This is where we electronics engineers would need the same immunity from litigation or criminal prosecution the manufacturers of these deadly and crippling "vaccines" have got. Like Orwell so aptly stated: "Some animals are more equal than others". I once scrapped a really interesting product I had designed when I found out that its users could kill themselves with it unless paying attention all the time (no, it wasn't a helicopter, it was an electronically controlled mixed gas rebreather for SCUBA diving or work in hazardous atmospheres, I even got a patent granted for one of the solutions in it).

4.  I was alerted that there is a shortage of PT-65B at some of the common distributors. No wonder, the industry still uses them, and they are not technically obsolete, so with the current supply chain disruptions it's no wonder that they may be out of stock. But we have some alternatives ! And I found a place which has them in stock in large quantities (www.jameco.com). I checked today and they have the PT-65B and the RT-65B  (the one with full, safe metal enclosure) in stock. They also have the smaller and less powerful PT-45B which can do 5V/3A and actually is a better match for the Apple-1 than the PT-65B, which is stronger and a bit higher (but otherwise built on the same PCB). So I could not resist to order a PT-45B, too, and test it with a "direct feed" Apple-1, and it worked fine ! The only bad surprise I had was that after putting the switchers into the "shopping cart", a "tariff" adder did appear which increases the price by 17.5% ! The outcome is as follows (prices I saw today + "tariff" = final, and I assume no liability for errors of any kind):

PT-45B  $16.10 + $2.82 = $18.92

PT-65B  $17.55 + $3.07 = $20.62

RT-65B  $20.70 + $3.62 = $24.32

Not too bad despite of the "tariff" adder. Also note that the nice RT-65B with the full metal enclosure (which is electrically much more safe than the open frame PT-65B or PT-45B) costs only $3.70 more than the PT-65B which has the same power rating. The RT-65B has no Molex connector pins for which you would need MTA connectors (which may cost you $0.50-$1 each and also are affected by shortages), but has a connector rail with screws instead. You need U-shaped fork cable terminal lugs for that, but each of those costs only pennies. And it looks more "vintage" ! So if you consider the work and effort with making a nice isolated enclosure for the PT-65B, the RT-65B appears to be the smarter choice.

Comments invited !  

In post #26, macintosh_nik asked if the mounting method for the TO-3 regulators and heatsinks in this example of a "Bruker" power supply "Made in Germany" is better than the mounting method I propose --- see my post #3 in this thread:

https://www.applefritter.com/content/mechanical-tips-tricks-apple-1-builders

I don't think the method they used in their power supply is any good. Note they they have used a sort of brass cylinder with an inner thread as a mounting and contact device for the whole heat sink and the regulator. The brass cylinder is soldered in --- a reliability concern as a solder joint of this type is not strong enough to take all the forces of the parts attached to it. Vibrations or thermal cycles (expansion and contraction) will make this solder joint fail after a while. And when the regulator loses its ground connection it will stop to regulate but pass its input voltage to the output - possibly destroying all the TTLs living there. It seems that the plastic washers could act as a "springy" element to mitigate this problem but plastic will deteriorate and lose its plasticity over time - especially near heat sources. To me this looks like a botched mechanical design.

The mounting method I propose in the above post uses split ring washers to provide plenty of long term reliable springyness and there is no soldering in the ground connection. However, my method may be no good for TO-3 power transistors where a lot of current flows though the screws. More brutal mounting methods may be needed for those power transistors but if the torque is wrong they will cause problems.

On the Apple-1 PCB there is a large exposed copper area (no solder stop mask) below the heat sink. I recommend to use thermal paste not only between the regulator and the heat sink, but also between the heat sink and the PCB (do not use too much paste or cleaning up the mess when it gets squeezed out of the gap by tightening the screws gets more nasty in this case). The idea here is to help the heat sink to cool that section of cooper foil which helps to cool the rectifier diodes, which may run quite hot. This cooling effect of course is not optimum, but why not use it.

