Jump to content

Photo

RF Problems? Poor picture quality.


57 replies to this topic

#51 Keatah ONLINE  

Keatah

    Quadrunner

  • 17,310 posts

Posted Wed Mar 15, 2017 1:26 PM

Maybe. But with what? Equally aged chips?

 

Or test for ringing and crosstalk on address/data lines and filter that out? Maybe.. Needs on-bench testing to determine the cause.



#52 AtariLeaf OFFLINE  

AtariLeaf

    Quadrunner

  • 9,943 posts
  • Location:Ontario Canada

Posted Wed Mar 15, 2017 4:11 PM

Maybe. But with what? Equally aged chips?

 

 

 

True, so just living with it is probably the best to hope for. As long as it isn't causing damage to anything I guess it's simply the result of 40 year old parts.



#53 Keatah ONLINE  

Keatah

    Quadrunner

  • 17,310 posts

Posted Wed Mar 15, 2017 4:37 PM

Pretty much I guess. It is important to understand how chips age. In semi-layman's terms:

 

Once a chip is fabbed and packaged for sale, and eventually put into use, it operates at X voltage and Y current. Over time these power requirements may increase slightly. A chip may overshoot logic high and ring a little, or it may be slow to rise. It may need a slightly higher trigger point. But there is a limit as to how the specs will drift before stabilizing for many decades or even centuries.

 

Good design will have anticipated and compensated for these expected changes on the drawing board. Good design will allow the variances to happen with no visible effect to the end-user. So I'm saying that parts may have drifted out of tolerance enough to cause interference. The design of the VCS may not be "tolerant" or "generous" enough to allow for 40 years of cumulative drift.

 

Who knows, it may stabilize at the 50 year mark and the interference will never get worse. Or it may continue. In any case, simply replacing parts of the same spec and of the same age/usage isn't going to fix anything. Not unless the replacement part is closer to what was specified at the factory.

 

Additionally, the design can be changed. And hobbyists are trying that with video mods, regulators, and new caps. But I've not seen anyone do an analysis of what fell out of specification and why. Caps are easy, you visibly see a failure, you can measure it instantly with a DMM. But how about those logic lines? Anybody poking around there? I didn't think so..

 

Design changes are going to be needed to accommodate what is effectively a different TIA today than what it was 40 years ago. This might include different trace spacings. Extra buffers between the 6502/RIOT/TIA. Additional pullups. Traps and filters that resonate at a wider range of frequencies, a wider band/notch filter for example. Different methods of couplings.

 

Compare it to a high-mileage engine, the spec of the oil may need to be beefed-up to meet different requirements.



#54 DrVenkman ONLINE  

DrVenkman

    Stargunner

  • 1,954 posts
  • Back off, man! I'm a scientist.
  • Location:KMBT

Posted Wed Mar 15, 2017 4:49 PM

To make matters more confusing, stop and think about how many chip fabs made 65xx chips under license, and for how long? MOS, Rockwell and Synertek all made them by the millions in multiple facilities and for a very long time. I have Rockwell-made 6532 RIOTs fabbed in Mexico as late as 2005 or so inside two of my early/mid-80's Woody and Vader machines. Manufacturing equipment ages, test equipment ages, people come and go, materials vary from year to year to decade to decade ...

 

We ought to just be glad it works at all. It might not keep working forever. :) 



#55 AtariLeaf OFFLINE  

AtariLeaf

    Quadrunner

  • 9,943 posts
  • Location:Ontario Canada

Posted Wed Mar 15, 2017 5:05 PM

Pretty much I guess. It is important to understand how chips age. In semi-layman's terms:

 

Once a chip is fabbed and packaged for sale, and eventually put into use, it operates at X voltage and Y current. Over time these power requirements may increase slightly. A chip may overshoot logic high and ring a little, or it may be slow to rise. It may need a slightly higher trigger point. But there is a limit as to how the specs will drift before stabilizing for many decades or even centuries.

 

Good design will have anticipated and compensated for these expected changes on the drawing board. Good design will allow the variances to happen with no visible effect to the end-user. So I'm saying that parts may have drifted out of tolerance enough to cause interference. The design of the VCS may not be "tolerant" or "generous" enough to allow for 40 years of cumulative drift.

 

Who knows, it may stabilize at the 50 year mark and the interference will never get worse. Or it may continue. In any case, simply replacing parts of the same spec and of the same age/usage isn't going to fix anything. Not unless the replacement part is closer to what was specified at the factory.

 

Additionally, the design can be changed. And hobbyists are trying that with video mods, regulators, and new caps. But I've not seen anyone do an analysis of what fell out of specification and why. Caps are easy, you visibly see a failure, you can measure it instantly with a DMM. But how about those logic lines? Anybody poking around there? I didn't think so..

 

Design changes are going to be needed to accommodate what is effectively a different TIA today than what it was 40 years ago. This might include different trace spacings. Extra buffers between the 6502/RIOT/TIA. Additional pullups. Traps and filters that resonate at a wider range of frequencies, a wider band/notch filter for example. Different methods of couplings.

 

Compare it to a high-mileage engine, the spec of the oil may need to be beefed-up to meet different requirements.

