Mechanical Tips & Tricks for Apple-1 builders

Of course, you could just slap them in and solder them in place. It will work. Most of the time. And it will last for a while. But can it be done in a better way ?

You may have noticed that in all the higher quality electronic equipment, the big electrolytic capacitors always are held in place by some metal strap, or some zip-loc tie(s), or some glue, etc. The reason is, a solder joint is not a good, durable mechanical fastener, and these big electrolytic capacitors, despite they appear to be lightweight, actually are a bit too heavy to be supported and held in place by their solder joints only. To enjoy a long, reliable service life, they must be attached to the circuit board or the chassis by some additional mechanical fastener other than their solder joints, which otherwise soon will develop cracks and intermittent contacts.

As already mentioned, many different types of fastening methods exist. We want to keep the authentic Apple-1 looks, so anything visible like metal bands and zip-loc ties is not desired. Glue works, unless it's the wrong sort of glue, such as the glue out of a heat gun. A little bit of rattle, and it will break loose from the PCB or the capacitor's plastic foil. Two component expoxy glue certainly will hold the capacitor in place, but it will do so forever and your chances are slim to get a bad capacitor off the board again without using a lot of brute force, possibly causing damage to the PCB.

Now, consider this option: 3M makes a wonderful, squishy, double sided, self-adhesive tape you can find in almost every Lowe's or Home Depot or any other hardware store:

CapMount1.JPG

IIRC they offer two versions, one for indoors use, one for outdoors use. I suspect the outdoors version might contain some nasty additives to keep bacteria and fungi from feasting on it, but I prefer it anyways, as only very small amounts of it are used in the Apple-1 build:

CapMount2.JPG

Then, the red tape needs to be peeled off, and then the big blue capacitors can be mounted, carefully. Don't compress that 3M tape before you are certain the capacitor is properly positioned. Once you compress it, it sticks like hell. But unlike some other products of this kind, if anything went wrong, you still can pry the capacitor loose from it again, by applying a gentle, but increasing finger force, over some seconds, on one end of the capacitor, trying to lift it up from the PCB. That 3M tape notices your effort and slooowly will let go of the capacitor, leaving no residue on it. It stays on the PCBs (at least for me). From there, you can sort-of rub / roll it off, as if it was old chewing gum, which did not yet harden. It leaves no residue on the PCB either. I can't tell how long this magic works like that. But I had to replace one failed capacitor after six months, and that 3M stuff still behaved in the same way as if it was new.

It has immense holding powers. I think, if that tape were used to cover gloves and shoes, Spiderman crawling on a ceiling could become a reality. Just look at this photo, I'm holding the PCB upwards down against the window in my lab, and the capacitors don't fall out:

CapMount3.JPG

This is very nice. They just stay put. So you can work on the backside of the PCB, snipping off the leads, and then soldering them in place, without any risk to find the capacitors dislocated afterwards:

CapMount4.JPG

Do not bend the leads, keep them straight. This is only possible due to the sticky tape, and helps a lot to avoid PCB damage if the capacitors ever need to be replaced.

Now, here is a little trap / pitfall that also is dodged by using this 3M tape to mount the capacitors:

CapMount5.JPG

It's difficult to show the problem in a photo, but maybe you can see that the 3M tape keeps the capacitor a little bit away from the PCB surface. Now look at the solder bead that develops on the top side of the PCB, when the capacitor lead is soldered in. Note that the distance of the solder holes for these capacitors is quite narrow, I would say, too narrow, such that the leads will be bent inwards, towards the capacitor body. Very close to the capacitor body. Regardless what you do, the clearance between the capacitor lead and the capacitor body will be very small, and without the capacitor being elevated a little bit by that tape, the solder bead may be able to touch the capacitor body, melt through the blue plastic foil, and, on the positive side, make a short to the capacitor's aluminum can. This short will kill the rectifier diodes instantly, and they typically fail not as an open, but as a short, and the next thing you can see is smoke coming out of your transformer, which will be very very hot. (Don't ask how I know).