The LM309K regulators with the "golden base" are a nice find. Note however that they are a little bit too weak to supply the Apple-1 populated with TTL ICs. This machine needs about 1.4A average from the +5V rail but the peak currents are higher than the 1.5A the LM309K is specified for. Its internal overcurrent protection may kick in at 1.7A peak and this is instantaneous (according to the measurements I did on much the same LM309K you show in your photo - I have maybe a dozen of the "golden base" ones). So there way be local "brownouts" in the +5V rail you best can see with an oscilloscope. Whether the Apple-1 will run robustly under these conditions can't be guaranteed. More likely, not. There are two remedies:

1. You can partially populate the Apple-1 with LSTTL to reduce the load current on the +5V rail

2. You can add a resistor across the Vin / Vout pins of the LM309K on the backside, which is sized to just provide a few 100 mA by itself. The LM309K then will add the current which is needed to get the +5V. Said resistor must be a power type (typically 2W) and it will get hot. It may burn a hole into your tablecloth if your Apple-1 does not have feet.

I have found a super nice solution for Apple-1 feet I use on all my machines. These are rubber feet and they sit snug and don't scratch the surface they are sitting on. They are mounted with screws, not glue, and provide just the right clearance from the table surface to the PCB. If anyone is interested I will post more about these rubber feet in the above mechanical tips thread.

- Uncle Bernie

In post #28, macintosh_nik wrote:

"I think you should post a picture of your screwed up pins, someone will be interested for sure."

"I put the LM309K on one of my boards, with usual chips ....  The result was very unexpected: through the Juke-Box card I tested each memory block for about an hour with the Mike Willegal test and ran the Integerm Basic program."

"Should I paint my radiator with spray paint for electrical reasons ? "

Uncle Bernie answers:

- I put info about the rubber feet into the thread where it belongs:

https://www.applefritter.com/content/mechanical-tips-tricks-apple-1-builders

... see my post #9 in that thread.

The LM309K you have may have a process tolerance depending higher overcurrent shutdown limit than the mine have. On the lab bench I found out that the mine quit at 1.7A of load current. Which is dangerously close to the 1.5A rated load current. This limit was not trimmed on these parts so I suspect the mine are marginal and close to the lower acceptance bound. So like with everything, you mileage may vary !

As for painting heat sinks, I found the outcome looks terrible and never gets anywhere close to the nice eloxated surfaces you see on newer heat sinks. I once make the mistake to run a very dirty salvaged heat sink though the dishwasher and when it came out the surface looked dull and grey. After I spray painted it black it looked so unprofessional I threw it in the trash. There is no electrical reason to paint heatsinks black as no paint provides any reliable insulation. They use the eloxated black color to improve the heat transfer. So a grey one will have a slightly worse cooling efficiency than a black  one.

In post #32, macintosh_nik wrote:

"I finally tried this connection method, it works great and the starting grid looks pretty normal.

I made a little video on how this can be done, hope this is helpful."

Uncle Bernie's comment:

Nice work, macintosh_nik ! Very nice, professionally made video !

A WARNING.

RISK #1: do not use the same color for different voltages

Here is a warning to anyone who wants to do this: In the video, I can see that you did use wire colors which can be confused. For instance, there are two red wires leading to  V3 and V2 which can be confused, and if this happens this will blow up ICs  in the Apple-1. DON'T DO THAT ! Instead, use wires with a different color for the different voltages, as I do. (I guess that in Russia, it may be difficult to get the same colors I recommend, but pleeeese, don't use the same color for two different voltages !).

RISK #2: add a placard to avoid confusing the power supply requirements

Another risk how you can kill your Apple-1 with this method is the omission of the "DC ONLY" placard on the Apple-1, near the J1 connector.

This placard is imperative to have to prevent any confusion of the two possible power supplies - transformers (AC) vs. the switchmode supply (DC).

I've attached a few photos which colors I use and the placard.

PDRM1825.JPG

The same wire colors are used on the backside of the Apple-1:

PDRM1850.JPG

Here is the connection of the cable to the RT-65B (the DC cable are the four wires on the right hand side):

