Thanks Josh, I'll verify this. I was thinking the processor was not addressing the RAM because it is not completing the ROM addressing, but i don't know if this is the case. I'll connect the LA and verify if the ROM addressing is stopping early or the RAM decoders are not providing the proper sequence...
BTW: Brian do you have a LA so we can compare the same signals on the ROM and RAM addressing/enabling on the working and not working unit you have? I agree with Fugu, I'll replace the capacitors only if I'll have a reasonable suspicion and all the other measure make sense. Thank you Mauro Il 15.12.2017 04:04 Josh Malone ha scritto: > You mentioned the CE line. This sounds similar to the problem that I had with my 102 ram address decoder circuit. I had to replace the demultiplexer. If your lowest ram is not responding properly this might happen? > > On Dec 14, 2017 8:18 PM, "Fugu ME100" wrote: > >> If your voltages appear stable and in spec I would leave the caps, unless you see any that have leaked. A white or greenish residue under the cap is usually a good sign it has leaked. I think the green tint is because it also etched the copper under the cap. >> From: M100 on behalf of Mauro Pintus >> Reply-To: >> Date: Thursday, December 14, 2017 at 5:10 PM >> To: >> Subject: Re: [M100] diagnose >> >>> Really appreciate your effort!!! >>> I'll try to measure the suggested signals and get out my bag of GOOD caps :) >>> Mauro >>> >>> Inviato da iPhone >>> >>> Il giorno Dec 15, 2017, alle ore 01:55, Fugu ME100 ha scritto: >>> >>>> It would be interesting to see if the Address Lines fail at a specific address all the time. Might give a clue as to what is causing the issue. In the old days it was possible to watch the address lines walk through a count sequence if you had a bad ROM or RAM. Although this one appears to keep retrying something. >>>> Monitor them at the ROM pins if you can that way it will test the CPU to ROM buffers. >>>> On the Electrolytic caps I always assume they are dead - either dried out or leaked out ….. and yes as a matter of course I just replace them all, I have bags of all the ones used on the M100 and T102s. I also have a big bag of removed caps LOL. It takes me about 1-2hours to do a complete replacement of caps, if they come out cleanly. However your voltages seem to be pretty good and the CPU is booting. Usually if the +5V or -5V is off spec the caps are bad or if the LCD screen is modulating then the -5V is bad. At least that is my experience of the M100/T102. I just had to replace the power caps on a PC8201A with a bad LCD screen now it looks good - so it is a common issue regardless of the manufacturer. >>>> From: M100 on behalf of Mauro Pintus >>>> Reply-To: >>>> Date: Thursday, December 14, 2017 at 4:31 PM >>>> To: "[email protected] [30]" >>>> Subject: Re: [M100] diagnose >>>> >>>>> Actually the reset is moving only at power on and when I press the reset button, otherwise it is stable. I tried to follow the reset signal after the Reset out of the CPU through couple of buffers but it is consistent and not cycling. >>>>> As LA I only have a Saleae 8 clone, but should do the job. >>>>> I'll try the address and data lines as suggested. >>>>> Any other signal of interest I should trace? >>>>> Mauro >>>>> >>>>> Sent from my iPad >>>>> >>>>> On 15/dic/2017, at 01:12, Fugu ME100 wrote: >>>>> >>>>>> What is the RESET line into the CPU doing during the loops? Is the RESET stable? >>>>>> The CPU appears to be RESET, heads to the ROM for data but for some reason fails to complete and RESETs itself. No activity on the RAM or LCD would indicate the CPU is not completing the ROM boot. At this point I would usually hang on my Logic Analyzer to see where the Address and Data lines are heading. >>>>>> Have you checked out the Address Lines and Data Lines to make sure none are stuck at either 0, 1 or some other state? They all should show some activity during a RESET. >>>>>> From: M100 on behalf of Mauro Pintus >>>>>> Reply-To: >>>>>> Date: Thursday, December 14, 2017 at 3:55 