Amiga 2000 Reset Circuit Behavior

[jasonsbeer]

I am looking at a gray screen Amiga 2000, Rev 4.3. It has quite a bit of water damage.

I checked the reset voltage upon startup on pin 18 of the 68000. It stays at zero for about 1-2 seconds after the power supply is turned on. It then goes to 3.45v and stays there. Does this sound correct?

Looking at the schematics, reset is labeled _RST. In that case, reset should be low and "regular operation" should be high. I am curious about the 3.45v. I expected it to be 5v.

Thanks

[solarmon]

The _RST signal is active low. It will pulse low to initiate the reset of the chips it is connected to.

This _RST signal should be normally pulled high to VCC via a pull up resistor, so check that this resistor and trace going to it is OK.

[solarmon]

From the schematics at:

https://www.amigawiki.org/dnl/schematics/A2000_R6.pdf

(I could only quickly find rev 6)

The pull up resistor for _RST is on resistor pack RP101

[jasonsbeer]

RP101 is shown as 4.7k on the Rev 4.x schematics and 2.7k on the Rev 6.x schematics. Not sure if that is a typo or if is really different. I cannot make out what is written on the resistor pack to determine the value. The traces are all good. I spent a ton of time checking traces on the CPU and custom chips. I have not found a bad one yet.

I've come to the conclusion the startup reset is working as expected. 5V through a 4.7k resistor...3.45V seems reasonable.

If someone happens to have a 500 or 2000 on the bench, maybe you can check the voltage on pin 18 of the 68000 for me as a comparison.

[Avrovulcan]

I have a rev 4.4 machine if it helps I can have a look at what value RP101 is on mine.

Regards
Andy

[demolition]

Originally Posted by jasonsbeer

I've come to the conclusion the startup reset is working as expected. 5V through a 4.7k resistor...3.45V seems reasonable.

It all depends on how hard it is being pulled the other way, but for TTL, any voltage >2.7V is a normal high logic level, so your reset signal sounds good to me.
When you say grey screen, do you mean that it starts at black with _RST low and the proceeds to a grey screen? Or is it actually stuck at the black screen (with the brightness on your monitor set too high)?
If you have an oscilloscope, set it to trigger on the chip select signal on the ROM to see if the CPU starts reading code from it.

For any diagnosis like this, I would recommend to burn a DiagROM to help with the diagnosis. It can run without functioning memory or several of the custom chips if you read its output from the serial port.

[Avrovulcan]

The DiagROM is a worthwhile tool to have then??

[jasonsbeer]

Originally Posted by demolition

It all depends on how hard it is being pulled the other way, but for TTL, any voltage >2.7V is a normal high logic level, so your reset signal sounds good to me.
When you say grey screen, do you mean that it starts at black with _RST low and the proceeds to a grey screen? Or is it actually stuck at the black screen (with the brightness on your monitor set too high)?
If you have an oscilloscope, set it to trigger on the chip select signal on the ROM to see if the CPU starts reading code from it.

For any diagnosis like this, I would recommend to burn a DiagROM to help with the diagnosis. It can run without functioning memory or several of the custom chips if you read its output from the serial port.

It goes straight to gray on the monochrome output. No aparent change in shades. So, probably black screen. There is a video signal. The GBS8200 I ordered has been stuck in shipping for a long time. I'm trying to rule out anything obvious in the meantime.

I'm hoping to avoid buying an oscilloscope due to cost. I have a logic probe, diagrom, and null modem on the way. I'll start checking for actual activity on the chips when the logic probe arrives.

[jasonsbeer]

Originally Posted by demolition

If you have an oscilloscope, set it to trigger on the chip select signal on the ROM to see if the CPU starts reading code from it.

I'm still waiting on my logic probe and am using my multimeter. I wanted to follow up on this.

_ROMEN (enable) is pin 12 on the kickstart rom. It is fed by pin 21 on GARY. When I turn on the power, this pin goes to 5.0V for 1-2 seconds and then drops to 3.9V. This makes my hair tingle a little. This is not correct. That voltage is too high for an active low signal to trip enable. Could this indicate a bad GARY?

[demolition]

Sounds to me like there is some action going on on that pin (which there should be until it has finished running the code in ROM).

_ROMEN will only be pulled low while the CPU is reading from it so it would be high most of the time. If you're just using a voltage meter, then the voltage you see would be some kind of average depending on how the meter handles high frequency signals. Imagine _ROMEN is a pulse train which is low 20% of the time, then the average voltage would be ~4V.

[jasonsbeer]

Originally Posted by demolition

Sounds to me like there is some action going on on that pin (which there should be until it has finished running the code in ROM).

_ROMEN will only be pulled low while the CPU is reading from it so it would be high most of the time. If you're just using a voltage meter, then the voltage you see would be some kind of average depending on how the meter handles high frequency signals. Imagine _ROMEN is a pulse train which is low 20% of the time, then the average voltage would be ~4V.

That makes sense. I just need to be patient and wait for my logic probe.

BTW - I did get the diagrom today. I put it in and still black screen. Null modem is also on the way.

[jasonsbeer]

Life! Maybe...

Observation: Amiga 2000 goes straight to magenta screen on startup. No other colors or color changes. This behavior does not change with KS 1.3, Diagrom, or even no ROM in the socket. There is nothing on the serial port from Diagrom.

