exxos blog - random goings on

[mikro]

It's not the first time this kind of thing has happened for a Falcon recently, I remember Mikro fixed a Falcon by putting the motherboard in a dishwasher.

To be more precise: it fixed it at some point because I had a solder blob left on the Combel so the washing process moved it by some micrometres. But in the end I had to find the solder blob and remove it: viewtopic.php?p=83446#p83446

[exxos]

Been doing more tests this morning with the clock patches. Long story short I cannot really find any difference between any clock patches or REV A , C/D Falcon. Of course I cannot really spend the time and testing every single piece of software which may malfunction, so my tests could may well be inconclusive.

In general though ,everything behaves as expected and I still cannot conclusively say why there are so many versions of the clock patch because long as the signals are clean and stable and I really does not seem to be any issues. At least with DSP2SDMA and GB6 stability tests.

My version 3 inverting and version 4 noninverting clock patch doesn't seem to make any difference. Though I think technically my V3 patch may be better than the V4 in terms of signal quality. But they tackle and solve problems in different ways.

So i'm afraid I cannot really say why some people have nothing but trouble with clock patches and others don't. But it is entirely possible that the problems they are having are completely unrelated like the reset problems I was having which seemed to manifest itself as unstable clock patches.

Of course my sample set of two falcons isn't really good enough to form any solid conclusions but my V3 patch should work best going by my current tests on @JezC REV A Falcon and my own more common rev C/D.

I do really ponder why the SDMA chip was changed in @JezC Falcon. I'm not aware of there being any early or bugged chips ?

So I think I probably fit the V3 back into @JezC Falcon, then leave it running all day to make sure its all stable, then it can be shipped back. Don't know if your passing anywhere near on the M1 @JezC as if your basically going past at any point, be safer to pick it up than ship it..

[JezC]

Not sure when I will be cruising down the M1 before early November yet @exxos but makes sense not to ship it unless we have to...

Let me get back to you in a few days if that's ok?

Fingers crossed she behaves herself for you ( and doesn't draw any more blood)!

[exxos]

Let me get back to you in a few days if that's ok?

Yeah no problem it would probably take me a couple of weeks to get around to boxing it up anyway...

[JezC]

Let me get back to you in a few days if that's ok?

Yeah no problem it would probably take me a couple of weeks to get around to boxing it up anyway...

Ok, probably going down & up the M1 on the 8th November (in a large van...) as we help one of the kids move into a new flat...but there's a slim chance I'll be down around Mansfield before then (in my last couple of weeks of gainful employment)...

[exxos]

Ok, probably going down & up the M1 on the 8th November (in a large van...) as we help one of the kids move into a new flat...but there's a slim chance I'll be down around Mansfield before then (in my last couple of weeks of gainful employment)...

Weekends and wednesdays I am normally in derby.. So about 10 mins off the M1. Other days could be tricky. There is a shed outside the flat where stuff gets dumped for deliveries , could pickup anytime/day from there, but its not very secure.. so not a good option really. So you have to let me know on days to see if we can make it work.

You could pickup from Uttoxeter almost any time/ day, but that's going to add a hour easily onto your traveling.

[exxos]

My UK PCB assembler and I need some additional data which I haven't dealt with before. Mainly they want to know circuit size, the panel size, the edge rail size, and where the fiducials are. Working out the sizes is a bit time-consuming, and just a little bit annoying to manually working out where the fiducials are as its a bit more involved.

So I thought why am I even doing this in the first place went AI can figure all this out for me

I explained the problem and all the file structure which I get from JLC, gave it some example files and what I wanted to extract information wise....

info.PNG (30.71 KiB) Viewed 129 times

So now all I have to do is run the script on the zip file and it figures everything out for me now! How cool is that ! Another AI win

[exxos]

I asked Grok to describe what it saw in this image..

