I recently bought a loose MITS Altair 8800 CPU board plus a couple of other related boards. These boards, when bought in a kit, were assembled by the purchaser. These boards would have been assembled in 1975 and I have never seen this type of assembly. Was this done at the time or a flourish? I've never seen this before and I have lots of vintage computer equipment of the era. Is there any advantage or disadvantage of doing this? I am guessing it's not necessarily a good thing. This was done with diodes on two boards.
Only reason I can think of is that they didn't want to risk cracking the glass body *shrug*
No, it was not common. I do not remember ever seeing that except on prototypes.
Preventing cracking of the glass package is what occurs to me also; I had that happen before I made it a habit of using a lead bender.
I have occasionally done the same thing though to make test points.
I suppose it could be also to make leads longer and let them act as a heatsink.
Also thought maybe they used some plastic or wood something to keep diodes in place by having some stick or needle through the loops to keep diodes at certain height, while circuit board goes through wave soldering machine.... but there are lead forming machine which can form leads to keep diodes at certain height.
Another thought would be those tiny loops will act as a couple of very tiny inductors reducing component count.
Making the leads shorter not longer would help the diode cooling since the PCB would act as a better heat sink than the excessively long leads. I built a number of kits from MITS back in the day and also owned two Altairs, one factory assembled, but I've never seen the leads looped like that. Did they put loops in any other leads than the diodes?
Did they put loops in any other leads than the diodes?
I don't have enough light to take pics but in a word, "no". Just diodes. This was done the the Alir 8800 CPU board and the Altair RAM board. There is a third Tarbell Cassette Interface board but it doesn't have any diodes so no loops.
I suppose it could be also to make leads longer and let them act as a heatsink.
The leads need to be as short as possible to lower the thermal resistance to the printed circuit board.
Another thought would be those tiny loops will act as a couple of very tiny inductors reducing component count.
On a slow 1 watt zener diode? Not a chance.
Being scared to break the diodes was on my mind as well.
It still crosses my mind with 1N4148.
I suppose it could be also to make leads longer and let them act as a heatsink.
The leads need to be as short as possible to lower the thermal resistance to the printed circuit board.
Another thought would be those tiny loops will act as a couple of very tiny inductors reducing component count.
On a slow 1 watt zener diode? Not a chance.
I've seen it done on 5 & 10 watt resistors on PCBs .
These devices, if mounted close to the board will "cook" it, as they are designed to dissipate heat in free air.
The PCB looks to have a greater surface area, but the heat is closer to the solder joints which will degrade over time.
Eventually a point of "runaway" is reached, ultimately causing delamination of the board.
This isn't just theory, I have had to repair many such "cooked" PCBs over the years.
The loops radiate heat into free air, & do help, but the biggest help is to get the big hot devices clear of the board.
Used on diodes, it was probably due to concern about cracking the bond between the lead & the glass envelope.
I remember being shown this technique in trade school.
We were told it was used for strain relief, so if the board flexed due to vibration or g-loading, the leads wouldn't break so easily (Instead of the bending being confined to a 90deg bend, it is spread over the 360deg loop) or pull on the component and cause damage.
It reminds me a lot of the old looped brass petrol tubes on ancient motorbikes. Those were intended to mitigate vibration damage too.
I've seen it done on 5 & 10 watt resistors on PCBs .
These devices, if mounted close to the board will "cook" it, as they are designed to dissipate heat in free air.
The PCB looks to have a greater surface area, but the heat is closer to the solder joints which will degrade over time.
Eventually a point of "runaway" is reached, ultimately causing delamination of the board.
This isn't just theory, I have had to repair many such "cooked" PCBs over the years.
The loops radiate heat into free air, & do help, but the biggest help is to get the big hot devices clear of the board.
I have run across that many times with power resistors but diodes should not be operating hot enough to scorch a printed circuit board.
Unlike resistors which can operate reliability at higher temperatures, axial power diodes depend on conduction through their thick copper leads into the printed circuit board for power dissipation. It almost never matter but for where it does, there are application notes discussing junction-to-ambient thermal resistance of diodes based on lead length and copper pad size.
I've seen boards like that where all the signals you are likely to probe with a scope have been looped to make it easy to attach a probe clip.
Here's a part of the second board. A MITS Altair memory board. Same thing. Diodes only. Three are shown with a red arrow.
I vote for heat dissipation.
Did those older boards even have the massive ground planes that more modern devices have? As I recall, mounting parts such as diodes and resistors away from the board to improve air flow and heat radiating ability was common practice in those times and a reason for those parts to have heavier leads in the first place.
Style and convenient attachment points for probes would be a side benefit, but I doubt that was a primary purpose.
I built and debugged an Altair 8800 in 1976 and remember the various boards but not having to loop the leads like that. I do something like that when prototyping circuits so I can probe, trim or snip more easily but I've never seen any lead kinks on production equipment except for power components.
