3 posts / 0 new
Last post
Last seen: 10 hours 6 min ago
Joined: Apr 1 2020 - 16:46
Posts: 113

After having fixed the nastier quirks of the Apple-1, which did cause it to crash all too often, your "Uncle Bernie" proceeded to fix some of the less pressing issues, such as the video signal, which in the original design is wrong, as it does not conform to the NTSC standard, and makes finding a monitor that works with the Apple-1 more difficult than we want it to be.


As with the other quirks of the Apple-1, this video signal issue did not deter the two Steves from selling their cute novelty, because it almost worked, or it worked most of the time, and so none of the many quirks it had - like dogs have flea - was a "show stopper".


We can't demand any perfection from any novelty product, and the Apple-1 indeed was the world's first microprocessor and DRAM based single-board computer complete with character video output and keyboard entry, and a BASIC interpreter. Nobody else had that in 1975 !


To be more precise about the timeline, in late 1975 the handwired Apple-1 prototype was shown, only a few months after the 6502 became available in mid 1975 (with the ROR bug), but since then, the race was on ! The two Steves had no time to lose and when the PCB based Apple-1 was for sale in mid 1976, they still were a bit ahead of the competition. The history is full of examples where inventors / innovators were still perfecting their product, losing time, while the competition already got rich with inferior, less perfect products of the same kind. This perfectionism for novelties is a trap that needs to be avoided. So the two Steves did the right thing. The industry calls this "good enough": if the customers buy a product and don't get furious enough to make death threats, or to become members of a class action lawsuit, then the product is "good enough" to be sold. Another example for this mindset: American made cars. Juuuust "good enough". If you want more perfection, buy a Japanese or German made car, but keep in mind that inferior parts "Made in China" began to appear in their supply chains, too. 


As to the Apple-1's video signal being "good enough", most of the primitive B&W monitors or modified TV sets from back in the day could take the abusive, non-standard video signal produced by the Apple-1 without acting up too much. On my old Shibaden Model VM-903U monitor, probably made in the late 1960s, the only ill effect caused by the non-standard video signal was a character field sitting too far to the right, so the last column of characters was outside the visible area, until I adjusted the magnet rings on the CRT to move the scan field off center. This operation on a monitor under power certainly is not exactly "consumer friendly", but back in the day, almost anyone who ventured into these first microcomputers was an electronics engineer or such, and could do this (or other) mods. Some daring individuals not afraid of death even wired a video input into their hot-chassis type TV ! (Don't do this, it may turn your Apple-1 into an electrocution machine that can kill anyone who touches it). These were the wild times when even 12 year old kids could play with matches and gunpowder without being arrested as "terrorists". Today's "laws and regulations" would have killed the fledgling hobbyist computer "industry" on the spot. Remember how the FCC nearly succeeded with that in the late 1970s, once they woke up and passed ridicolous regulations which forced Apple to redesign the Apple-II, and forced Atari to put several pounds of die-cast EMI radiation shields into their Atari 400 and 800, greatly increasing costs ? A few years after that the FCC had to roll back their excessive regulations because they were crippling the whole computer industry. Then, all of a sudden, cardboard shields with thin aluminum foil were "good enough" ! Government at its best ! But back to the Apple-1:


The root cause for the character field not being centered is that the Apple-1's HSYNC pulse is 10us wide, and not 4.7us as specified for NTSC (and PAL), and the unwanted additonal 5.3us come too early after the character field. If you study the Apple-1 video circuitry long enough you will find out why Woz made this compromise: the wicked way the horizontal counters work does not offer any low gate count way to make a HSYNC conforming to the NTSC standard by digital means. Doing it right might have cost at least one more TTL package. Typical analog, CRT based TVs or monitors of the time could digest the non-standard Apple-1 video without throwing up, except for a misplaced character field in some cases.

But this more modern little B&W TV I bought off Ebay, used, for cheap, could not take the Apple-1 video signal without refusing to work properly:



It's a "Personal Black & White TV with AM/FM radio", Made in China (what else did you expect for the $39.99 price tag it once had ?), from an unknown manufacturer who, despite being Chinese, was too ashamed to put his name anywhere on their not-so-fine "product". Maybe they wanted to avoid death threats from desperate customers, who knows. (You might notice I have a lot of dark humor ... this inevitably comes with age, when you witnessed 50+ years of mankind's devolution and dumbing down).


As a cheap disposable TV and Radio, it may be "good enough", although my expert eye trained by many years of designing ICs for TVs (until the very last CRT based generation) shows me that it's guts are crap, and not due to age, but by design. Buyer beware, you may get what you pay for, and never expect to get more value than you pay. These little TVs are just too cheaply made and for a little more money you could get a much better one. But in this case, its refusal to display the Apple-1 video is justified, and not due to it being too cheap. Many more modern TVs and monitors, regardless of price class, would show similar symptoms.


The reason for the distortion on the first few scan lines is that the Apple-1 does not produce a serrated VSYNC as required by the standard, which would have cost one XOR gate with two inputs. Without a serrated VSYNC, the horizontal oscillator (or its digital equivalent) may lose synchronisation during the vertical sync period, and then it may not be able to regain synchronisation fast enough to be able to show a good picture. On some TVs / monitors, you may get a mostly stable picture, but with some ugly distortions of the beginnings of the first few live video lines, which shows the lock-in process for the horizontal oscillator. On other TVs / monitors, the horizontal oscillator never acquires lock, due to the non-standard video signal, and the whole picture falls apart (or "crawls" around).


Now, the Apple-1 has no XOR gate anywhere in its circuitry (except within the 6502, and those we can't use to fix the video). Also, like with all of my mods to fix the Apple-1, I did not want to add ICs nor cut any PCB traces on the motherboard. These requirements come from my desire that my mods also should be acceptable for Apple-1 originals, and removable by just de-soldering the added components, without leaving any trace.


