ZX81CCP video output for ZX81

[sirmorris]

Not tested yet, just soldered. I am too shy. Maybe something's wrong.

haha I understand this feeling very well

[PokeMon]

Well the good news, it is working and I get a good picture without adjusting the trimmers.
Unfortunately there is something wrong with the standby sync generator.
If I switch it off - everything is fine, if I switch it on with normal display after 60 seconds or so my screen loses picture.
If I turn it off via dip switch picture is coming back.

So 50% is working. And the signal from the drivers is perfect, a little bit to strong with 1.3 V from sync to white.
Maybe I have to adjust the two output resistors a little bit.

And my distributor is today able to ship the missing 74LVC1G386 (for the inverted picture).

[PokeMon]

Well now it's 90% working - just the XOR gate 74LVC1G386 missing but should be here monday or tuesday I think.

First I had problems with the timer NE556.
I changed to have the voltage of the hsync capacitor as reference for faster reaction when loosing sync instead of the vsync capacitor.
This has not been tested but thought this would be possible.
Well in fact I have the problem that during longer vsync the hsync pulse from the ZX81 is blocked and hsync comes from standby sync generator.
This could be solved easily through changing the components for hsync and vsync back.
So now the picture was stable even with standby sync active.

The next problem was the wrong timing of NE556 vsync, with about 15ms instead of 20ms.
I found some misterious swinging voltage when charging the capacitor for a period of several ms.
This could be solved with a power supply capacitor of 100uF at 5V power supply.
The NE556 is very sensitive for power supply ripple.
Didn't expect this. After all, it was running with the exact frequency and the swinging has gone.
I don't have a capacitor at the board, maybe could place a small one with about 4.7uF or so at power supply pins.
I have to check what capacitors are available not to big (0805 or 1206).

And I had problems with the 3.3V power delivered from the simple diode BAV199 for the LVL gates.
Had a significant ripple after vertical sync pulse, when drivers have to deliver current for the video output.
The LVL are very sensitive about power supply voltage ripple or noise.
Could stable this with a capacitor 100uF immediately with no ripple / swinging.
At the board I had place for a 100nF capacitor but has to be increased to a few uF I think.
I have to look for maybe same for the 5V.

For picture during LOAD I had to increase R11 from 22k to 39k.
Otherwise had some disturbing pulses at the output - not stable sync during LOAD.
With this change picture is perfect in normal mode and during LOAD.
I will add a video of it when finished with XOR gate.
Nice to see.

And I corrected the output resistors.
The driver output of the low voltage gates (LVL) are very good and impressive in comparison to general TTL logic.
Nearly no voltage loss at high output (when delivering about 10mA) and nearly no voltage loss when sourcing down to GND.
So high output is somewhere of 3.1V and low output is about 0.1V (with load !) with a power supply of 3.3V.
I changed R9 to 390R instead of 220R.
Output now is at 0.1V for sync level, 0.4V for black level and 1.1V for white level. Perfect.

Picture looks very good, very stable, very sharp.
I think better than my first circuit (ZX81CCP).
And power consumption is about 7mA with no load and about 16mA with 75R load at output.
Could increase max. 1mA with the missing XOR gate.

All changes are made now in the schematic.
Fortunately no changes for the board.

ZX81SCP_sch.gif (20.84 KiB) Viewed 6967 times

[Rink]

Cor. Debugging video circuits looks a lot like hard work.

[PokeMon]

Well - there is always a difference from a prototype to the produced circuit.
It took several hours to measure out all problems and find the reasons.
But that's fun too, especially when you find a way to get it working.

I always place a usual power supply capacitor of 100uF on my developement boards as there is plenty of space for it.
The behaviour of the timer with spikes on power supply is very sensitive - I didn't expect that.
It seems that the trigger circuit is not a good trigger when producing a fast swing on rising voltage when loading.

This is gained internal and not just the ripple. From logic when it switches from discharging to loading it should
discharge again only when capacitor is loaded with 2/3 of Vcc. But it is pending than very fast between loading and
unloading with a frequency of about 100kHz or more. This is not what I understand as a trigger. The trigger gap
voltage is 1/3 Vcc and the swing is about 1/6 Vcc but never reach the 2/3 voltage. But I am not very convinced
of this cheap universal timer, called often a precision timer from some manufacturers. The internal reaction
time of 0.5 us for switching off the discharger is something I would expect for DTL or DDL.
I wonder how difficult to build this timing when copying this circuit from other manufacturers.
As example, the 74LVC1G07 open collector could be used as discharger (and is used) and is with 4ns typical
switch on/off time about 100 times faster.

I didn't test the swap of hsync and vsync timer, this was just an idea. But this was uncritical because you can change
this any time while swapping the components when soldering.

