The Load Capacitors For The Crystal Are Size Incorrectly.

[dennowiggle]

Experiment:
I took some measurements of GROMCLK on my recently built V1.0 2068-Z80-TMS9118 VDP board. GROMCLK runs at OSC/24 so is a good way to measure the crystal frequency without loading the crystal with the capacitance of a probe.

This is the specification taken from the datasheet I have, section 5.4 "Timing Requirements Over Full Ranges of Recommended Operating Conditions", specifically Fext.

Specification:
Target Frequency GROMCLK PPM Spec
10,738,635.00...... 447,443.13..............0
10,738,098.00...... 447,420.75.........-50
10,739,172.00.......447,465.50.......... 50

Setup:

• For measurements I used a precision frequency counter that has worked reliably for many years.
• The precision frequency counter was powered for a minimum of 30mins before measurements for accuracy.
• The crystal used was the recommended 30ppm one from Digikey.
• The Retro Z80 program GREEN was run and a green screen was shown on the monitor.
• The board was left to run until it warmed up.
• Frequency was measured at GROMCLK and then multiplied by 24 to calculate the crystal frequency.

Test Results:
Here are the measurements indicating that 43pF is nominal for the two load capacitors on my board.

Capacitance......GROMCLK .........OSC Frequency...........PPM
27pf....................447,482.10..........10,739,570.40...............87
33pf....................447,466.70..........10,739,200.80...............53
43pf.....47..........447,436.00..........10,738,464.00.............-16
47pf................... 447,435.40..........10,738,449.60.............-17
39pf....................447,451.50..........10,738,836.00...............19
39+4pf..............447,441.90..........10,738,605.60................-3
Note : "43pF 47" indicates that while a 43pF was used it in fact must have been 47pF. Problem with kit, Ugh.

Check:
The schematic for the NABU is available and it uses a similar VDP IC. That schematic shows the use of 33pF and 47pF load capacitors with an additional 4-40pF variable capacitor for tuning in parallel to the 33pF. From this I guess 47pF must have been the initial target.

Conclusion :
Change C4, C5 to 43pF.

Ideal Situation:
Would be nice to test a crystal that has been calibrated at the factory with data provided and match the load capacitors to that. Probably not gonna happen ;)

[dennowiggle]

We could also play some guesstimates to check another way:
The Crystal spec from Digikey shows the Load Capacitance of crystal MP107-E = 32pF.

So you could guesstimate C4,C5 from this equation to make sure we are in the right ballpark:
C4(or C5) + C(holes) + C(traces) + C(VDP Cpin =10pF max) = 2 x (32pF - C(VDP Cpin-pin) )
43 + 2 + 4 + 5 = 2 x (32 - 5)
54 = 54

Or

47 + 2 + 4 + 5 = 2 x (32 - 3)
58 = 58

Disclaimer = guesstimates are made up numbers

[johnwinans]

43pf? I am ordering some parts today, I'll get some and see if the color improves.