PDRM2910.JPG

Note that "BLACK" and "RED" use two wires each in this cable. This is NOT due to avoid voltage drop but to decrease the current burden on each of the contacts in the J1 connector. These MTA connectors (or as I call them, "CFH - "Connector from Hell") are only rated for 50 plug / unplug cycles - which they can take - but after that the contact resistance goes up quickly with evey further cycle and on the +5V/GND contacts this may lead to a large voltage drop and to overheating of the contact spring, until the connector plastic starts to melt. I have seen this on some of my burn-in rigs which run 24/7 to produce my famous IC kits. These, of course, get plugged/unplugged often to 'harvest' the ICs after the burn-in. So I know that these connectors do and will fail if plugged / unplugged too often. About half of my burn-in rigs run with the "direct feed" and half run with transformer feed. The 'direct feed' ones are much more sensitive to increased contact resistance because there is no further voltage regulation downstream of J1 (the on-board regulators of the Apple-1 are bypassed).  Using two wires / contacts for +5V/GND increases the lifetime of these connectors in this mode of use very much. But for those Apple-1 users who have a stationary set-up and don't plug / unplug the power supply cable all the time, this is not an issue. Some users, however, put their Apple-1 rig away after a demonstration and take it out ... rinse and repeat ... every time the J1 connector gets plugged and unplugged, which is no good. Expect that header on the end of the cable to last for ~70 such cycles if only one wire / one contact is used, and ~ 100 cycles if two wires / contacts are used for +5V/GND. The +12V and -12V supplies are not that critical because the load is only 150...200mA.

RISK #3:

There is a reason why I did not publish and promote this "direct feed" method to an uncontrolled audience. When I developed the method, I did a lot of lab experiments to assess that the method bypassing the regulators is safe for the type and manufacturer and lot# of all the regulators (LM323K, 7812 and 7912) which I had in stock for making my famous IC kits. All were made from ST. But then that stock was depleted and I only could get other regulators from other manufacturers. I did not want to repeat the tedious lab work to qualify the direct feed method for all the new mixes of regulators I had after the original stock made by ST was depleted. This work really is tedious, see here:

PDRM1766.JPG

This photo was taken mid August 2021 - and you can see that the regulators had no heat sink (they don't regulate anymore, due to direct feed) and could be quickly changed if they would have been blown up by the mighty switchmode power supply (well, not super mighty, but sure powerful enough to blow up a regulator). What you don't see in the picture is the massive amount of testing that was done with all IC sockets empty before this picture was taken. For this test with no load, there must be a minimum load resistor on the +5V (the big resistor you can see in the picture). And this, dear reader, is one of the precautions I'd recommend if you experiment with the 'direct feed' method. If you decide to do it (at your own risk), try it first with the regulators in and all IC sockets empty. And then see if any of the regulators gets hot or blows up. This limits the damage that can happen. Like always, do any such experiments at your own risk ! (You also might notice the little safety PCB with quick blow fuses I've used at this stage of the work, the minimum load resistor for the +5V rail also is there). Note that I did use the open frame PT-65B at this stage of the work, which is dangerous in itself as it allows touching high voltages by curious fingers (the mine are not curious with high voltage things as I had the safety training). Also, my whole lab runs on an isolation transformer so there is no hazard touching anything with only one finger. Downstream the isolation transformer there is no "hot" wire anymore which could electrocute  anyone who touches it. But you household line voltage outlets do not have these safety features and so the PT-65B could electrocute you if handled carelessly. The RT-65B with the full metal enclosure costs only a few dollars more and is much safer. You could also use the RT-50B which is cheaper, compensating the cost added by the metal enclosure. The RT-50B is weaker but still plentyful for the Apple-1. The RT-65B can run three Apple-1 at once ! (Which I do with my burn-in rigs).

CONCLUSION

The 'direct feed' method is a way to dispose with the awkward transformers and to run the Apple-1 with a switchmode power suppy. But it has inherent risks if wires are confused due to improper colors being used. It also has unknowns, like how the regulators would react to the 'direct feed' bypassing them. Not all regulators from all manufacturers may react in the same way and not all may be suitable for 'direct feed'.

These risks can be mitigated by proper procedure, use of proper color codes, and by careful assessment of unknown regulators in a lab setup under controlled conditions. But the risks are there, and any mistake can blow up your Apple-1. This is why I did give the 'secrets' of 'direct feed' only to select Apple-1 builders. Now, the beans have been spilled without my fault, and so please heed my warnings to avoid trouble and damages. Again, use all this information at your own risk, and don't do it if in doubt !

- Uncle Bernie

Hi fans:

We do now have a much better method to run the Apple-1 off a switchmode power supply. With the new method, NO modifications to the Apple-1 motherboard are required. And the issue with hand selecting / qualifying the regulators for use with 'Direct Feed' is also off the table.

