Electronics > Beginners
Took the Plunge - GoldStar OS-7020 Oscilloscope
Whales:
--- Quote from: rsjsouza on February 03, 2020, 11:43:19 am ---One question I have: does the process of removing the background shade performs yet another decompression/recompression? I suspect not, but wanted to ask.
--- End quote ---
Yes it does.
Original jpegs -> decompress into raw (in-memory) pixels -> adjust levels to remove background ghosts -> compress into new jpeg
There are tools that can edit JPEGs directly (ie without decompression), but they're pretty rare and would be limited. It should be theoretically possible to keep most of the JPEG DCT blocks and throw out just the ones you don't want, ie 'lossless' editing of a lossy format, but I'm not aware of any easy ways.
The only mainstream/common lossless JPEG editing available is image rotation, but even then some tools stuff it up and re-compress the image anyway. You only need to edit some metadata for this one.
The saving grace is that I destroyed a lot of other detail anyway, so having an extract decompress + recompress really didn't do much harm :P
--- Quote ---In my opinion, however, Brumby's grayscale version looks much more appealing to my eyes than your two versions above.
--- End quote ---
I agree as well. I quested for a compression method small enough to attach the PDF on the forums, too much was thrown out.
In 2020: a 20MiB PDF is harmless to have on disk, just annoying to transfer over some people's internet connections. Getting it down to the size of a floppy isn't really necessary unless you face arbitrary rules where you want to upload it.
--- Quote ---One thing I may do to this and another manual I posted there is the removal of the shading of the other pages - a brightness setting on the scanner would have minimized/solved this.
--- End quote ---
N.B. 'brightness' on different devices means different things. Sometimes it adjusts both the black and white points, others only the white point, in some it even adjusts gamma instead. As a general rule for monitors and scanners: ignore what the controls are named, fiddle with them and work out what they mean yourself.
The 'Xsane' scanning utility (Linux) is getting old, but it still works and provides lots of level adjustments & preview options. It was my choice the last time I did some mass scans, but that was a few years back.
Whales:
--- Quote from: Brumby on February 03, 2020, 12:33:04 pm ---This is why I avoid threshhold processing for text. The intermediate intensity pixels add clarity when viewed by our eyes, which - in conjunction with the human brain - have developed algorithms to see things much more clearly.
--- End quote ---
Yep.
In some situations we can cheat: really high res vector art or vector text can be saved in a thresholded format, then when it gets downscaled onto your screen by your image viewer it looks nice and smooth again. Unfortunately that's not really possible in this situation.
--- Quote ---Reducing the ghosting from the back of the pages would have been useful
--- End quote ---
N.B. I had a hard time getting this (levels/curves adjustment) to avoid damaging the greys around the text itself. Text didn't look anywhere as nice, even with some gamma tweaking.
There are some smarter tools out there like unpaper. Not perfect and the output has sections whited out (read:ugly), but pretty good at preserving the text you want.
--- Quote --- - so would converting from a colour palette to greyscale.
--- End quote ---
This may not save as much data as you expect. JPEG already heavily subsamples the colour out. Not sure of exact %.
olafwagner:
I have been quiet for a couple of days waiting for my 3M resistors to arrive.
1. Inspecting the PCB
While waiting for them, I took a closer look at the PCB, and discovered a problem. Firstly I removed the 1M resistor, and when I removed it I saw the pad for the original resistor was missing (probably why it was hacked onto the bottom). Also, I saw that the connector that goes to the 'Focus' pot was loose, and had a broken trace. This immediately gave me a good theory on why the HV rail was first too low, and subsequently too high. My assumption was that focus is connected directly to the HV rail (through the aforementioned R697 in yellow on the attached schematic), and when this resistor was 1M it was 'draining' too much of the HV rail through the focus pot. I believe when I removed the PCB the first time, it may have caused the trace to get damaged (I was so gentle I assume it was already somewhat damaged, but me removing it broke it, and as a result there was no longer ANY current running from the HV rail through R697 and through the Focus pot. Hmmm....
2. 3M Resistors (Part 1)
Eventually the resistors arrived, and I saw they had '3M' written in pen, but fortunately I double-checked with my DMM and they turned out to be incorrectly labeled. They were actually just 3 Ohm resistors (that would have beeen fun had I installed them). I looked under a magnifier and saw the colors were Orange, Black, Black, Silver, Brown, which backed up the DMM. (3 (Orange) 0 (Black) 0 (Black) * 0.01ohm (Silver) = 3 ohm (with a +-1% tolerance (brown)). I contacted the Ebay seller and to his credit he immediately sent me the proper resistors. Just more time to wait...