The problem with the low voltage logic (LVL) in comparison to standard TTL with sensitivity about spikes surprised me.
The gates have a power supply current of approx. 200 - 500uA and I thought a 100nF capacitance was enough.
Of course the diode BAV199 plays in this game too and it came to swinging when continue with high output voltage after
long vsync pulse. Couldn't measure anything for hsync. Maybe not important as there is a picture on the screen but not
nice and unwanted in my scope picture.

I didn't use LVL on my prototype board as these are available in SMD only and difficult to solder. So I used TTL logic for
test and in reality used the LVL. All data only read from datasheets. That gave the changes for some resistors as the
reality is a bit different from datasheet, especially when only min and max values are given and the typical have to be
guessed.

Anyway was fun to develop this board - otherwise wouldn't do that.

[1024MAK]

555 timers are known for causing problems with the supply (same for 556 chips). Normally it is recommended to have a capacitor close to the IC, the minimum value being at least double that of the timing capacitors.

Mark

[1024MAK]

555 and 556 timers are now very old designs.
It would be interesting to see how well a slightly more modern CMOS version would perform. The snag being, the value of the timing components may need to change.

555 and 556 timers can perform a bit better if you connect a small value capacitor to the CV pin (pins 3 and 11 on the 556). By small I mean 10nF to 100nF.
This decouples the internal reference voltage divider.

And yes, 555 and 556 timers do not have good noise immunity when operating from a low supply voltage.

Good that you are making progress. Keep at it

I'm looking forward to the final version

Mark

[PokeMon]

I will add two 10uF ceramic capacitor in 0805 case for either 5V and 3.3V.
I don't want to change the board nor want to solder some wired components on it.
it's about factor 100 more than the 100nF capacitors and should do the job.
I hope so.

Even it would be possible to add a bigger capacitor at supply cable.

[PokeMon]

Now the board is finished, tested and working 100%. The 74LVC1G386 arrived finally.
The picture quality is awesome, very sharp and video out signal is a bit more perfect than the previously designed ZX81CCP and it restists any shortening at the output, which delivers exactly 1Vss at 75R. 0.7V picture level (black/white) and 0.3V sync level. The picture looks very perfect either on scope or an a real modern Plasma or LCD TV.

The board is supplied with 5V from ZX81 board and takes only 7mA without load or 16mA with load of 75R when connected to a video input. This is quite less than my first board (ZX81CCP) thank to the LVL gates with less power comsumption and with more functionality. I think the old board could take up to 60-70 mA. Additional you need only signal from pin 16 of ULA.

The board ZX81SCP has now a standby sync generator which keeps valid TV signal in FAST mode (of course with empty screen) and during LOAD when awaiting audio input at EAR and shows the well known stripes at a modern Plasma or LCD. I think this is a novum for all new TVs. The standby sync circuit prevents modern TVs from automatic switching off or switching to another source or channel when ZX81 goes to FAST mode or enters the LOAD command.

The board supports an invertation of the picture (white letters on black background) and inverts the picture in FAST mode too (black screen but with sync). The backporch is added for a picture with good contrast regardless if delivered from ULA or not and regardless of old (C184) or new (C210) ULA. The board can be adjusted via two small trimmers, one for sync and one for pixel voltage.

Function of DIP switch:
DIP-1 (oben) ON=normal picture OFF=inverted picture
DIP-2 (unten) ON=standby sync switched off, OFF=standby sync active

The board hat 29 x 21 x 5 mm (1.15 x 0.85 x 0.2 inches) and could be placed in the modulator case. It's designed as single layered SMD board and components only at one side, the other can be fixed with a tape somewhere in the case of ZX81. No high temperature due to low power consumption with 80mW (5V / 16mA).


here to see dimensions in comparison to ZX81 case:

IMG_4987.gif (142.59 KiB) Viewed 6749 times

here scaled up a bit more in comparison to ZX81 logo:

IMG_4988.gif (103.96 KiB) Viewed 6749 times

and here the board itself
(dip switch a little bit scratched from screw driver while testing)

IMG_4991.gif (161.73 KiB) Viewed 6749 times

The trimmers can be adjusted with a small screw driver with 2mm. I propose an isolated screw driver as a metal one affects picture when holding (human body). Could be used but maybe have to readjust a little bit.

[PokeMon]

Here some pictures form ZX81CCP, picture quality is same or let's say better with the new ZX81SCP.



picture via the RF modulator as known with poor quality, missing sharpness and contrast





picture with the new video board ZX81SCP





and inverted picture (via dip switch selected)





and zoomed for better distinction - the bad F signal:





the new signal with ZX81SCP, sharp and good contrast





and inverted by the video board