Keyboard Shortcuts
ctrl + shift + ? :
Show all keyboard shortcuts
ctrl + g :
Navigate to a group
ctrl + shift + f :
Find
ctrl + / :
Quick actions
esc to dismiss
Likes
Search
TDS3032 does not boot. LCD backl-ight and fan working.
|
This is going to be much better with a schematic.
toggle quoted message
Show quoted text
Typically, given a "dead" processor, you check the following: 1) power 2) clock 3) state of reset line 4) state of interrupt lines (especially the NMI) 5) state of the read/write line and the valid address (if any) line 6) variously, look at the address lines 7) look at the data lines for valid data, correlate to read/write pulses. 8) look for ROM and RAM chip selects. You'll need to look up the chips for pin connections. If you have a 2 or 4 channel oscilloscope, you're ok (4 is better than 2), and if you have a logic analyzer, that could be even better. Exactly what these lines are supposed to do, for instance, R/W and data/address are found in the processor hardware manual. Harvey On 1/12/2023 9:21 AM, Paul Carew via groups.io wrote:
I'm taking another look at this board. |
|
Hello Harvey,
Yes a schematic would transform things, but unfortunately, none seems to be available to us mortals. 1) power -> Checked. 3.3v, +5v -5v operating. 2) clock -> This seems to be an issue. Still investigating. The oscillators are running. The 75Mhz looks like it gets divided by 3 and fed to a buffer chip. The supposition is that the output from this buffer chip is the clock for the processor. However, the OE is low, thus no output. Having tried to trace the OE line on the board (it's a dense board), it disappears into a "via" underneath the processor BGA.... not sure, without pulling the BGA, what this is then connected to. 3) state of reset line -> Verified that reset control chip is operating correctly. 4) state of interrupt lines (especially the NMI) -> Not verified at this time 5) state of the read/write line and the valid address (if any) line -> No address line or data line activity seems to be occurring. I've checked at several Ram chip address / data lines etc 6) variously, look at the address lines -> See above 7) look at the data lines for valid data, correlate to read/write pulses. -> see above 8) look for ROM and RAM chip selects. -> Haven't checked the ROM CSs yet Paul. |
|
Interleaved:
On 1/12/2023 1:04 PM, Paul Carew via groups.io wrote: Hello Harvey,Good. 2) clock -> this seems to be an issue. Still investigating. The oscillators are running. The 75Mhz looks like it gets divided by 3 and fed to a buffer chip. The supposition is that the output from this buffer chip is the clock for the processor. However, the OE is low, thus now output. Having tried to trace the OE line on the board, it disappears into a via underneath the processor BGA.... not sure without pulling the BGA what this is then connected toI adore BGA chips, so easy to service, replace, diagnose. </sarcasm> If there is no clock drive, the processor is either on internal clock (happens with some of them), or it doesn't have a clock on it, and therefore will do nothing.? Many OE lines are actually active low, so there perhaps should be an output.? Depends on the driver. 3) state of reset line -> verified from reset control chip is operating correctly.Then the chip is not being held in reset, and neither is the rest of the board. 4) state of interrupt lines (especially the NMI) -> Not verified at this timeIF you could find them.? These become more important later when the system gets hung. 5) state of the read/write line and the valid address (if any) line -> No address line or data line activity seems to be occurring. I've checked at several Ram chip address lines etcNo activity is consistent with the lack of a clock. 8) look for ROM and RAM chip selects. -> Haven't checked the ROM CSs yetIf you don't have any valid address lines, then the CS part doesn't matter. The logic goes like this: 1) no clock = no nothing, processor is stalled. 2) processor trying to boot can be stalled by the reset line low. 3) if the processor is trying to boot, we should see address to a fixed location (typically, either the lowest or highest location in memory), and attempts to read memory. 4) If it can, then you can see a sequence of fetches as the processor tries to access memory and execute code.? NMI down would force the processor into a loop that kept executing the NMI code. 5) hardware failures would possibly get stuck in some loops trying to read bad hardware.? This would depend on the paranoia of the programmer as to what would happen in that case. No clock is possibly the real problem so far.? BGA is not convenient. I'd investigate the clock buffer at the moment. Harvey |
|
Hello Harvey,
Looking at the 7Z26 data sheet for the clock buffer ( ), it looks like the OE is active high. The signal on my board is currently held low, but as I mentioned, it is not obvious what's supposed to be driving it as the signal disappears into a via under the BGA. However, your comment about internal clock is interesting. I'll take a look at the MPC860 datasheet to see if that is the way the processor starts up. It maybe that the external clock is actually controlled from the processor itself (hence the via under the BGA) Paul. |
|
What it *looks* like is that this is a microprocessor more than a microcontroller.? So no internal ROM and RAM, although it does have JTAG interfaces.? I wasn't able to find the clock description on the data sheet (which was mostly hardware interface).
toggle quoted message
Show quoted text
I'd like to see a diagram of the clock tree and some words on the boot-up sequence. Harvey On 1/12/2023 1:59 PM, Paul Carew via groups.io wrote:
Hello Harvey, |
|
This is a fairly standard, but limited (it's old) clock tree. There is no internal oscillator.? Clock is from two possible sources, either the crystal oscillator (if there, look for a crystal) or an external oscillator.
