SE/30 scope debugging: what happens right at power-on?

[apm]

Question for the experienced debuggers here: on the SE/30, what are the first things the 68030 does when powered up, and what are the best signals to test to work out early startup problems?

A while back I bought an SE/30 logic board (socketed) which had been recapped but was still nonfunctional: horizontal stripes, no signs of life. I'm testing it in a machine with a known working analog board. There's nothing obviously wrong with the recapping job on visual inspection. Since UB11 is a known failure point, I removed it. No change. I also swapped the ROM SIMM with no change. 

At that point I started poking around with a scope. I can confirm the RESET/ line is high and the 16MHz clock is present on the 68030. But I couldn't find any activity on any of the address or data pins, though admittedly I didn't test them all.

Rather than stabbing around randomly, can anyone explain the very first signs of activity to look for on the logic board?

 

[techknight]

it would be far easier to test the Address and Data lines between the ROM and CPU. they commonly get broke. 

 

[mrpippy]

Have you reseated the 68030? My socketed SE/30 just had horizontal stripes at first, and even after recapping/cleaning would still occasionally do stripes and/or reset while running. Popping the 030 out and reseating it seemed to fix the problems at last.

Does the ROM Chip Select line ever go low? And what's the resistance between the terminals of the reset switch? I saw it was 500 ohms and thought my reset switch was bad (and pulling down RESET), but now I'm not sure this was the case.

My thread might be helpful: https://68kmla.org/forums/index.php?/topic/23538-almost-fixed-my-simasimac-se30-booted-once/

 

[apm]

Good thoughts. I tried reseating the 030 and the ROM SIMM, with no change. Address and data lines from the CPU to the ROM SIMM socket all test good. I don't have the scope here at the moment so I will have to test the ROM CS/ line later.

Weirdly, I think the reset switch has failed but in the opposite direction: pushing the button does not close the contacts. However since UB11 has been removed this shouldn't be the cause of the problem.

I could continue to test traces but I'd be interested to take a more directed approach. Given that the CPU is not held in reset and is getting a clock, what should be its first observable actions? Are there any other pins which, if held in the wrong state, would prevent it from initialising normally?

 

[techknight]

the RESET needs to be toggled on power up or the CPU wont start. 

Also I have seen 1 RARE case that the CPU itself has failed. A good way to check that, is remove the ROM and scope an address/data line, the CPU should be walking the bus. at a full 5V swing. the Address bus should effectively be a "program counter" at that point with each square wave 1/2 of the next bit in line.

Insert the ROM, and if the square waves fall to 2.5V or under, the CPU is bad and cannot drive the bus properly. 

I have had this happen on JDW's unit. 

 

[apm]

the RESET needs to be toggled on power up or the CPU wont start. 

Does that mean that booting without UB11 doesn't work then? I removed it in case it was holding RESET/ low, and verified that the line is high. Do I then need to bring it low and release it for the CPU to start?

I do have a spare 68030 somewhere. If I can find it that's an easy test.

 

[techknight]

Well i would see if the 030 is walking the bus properly with the ROM removed before you do anything. 

But yes I am fairly certain the system requires RESET to be toggled before anything can happen. 

 

[mrpippy]

I was under the impression that UB11 was the chip that actually drives RESET (assuming the reset button isn't being pushed). UB10 is an identical chip though, and it outputs its reset signal (on pin 5) to a test pin--maybe wiring that to the board's reset signal would be an easy test.

 

[techknight]

Well yea it drives the reset. I dont think anyone has said anything otherwise? 

But it was mentioned the chip was pulled so nothing is toggling reset anymore. 

 

[mrpippy]

Yeah, just wasn't sure apm understood that the board is not gonna boot with UB11 removed. Despite being the sound chip, it is necessary.

 

[uniserver]

Also cant dead burns filter cause a variation of this?

 

[techknight]

Possible, but I have never personally seen it. Of course that doesn't mean it cant happen. 

All a bourns filter would do is cause the IC to hold one of the interrupt lines before the interrupt vectors are initialized, and that would result in that. You can scope for stuck interrupt lines or interrupts that are firing too early. 

 

[apm]

Well I think I found the problem. See if you can spot it.



It's a bit embarrassing how many steps I went through before getting there. Clock was present, 5V was present, no interrupt lines held, toggling reset manually produced no result. With or without the ROM, the address lines were all over the place between 0 and 5V and just sitting there.

This is the second time I removed the CPU from its socket (first time to reseat it) and I didn't notice anything the first time. It's possible I did this-- the socket is stiff-- but I doubt it. I got this board from eBay, advertised as not working post-recap. The new caps look fine but the socket has a few marks on it, and it has scoped the same way since day 1. I guess these CPUs are more fragile than your average DIP.

Now to see if I can find that spare 68030.

 

[CC_333]

That definitely looks like a crack!

I would guess that, assuming nothing else is wrong, that board will work perfectly with a new CPU.

Good luck finding your spare!

c

 

[techknight]

Well with a good logic analyzer, or even a dual-channel scope without the ROM in place, a typical 68K machine will walk the bus. 1 by 1. So the address lines look like a binary counter. 

If you didnt get that far, well. CPU be bad.

And yes, the CPU is ceramic. Pry up on a ceramic dinner plate and see what happens. CRACK

 

[apm]

Yep, that was it! Found the spare CPU and it booted right up.

For what it's worth, it boots fine without UB11 installed. It even has sound, though presumably only on one channel.

I bought this board because it was socketed. I really wouldn't have expected that, indirectly, that was also the reason it didn't work.

 

[uniserver]

I wonder if the dry cycle in the dishwasher can cause the cpu to crack? I know people like throwing their pcb's in the dish washer, for some reason.

 

[uniserver]

Dupe