Daedalus2097

Not especially, no. I would have the box ventilated as any PSU of a similar power (add the various power ratings together to get an overall figure), but those transformers themselves don't generate a lot of heat unless they're being pushed hard. I would keep all the outputs in phase, just in case some of them end up connected in parallel, but I expect that's not generally possible.

Yep... But if you're building a custom PSU into an enclosure anyway, why not just fit as many discrete transformers as you like? Twin secondary 9V transformers are available off the shelf (for example, 2x9V 1.5A secondaries), so for 4 isolated outputs you only need 2 transformers. Add a Mean Well module (for example, 5V 4A & 12V 2A), and there you go - high quality 5V and 12V DC outputs, and multiple, isolated 9V outputs.

Just a thought here, but what about putting a switch between the capacitor and the RPi? That way, when you want to actually turn off the machine, you can isolate the RPi, leaving the charge in the capacitor there so it'll take less time to fully charge the next time you use it. Might seem a bit of extra faff, but could help if the half a minute charge time becomes an issue.

AC/DC differences alone won't make the difference here. If it's possible to have multiple devices connected individually with their own PSUs and have no ill-effects, it will be possible to power them from a single PSU. It might need some additional consideration, like separately switched outputs, but it will be possible with a small amount of work. Worst case scenario, an AC supply with multiple isolated outputs is pretty trivial to build. Yeah, I'm using a couple of official raspberry Pi power supplies for my Atari 800XLs, Working great! PC PSUs are massive overkill for 8-bit stuff. Especially newer ATX PSUs, which might be dangerous to use with such low loads as they're built to expect a minimum load of dozens of watts. There are plenty of cheap industrial PSU modules by the likes of Artesyn, Mean Well etc. that would be better suited to lower loads, as well as being smaller.

Yep, it's pretty straightforward to create your own components in Eagle too (or, at least 7, which is the latest version I've used).

Having recently rescued a couple of 800XLs from my parents' attic, the first thing that immediately came back to me was the poor keyboard connections, which I remembered gave me a lot of grief as an experimental child. And, sure enough, my original machine shows where the mylar had cracked, and where I had bypassed the bad traces with some crudely soldered wires. Naturally, it was unreliable at best, and it means my function keys don't work most of the time. I was actually thinking of replacing the connector with a pin header, and the cable with a proper ribbon cable running to another pin header on the keyboard itself. I can keep the original friction connector in case I ever feel the need to replace it in the future...

The numbers at the edge of the box will be the pin numbers of the chip. Schematic diagrams will order the pins in a logical layout, rather than a physical layout. In fact, the entire schematic will also be a logical layout that doesn't represent the physical layout of the circuit. This makes it easier to follow signals and understand the operation of the circuit. The labels inside the chip represent the signal names. A0-A15 are the 16 address lines, D0-D7 are the 8 data lines, Vcc and Vss are the power supply, and then we get more specific. NMI is typically Non-Maskable Interrupt, RST is Reset, R/W is whether the bus is being read or written (usually CPU relative), HALT is an all-stop signal, usually because of a serious fault detected, unhandled interrupt, taking control of the bus from the CPU or similar. RDY is Ready, usually used to indicate that the bus is ready to start a requested transfer, or to move onto the next step of a transaction. AN0-AN2 are the three signals of a special ANTIC bus on Atari 8-bits, which are used for direct transfers between the custom chips without getting the main CPU bus involved, and are typically used during the building of the display. 0O and F0O are clock signals. A bar above any signal name indicates that it's active low, i.e. it is asserted by dropping the signal level to close to 0V. The general descriptions above could be improved with some specific context. I'm not particularly familiar with these systems so I can't offer much more than that.

Yeah, the audio caps are installed backwards on many CD32s, but the other electrolytic capacitors in general will all need to be replaced at this stage, since they're the awful SMT types and will leak and damage the board. The keyboard port (aux) on the CD32 itself will only work with Amiga keyboards - the A4000 one has the same connector, all of the others can be connected with a simple adaptor). The SX-1 can take a normal, old-style PC keyboard - either AT directly or PS/2 using a simple adaptor you can find for a couple of quid. This is by far the cheapest option if you don't already have an Amiga keyboard lying around the place, though the key mapping won't be 100% since there are a couple of differences (e.g. no Help key on a PC keyboard). In particular, it predates the introduction of the Windows keys so I'm not entirely sure how you can access the Amiga qualifier keys, though it's mostly applications that will use them.