Whether you adopt this mounting method or not, you should be aware of this trap and keep an eye on it. This is one of the typical bugs that would be sorted out in an industrial environment during ongoing product improvement. They would notice the leads to be dangerously close to the capacitor body and fix that in the next revision of the PCB. They also would have added some more bypass capacitors and maybe, even the terminators on the multiplexed address bus. Then, that Apple-1, Rev. B, would have been a fine, robust machine without any major quirks left. But with that fledgling garage company Apple was back in the day, no such product improvement ever was done on the Apple-1. They just orphaned it and went on to make the Apple-2. Smart move. Otherwise, they probably would have gone out of business, and there would be no Apple-1 cult, and you could not read my drivel on this forum, because it would not exist. What would we do without our cult ? Life would be much more boring !   

Here are my tips on how to mount the LM323 in an Apple-1.

There is the industry standard mounting procedure for TO-3 power semiconductors, which is wrong for us hobbyists building Apple-1 clones, and there is a mounting procedure I have developed and tested on my Apple-1 builds, and it works well.

Let me explain:

The typical industry standard mounting procedure to mount TO-3 packages involves all stainless steel mounting hardware, which is good, and we should do that, too. But in the industry standard mounting procedure, split ring or star lock washers are avoided. Instead, they use Belleville washers to get the required force. Their typical stacking is: two 6-32 pan head machine screws go directly into the TO-3 mounting holes, and on the bottom of the PCB the sequence is: No. 6N Flat washer, No. 6 Belleville washer, another No. 6N Flat washer, then a 5/16" Full Nut, or, alternatively, a so-called "Sync Nut". The nuts are tightened alternately to avoid bending the TO-3 base plate. When the nuts are snug, they are  torqued to 6 inch-lbs (+/- 1) using a small torque wrench. According to the Harris application note I got this procedure from, this torque causes compression of the Belleville washers to a force of 150 to 300 pounds, each ! The mounting hardware must be spanking clean and not be contaminated with the heat grease or oil, otherwise the torque will be wrong. Properly installed, this yields a very reliable TO-3 mount that is robust against temperature cycles and vibrations.

Alas, most hobbyists can't do this procedure properly, the first obstacle being lack of a small torque wrench, and you can't get any high quality stainless steel mounting hardware at places like Lowes, Home Depot and such, as those either don't have these rare speciality items, or they have Chinese made knockoffs which may work, or not. The long-term springyness of the split ring and star washers under many thermal cycles are the most doubtful properties when using Chinese made mounting hardware. Hmmm ... the PCB of your Apple-1 probably also is "Made in China" using the cheapest possible base material.No way to safely use those Belleville washers with their "up to" 300 pounds of force ! Most likely, the cheap PCB would be permanently damaged and it may delaminate in the vicinity of the screw holes. Still, I would like to try this industry standard mounting method as described above on an original MIMEO PCB that, so it is said, was made using mil-spec base material by a company also working for the MIC. Alas, I never saw one of those MIMEO boards.

Here are my thoughts on the LM323 mounting  issue. On our cheap PCBs (I'm guilty to be a cheapskate, too) we can't use the industry standard mounting method and we certainly can't use Belleville washers torqued to produce 300 pounds of force.

The question to be asked is why the industry standard mounting method tries to use such forceful mounting components. They literally bolt those TO-3 packages down using brutal force, and only the skillfully applied torque wrench saves the PCB from destruction. The reason is as follows: on typical TO-3 power transistors, the collector, which carries a lot of current, is the TO-3 can itself, so this current must flow through the bolts, and from there, into the PCB foil. It is understood that these connections must be very tight, preferable gas tight, and have a well defined, large,  contact area, and shall not loosen under many thermal cycles. This explains why the industry uses these sophisticated mounting methods for TO-3.