According to the startup reference sticky on this site, a magenta screen indicates "single-task or cold-start initialization failed".

I'm hoping someone might have a thought on where to go from here.

What I have done up to now:

1. Removed battery and cleaned area with vinegar and rinsed with distilled water
2. Cleaned the board to eliminate mineral and rust deposits as much as possible
3. Replaced corroded sockets for both CIA's, Agnus, Paula, Denis, Gary, 68000
4. Desoldered keyboard, mouse, and joystick connectors (to be replaced)
5. Reflowed a lot of solder joints where corrosion was present
6. Tested traces among custom IC's, all traces for 68000 and KS. No bad ones found yet.
7. Confirmed all custom IC's are working
8. None of the chips are hot
9. There is voltage on the 7mhz clock, but I don't have a way to test the frequency

[demolition]

Without the KS ROM installed, the list of error colors does not apply. Does it stay magenta when you hold _RESET low? And is it bright magenta, i.e. something like color #FF00FF?

I am wondering if the CPU experiences a double bus fault? You can tell this is the case if both _BERR and _HALT are low. If there is a problem with the bus from the CPU it could end up in this state which stops all activity.

[jasonsbeer]

Finally getting back to this machine. Got a Rev 6.3 up and running in the meantime. I had to set this board aside for my own sanity. Here is an update. I now have a o-scope to help me out.

I revisited the RESET and HALT pins. I have attached a pic from the scope. The image is identical for both the HALT and RESET pins.

1) When both the HALT and RESET pins are in circuit or manually pulled high, they both oscillate in the same way (see pic).
2) When the HALT or RESET pin is bent up (not in circuit) the other pin is high. The bent up pin is pulled low...suggesting the CPU is trying assert a halt and/or reset?
3) When both HALT and RESET pins are bent up (not in circuit) both are low. Again, suggesting the CPU is asserting a reset?
4) When the HALT and RESET pins on the CPU are not in circuit, the circuit is high for both (normal function). These are pulled high by RP101.
5) BERR is always high.

So...best I can tell...the CPU asserts reset...comes out of reset...and immediately goes back into reset. I am at my wits end with this board. I believe this may be some sort of bus contention, but I have been unable to determine where it is coming from. Buster seems to be signalling fine. I have removed and tested U103,104,105, and 106. They test OK. Only thing I've noticed is the enable signal (_OEB) on U103 and 105 is stuck high. I'm not sure if that's a cause or a symptom.

Short of desoldering and testing every LS logic buffer, I could use some suggestions if anyone has any. Does this sound like a bus contention issue?

[cdh]

Over the summer, I repaired an A2000 rev 4.4 motherboard. It had similar symptoms to yours (black screen), but I can't say whether the RST line stayed at 3.45V. Didn't test that.

I saw corrosion all the way to the ROM socket. After removing the battery, both the CPU socket and the ROM socket had to be replaced. A number of traces were damaged between them. Even though it was tedious, I found the best way to deal with that was to walk each pin from the CPU socket with an Ohm meter. Depending on your meter, a continuity check might beep even with 10 Ohms or more of resistance. First to the right (pins the ROM socket), and then to the left (fingers on the CPU slot). The CPU slot fingers were a pain, but I luckily had a 68000 CPU slot relocator that I could insert in the slot, which meant that most of the pins matched up exactly with those on the motherboard's 68000 socket. There are some fingers on the CPU slot which are buffered, so you need to go to other ICs on the left to verify those.

[jasonsbeer]

Thanks for the reply, CDH. You are not far off. Andrew from Amiga of Rochester gave me a call and we tracked it down to a "weak" trace that _KBRST occupies. It toned out great, but it was unable to keep pin 5 on Gary high. The result was that it was constantly resetting, as if you were doing a keyboard reset over and over. I ran some jumper wires and now the reset problem is solved!

That problem is fixed. Computer still doesn't work. Gotta keep on keepin' on!

[cdh]

Congratulations on solving the issue with _KBRST!

As a follow-on, you might want to probe pin 12 (_OE) on the Kickstart socket from cold power-on. If you see _OE pulse, then you can have some confidence that the CPU did an initial fetch and that Gary decoded the address.

[Tandrews-ny]

Originally Posted by cdh

Congratulations on solving the issue with _KBRST!

As a follow-on, you might want to probe pin 12 (_OE) on the Kickstart socket from cold power-on. If you see _OE pulse, then you can have some confidence that the CPU did an initial fetch and that Gary decoded the address.

After looking at the board I found a pin was shorted, fixing that got the board booting but we're now encountering another issue I have been unable to sort out. It boots but the system just stops responding (no CPU activity, ram activity, etc). I spent a week testing different things and could not sort it out sadly. I am keeping the board and sending another fully working board so I can continue trying to fix it over time.

[B2k_ad]

HI
since you asked half a year ago if the value of RP101 changed.
Yes it did.
Rev 4.3 would have 4.7K but could and should be changed to 2.7K
Actually those resistor pacs near the RTC battery are prone to fail.
Since you have a 4.3 it should be populated with RP900 and RP901, if it still is... theese have to go.
Its called "the buffer fix"
RP 900 and 901 are pullups, basicly the fix sets them closer to the actual buffers U603 and 604
and they become pulldowns.

I really should finish my Documetation of all the fixes from 4.1 to 6,4
They work... this is witten on an rev 4.3