A bit late to this, but I don't remember there being any such control in the Amiga version. I do remember that J was the key for the in-system jump (between when you entered a system from hyperspace and when you are within range of the planet). I seem to remember reports that the Amiga and ST versions were written from scratch without the original source code, so that could explain the missing feature.

Yeah, if you can measure roughly 4.7K (it will never be *exactly* 4.7K, just needs to be the same ballpark) between the Halt pins on the CPU or Gary and 5V, then the resistor pack is fine. Hmmm, there should be a brief blip where the Halt goes high, like the Reset line. Can you see the blip as it goes high on the known good machine? You've tested the CPU elsewhere too I take it?

As for the Amiga portion of the question, it didn't have the same general purpose capability as SIO, but the Amiga's external floppy drive port allows daisy-chaining of floppy drives up to a limit of 3 extra drives. As said, other peripherals were taken care of by serial, parallel, and later SCSI (and later still IDE) interfaces.

Heh, IIRC stands for "If I Remember Correctly", so it's not a part But with absolutely nothing happening on the data or address lines, the DiagROM isn't going to run. Halt being stuck low could be a pull-up issue on the board, a fault with either Gary or the CPU, or a fault detected by Gary or the CPU that is so early the bus activity is stopped before it starts. With the power off, check the resistance between Halt and ground, and between Halt and +5V. It should be around 4.7K ohms in both cases. A little different from that won't matter, but if it's below 1K or above 10K on either side, that could cause an issue. Too low on the ground side or too high on the 5V side will cause Halt to stay low, preventing the CPU from starting. The pull-up resistors are a resistor pack that's near Gary I think, and could be affected by corrosion, a broken track in the area, or could simply have failed. If the pull-ups are ok and you try to boot with Gary removed and the Halt line stays constantly low, then the CPU is potentially the issue, though if you've tested it in a known good machine then that shouldn't be the issue.

Well, I received a new GTIA chip from tf_hh today. Stuck it in and... It works Proper logic levels on the bus and a READY prompt straight away! Thanks very much!

In theory you could change to a different keyboard layout where the U is in a different location, for example Dvorak. You could also copy and paste the letter (or the entire command name from e.g. a directory listing) if it's a Shell window launched after Workbench has booted: highlight the text to copy, use Right-Amiga-C to copy and Right-Amiga-V to paste. But really, it's likely the membrane of your keyboard is damaged. They can be replaced with modern reproductions in most cases.

Hmmmm, I would try setting them all to 0 anyway, but it might not make any difference if the processing stage is permanently enabled and so keeps the lag. That's a shame.

The inductor should be easy to test - it will have a low DC resistance, and if it's directly in the incoming path, it can simply be bypassed temporarily to check if it's the issue, provided the power supply is nice and stable. The diode (yep, it looks like a diode) should also be simple to test, most multimeters have a diode check function which should give a forward voltage drop when the probes are attached one way, and no reading when attached the other way. If it's a zener diode then it might have a further fault beyond that, but to me it simply looks like a rectifier diode to protect the system from a reversed polarity supply. The last part up there is the small ceramic capacitor. If that's across the power rails, it's most likely not critical to the operation of the system (like the inductor, just used for filtering the power supply), so once it has more or less infinite DC resistance it should be fine. Are there schematics for this board available? If so, they'll be able to tell you the exact function of these parts, and often their values or part numbers.

While HDMI itself won't introduce much lag, quite often the image processing carried out by the TV does, and that can be applied to any input - analogue or digital. Things like dynamic contrast ratio (ugh) and similar effects need to analyse an entire frame, then adjust the TV accordingly before actually displaying the frame. Check if the TV has a "gaming" mode in amongst the colour and contrast settings - many do, and it removes steps from the processing, resulting in certain features being disabled but reducing the lag to the lowest possible using that display.