So, hereby, I officially declare (as the developer of the 'Direct Feed' method):

DIRECT FEED OF APPLE-1 IS OBSOLETE / DEAD !

DON'T make that mod to your Apple-1 anymore !

Instead, use the modified PT-45B or PT-65B switchmode power supplies which were announced in these two threads:

https://www.applefritter.com/content/mods-mean-well-switchmode-power-supplies-power-unmodified-apple-1

https://www.applefritter.com/content/uncle-bernies-power-supply-mod

The latter thread announces a commercial offering of readily modified "Mean Well" power supplies, in a nice isolated enclosure, so you don't need to venture into the realm of working with potentially lethal line voltages all by yourself.

IMPORTANT  NOTE  FOR  CURRENT 'DIRECT FEED' USERS

If you currently have implemented 'Direct Feed', you can remove the three  wires from your Apple-1 motherboard and then modify the PT-65B you already have to the new method (PM me for instructions if you want them). This is a much cleaner and better way to power your Apple-1 from a switchmode power supply. ('Purists' of course still would use the original "Stancor" brand transformers, but IMHO this makes only sense if they can actually be seen, and then, the new ones you still can buy today are stamped "Made in Taiwan" and not "Made in the USA", like back in the day).

SOME  BACKGROUND  ON  THE  DEMISE  OF "DIRECT FEED"

(and why I never officially released the 'Direct Feed' method to the general public)

I had developed "Direct Feed" to be able to run more burn-in rigs on which my famous Apple-1 IC kits were produced (now sold out). Because it worked so well for me, I  considered  to make it public after a thorough beta test of the instructions by third parties / volunteers, and depending on the outcome of this beta test phase. During that time, all my IC kits came with 7812, 7912, and LM323K regulators which I had painstakingly qualified to be able to sustain the special conditions under 'Direct Feed'. Not all regulators of all manufacturers qualify, as some may draw excessive current and die (depends on the particular bipolar process technology these regulators are made with, which is different from manufacturer to manufacturer).

And then I announced "Direct Feed" here on this thread, with the intent to find beta testers:

https://www.applefritter.com/content/running-apple-1-one-cheap-switchmode-psu

Only a few select people from around the world ever got the 'Direct Feed'" modding instructions from me, as beta testers, under the condition to keep them confidential. But one of them 'spilled the beans' and made a youtube video about how to do the "Direct Feed" mod to the Apple-1, without my authorisation (see his post #32 on this thread). He used confusing colors for the wires and did not provide enough information how to get it right.

I was alarmed as I could foresee the possibility of great carnage and "loss of life" of innocent Apple-1 motherboards, due to the danger that people trying the mod with incomplete information seen in the video (confusing colors for the wires, no mention of the requirement to use specially selected and qualified regulators, ...), so I published a few photos on this thread (see post #33) from my own modded Apple-1 using 'Direct Feed', just to prevent disaster for those who saw the video and decided to try the mod regardless. Hoped this helped to prevent damage to innocent Apple-1.

Back then I also did put a warning in the "Tips & Tricks" pdf which every Apple-1 builder using my kits could get for free:

------------- copy from "Tips & Tricks for Apple-1 builders ----------------

"UPDATE / WARNING (27th October 2022): one of the beta testers has spilled the beans and made a youtube video about the 'direct feed' method which I never released officially. Alas, he used confusing color codes for the wires, two red wires which carry different voltages. No good ! There is a devastating mishap lurking there, blowing up most of the ICs on the Apple-1 motherboard. Here is the warning that came with my earlier “Tips &Tricks”:

This “direct feed” method is still in beta test phase and not “officially released” to the public. Do not try to reinvent it yourself ! Not all information needed to do it successfully was given above ! Be warned ! So after the beans have been spilled with no fault of my own, I had to write a rebuttal and a warning in the “direct feed” thread on Applefritter, see my post #33 in this thread:

https://www.applefritter.com/content/running-apple-1-one-cheap-switchmode-psu

And to prevent things going wrong, due to the confusing color codes in the youtube video, I've also provided a few photos there with a wiring using color codes that should prevent mishaps.

