How to get the Apple-1 crystal oscillator on target frequency !

While developing a color graphics card for the Apple-1, I discovered that the Apple-1 crystal oscillator typically is greatly off target, meaning that it does not run with the 14.31818  MHz frequency stamped on the crystal.

This new thread is meant to show you how to get the Apple-1 crystal oscillator on target frequency, and it is not meant to start yet another discussion as to the reasons why the oscillator circuit in the Apple-1 fails to run at the proper frequency regardless which crystal is used (both "series mode" and "parallel mode" crystals run off target, the latter even worse, but even the former, which under ideal conditions should be on target frequency, are still off target by far too much, as evidenced by peo2000's post #9 in this thread:

https://www.applefritter.com/content/wanted-apple-1-owners-frequency-counter

So I spent a few days in the lab to work out remedies. Since I'm still shipping Apple-1 kits (they are slowly running out) I had to find a solution quick. Here it is: all my kits shipped with the Abracon crystal from this day on will have a small ceramic trim capacitor in them which can be soldered into the Apple-1 in lieu of the C4 capacitor.

This is how it looks:

PDRM3674.JPG

It does not look too offensive, I think. The only problem is with all PCBs based on the "Russian Gerbers", the through holes are a bit too small for the leads of this trim capacitor. So you either have to drill a slightly larger hole for both leads of the "C4" and then solder both on top and bottom side of the PCB (as the through-hole connection gets lost in the process), or you use a Dremel tool with a diamond grinding tip to take away some material from the leads. "Newton" PCBs do not have this problem, as their through holes are large enough.

The frequency trim range for this trim capacitor (2pF ... 34pF) is large enough to get the "Abracon" crystal on target. Here are the extreme ends of the trim range (measurement is done after the first divider stage, on pin #2 of the 74161 at location D11 - this is the "DOT CLOCK" in the "Terminal Section" schematic:

OSC_Fmin.JPG

The above picture shows the minimum trimmable frequency: 7.156417 Mhz (your mileage may vary). Here is the maximum frequency at the other end of the trim range ( = 7.163402 Mhz):

OSC_Fmax.JPG

At about the half setting of the trim capacitor, the frequency is on target (ideally, 7.1590909 Mhz, but don't expect to ever get that exactly right):

OSC_Fntsc.JPG

In the above picture, the oscillator runs 20 Hz too slow, which is OK. If you don't have a frequency counter, just set the trim capacitor to this position:

PDRM3675.JPG

Once you have one of my color graphics cards (don't hold your breath though, it's still in development) you can always fine tune this trim capacitor for best results. Do not worry about the "color subcarrier beat frequency" issue brought up by CVT in his post #15 in the above thread. For healthy color TVs/monitors it does  not manifest itself, because they typically run their local crystal oscillator a little bit off the 3.57954545 MHz NTSC color subcarrier frequency specified by the standard (i.e. SMPTE 170M-1999), to avoid the beat effect.

Said standard calls for Fsc = 5 Mhz x (63/88) = 3.57954545... +/- 10 Hz, which the TV stations had to meet, otherwise they would be in trouble. But color TVs as such are much more forgiving and are willing to "eat" color burst frequencies which are greatly off target, but not too much, else they will turn off the color and revert to black-and-white. CVT took some measurements and posted his findings for various monitors and TVs in his post #13 of the beforementioned thread. Do not forget to divide his numbers by 2, so you can compare and see where you are. 

This is the remedy for the "Abracon" crystal in my kits. You can still buy them at Mouser, they are the "AB" type (Mouser # 815-AB-14.31818B2), although those in my kits were made 24 years ago. But I have measured the crystal parameters and I am confident this is the very same type of crystal with the same specified load capacitance of 18 pF. 

If you dislike the looks of the trim capacitor in the Apple-1, please tell me. There are other ways to get the frequency on target, but then you need to cut a trace on the solder side of the Apple-1 motherboard and solder in a small (0805) SMD capacitor that can barely be seen. Tell me if you would prefer this "invisible" solution over the trim capacitor - I am currently in the lab to work out that solution, too. But I think the most pragmatic solution is the trim capacitor.

Keep in mind you do not need to have the Apple-1 run at exactly the "right" frequency. It runs just fine with the standard C4 and any crystal. But if you ever want color graphics, or do exact timekeeping (i.e. a software time-of-day clock), then your oscillator must run on the proper frequency. And why not put that trim capacitor in while building your Apple-1, and not fix the problem later, which always involves risky surgery on the motherboard. ("Get it right the first time.")

Comments invited ! (But please avoid technical questions on the why, ask these on the other thread mentioned above. This new thread here is strictly meant to show how to fix it, and to discuss the "looks" being acceptable or not.)

- Uncle Bernie