The LM323 regulator, however, is an entirely different animal. Its TO-3 case is the ground reference, and very little current flows there. All the load current flows through the two pins, which are soldered in. The small ground return current, some 10 mA, but not much more, does not require any brutal bolt-down mounting methods. The ground connection must be reliable, however, and under no circumstances it may get high ohmic or an open circuit. Because, if this happens, the regulator will stop regulating and just pass its input voltage to its output, possibly killing all the ICs connected to its output.

So what we need, is a reliable ground connection, one that does not use huge forces, but still can take lots of thermal cycles (in which the bolts and all the other materials in the stack expand and shrink, a very tricky affair to calculate the forces involved). The simplest solution to this problem is to add a spring element which can compensate for all this expanding and shrinking. Most likely, a star lock washer alone is not sufficient for that, so I propose the use of two split ring washers which go under the heads of the 6-32 pan head screws, on top of the TO-3 package. European builders take note that the original TO-3 package was specifically designed for these American screws, and not for the metric M4 machine screws, which juuuust will not fit, because they are a little bit to large in diameter. And a M3 is too small to properly mount a TO-3. Some European semiconductor companies made slightly modified TO-3 packages with slightly larger holes, to take M4 screws. You can check out what you have got and see if a M4 fits. If it does not, you could take the LM323 to your drill press and drill slightly larger holes in it (a 4 mm diameter drill bit will do the job, but use a vise to keep the LM323 in place). American builders of course use the local hardware, not the metric ones.

On the bottom side, we use a No. 6 star washer (to bite into the PCB foil), then a No. 6N flat washer, and then a 5/16" Full Nut, in that order. This assembly then is alternately tightend by use of plain finger force only. I have found that some of these 12-point 5/16" sockets have a recess which nicely centers the washer stack, if that socket is held snug to the PCB while turning it with the fingertips - do not use a wrench. A screwdriver on the top side will keep the screw from turning, but is not turned by itself, to avoid slipping and marring the screw.

Alas, assembling the TO-3 with its heat sink and its mounting hardware is one of the operations where you'd wish to be an alien from outer space having four arms and four hands. I found that fastening the PCB in a vise (with rubber/plastic protectors on the jaws to avoid marring the PCB) such that its stands perpendicular to the work bench surface can greaty help to reduce the number of arms and legs required for the installation, but it's still somewhat tricky, as I still would need three hands.

Here is a photo of the mounting hardware that is needed:

TO3_moun_HW.JPG

And another one showing the sequence how they reside in the assembly:

TO3_order.JPG

What still needs to be discussed before the assembly process can be done is the topic of which thermal grease is to be used, and how much, and where.

Those who build PCs probably know about this "magic mystery ceramic thermal grease" peddled by various snake oil sellers in microscopic quantities at usurious prices. The 1970s equivalent is the "oxygen free magic copper cable with gold plated connectors for superior HiFi sound". Don't waste your hard-earned money on hucksters !

I recommend the "Thermalcote I" product from AAVID THERMALLOY which comes in 2 oz tubes for a good price and is perfectly fine for the task at hand. This 2 Oz tube will last you many years. Use a small brush with stiff bristles to sparingly apply it to the bottom side of the TO-3 package. You still should be able to see some of the metal through the grease, like this:

TO3_grease.JPG

Avoid to get grease on the pins. Some people put heat shrink tubing in the pins to keep them clean, and later pull the tubes off. I just use Q-Tips afterwards to remove any unwanted grease, and to clean the mounting holes from grease. You do not want to have any grease on the mounting screws.  

The next (somewhat controversial) place to put thermal grease is the backside of the heat sink, which mates with the PCB. On the Apple-1 PCB, there is a metal area below the heat sink, which is circuit ground, and it connects to various TO-220 regulators and rectifier diodes both electrically and thermally. I think that it may be desirable to have a good thermal contact between this ground area and the heat sink, to help keeping these regulators and diodes a bit cooler. So I put a thin layer of thermal grease on the backside of the heat sink, too. But the efficiency of this measure has yet to be confirmed by taking temperature measurements. It could be wishful thinking and not have much effect. But for me, it also helps with the assembly process because it makes the heat sink to stick somewhat to the PCB, and prevents it from wobbling around and falling off. Keep in mind that I'm no alien from outer space equipped with four arms  !