BTW, I don't know if the CPU board will work without a front panel board as a few signals are sourced there. Certainly the front panel is needed to write to memory unless you can find a PROM board to get some sort of monitor running. It took about four months to find a defective chip that prevented the CPU from reading the front panel switches. The assembly instructions were OK, but the test instructions and circuit description were, um, lacking. I learned a lot though.
Many people had trouble with the original 4K DRAM board. The layout wasn't so good (Ground plane? We don't need no stinking ground plane!) Event timing was determined by one-shots (monostable multivibrators) and wasn't stable. MITS came out with a synchronous version of the board later on so they must have been aware of problems.
Cheers,
I see people have mentioned heat dissipation, and mechanical stress relief, but it is my understanding that these loops are basically for a combination of these two effects: reduction of stress from thermal expansion.
Imagine if the circuit is switched on, the diode starts dissipating power, and warms up. Due to thermal expansion, the diode and the horizontal portions of the leads will try to elongate. However, they cannot, as the ends are anchored to the PCB. The PCB will be at a lower temperature, so the pads will still be the original distance apart. The difference in length must be taken up by bending of the diode leads, the solder joints, and/or the PCB. Repeated temperature cycling caused by powerup/powerdown cycles may eventually result in a fatigue failure.
With the loops in place, any elongation of the diode simply results in the loops getting slightly larger, and the resulting mechanical stresses are considerably reduced.
That is the theory, anyway; I don't know of any quantitative data on the generated stresses. The thermal expansion coefficient of FR4 is likely significantly different to that of the diode, so it may be that, depending on temperature gradients, the PCB actualy grows more than the diode. The loop should work in either case though.
Making the leads shorter not longer would help the diode cooling since the PCB would act as a better heat sink than the excessively long leads.
on a modern multilayer pcb with solid planes sure, on a single/double layer pcb with just traces not so much
I've seen such kind of diode assembly in a Datron 1061 DMM. It is used on the reference diodes to reduce mechanical stress on them. This was for sure done to keep them from drifting due to mechanical / thermal stress, but I don't see a reason to do so on computer boards or vanilla zener diodes.
This was done to keep the electrons happy. Kind of like a Six Flags amusment park ride.
It's used for the same reason that 90° corner traces in high speed PCBs are forbidden.
The electrons would fly off the sharp corner of the bend!
It's used for the same reason that 90° corner traces in high speed PCBs are forbidden.
The electrons would fly off the sharp corner of the bend!
And here in Switzerland flying off the bend could mean flying off the side of a mountain!!!
This thing you ask about is a method mostly used in flight HW and some other high-reliability systems assembly back in the late 70's through the advent of SMT devices. SMT is so much more reliable than through-hole that all the flight or life critical kind of systems went to that as soon as they could, so this kind of thing died out much more quickly than did the "rest" of the through-hole PCB assembly market.
It was indeed (as pointed about above by someone who said they saw it in tech-school) for strain-relief. Flexing of the board, thermal expansion, etc. was prone to cracking of the leads for resistors (esp. power resistors), diodes, and other axial lead through-hole components.
You see this a lot in 70's-80's era test equipment (esp. RF test equipment) from quality manufactures like HP. It came about during space-program testing, and from field repair learnings from early jet fighters, etc. It was a mil. spec. when I first learned of it. I used to have a copy of that mil. spec. (on actual slaughtered tree media, not electronic). :-)
Gee. Does that date me? Ha ha.
Well if not, the fact that the first computer I used (and coincidentally the first one I owned) was a Kaypro-II made by Non-Linear Systems. Precursor to the modern lap-top. It was "portable" in that the keyboard snapped onto the machine's front-panel and formed the base when in transit (there was a handle on the back) and that covered the floppy drives and the 9" green phosphor mono-chrome monitor to protect those and the keyboard while lugging the 25 lb. boat-anchor around with you. :-) (It had all the RAM one could address in a CP/M OS machine running on the 8-bit Zilog Z-80 CPU... 64KB That's right KB, two digits. :-) And folks thought 640K was all the RAM you'd ever need. That's because only a few years before, all they could get was 64K at a time. There were machines that had 128K, but they implemented a clunky form of paging and could only address the upper or lower half at a given time.
https://en.wikipedia.org/wiki/KayproEnjoy
This thing you ask about is a method mostly used in flight HW and some other high-reliability systems assembly back in the late 70's through the advent of SMT devices. SMT is so much more reliable than through-hole that all the flight or life critical kind of systems went to that as soon as they could, so this kind of thing died out much more quickly than did the "rest" of the through-hole PCB assembly market.
Early SMT device had a terrible record for reliability, and most people needing high reliability stayed away for several years. The exception was the military, who sponsored most of the early development of SMT. They lived with the reliability issues, worked on them, and things matured. The key issue was differential thermal expansion rates, for which there were some bizarre short term solutions, like multi-layer ceramic PCBs. These were made from the same ceramic as high reliability IC packages, which avoided differential expansion issues at the cost of a ludicrously brittle board. Fun days.