Within these requirements, alas, even I can't find a perfect fix for the video problem. But I have developed a mod which both adds the serrations in the VSYNC and shortens the HSYNC to conform better to the standard and centers the character field on the screen.


Here is the schematic of the video fix mod:


The five parts for this mod cost 10 cents, and if you paid 25 cents (US "quarter") for them, you got ripped off. All you need is a PNP transistor, a 47pF capacitor, and three resistors, of the values 3.3K, 5.1K, and 100K. Any small-signal PNP with a hfe better than 125 will do (and this means, almost anyone, TUP = Transistor, Universal, PNP). The tolerances on the timing components (47pF and 5.1K) should be tight (1-2%). The 1% tolerance on the 3.3K resistor is to avoid adding yet another widely moving target to a node where other components (the R1, R2) already have introduced large tolerances. The tight 1% spec can't cure these, but it will reduce the probability of the mod not meeting target criteria. If you seek perfection, you could replace this 3.3K resistor with a 5K trim pot and adjust it until an oscilloscope shows the same voltage level just before the (shortened) HSYNC as the black level voltage in the live video portion. Then put in a fixed resistor of a close enough value. But I don't think such a manual adjustment is necessary, as typical TVs and monitors are quite tolerant to what happens just before the HSYNC, and if there is a little step there to either side of the black level, nothing bad will happen. If the step is upwards, there may be a faint grey bar outside of the visible portion of the screen, and if the step is downwards ("blacker-than-black"), you can't see that anyways.


This is how the mod looks installed in one of my Apple-1 built on the "Replica-1" PCBs sold on Ebay by "Pinguso". Other PCBs not based on the open-source Gerbers may look slightly different, but the general location of the solder pads and traces should be the same:



The yellow cylindrical capacitors belong to the reliability mods, and not to the video mod.


Here is a closeup of how the five added components are soldered in place:



Here is a screenshot from the same little TV as above, after the video mod was put into the Apple-1:




The picture is now stable, and the character field is centered in a better way.


All this was achieved by adding just one transistor and four passive components costing maybe 2 pennies each. Woz could have designed in those 10 cents worth of components to have a less troublesome video signal closer to the standard. Instead, they recommended a certain type and brand of TV / monitor which worked better with the non-standard signal than others, or so the story often is told. Sigh ! ~45 years too late, I have added my proverbial "10 cents worth of advice" !   


Those who are younger and have better eyes than me may want to try to implement this mod using SMD components. These "bird feed" parts are so small nowadays that the mod would be barely noticable. But in any case, make sure not to allow any leads to touch PCB traces, despite the traces have solder mask. Solder mask is thin and no reliable insulator. You don't want to produce a hard-to-spot short below a component.


Here is an oscilloscope screenshot showing the now serrated VSYNC:



The red circle is at the end of the VSYNC and the "needles" seen there are the serrations added by the mod, which are short positive pulses at the HSYNC rate, and these keep the horizontal oscillator in the TV or monitor synchronized. The original Apple-1 VSYNC lacks these serrations, which makes many TVs or monitors unhappy.


And here is the now somewhat shortened HSYNC:


... which now is ~6us wide, still wider than the standard, but I found this may be the best compromise of HSYNC width and the position of the character field on the screen (if you want to experiment with that, just vary the value of the 5.1K resistor).


Note that the original Apple-1 HSYNC started at the time of the narrow glitch pointed to by the red arrow. This glitch is produced by the added circuitry, because the PNP starts to inject current into the video summing node before the combined sync signal has traveled though a D-Flipflop within the 74175 at location C13, which is clocked at 14.31818 MHz, so the glitch is quite narrow, and hence, would produce a thin, vertical, grey line on the far right side of the TV screen if this place would not be located in the "overscan" region of a typical TV or monitor, where it is invisible on the screen. If you can see it, your TV / monitor is out of adjustment. There is no way to avoid this glitch within the limitations of this super simple five components analog circuit. "Good enough" as discussed above. Don't make death threats against me and don't start a class action law suit ;-)


I really like this little circuit, as it proves (again) that analog designers can do a lot of things with just one transistor for which digital designers need many gates and many dozens of transistors. Remember the three transistor walkie talkie of the 1960s ? Compare that to the gazillion of transistors in your cellphone. I think it's called "technical progress". It took us half a century to get there.


Comments invited !  - - -

Reply to this post and tell us your results on which type / brand of monitor before and after putting in this fix !

macnoyd's picture
Last seen: 3 hours 48 min ago
Joined: Oct 15 2012 - 08:59
Posts: 668
Well done UncleBernie!

Well done UncleBernie! ... And thanks for sharing.

Last seen: 1 month 3 weeks ago
Joined: Oct 3 2014 - 22:06
Posts: 70
Serrated vertical pulse

Bernie thanks sharing the details of your clever serrated vertical sync fix.

It is nice to see people that understand analog video are still out there :)

I am lucky I did never encounter this problem when I use  an Apple II monitor.

When I built my first Apple I about six years, I noticed Woz took some liberaties with the video standards.

1) The horizontal sync rate matches the  NTSC color standard rate of 15734.262 hz.

2) The vertical sync rate is slightly off at 60.05 hz.  NTSC color standard vertical sync is 59.94 hz and NTSC BW standard vertical sync is exactly 60 hz.This probably explains why early Apple monochrome monitors (i.e. Apple II) had external controls for vertical hold, but not for horizontal hold.

3) NTSC is interlaced video, while the output of the Apple I is not. 

4) Vertical pedestal was not serrated with horizontal pulses as you mentioned.



Log in or register to post comments