After everything is put together in the proper order, and the nuts have been finger tightened with a 5/16" socket to be snug, I put the socket in a little wrench and tighten the nuts by another 90 degree turn, but not more. The split ring washers must be fully compressed.

Then comes the check for shorts, before soldering the two pins. Use a multimeter to check for shorts between the TO-3 case (at the circuit ground) and either of the two pins. Most multimeters have a "diode" check setting, which for this check is preferred over a resistance setting. Seeing one diode forward voltage (some 300mV-500mV) is OK, but seeing "0" is bad, which means a short, and you need to take the whole thing apart again. Note that discharged electrolytic capacitors act as a short for a while, so be patient. Same thing with a circuit beeper used in lieu of the multimeter: if it beeps briefly, all OK, no short, if it does not stop beeping, it's a short. But I prefer the multimeter.

After the check is done, solder the two pins. Then, do the check again. If all is OK, use Q-Tips to wipe all the unwanted spots of thermal grease away, and use a throwaway rag to clean all your tools that have been used in the process. Otherwise, after a while, you will have little, white, nasty specks of thermal grease everywhere.

This should be the last assembly step of the Apple-1 prior to first power up, which is done with all the IC sockets still empty. See post #1 of the following thread on tips and tricks how to proceed from this point:

https://www.applefritter.com/content/how-populate-empty-apple-1-ic-sockets

I was at Lowes today for some other things and looked that tape up.

3M SCOTCH MOUNT EXTREME 1" x 60" for $8.98

Lowes Item #394705

On the original pack no SKU to be found but a "414H" above the barcode.The barcode is 5114191976 (not certain, there may be hidden numbers like the 666 hidden in every barcode)

The good news is that it is "outdoors" and "indoors", so no nasty toxic agents, hopefully.

Sidenotes:

As a general rule, the outdoors version of any product tends to be much more robust against deterioration from bacteria and fungi, so I prefer those. These nasty germs eat everything. I wonder what happened to these rubber drive belts in cassette recorders.

The drive belt of my cassette recorder from 1968 still was good 40 years later, then the capstan drive rubber wheel had turned to goo. But all the drive belts of ALL the cassette recorders I bought to work on the improved ACI were deteriorated, even for a cassette recorder made in the early 2000s and still sealed in the factory plastic box. ALL the replacement drive belts I bought off Ebay from various vendors had microscopic cracks already which can't be seen with the naked eye. Useless. So I got stuck with this project.

Maybe these deluded tree huggers who also gave us the curse of lead-free solder and toilets that don't flush properly, and the showerheads that turn a 5 minute shower into 15 minutes to get the the soap out of the hair, etc., had some nasty, but effective poison banned which protected that 1960s rubber belts from being eaten by microbes. Who knows.

The same type of morons however seem to have nothing against laws and mandates to have all these nasty flame retardents in matrasses, furniture upholstery, bedding and pyjamas, especially for kids. Most of these agents are neurotoxic. Want to have some really autistic kids ? Put them to sleep in these toxic beds. It's not only the vaccines. And these flame retardents are in stuffed animals, too. And in plastic bags. Everywhere !

The 44-pin daughter card connector J3 on the original Apple-1 had no "ears", but "earless" connectors of the correct green color are hard to find nowadays, which is 45+ years later. Those with "ears" still are easy to find. There is a technical reason: the electronics industry quickly learned to avoid the "earless" connectors, same as they learned to avoid the unreliable TI made IC sockets that were used in the Apple-1 originals. Maybe all these much disliked components were already unwanted surplus even back in 1975, and so they were cheap and hence, attractive for the fledgling Apple Computer Company.