PM >>>>>> To: >>>>>> Subject: Re: [M100] diagnose >>>>>> >>>>>>> Thank you Fugu and Brian for the feedback. >>>>>>> Yes I tried the beep BASIC test with no luck and also the cold start (the M10 and the M100 are almost identical...). >>>>>>> >>>>>>> I made few other measurements trying to replicate what Brian is doing and for sure the unit is in a loop... >>>>>>> In the linked google photo album I added another picture (the first one) showing the ROM CS (Pin 20 yellow trace) and the IO/M of the main processor (Pin 34 blue trace) >>>>>>> https://photos.app.goo.gl/8YSbaAxZBuXu5rjC2 [21] >>>>>>> As you can see at the beginning of the repeted sequence, I get only 5 pulses on the IO/M >>>>>>> The strange thing is that there is so little activity on the RAM CEx (chip enable), like nothing is loaded. >>>>>>> Finally I measured also the CSx (chip select) on the LCD connector and none are moving, confirming that on the LCD nothing can be displayed. >>>>>>> Mauro >>>>>>> >>>>>>> On Dec 13, 2017, at 7:01 AM, Fugu ME100 wrote: >>>>>>> >>>>>>>> Vee at -4.9V is good. >>>>>>>> You may have already looked at these options, so apologies if you have… >>>>>>>> The repetitive nature of the ROM CS cycling could be the CPU in a software loop waiting for keyboard input. Have you tried the BEEP test from BASIC to see if the motherboard is actually running? >>>>>>>> Have you tried a cold start? On the M100 press ctrl+pause while powering on, not sure about the M10. It could be the memory is corrupted with the loss of backup power. Always a good idea when first powering after switching off/on the backup power.. >>>>>>>> From: M100 on behalf of Mauro Pintus >>>>>>>> Reply-To: >>>>>>>> Date: Tuesday, December 12, 2017 at 5:24 PM >>>>>>>> To: >>>>>>>> Subject: Re: [M100] diagnose >>>>>>>> >>>>>>>>> Hi, >>>>>>>>> I have a NOT WORKING Olivetti M10 and a Rigol DS1104Z oscilloscope, so I made few measurement to try to help Brian and in the mean time try to troubleshoot my unit. >>>>>>>>> My unit was working properly and stopped as soon as I flipped the NiCd battery switch off while the unit was ON (done by mistake with the cover open). >>>>>>>>> Now the screen turn on but I can regulate just the contrast and no text appear on the screen. >>>>>>>>> I also made a small arduino circuit to test the LCD alone, and I'm able to write text on it, so the LCD should be fine. >>>>>>>>> BTW I think Brian is using way too long ground leads on the oscilloscope probe. >>>>>>>>> I made my measurement clipping the alligator clip to the metal case of the main crystal that in the M10 is grounded with a pice of bare wire. >>>>>>>>> Is your scope grounded? I can see in one of your pictures the usual US plug with only 2 terminals... >>>>>>>>> Here are some screenshots of my measurements. >>>>>>>>> https://photos.app.goo.gl/8YSbaAxZBuXu5rjC2 [16] >>>>>>>>> Pictures: >>>>>>>>> 1) CLK pin 37 of the 8085 >>>>>>>>> 2) Pin 7 of the 8085 >>>>>>>>> 3-4) CS Pin 20 of the ROM it seems to cycle over and over >>>>>>>>> 5) Reset * at power on >>>>>>>>> VEE is 4.9V >>>>>>>>> V across the battery is 4.02 when the unit is on and 3.9 when off >>>>>>>>> Tomorrow I'll continue hoping Brian or some of you can confirm/compare some of my measurements. >>>>>>>>> Have a nice day >>>>>>>>> Ciao >>>>>>>>> Mauro >>>>>>>>> NOTE: I think I will measure all the pins on the system bus so anyone with an oscilloscope can do the same without opening the unit case... >>>>>>>>> >>>>>>>>> On Dec 11, 2017, at 9:28 AM, Fugu ME100 wrote: >>>>>>>>> >>>>>>>>>> Ah the "special tool" requirement always hated to see that in any step….. >>>>>>>>>> What did the -5V measure on the meter? If it is not within spec neither the LCD or buzzer will work. >>>>>>>>>> From: M100 on behalf of Brian White >>>>>>>>>> Reply-To: >>>>>>>>>> Date: Sunday, December 10, 2017 at 11:30 PM >>>>>>>>>> To: >>>>>>>>>> Subject: Re: [M100] Fwd: Re: diagnose >>>>>>>>>> >>>>>>>>>>> Thanks for that tip. >>>>>>>>>>> And oh like some GM repair manuals where replacing an entire transmission might be 11 steps, but