bus/rs232: Add the Cricket external sound box for the Apple IIc.

[robjustice]

Emulation of the Street Electronics the Cricket for the Apple IIc. Disks and rom dump is pending to be added to the A2DP. Can be grabbed from here as well: https://archive.org/download/cricket-6801-03-801-00/Cricket_6801_03-801-00.bin

[rb6502]

You'll need to change the short device name to fix "mame -validate" since "cricket" is already taken. Maybe "secricket" or something similar.

I don't know how the clock works (looks like it's stored in the 6801's internal RAM?) but it might be interesting to also inherit device_rtc_interface so it can be auto-set to the host machine's time/date on startup. I won't hold up a merge for that though.

[robjustice]

I have updated that, thanks.

That would be nice to set the clock. I'm assuming its stored in 6801's Internal ram as well. Must be some sort of flag generated or set to say its been updated, would just need to work that part of the puzzle out.

[cuavas]

You seem to be using m_data for two things – the value being driven onto P4, and the value seen on P4. This inevitably leads to issues.

[robjustice]

It has the data bus for the three devices connected to port4. Its setting the direction/ddr appropriately for the port and either reading or writing the data bus value before then driving the device control lines via the port3. I had it like this as this is how it would be working for the real hardware. Is this something that needs separating more for how mame handles things?

[cuavas]

It will only “work” like this if you assume the program is well-behaved and does things in a particular order. In reality, the logic levels visible on the port is some combination of the levels driven by all the devices driving the lines, which may be more than one if the software is badly behaved.

You also need to consider that the levels driven by the peripherals won’t “stick” when the peripherals are deselected. It’s a lot easier to keep the levels driven by each device separately, and AND them all together when you need to get the overall levels visible on the bus.

[rb6502]

I understand what you're saying but there is one piece of firmware running this thing that doesn't change and it works right now. And what it's trying to do (read and write the AY registers) is discrete and well defined. There is no need for the whole AND thing.

[cuavas]

You really need to detect active edges on these data bits. If the software writes the same value to the bit twice in a row, it won’t cause an additional write. For example, just looking at the TMS5220:

    ...

    if (BIT(~data & m_p3_out, 0)) // 5220 read
        m_tms_data = m_tms->status_r();
    else if (BIT(data, 0)) // 5220 deselected
        m_tms_data = 0xff;

    if (BIT(~data & m_p3_out, 1)) //5220 write
        m_tms->data_w(/* calculate actual levels from all devices driving the lines here */);

    m_p3_out = data;

Also, what exactly are bits 4 and 5 connected to? The AY-3-891x chips don’t have a chip select input as such. They have BC1, BC2 and BDIR inputs, but the only time the chip doesn’t do anything is when BC1 and BDIR are both low. If either BC1 or BDIR is high, the chip responds. So I don’t see how the logic here can be correct, unless there are additional AND gates to force the corresponding BDIR low if bit 5 or bit 4 is low.

[robjustice]

There is a chip select input on the AY8913 version of these, and the Cricket is using the AP8913 variant. Here is the not quite finished schematic for it. It has a few wrong symbols for caps and needs the power supply checked, but otherwise its fine. Will give you an idea how its hooked up.
cricket.pdf