Yep, thanks very much, and in a couple of days I'll let you know whether that sorts the issue or if something else is going on as well. Interesting stuff... I wonder if it's an age issue with the chips themselves, or a secondary effect of something else. For example, as the capacitors in the PSU or main board become less effective over the years, they will be allowing a higher ripple in the supply, and won't be filtering the logic-induced noise out. Both of these conditions could be putting extra stresses on the I/O stages of the chips, which could be accelerating their demise.

Just to add about compatibility with other Amiga games: Yes, essentially the CD32 is an Amiga 1200 minus a keyboard, plus a CD drive. As a result, you can play pretty much any game on the CD32 if you can get it onto the machine - in fact, most CD32 games are simply older Amiga games stuck on a CD with occasionally some extra animations or CD music. Many were even made for the older Amiga models and so don't even use the enhanced graphical capabilities of the machine over the 1987 A500... Now, that's the tricky part since it doesn't support a floppy drive without additional hardware. But there are some people who make up compilation CDs containing dozens of floppy disk games on one disc, and that's probably the way to go. Bear in mind however that many Amiga games also needed a keyboard, so you might find it good to also get an A4000 keyboard (which plugs into the CD32's Aux socket), or a PS/2 to A4000 converter and use an old PC PS/2 keyboard. If you did decide to spend some money on it, there are new homebrew expansions like the TF 328 that give the CD32 an internal IDE port so you can connect a compact flash card that acts as a hard drive, and some extra RAM (which increases the speed of some games and allows the loading of the OS as well as games from the hard drive). It's not cheap, but it makes the CD32 into a pretty capable all-round Amiga.

For board 1, if all the traces are 100%, the next thing to look at could indeed be the Agnus chip. If you have issues with the 74 logic in the RAM bus, you'll normally get a green screen, but this sounds like the CPU's not starting. You don't even get a very short burst of bus activity, no? I would check the ROM socket, including the alignment (the ROM socket should have an empty row at the pin 1 end. What is the halt line doing? That should be similar to the reset lines, briefly going low before staying high. If it stays low, the CPU won't try to run. IIRC, Gary also connects to the halt line, so double-check continuity all the way across. If halt, reset and CPU clocks are all ok, you should be getting bus activity as it tries to read the ROM. I've found that some Agnus chips are stuck into their sockets with a double-sided foam pad. Why that is, and others aren't, I've no idea. But I've been able to remove them by pushing two chopsticks through the holes on the underside of the board until it eventually works loose. Not for the faint-of-heart though! Corrosion can and does reach the Agnus socket to cause problems, so it needs to be checked. As for board 2, if all the traces are ok, I'd sooner suspect the 74 logic than the RAM chips. They can get damaged internally by corrosion that creeps up their legs, even if they're cleaned up. In particular, U10-13, 32, 34 & 35 can all be affected by leakage. Also, I once had one that had a damaged trace right by one of the legs of one of the U10-13 chips. It was such a fine fault that it wasn't visible, and pushing the multimeter probe down onto the chip's pin was enough to close the break and make it look like the continuity was fine. Had me stumped for a while until I happened to put pressure on the pin a different way and found an unreliable connection. DiagROM should be able to help you pin down the bad bits on the bus, which you can then check with a scope while the memory test runs to get an idea of where the issue is.

Indeed, and there appears to be a minimum order of $50 from the US site too. I'm interested - how much for one chip shipped to the UK?

Cheers guys, not to worry. So Best Electronics would be where you'd order from anyway? I might give it a shot... Maybe a group order could be an idea if anyone else needs anything from them?

Okay, so I was having another look at this today and couldn't find any decent sources for the suspect GTIA chip in my machine (C014889-01). In fact, the only place that looks promising is the wonderfully Geocities-looking ball of internet nostalgia that is http://www.best-electronics-ca.com. Is this the place to go for these things? For simplicity, I prefer to order within Europe whenever possible, but it looks like that's not an option.

Good to know, thanks!

Thanks very much for that, I'd found the official Atari service manual but the Sams one is indeed much more useful, as are those schematics. Yeah, the address lines are common between the RAM chips, but one chip is used for each data bit so only one bit can be affected by each chip. I do feel like the GTIA is the problem (or at least one of the problems) - I've seen chips do that sort of bus loading before and causing all sorts of issues across a board.