Let me emphasize that I did not officially release the “direct feed” method yet. But now, with the beans having been spilled without my consent, people will experiment with it against my will. If you decide to do so, do these experiments at your own risk. I have warned you, and I shall not be held liable for any incidental or consequential damages of any kind, if you experiment. My own long-term tests with running half of myApple-1 burn-in rigs using “direct feed” are not completed yet.

Update March Y2023: all my burn-in rigs have been retired as I'm running out of ICs. There are a few complete kits left, to be sold on Ebay as long as supply lasts, and a few incomplete kits are left lacking critical ICs which are only available at usurious prices anymore, and I won't buy anything at usurers. So my IC kit production has stopped already. But over all these years, none of the Apple-1 running 24/7 with “direct feed” blew up.

--------- end of copy ------------

It was that incident with the release of the video which nudged me into never releasing the 'Direct Feed' modding instructions to the general public. I feared that my writeup might get into the wrong hands of people who do not read all the text in any instructions because they think they are smarter, and then, possibly, also would use unqualified voltage regulators from random manufacturers or manufacturing lots for their attempt to implement 'Direct Feed'. Which may blow up. I had one LM323K that was not specially qualified for 'Direct Feed' which happened to be in one of my earlier Apple-1 builds which was later converted to 'Direct Feed', and that LM323K got damaged. Nothing 'blew up' and the Apple-1 continued to work (except that the LM323K did not regulate anymore after that 'Direct Feed' was removed to use the new, modified PT-65B solution).  So the risk is real, you simply can't use random regulators for the 'Direct Feed' method. And all my stock of the qualified ones is long gone.

So this is the end of 'Direct Feed'.

It served us all well as long as supplies of qualified regulators lasted.

Now we have a better way to power our Apple-1 from a switchmode power supply.

(It's called 'technical progess', folks !)

- Uncle Bernie

In post #37, Macintosh_nik wrote:

"At the moment we don't have a new way to connect power other than the announcement and the offer to buy a modified power supply for 130€."

Uncle Bernie answers:

What do you mean with "we" ? Those who are too lazy to send me a PM (see: "send PM" button) for the PT-65B/PT-45B mod instructions ?

I'm just finishing the last edits (such as putting voltage ratings of the capacitors into the photos) but anyone who is willing to contact me by PM here on Applefritter (or by my appleonedoc email known to builder of my kits) can get the pdf with the instructions free of charge. It's a beautifully made pdf with 36 pages. I did not spend all the work on it to keep it uncirculated. But you have to ask me personally for it. I may relax these rules later, but still want to keep it away from idiots who would not follow these instructions exactly. This is NOT child's play. You can get killed or set your house on fire if you do it the wrong way deviating from the path I can show you. Or you may lose an eye of you use the wrong electrolytic capacitor because you can't (or are unwilling or too lazy) to order the right one from Mouser. Other than these concerns to prevent the instructions from being abused by the wrong people they are ready to go. Beta test phase is over. The instructions for the RT-50B and RT-65B have been delayed, however. I don't know if I ever will make them available. There are some concerns with them which do not apply to the PT-45B and PT-65B. These are the easiest and safest to modify, despite they are "open frame". Part of the concern with the RT-50B and RT-65B, which I would prefer because they come in a metal cage that prevents touching the line voltage, is that "Mean Well" used a different overvoltage protection circuit in them, which does not support the added overvoltage protection channel for output voltage V2 which is part of the mod that well. There is one failure mode which renders this added circuit ineffective in the RT-50B and RT-65B. Whereas in the PT-65B and PT-45B, the added overvoltage protection always works, if it was put in correctly as per the instructions.

- Uncle Bernie

In post #41, Macintosh_nik wrote:

" I just don't really understand why, if you want to give something, why do you want to be asked ? "

Uncle Bernie answers:

You probably can't understand this because you don't live in the USA where parasitic tort lawyers are lurking under every stone.

So I need to follow a certain procedure which protects me from a lawsuit, if anything bad happens.

Same thing when you rent any sports gear here in the USA (like skis, motorboats, ...) or when you rent a place on a shooting range. You always first have to sign a "waiver" which is legal paperwork that explicitely states that you do this and that solely at your own responsibility and your own risk, and you shall not sue the furnisher of the equipment regardless what happens to you (or others).  Remember the 10's of millions of US$ lawsuit against McDonalds for coffee being too hot and possibly, if spilled, may cause burns ? The plaintiff got burned and won the lawsuit. Ridicolous !