3. Checking the Caps 'In-circuit'
While I waited I thought I would turn my attention to all the relevant capacitors in both the HV and blanking circuits. However, I didn't want to remove each and every one to test them, so I thought I'd do a 'cursory' in-circuit ESR test to see whether anything obvious pops up. The problem is that my ESR meter only offically measures down to 1uF (I have a Signstek MESR-100 v2), and many of these caps are much lower, but regardless I soldiered on and marked those that could be measured (and measured good) in green, and those that came back with no resistance measurement at all in yellow) Unfortunately this didn't highlight any caps as being problematic (I guess that doesn't neccessarily mean that they aren't part of the problem, just not obviously so - or are in-circuit ESR measurements somewhat dependable?)
4. 3M Resistors (Part 2)
The correct 3M resistors arrived today, so I immediately installed one (as well as a small bodge wire to address the missing pad and broken trace). I also replaced both transistors Q621 and Q622 (might as well while I have the board out), and then reinstalled everything and powered up the scope.
At last, I can report a 'baby-step' of progress. The HV rail now reads -1,821v, which now falls WITHIN the spec range of -1,995v to -1,805v (only just, but I'll take it). :-) OK, so my theory above on the wrong resistor value and broken trace turns out to be correct.
Unfortunately, all my other issues remain, so there is still some work to do. :-(
Two of the issues which I still feel must be related are:
- Unblanking signal (which I can see at "P1") does not make it to "P2"
- The INTEN Pot (or internal trimmer) are unable to reduce the signal to a blank screen.
5. Next Steps
So I have 3 theories (options?):
Theory #1: Capacitors: There are only 3 caps in this area (C659, C650 and C649.) (While my schematic shows C671 it is not physically on my PCB nor is there a silkscreen marking for it) I could pull those 3 caps from the board and test them properly out of circuit. I don't feel very strong about this because all 3 tested fine in circuit.
Theory #2: Transistors: There are 2 more downstream transistors: Q632 (2SB648A which is a PNP Hitachi Low Frequency High Voltage Amplifier) or Q623 (2SD668A which is the complementary NPN Hitachi Low Frequency Low Voltage Amplifier) Does it make (theoretical) sense that if either of these transistors had a problem it would affect Q622? Doesn't seem logical seeing as the emitter of Q622 goes both to the base of Q623 as well as -12v (through R691), so I would expect Q622 to function regardless of whether its output signal is amplified or not.
Theory #3: Diodes: Another very weak theory, but I am unable to test these diodes in-circuit with my DMM, I just don't get a 'beep' in either direction on ANY of the diodes (my DMM just doesn't crank out enough volts I guess). Is there a safe way to test these diodes in circuit? Or should I pull them and plug them into my component tester? (there are about 4 or 5 in the vicinity).
Those are pretty much all the options I can think of. The only other components are resistors, and I have measured all of them in-circuit and they all read very very close to their expected values.
olafwagner:
I was poking around the board some more taking measurements, and I got 'something' at "P2" by turning the intensity ALL the way up, but what I see doesn't make sense.
I left the scope running, had some dinner and when I came back I switched it off, and heard a lot of static 'pops'. I switched it on again, and while measuring some more, I heard more pops that sounded like they were coming from near or around the CRT. As I switched off the scope and was staring at the CRT I saw a few tiny sparks (accompanied by pops) on the power supply board (on the negative pin of one of the big caps) I immediately discharged all the caps again and removed the power supply board.
When I inspected the board, I saw yet another damaged trace at capacitor C821, which sits between the 160v and 180v rails. I measured the ESR on the cap (and it measured good), so at first I just wanted to fashion another bodge wire, but I thought I am so close, I am going to remove this cap and test it properly.
I heated up the soldering iron, and as soon as I touched the negative pin (where the broken trace is), I heard the classic 'simmering' sound, and I knew, this cap has leaked. I then removed the cap entirely, and it is nice and wet on its underside (see attached pics).
Also when I removed it, (of course), the pad immediately broke loose. The broken trace also means that the 160v rail will not work (because that pad sits inline), so while I didn't retest my 160v rail recently, I assume that it couldn't have worked (or at least not properly)
For giggles I hooked it up to my LCR and it still measured 763uF (should be a 1000uF cap) which is more than I was expecting, but certainly far out of 'spec'.
Now I need to order a replacement (I don't have anything close to this large in my stash). I am curious that this is only a 50v cap, when it sits between the 180v and 160v rails. (even the schematic calls for a 1000uF 50v cap) Is this because the delta is just 20v? Or should I try to find a cap with a higher voltage rating?
AVGresponding:
The cap rating of 50V is because in normal operation it doesn't see more than 20V (though if one of the rails it's connected to fails to 0V, it's not gonna be happy!), and a 1000uF cap with a say, 200V rating would be big and expensive.
Since its bum is still wet, I'd guess it's a recent failure.
One problem with testing caps for leakage failure is that it often doesn't manifest until you get close to operating voltage, or at any rate, well above the voltage that most LCR meters put out.
EDIT: Also, :wtf: on those 3M resistors! Good thing you did check them before using them!
Navigation
[0] Message Index
[#] Next page
[*] Previous page
Go to full version