content.jpg (205.61 KiB) Viewed 75 times

The image depicts the screen of a Rigol digital oscilloscope in "STOP" mode, capturing a single waveform trace. The display shows a cyan-blue signal line on a black gridded background, representing a voltage-over-time measurement.
The waveform appears to be a high-frequency oscillating signal with characteristics of ringing or damped resonance, commonly seen in electronic circuits (e.g., due to inductive or capacitive effects in response to a pulse or switching event). It starts on the left with low-amplitude noise or small oscillations, then transitions into a series of peaks and troughs with increasing then decreasing amplitude—peaking around the center of the screen before tapering off toward the right. The signal crosses the zero-voltage line multiple times, suggesting an AC component or underdamped response.
Key on-screen measurements and settings include:

Vmin[2] = -368mV: The minimum voltage measured on what seems to be channel 2.
Vpp[2] = 600mV: The peak-to-peak voltage amplitude.
**Freq[1] = **********: Frequency measurement is unavailable or errored (displayed as asterisks).
Vertical scale: 200mV per division (indicated at the bottom left, with the grid divided into 8 vertical divisions, implying a full-screen range of about ±800mV around the center).
Horizontal scale: 20.0ns per division (time base, with the grid spanning 10 horizontal divisions, covering roughly 200ns total).
Trigger position: -8.000ns (marked with a "T" icon on the screen, indicating the trigger point slightly left of center).
Additional reading: "F 2 232mV" at the top right, possibly a cursor or function measurement (e.g., frequency or voltage at a specific point).
Cursors: Two vertical cursors (orange lines labeled with triangles) are placed across the waveform, likely for manual measurement, along with faint horizontal cursor lines near the top and bottom.

The oscilloscope is labeled "RIGOL" at the top left, and the overall setup suggests a captured transient event in a lab or testing environment, with no other channels or overlays visible. The photo of the device includes part of its beige casing and a slight reflection or glare on the screen.
125 MHz
To arrive at this: The oscilloscope's cursor measurement shows a time delta (Δt) of 8 ns between points one period apart on the waveform (from trough to trough, ignoring the negative sign which indicates cursor order). The period T is thus 8 ns, and frequency f = 1 / T = 1 / 8 × 10⁻⁹ s = 125,000,000 Hz = 125 MHz.

[exxos]

Needed to test some DRAM today. So as my H5 was setup anyway.. I plugged in the RAM test board (with the ST536)

IMG_4155.JPG (375.69 KiB) Viewed 27 times

BUT it went totally nuts ?!

IMG_4156.JPG (286.58 KiB) Viewed 27 times

Pressing down on the DRAM board caused it to crash..

IMG_4158.JPG (228.83 KiB) Viewed 27 times

But not touching it...

IMG_4160.JPG (258.03 KiB) Viewed 27 times

It is interesting that somehow TTram fails when ST-ram is acting up.. That's a bit odd...

So putting back the SIMM...

IMG_4161.JPG (329.93 KiB) Viewed 27 times

IMG_4162.JPG (245.34 KiB) Viewed 27 times

All is fine again..


Put the DRAM board back.. resetting like crazy again...


Took the ST536 out and just used a DIP CPU..

IMG_4163.JPG (302.23 KiB) Viewed 27 times

Now just get 3 bombs with TOS104..

IMG_4164.JPG (167.09 KiB) Viewed 27 times

Put the SIMM back in.. and all fine again..

IMG_4165.JPG (190.02 KiB) Viewed 27 times

I mean the DRAM board was working fine last time it was used, so

[exxos]

Time for the diagnostic cartridge.

IMG_4166.JPG (308.83 KiB) Viewed 36 times

Luckily I labelled up the DRAM chips on the test board with the correct bit numbers

IMG_4167.JPG (197.62 KiB) Viewed 36 times

It would be somewhat unlikely that more than one chip would fail like this...


Time to investigate......

It looks like one of the socket pins on the ZIF isn't behaving correctly so far..

Also I swapped chip 8 and 15, and the fault is still showing on bit 8.. So its not the IC's themselves...

Tried a different RAM board and its fine..

IMG_4170.JPG (376.78 KiB) Viewed 29 times

IMG_4171.JPG (187.02 KiB) Viewed 29 times

So time to resolder my DRAM test board I guess....