In case of modifications of switchmode power supplies, we are not dealing with child's play, it is a device that has potentially lethal voltages lurking in it, and even at the only 65 Watts or so it has, enough small components which must be carefully chosen to to avoid overloading which may lead to a fire or explosion hazard. 80% of the text in the mod instructions deal with avoiding these hazards, otherwise I would not have used up 36 pages !!!

And if you remember our private, non public email discussion about your attempt to  replace  the prescribed electrolytic capacitor with some inferior type which may be unsafe and may cause said fire / explosion hazard, it actually was  y o u  who gave me second thoughts about publication of these instructions. So I re-wrote them with stern warnings about almost everything which could go wrong. The first versions of the instructions had 3 pages, not 36. But somehow I must protect myself from lawsuits, you certainly understand. It's not that my instructions are unsafe, as the mod was carefully developed and assessed in my lab, and I have the required expertise (worked for a switchmode power supply IC company), but it's the p e o p l e who do the mod who may be unsafe or not. There is a certain type of personality out there who never follows instructions and never reads user manuals, and these people do everything  their  own way because they think they are smarter and can cut corners. Some win the 'Darwin Award', and rightfully so. And then their family sues. This is one reason why the whole country is devoid of entrepreneurial spirit and everybody just seems to lay flat waiting for the lawyers to die or go away. So they can start their own business without the permanent threat to get sued into oblivion. IMHO, the lawyers will not go away, so the U.S. economy will never be great again. The entrepreneurial spirit this once great nation had back in the early 20th Century is now all but gone, most of the industry has moved offshore (where they don't get sued) and the will of the people to be productive has been broken, everybody appears to just want free handouts, free internet, and mind altering drugs. This is not looking good. And it is lawyers and lawmakers who are responsible for this decay, which I think is accelerating, based on my own observations.

- Uncle Bernie

In post #43, Macintosh_nik wrote:

"So in this case, we have to wait a little longer for you to finish your manual, right ?"

Uncle Bernie answers:

The manual is finished except for the last APPENDIX III, which I intend to add when I have more time available. This appendix is not needed to do the mod. It is meant to help (qualified) readers to understand my lines of thought and maybe it will help them to develop similar mods for other switchmode power supplies.

The mod instructions for the PT-65B and PT-45B  are available right now, for anyone who contacts me by PM here on Applefritter or by my email known to all builders of my kits. But I will not publish them to the general public at this time. I first want to see how all works out, and reserve the right to deny access to the pdf to candidates which I deem to be unsafe with such work on line voltages. On other words, I want to keep the proliferation of these instructions under control, at least in the beginning.

Let me also clarify that a "Disclaimer" - which you did understand in the right way, what it does - will NOT protect an author or a publisher from a lawsuit. Here in the USA anyone can sue any targeted person or company for any reason, at any time, for any amount of alleged "damages" (Millions or Billions of US$) in a court of law. No proof of any actual damages is needed. The problem is that if a lawsuit is opened, every party must pay their own lawyers (to the tune of $600-$1000 per hour) regardless of the outcome of the lawsuit (meaning: who wins). So "rich" people (or corporations with deep pockets) can  financially ruin any adversary they target. In many other countries, other than the USA, the party who loses the lawsuit must pay all the costs, including the fees of the lawyers of the other party. Which is a deterrent to start frivolous lawsuits.

The "Disclaimer", if properly worded, will indeed prevent to lose the lawsuit. No "damages" need to be paid, if the court finds the "Disclaimer" to be valid. But if I get sued, I would need to pay my lawyer's fees in any case, all by myself. This may be some 100000's of US$. Which I don't want to lose.

So you understand now that I must be very careful with publishing modding instructions for electronics running on line voltages, which are potentially lethal. I do not want to "lure" random people who may lack the background and experience to do such work into trying it regardless. It's the same thing as giving a scuba tank with regulator to a newbie without the training. This person will most likely die during the first few attempts diving with that equipment (lung rupture, 6ft / 2m depth of a swimming pool is enough for such a fatal outcome). You also don't show a kid how to fly an airplane. Or how to load a handgun. You just don't do that, to protect them from injury and death, if they would try to do that without supervision. So my policy is to screen the people who will get the mod instructions.

- Uncle Bernie