HP-65 1333A02242 display missing all the decimal dots

[teenix]

(11-09-2020 09:49 PM)albertofenini Wrote:  We have also uploaded the waves for disp0, disp1, disp2, disp3, disp4 and step at the anode driver ...

The waveform for your "step" looks like it is the "E" waveform on my diagram. The others look ok.

I would expect the step trace to show 14 pulses spaced 20uS apart and one spaced 5uS apart, all over a 280uS period. These pulses step the cathode driver to the next digit on the display.

Interesting stuff follows - maybe not :-)

The cathode driver is essentially a 15 stage counter which activates each output in sequence with the step pulse. Normally these type of counters would reset to the first output after the maximum pulse count is received and scan again, but this approach will only work in the Classics if the anode driver was synchronized to this event, but it isn't.

It is the job of the cathode driver RCD input to reset the counter back to the initial output. This pulse comes from the ARC chip and is synchronized to the 5 display data pins between the ARC and anode driver and in this way, each digit will receive the correct segment information.

It takes the CPU 280uS (nominal) to execute 1 instruction and the 15 display digits are fully updated during this time with 20uS available to update each display digit. Timing is critical with this setup as the display could show garbage or information could bleed into following digits, or the driver transistors in the IC's could be damaged.

Now those with a mathematical mind might say something is wrong here.
15 digits x 20uS = 300uS, not 280uS.

The answer lies with the final decoding circuits inside the anode driver. When a decimal point is to be displayed, the anode driver issues an extra step pulse and excites the decimal point coil in the last 5uS of the 20uS time slot. The decimal point is small compared to the other LED segments, so it requires less energy to appear at the same brightness. The coil for the decimal point LED is energized and disharges over a smaller time and also has about half the inductance as the others (130uH vs 68uH).

The display 7 segments are controlled by the data in the A register while digit blanking and the decimal point are controlled by the data in the B register. Each register has 56 bits split up into 14 x 4 bit nibbles. These can hold binary values from b0000 to b1111 (0 - 15) with bit 3 on the left down to bit 0 on the right.

The ARC normally sends the required 5 bit data to tell the anode driver to step the cathode driver and display the digit from the corresponding A register nibble.

If the B register nibble [bit 1] is set to 1 (ie. 0010) then extra data is encoded into the 5 bits so that after the digit is displayed, the anode driver steps the cathode driver again and lights the decimal point. Any number in a B register nibble with bit 1 set would work, ie 2 3 6 etc., however bit 3 cannot be 1 or no LED segments in the digit will light.

Because multiple decimal points can be displayed, the final digit count could well exceed the 15 digits available. If the cathode driver reset itself after the 15th digit, these decimal points would cause other digits to be displayed starting from the beginning of the display scan resulting in garbage being shown.

The logic in the cathode driver stops this happening by not resetting itself if more than 15 step pulses are received before a reset pulse.

This is shown in the Classic Notes on page 12 which has a HP-65 running a small program that places some 2's in the B register which then shows multiple decimal points while the display is active during running programs. You can see all the double stepping occurring in the oscilloscope image.

cheers

Tony