New project: An ACT Apricot F1 "barn find"

[stephen_usher]

Given that I'm seeing the MFM floppy data on the pin of the FDC (WD2797A) I'm wondering if the disk controller is bad and not seeing the data internally. I've ordered one via eBay. Hopefully it won't be a fake or dead chip. Thankfully it's socketed on the motherboard.

[rubber_jonnie]

I've been following with interest, as it reminds me of my first Mega repair, and a bit like my ongoing BBC B repair!!

Am I right in saying that the screen is all good now, it certainly looks that way in the photos?

[stephen_usher]

Yes, everything's working. The screen issue was caused by the brass wipers in the molex power supply collector being heavily tarnished and hence either not making good enough contact and having a high resistance. After firstly filing the surface (the tarnish was too hard for wet & dry sandpaper) and then polishing with sand paper that problem's fixed.

I have found a floppy which will work with it now, at least for the BIOS diagnostic. It's an old NEC drive which can have pin 34 set as READY rather than disk change. I still have to manually pull the motor-on line low though otherwise the drive fails to respond the way that the BIOS thinks that it should.

Oh, and I was well outbid for the Apricot Portable. It went for a silly £71 in the end as "Parts, not working".

[stephen_usher]

This evening I've not been hardware hacking, I'm software hacking.

I've written the basis of the virtual IR keyboard code on an old Raspberry Pi.

I'll be attempting to use the SPI interface synchronous serial output to drive the IR circuit and have written the encoding and packet output code using the bcm2835 library. The main() function merely initialises the SPI interface and then calls a function which generates the packet sequence required to set the Apricot's clock using the Pi's current local time as the source.

[stephen_usher]

Rushed home from work (to work from home this afternoon) as I'd failed to set the delivery address of an RS order to work. 99% of the time Parcel Farce deliver sometime between 12:30 and mid-afternoon... Typically not today!

This means that I'll have to work from home tomorrow when I wanted a full day in the office. It also means that I can't make up a better floppy cable for the Apricot. Still that's OK as I can work on the IR sender for the Raspberry Pi instead.

[stephen_usher]

OK, who's good at analogue electronics?

I'm trying to build an IR sender connected to the MOSI SPI line on a Raspberry Pi.

I've tried the simple 2222A transistor circuit posted on-line and that doesn't work. The input signal to the base gets squished to about 400mV and smeared out. Boosting the input by passing it through a couple of ports on an unbuffered hex inverter doesn't help. Of course, it doesn't turn on the transistor, but even if it did the output would be trash.

Here's the signal before the 10K resistor:

IMG_2292.jpg (68.39 KiB) Viewed 2654 times

And here's the signal at the base of the transistor:

IMG_2293.jpg (66.5 KiB) Viewed 2654 times

These are both on the same time base.

HELP!

[stephen_usher]

OK. I think that I had the SPI clock divider set wrongly, as in an order of magnitude too fast, so the transistor couldn't keep up.

[stephen_usher]

Actually, I was trying to run the clock at 3.84MHz rather than 3.84KHz!

Anyway, I can see the IR LED brightly in a camera, but only in a tight beam. Still, holding the IR receiver in front of it didn't allow me to get data to be understood. I wonder if the timings generated by the SPI are fully reliable without its clock when sending a packet, or the data bytes need to be sent little-endian.

[stephen_usher]

I think that I've found out why the communications aren't working, the documentation is wrong.

The service manual states that the optical link runs at 3.84KHz clock. However, I've just measured the input on the Z80 SIO CLKRXA pin and it's 2.24KHz. If the IR sender and the machine aren't in sync then the data will not be read correctly.

Unfortunately this means that I can't use the Raspberry Pi's SPI interface for this job as it can't run slow enough.

[stephen_usher]

I'm wondering is a MAS7838 chip might be the way to go, sending the data via TTL RS232: http://static6.arrow.com/aropdfconversi ... as7838.pdf

Driving it with a 4.91MHz crystal would give 19.2Kbaud asynchronous input with the ability to select 2.4KHz synchronous output.

The trouble is that the chip is out of production and there only seem to be Chinese sellers.