toggle quoted message
Show quoted text
When the processor boots up, I expect it to run off the crystal oscillator (generally lower frequency, the processor would like 60 Mhz or so internally). Making guesses here, looking at the following sequence: 1) external clock is turned off.? Crystal (not oscillator, likely) provides low frequency (32 Khz or perhaps 1-10 Mhz clock).? You should be able to see evidence of a signal at the crystal pins.? Processor boots at low speed. 2) boot software enables clock multiplier (PLL) and waits for it to stabilize. (say 1-100 msec) 3) once the clock is ok, the internal clock source can switch to the 60 Mhz (or so) from the PLL. Likely that the clock input from the crystal is lower frequency, and can be used for a low power mode. once the PLL is stabilized, switching between clock sources (PLL= 60 Mhz) or crystal (1-10 Mhz?) is easy enough.? Software timing loops need to know what's what, as do hardware timer settings. The system needs to make that kind of adjustment to figure out how to get the same delays and timing regardless of clock. Likely, low power mode does lots limited stuff and may not need to do all this. Harvey On 1/12/2023 5:51 PM, Paul Carew via groups.io wrote:
Hello Harvey, |
|
Paul-
toggle quoted message
Show quoted text
Can you test the clock buffer out of circuit? Could it have failed such that it's enable input is shorted low? Could something have happened to a pull-up on the scope board that's now ineffective? Dave Casey On Thu, Jan 12, 2023, 5:45 PM Harvey White <madyn@...> wrote:
This is a fairly standard, but limited (it's old) clock tree. There is |
|
Hello Harvey and David,
Thanks for your thoughts and insights, it's definitely helping to advance my thinking here. On one of the EEVBlog forum pages yesterday, a contributor known as "Sicco" from the Netherlands, just posted some help ( (3054b)-possibly-processor/msg4634896/#msg4634896 ) He has a board where he has pulled off the BGA processor and has discovered that the clk signal is actually the 25Mhz signal that is input to the 7Z26 buffer. Paul |
|
Hmmm, which indicates that the processor does not gate the clock to its own clock.? So it's getting a clock signal.? It's supposed to have a clock out signal somewhere, Hopefully, the clock out is available somewhere.? If this were not a BGA, you'd have better access to the pins.
toggle quoted message
Show quoted text
Harvey On 1/13/2023 11:25 AM, Paul Carew via groups.io wrote:
Hello Harvey and David, |
|
Hello Harvey,
Yes agreed, it would be different if this wasn't a BGA. I'll try and see if I can get any life out of the Debug port as suggested, but I'm thinking that it is likely that I'll have to try and pull the BGA, or at least 're-flow' it as a first step.. I'm going to see if I can find some scrap boards with BGAs on them to practice first :-) |
|
Agreed.? You need a JTAG debugger, I think.? This talks to a separate set of hardware with its own clock.? Not sure how it's going to work, never tried it.
toggle quoted message
Show quoted text
I'd try reflow first.? You have the equipment for that? If you're going to the effort of removing the BGA, you might as well replace it. I'd be looking into another board if possible, just because. No idea how easy it is to find the processor, though. Best of luck. Harvey On 1/13/2023 12:35 PM, Paul Carew via groups.io wrote:
Hello Harvey, |
|
The IR preheats the board to 150 to 200 degrees C, not at the melting point of solder.? It minimizes the thermal shock on the top chips.? It also minimizes the amount of heat needed to get the BGA chip off.? There's a definite limit for how long all this should take.? You should have a shroud to fit the BGA chip, helps minimize heat on other parts.? Be careful with the air flow, you can blow off small parts (don't ask).? The entire chip needs to be heated, unlike TQFP parts which only need the edges heated. (obvious, but thought I'd mention it).
toggle quoted message
Show quoted text
Definitely not an expert here. Harvey On 1/13/2023 1:57 PM, Paul Carew via groups.io wrote:
"..I'd try reflow first. You have the equipment for that?"I have a 'Hot Air' station and I just purchased a T862++ IR station, but no experience yet with IR. |
|
I'd at least try a debugger before pulling the BGA. Technically it uses BDM
toggle quoted message
Show quoted text
for the processor debugging. It might support JTAG for boundary scan (interconnect, etc), but good luck writing that netlist... BDM is available at the 100pin expansion connector. Keep reading those eevblog threads to find where others have already gotten in there successfully. Dave Casey On Fri, Jan 13, 2023, 12:50 PM Harvey White <madyn@...> wrote:
Agreed. You need a JTAG debugger, I think. This talks to a separate |
|
Well I know it has been a year and a half since I last posted on this thread, but I *finally* got around to removing/replacing the Motorola Processor MPC860 BGA.
I replaced the chip with a new one purchased from Motorola/NXP And.....drum roll.... **No Change** :-( :-( Same symptoms, blank screen with white back light and fan running, no other activity So after all this time, I think that was a 'red herring' Hmmmm.... |
to navigate to use esc to dismiss