ZX Spectrum Harlequin rev.H

Česky

28.07.2015 Tranlation with Google help - Sorry :-)

I have a certain idea, which I will try to meet.

The design will of course be based on the rev.G.

PCBs must be halved and be SMD.

Each IC should have its own suppression capacitor.

The possibility of supplementing -5V and +12V.

Cancel bridge rectifier.

Power connector will have a "plus" on the pin. It is a sin against the "norm" (original ZX Spectrum has a negative on pin)

but most power supplies (PS) in the market are so connected and most of them are switching PS, so that rectification is unnecessary.

Reverse polarity protection (what if someone used the original PS).

ZX-ROM will be in EEPROM and will be able to rewrite the program.

This is what the concept looks like.

Do not take inscriptions on 3D components seriously.

And this is what the prototype PCB looks like.

Update 3.10.2015

My colleague produced stencil on CNC milling machine.

The stencil.

Pasted.

Let's start place the components.

At work we have a fairly good equipment, so instead of handjob we fed the fitting machine with data. "Drudgery to the machines," says our chief.

Into the furnace with it.

Burned..

Design bugs.

I was critical to my predecessors, therefore it will be only fair if I'll be critical to myself.

In fact, no design is without bugs. These are then debugged with the next iteration.

Some are quite difficult to predict and many years of experience in the field will not help either.

Some bugs I would not do if I was more precise and for some bugs I am ashamed.

1) The worst mistake is the wrong case (footprint) for the U48 AD724JR.

Yes. I was cleaning the library so vehemently that I deleted SO16 with a width of 300 mil. Check out the photo of the PCB at the top and You will see.

Rows of terminals on the PCB are therefore much closer together. I helped myself so that I bent the leads under.

Fortunately for me, all pins are properly soldered during baking, so another action was not needed.

I note that all the bugs are continuously eliminated in the design.

2) LED description.

While optimizing PCB I swapped the LEDs but I forgot to swap their description.

+5V a +12V should be swapped.

3) Processor footprint.

From the picture, it is clear that although I used the footprint of a standard library footprint pads are too large.

It was possible the processor to fit just tightly. Pads should be smaller and more under the component.

The lesson is that not all IC that have the same number of the legs, the same pitch of the legs, the same name of the case and look the same, are the same.

4) RAM footprint

There it is again! This time, however, the component is slightly larger. A part of the pad should be seen in the front of the pin. Here it sits basically exactly.

Regarding the production it is not a problem, but the repairs would be made on it a lot worse. The issue is the same as with the processor.

Compare RAM and EEPROM sizes and You will see that the RAMs are a little bigger.

Because of the fact that the increasing of the footprint basically means rebuilding the entire board, I decided that I will live with this error.

5) Opening for speaker

I found out a little bit late that the speaker with an impedance of 50 ohm is probably a total extra of Sir Sinclair and that it will not be easy to get it.

The only commercially mass-produced type is that in the photo, and that is as you see significantly smaller than the intended space for installation.

If somebody built it, so be warned before using the type with a smaller impedance that the market is cluttered.

What would happen, for example, when fitting 8 ohm? In the case of programmer's error who forgot to turn off the H level at pin 7 IC7 (speaker control),

4V would approximately be on Q5 emitter. This represents a permanent current 500mA through the speaker, which is the power dissipation of 2W. This means

that strange smell will follow in a few minutes and then multicolored smokes. When fitting 4 ohm, the power supply would probably collapse on the first beep attempt.

Let's shorten the story. The speaker is small and the hole is large. Next time I will do a set of small holes.

6) Installation opening

Data of the mounting holes are taken from the original project, so I discovered the error of my predecessors after placing the PCB into the housing of Spectrum.

Installation opening will have to be moved 1.5 mm upwards. PCB will be drowned in the housing a little bit because of that. I still have to find out if it's okay with peripherals.

7) Keyboard connector

It is not so much about the mistake, but rather a necessity.

I ordered most of the components in the TME 30/06/2015, including the five-pin connector for the keyboard.

Unfortunately, that is still not in stock, it is suppoused to arrive until the end of November and it is the only missing component.

To be able to continue the tests I had to make it from one of the eight-pins.

Experienced eye notices that it is made from the two halves, but it works anyway.

Reviving

Well, let's forget, it worked well at first. Not even in a dream, what you think!

Among us, developers there is a saying, that "If it works on the first try so there's something wrong."

Vice versa goes out of it, if it does not work at the first time so it's perfectly fine :-)

After turning on consumption was OK, the voltage was OK, but the output was without video signal.

Of course, it was measured with an oscilloscope, after all at the very beginning I will not burn my TV.

So what about now?

Some impulses went there, but that was all.

Then I put a probe to the clock 14MHz and my instrument said that it measured 90MHz.

Well, I told myself that I have the fastest ZX Spectrum in the world, but without picture, it sucks.

I tried to replace the 74HC04 for 74HCT04 and ... 70 MHz.

After checking the datasheets I found out that my crystal has a much smaller capacity than in the original project.

This seems to cause oscillation on higher harmonics. The problem was eliminated by adding a 22pF capacitor.

Although the frequency was 13 994 kHz, counter checks confirmed a good frequency stability, so I decided not to do further experiments.

I left the HCT there too, exactly according to the rule "If it works don't touch it".

The moment of truth has come.

Connected via Video output to digital TV Samsung.

YES!!!

Connected via RGB.

YES!!! Success again

Then a quite boring and undocumented test of colors and sound followed.

Regarding the speaker is concerned, so the sound is clearly audible and therefore I declare that the used type is quite sufficient.

It was found out that the sound can not be used with the digital television, because before the image is coded into digital form,

It takes some time and that the sound has to be delayed to be synchronized. The consequence is a beep in the Spectrum and then only the beep on TV.

The delay is estimated about 200-300 ms. I am afraid that this could make a problem at games. TV shows that you are right before a mine,

but the Spectrum already thinks you stepped on it.

Another minor flaw is shaky image. The text slightly irregularly shivers in the horizontal direction.

It is certainly not because of the unstability of the oscillator, but the question is how the digital TV balances with greatly cut down PAL of ZX Spectrum.

I need to get conventional analog TV somewhere and do more tests.

14.12.2015

Finally I forced myself to clean and upload the project.

I had to split it into two pieces because Google yells that in one ZIP it's too large.

A colleague lent me an LCD analog TV (without DVB-T) and as I thought the picture was just fine.

I did the long pause because I was building DivIDE to be able to upload games.

There's a working sample based on Lotharek 2k14 on the image.

Of course I didn't play because I broke and throw away my joystick long time ago,

which I remembered after then I turned my flat upside down.

Mister Lotharek published the projekt DivIDE, but just for about 80%. He didn't respond to my email so I had to help myself.

Well I had to correct a little the wiring because I don't like the atrocity with scattered data and address bus.

Although in my early days I also used this trick too but the PCB was designed on square paper.

DivIDE will have its own essay. Christmas is coming so hopefully there will be a little time.

cztomeco@gmail.com