Blue, earless connectors like this one:

PDRM0285.JPG

are easier to find, and have been spotted in at least one Apple-1 original, so I'm not sure if they don't qualify for purists in their Apple-1 builds. Experts like Corey say that blue connectors in originals came from a replacement job done on the green connector, because the green one had failed. You can expect this fate for any earless connector sooner or later: after a certain number of insertion / removal cycles of the daughter card any earless connector will be on its last legs, and even the motherboard may have sustained some damage at that point. The only mitigation measure known to me (other than using a very, very thick PCB, which is no option for the Apple-1) is to use a connector with ears and bolt it firmly down on the motherboard, like this:

PDRM0750.JPG

... which involves drilling two holes into the motherboard. Yikes ! But if these screws are long enough to be fastened to a baseplate with two further nuts, then most of the forces from inserting and removing the daughtercard will be conveyed to the baseplate without going through the PCB, hereby avoiding damage to it. Did you ever observe how much the motherboard flexes and bends when the ACI is inserted or removed ? This is exactly the problem. If this is done only a few times in the life of the particular Apple-1, same as with PCI cards in PCs, it probably will be OK, but if the daughter card is changed often, such as in development work on daughter cards, the more robust mounting approach using a connector with ears and bolting it down, IMHO, is the more robust solution. Especially if the screws are finally bolted, via standoffs,  to a rigid base plate.

Otherwise, the "ears" must come off. The dreaded "ear job". But it must be done. Putting a connector with "ears" into an Apple-1 build, if the "ears" are not bolted down to prepare it for future hard work as a "lab rat", is considered inappropriate. It shows that the builder was too lazy or inept to to the "ear job", and this is no good sign as to the whole quality of the work.

To do a proper "ear job" you need a vise, a saw, and a file. But do not use small, tiny tools. Use a bigger saw with a blade made for cutting metal, like this one:

PDRM0766.JPG

The reason for using a bigger saw is that you can guide it much more precisely, using both hands, than is ever possible with any small saw, which wobbles around. The cut made with a bigger saw and proper technique yields an almost perfect cut that needs not much finishing work with the file. Of course, if you don't use a vise to hold the connector firmly in place, all bets are off. Use of thin, hard, cardboard (or some other suitable material that is not slippery) to protect the connector from being marred by the jaws of the vise is recommended. And don't overtighten the vise to avoid crushing the connector.

Using a blade made for cutting metal has the advantage that the teeth are finer, and so the resulting cut is much smoother than possible with a blade made for cutting wood, which typically have coarser teeth. Then, most of these connectors are made from glass fiber filled plastic, and a blade made for metal will not get blunt so quickly due to these glass fibers. But, be aware that these glass fibers are nasty, and will always take a toll on the saw blade. And on your health, if you inhale the dust, so take proper precautions to avoid inhaling it or getting it into your eyes (wear safety goggles and use a respirator with proper dust filter). Some experts say that glass fibers may be as nasty and dangerous as asbestos fibers, which are known to cause cancers, but so far the industry making glass fiber based products tries to downplay the risks (like the tobacco and the asbestos industry did for decades). Better be careful !

After the ears are sawn off, put the connector in the vise upright, and take a file (again, a bigger one) to it to smoothen the cut. The result should look like this:

PDRM0420.JPG

It takes some skill to get a good surface that looks as if the connector came out of the factory without ears. You may note some small imperfections being visible in the above photo. IMHO, these are inevitable. Because it's plastic.

I can do a perfect file job on metal, but on plastic things will happen. If you insist on perfection, better buy a dozen of these connectors and practice, practice, practice, until you eventually get a perfect one. I don't think you want to go to this extreme, but please don't accept a connector with an obviously botched "ear surgery". It can be spotted and once that botched connector is soldered in, your chances are slim to ever get it out again, unless you have professional desoldering equipment based on hollow tips and an electric vacuum pump. So you are better off using a connector with an "ear job" that has yielded an esthetically pleasing result. Maybe you buy two to have one spare, just if you botch the first one. I think if it looks like the one in the above photo, it's probably acceptable, although the example shown is not perfect.