one of them is to use the $12,000 machine no one outside of the factory has ever even seen. >>>>>>>>>>> >>>>>>>>>>> On Dec 8, 2017 1:38 PM, "Fugu ME100" wrote: >>>>>>>>>>> >>>>>>>>>>>> It might be better to use the CLK (37) output as a reference if it is needed. The CLK is derived from the X1,X2 XTAL inputs and should not be impacted by the loading of the scope probes. >>>>>>>>>>>> The M100 reference manual is quite good for troubleshooting some parts, although it does have a few broad instructions like "Check All IC's". :-) >>>>>>>>>>>> From: M100 on behalf of Brian White >>>>>>>>>>>> Reply-To: >>>>>>>>>>>> Date: Friday, December 8, 2017 at 9:55 AM >>>>>>>>>>>> To: >>>>>>>>>>>> Subject: Re: [M100] Fwd: Re: diagnose >>>>>>>>>>>> >>>>>>>>>>>>> I noticed something a couple days ago. Those noisy gnd and 5v lines were only like that because I happened to have probes on both X1 and X2 at the same time. >>>>>>>>>>>>> Later I was looking at a good M100, and happened to only be watching X1 and others but not X2, and everything was much better. But just by touching a probe to X2 while another was on X1, both gnd and 5v went exactly like those pics I posted from the bad M100. I can probe either X1 or X2 and it's fine, but not bith at the same time. >>>>>>>>>>>>> I don't actually see any special reason to monitor either of those signals except briefly to verify they exist, or to provide something steady to trigger on while probing others. Just pointing out something I saw that explains those ugly traces. >>>>>>>>>>>>> >>>>>>>>>>>>> On Dec 7, 2017 11:45 AM, "Josh Malone" wrote: >>>>>>>>>>>>> >>>>>>>>>>>>>> Yeah - there is something very weird with the ground on that Tandy. >>>>>>>>>>>>>> Maybe some problematic traces? Or a bad IC leaking current to ground? >>>>>>>>>>>>>> >>>>>>>>>>>>>> Can you measure the DC power consumption? It should be ~60-70mA IIRC. >>>>>>>>>>>>>> >>>>>>>>>>>>>> On Wed, Dec 6, 2017 at 10:14 PM, Brian White wrote: >>>>>>>>>>>>>> > After watching a youtube video about how to use an oscilloscope by sparkfun, >>>>>>>>>>>>>> > I switched all probes and channels to 10x and adjusted the trim caps in the >>>>>>>>>>>>>> > probes against the built in test signal to make the square wave look square. >>>>>>>>>>>>>> > >>>>>>>>>>>>>> > With gnd ref on the cpu gnd pin 20 instead of chassis gnd, and a couple >>>>>>>>>>>>>> > probes on X1 and reset_in, and one probe on chassis ground, now the x1 and >>>>>>>>>>>>>> > reset_in signals are steady, reset_in looks flatter, and only the trace for >>>>>>>>>>>>>> > chassis ground floats and bobs up and down now. >>>>>>>>>>>>>> > >>>>>>>>>>>>>> > So chassis ground simply floats around in relation to the rest of the board, >>>>>>>>>>>>>> > or at least in relation to the cpu. Not a bad afe in the scope. *chagrin* >>>>>>>>>>>>>> > >>>>>>>>>>>>>> > >>>>>>>>>>>>>> > -- >>>>>>>>>>>>>> > bkw >>>>>>>>>>>>>> > >>>>>>>>>>>>>> > On Dec 6, 2017 7:38 PM, "Brian White" wrote: >>>>>>>>>>>>>> >> >>>>>>>>>>>>>> >> This is the bobbing I was talking about. It's dumb for me to even be >>>>>>>>>>>>>> >> talking about it yet since I haven't even done a basic comparison check on >>>>>>>>>>>>>> >> some other reference yet. >>>>>>>>>>>>>> >> >>>>>>>>>>>>>> >> https://youtu.be/b9jAYsfTOzQ [3] >>>>>>>>>>>>>> >> >>>>>>>>>>>>>> >> And here's another look at a different signal hooked up a different way. >>>>>>>>>>>>>> >> Here, I have the ground ref hooked to the gnd pin 20 on the 8085, NOT the >>>>>>>>>>>>>> >> chassis ground. And CH4 is on pin 18, AD6. The offsets, trigger, and >>>>>>>>>>>>>> >> vertical scales are all visible on-screen. >>>>>>>>>>>>>> >> >>>>>>>>>>>>>> >> https://photos.app.goo.gl/mel4CKym2EtcV1352 [4] >>>>>>>>>>>>>> >> >>>>>>>>>>>>>> >> https://photos.app.goo.gl/K1AIj3Pmt15xXfR72 [5] >>>>>>>>>>>>>> >> >>>>>>>>>>>>>> >> >>>>>>>>>>>>>> >> -- >>>>>>>>>>>>>> >> bkw >>>>>>>>>>>>>> >> >>>>>>>>>>>>>> > Con Mobile Open 7 GB a 9 euro/4 sett navighi veloce con 7 GB di Internet e hai 200 minuti ed SMS a 12 cent. Passa a Tiscali Mobile! http://tisca.li/OPEN7GBFirma
