Troubleshooting H4 & H5

[exxos]

Will list some basic troubleshooting steps and update it as time goes on. I will lock this thread to keep it clean, but people can discuss in their own builders blogs anyway.

Firstly, all faults are likely going to be a bad solder joint or short somewhere. This also assume all the chips used are known working. Also assuming all the sockets and chips are located in the correct orientations. Also after 30 seconds of "ON" time place a expendable finger on every IC and make sure they are not burning hot.

Also assumption is all fixes are done in this thread viewtopic.php?f=78&t=2335 and you have checked power pins on all IC's to make sure they all have 5V on them.

Also assuming you have all the jumpers in place viewtopic.php?f=73&t=2229



RESET.

With a meter or scope. Check the RESET signal on one of the expansion sockets. When powered up, RESET should be low for approx 1.5 seconds and then go high and stop there.

Press the reset button and reset signal should go low for approx 1.5 seconds and go high again. Voltages should switch between 0V and 5V.

[exxos]

PULL UPS.

With the power OFF, check with a meter on resistance (10K range ) and go down every pin on a 68K expansion socket one by one, and connect other probe to VCC.

Every pin on the CPU (with exception to GND,VCC,E,CLK) Should measure 1K or 2.2K resistance.

[exxos]

BUS SHORTS.

Using a meter (MACHINE POWERED OFF!) on resistance, 200R setting, check for shorts on a 68K expansion socket. Start on pin 1 and pin 2 and then move the probes to pin 2 and pin 3 etc until you have gone around the whole socket. There will likely be a resistance there, but it should not be near zero. If its below about 10R then you have a short on those pins somewhere.



LS11 SHORTS

With power off, again check for shorts visually (under magnifier) and using a meter on 200R setting, check there are no shorts between any pins on the LS11.

[exxos]

FIRST POWER UP VOLTAGE TEST.

Using some bodge tape, tape the reset button to keep the machine in reset. Power up the machine and check all the voltages on a 68K expansion socket.

Every pin should have 5V on it. The exceptions are GND,E,CLK,RESET,HALT E & CLK will be switching and should be checked on a scope rather than a meter. A meter will probably measure around 2 volts on those pins.

[exxos]

CLOCK CHECKS.

Using the H4 schematic or IC pinouts Check all the MHz clocks on every IC with a scope.

Shifter should have 32MHz into pin 2 and output 16MHz on pin 39.

MMU should have 16MHz input on 5 and output 8MHz on pin 20, and 4MHz on pin 19.

GLUE should have 8MHz on pin 34. 2MHz on pin 54 and 0.5MHz (500KHz) on pin 43.

MFP 4MHz pin 35.

WD1772 8MHz pin 18.

BAD DMA 8MHz pin 39.

ACIA 500KHz pin 3&4.

CPU PLCC 8MHz pin 15, 68K sockets pin 15.

[exxos]

FIRST BUS CYCLE.

This part really needs a logic analyser or a digital storage scope at the very least. This is also assuming you have a stable white video frame being generated at this point.

You need to use the RESET signal as a trigger and test the whole address bus, all at once with a LA, or one pin at a time with a scope. One probe needs to be on /AS of a 68K expansion socket, and another probe for testing A1-A23 pins.

When RESET goes HI, the CPU will put its first address on the bus and set /AS low (0v) . This is "address zero", where every address pin will be at 0V. You will need to keep pressing the reset button to check every address pin one by one on a scope. You cannot just scope out pins on after another as the CPU will only (or should) only do that cycle ONCE, after reset. One bus cycle happens in typically in less than 4x 8MHz clock cycles!

Once all the address bus is confirmed to be at all zeros on the first bus cycle, and /AS goes low after rest, the GLUE will decode address "zero" and will switch the inputs to the LS11 low (at least one should go low during the first bus cycle). The LS11 being a AND gate, only 1 input needs to go low, and the output to ROM_CE (JP13 ). This will enable the ROM IC and it will place the first 2 bytes on the databus. At this point, some datalines will be low and some high.

A couple 8MHz clocks later, /DTACK will be set low by the GLUE and a 8MHz clock later /DTACK & /AS will go high. At this point the CPU has just executed its first instruction and will (or should) go on to executing the next bytes in the ROM and start booting TOS. Assuming you get that far, really the system starting to boot TOS should get far enough to boot the diagnostic cartridge.

Things are a bit more complicated than the above, but that should at least get people started.

EDIT:

For TOS104, the start bytes are $60 $2E. So the Binary data on the bus on the first cycle (after ROM_CE goes low) should be

01100000 00101110

It would be useful if someone with a LA could verify this on real hardware.

[exxos]

A new test to check address zero is decoding properly.

You will need a 68K pinout (print it out)

68k.jpg

Remove the CPU from the motherboard! Also assumption is the GLUE and other chips are fitted, but mostly this is a GLUE & ROM test anyway.

Then solder all these pins together with header pins and wire. It does not matter which IO port you do it on.

hl.JPG

IMG_0272.JPG

Assuming you have a properly programmed DUALTOS chip.. The DATA bits are like this..

1.JPG

2.JPG

$2E60 needs to be flipped to $602E which is %110000000101110

The arrangement of the databits on the CPU bus is:

Code: Select all

 (DATA 15..0)
 111111 
 5432109876543210
%0110000000101110

Now power up the H4/H5 (REMEMBER NO CPU FITTED!!)..

Measure the voltages on the DATA bus and write it down on the 68K paper you printed out earlier. Just the outer track of the motherboard for a easy 0V point for your probe.

For example :

D4 is 0V so write 0 on D4 on the printed paper.

IMG_0274.JPG

D3 is 5V, so write 1 on D4 on the printed paper.

IMG_0273.JPG

Then you *should* have this.

IMG_0275.JPG

Which matches the binary sequence of $602E %0110000000101110.

So this board has the correct start address. The address bus appears to be working , at least to the GLUE. Which is decoding the address zero, and enabling the ROM. The ROM is placing the correct data on the bus also.

[exxos]

I've done a simple PCB to do the above tests.

This will hardwire all the address bus low. So you can check on the ROM socket all address lines are low etc.

The data is shown on a bank of LEDs. HI and LO LEDs will light up. I did it this way as if one is floating , it could be causing rapid pulses on a single LED which may not show the problem. So having 2 LEDs driven stands a better chance of seeing errors.

It also will light 2 LEDs if the 8MHz clock is working.

Capture.PNG

They will be in the store in due time. All SMT assembled. It will only need the header pins soldering.

EDIT:
I may add a couple of jumpers so a few more addresses can also be tested.

[exxos]

The new the boards have arrived and I have tested one out and works great!

I did not get many of these made. They will be available in store in a few days time.

https://www.exxosforum.co.uk/atari/store2/#0289

sniffer1.jpg

IMG_2309.JPG