So far my hints and tips about the 44-pin connector. If you are interested in a more detailed description on how to bolt down a connector with ears in the right way, ask me. I don't think many builders may want to go that route, but if you want to do it, to get the the most robust mechanical solution, it must be done right, and the process has many pitfalls to avoid.

Hi fans,

here is another treat for you: nice rubber feet for the Apple-1. I've been using them for more than 2 years now and they stood the "test of time".

This are the reasons why you want feet on your Apple-1:

a) with no feet, the IC socket pins and leads sticking out on the solder side  will scratch your working surface. The screws for the LM323K regulator tend to do the same type of damage. For a while I mitigated that by putting 6-32 nylon "acorn" / dome head nuts they sell for maintaining pinball machines on these screws. But then the Apple-1 motherboard is tilted from the working surface. Looks ugly and the PCB will get warped over time.

b) with no feet, any metallic debris on the working surface may cause shorts. Which is bad.

If you have kept an eye on the various photos of Apple-1 posted on the web, people who want to avoid the above perils typically resort to mount hexagonal standoffs as feet. If these are metallic, bad things can happen as the upper left screw hole (near the video connector) is connected to circuit ground (0V) and on the solder side there is the fat +5V trace very very close to that plated through screw hole. A metallic standoff may damage the solder stop mask there and make a short. This can happen at any time. Solder mask is not meant to take such an abuse. The danger can be avoided by using a nylon washer between the standoff and the PCB. But then you can use hexagonal standoffs made from nylon instead of metal. And on the long run, they will ruin the solder mask there, too. And they don't last for decades. And they look ugly. And they are slippery. And believe it or not, they have sharp edges which also can scratch your working surface.

So I have looked into rubber feet, but alas, I found none which could be screwed on and still have a diameter that is small enough not to protrude from the Apple-1 PCB boundary - this looks ugly, too. Self adhesive small rubber feet exist in various shapes but they tend to fall off sooner than you think.

The best candidate I have found are these. These are not meant to be feet but can be used as such:

Rubber_Feet.JPG

These actually are vintage automotive parts which are still sold by many vendors. To find them, search the web with the following keywords:

6-32 Rubber Rivets - Threaded Inserts - Well Nuts - 5/16" hole .496" tall

They are not exactly cheap. The best price I have found is Ebay seller "nelsonaudio". His price as of today is $9.99 for 10. So about $1 per foot. As an Apple-1 needs five such feet, you need to build two Apple-1 to justify this investment. I will gladly sell you an IC kit for yet another build - as long as supply lasts - if you send me a PM ;-)

You also need five 6-32 x 1/2" stainless steel screws with round heads and five #6 stainless steel flat washers.

When mounting these feet, make sure that the convex side of the washers mate with the top side of PCB. And screw the feet on with fingers only. You can tighten the screws later when all feet are installed and the Apple-1 sits on the working surface on its new feet. Be very careful not to slip with the screwdriver. And don't brutally tighten them. Here is how they look when mounted:

Rubber_feet_A1.JPG

These feet work well. The Apple-1 won't slip and you can also lean it upright against a wall for storage and these rubber feet prevent it from slipping.

The only drawback I have noticed is that the screws tend to get loose over time. I just re-tightened them as necessary. You could use Loctite blue  thread locking compound to avoid the screws to get loose, but this is messy and costs more money.

I think these feet also would work well to mount the Apple-1 on a rigid base plate (such as in an enclosure). The PCB will warm up and cool down and hereby expand and contract. The rubber feet will help mitigate these forces. Note that bolting down a large PCB in all four corners with no wiggle room spells trouble. This is why you can find card cages with rails and those wobbly nylon snap-in contraptions on stacked arcade machine boards. This looks like a cheap and crappy solution but actually is needed. If a stack of larger PCBs is tightly and brutally bolted together, all sorts of problems will develop due to thermal cycles and reliability will go to hell.