 

Sounds like you know your stuff, you should pick up a 2600 or two and do some experiments and see what you come up with. I don't even understand half the words you used in your post :D



#56 Keatah ONLINE  

Keatah

    Quadrunner

  • 17,310 posts

Posted Wed Mar 15, 2017 6:41 PM

I have high confidence that healthy chips with no grown defects or impacts from static discharge will continue to operate for another 150 years. Most formally trained semiconductor engineers peg the lifespan of the early chips at 200 years.

 

We're coming up on 50, so far so good. Little bit of tolerance changes here and there, but the logic is intact. Early chip logic can work with a wide range of voltages, CMOS even more so than the old TTL. Early logic was not necessarily meant for controlling analog parts, certainly not in something as demanding as the precision required for video generation. But they worked!

 

Regarding static discharge. It's not all that different from a pot-hole in the road. It starts small, perhaps a kid sitting the street playing with a hammer, lucky he didn't get run over! Anyways, a tiny crack happens, and then gets bigger and bigger as it's agitated by passing cars and weather and thermal expansions from day/night cycles. Eventually the hole becomes too big. And the road becomes impassible.

 

Growing defects from a static hit have been observed under a microscope. Observed to get bigger and bigger with usage. That's why a gentle static hit doesn't kill the chip right away. But life is shortened. Electrons in the silicon start migrating away, the gap gets bigger, heat goes up from all the current flowing through a smaller and smaller pathway, the chip warms. And one day that exponential rise blows the chip. Poof!!

 

Furthermore, as a chip ages it may develop sharp radiative spots at certain points, from electron and material migration.. at random. The chip now has a microscopic interference transmitter. Blasting out RF noise through that new miniature antenna.

 

"Splashes" of material, like a rock and crater, can cause shorts, both immediate and latent. Just like a Giant Meteor crashing into a city, certainly gonna interrupt roads and sidewalks!

 

Not weird enough for you? Check the space agency's research into whiskers. This is where metal material grows like a plant. And eventually a branch touches something it isn't supposed to, like a nearby trace, and now you have a short. A very real problem in early satellites. It explains why some have been dead for years and all of a sudden, because of thermal cycling, the whisker breaks, and the circuit now works again. By the way this is visible in classic radios and other vintage electronics with nothing more than a jeweler's loupe. And it can be felt on the on the unpainted inside part of the Apple II baseplates.

 

Look. Here's a couple of photomicrographs I pulled at random from google image search or the articles listed below.

 

Here we have what looks like a big splat. Something blew and debris went everywhere.
E40Fig01B.gif

 

This circuit may continue to work for a while. But it can develop a hot spot where the hole is. And under maximum rated load it might blow. As the part ages, all the other pathways have enough material to last 200+ years, but not that spot. It might make it 10 years.

image1.jpg

 

In similarly damaged circuitry, current might leak, make a hot spot, and cause higher power drain. It may effect how other parts of the circuit work, like voltage-controlled oscillators, or electrically-tuned filters. It might even throw out negative feedback circuits and regulators causing them to operate at a different set-point.

image4.jpg

 

I like this one, it shows clear damage and a near-complete break in a pathway. If that's a non-critical pathway, it might radiate RF from those jagged edges and gig's up! Read the below articles to learn more.

main-qimg-a2a303f5bc843439e9f7cd7b5331b8f1.jpg

 

Now for the whisker. Strange little fuckers. But all you need to know is some metal combinations that include Zinc/Tin can grow these crystals. If the aerospace industry had problems with this, what does that say about 50 year old consumer electronics? Thankfully whisker resistant solder and metals are now standard fare in electronics, so this is less of a problem. This can happen both inside a chip, and outside in free air, or vacuum for that matter.

8529sci1a_5cxd.gif 855_13_34_figure_1new_1.jpg matte-tin-plated-IC-01-med.jpg

 

A gross example on a macroscopic scale. All it takes is one of these wisps to short a line. Out of sight, under a chip, on the back-side of the visible pins. Or maybe under the coating of a PCB. Rare, but something to keep in mind.

Zinc_whiskers.jpg 500x1000px-LL-5d55bd49_AgWhisk-PrimCont-after-fail1.jpeg

 

..and finally illustrating the importance of clean contacts and connectors between individual IC pins and other parts. Yup. I've saved cartridges and vintage interface cards by getting rid of such nastiness. No doubt that foreign materials like moisture, sugar, skin oil, residue from inappropriate cleaning materials and a host of other substances exacerbate this problem. Sometimes all that's needed is a good cleaning and coating to get another 30 years life out of vintage electronics.

SilverSulfideWhiskers1.jpg whiskersfig1.gif

 

The internet is chock-full of material on these topics! It isn't my intent to provide a tutorial, but rather bring awareness to unusual common modes of failure you may not be aware of. Why not conduct your own research beginning with some randomly picked links?

http://www.sourceres...r/ESD.cfm?emART

https://www.maximint...ndex.mvp/id/639

https://en.wikipedia...er_(metallurgy)

https://www.nist.gov...-whisker-growth

https://nepp.nasa.gov/whisker/

https://www.maximint...dex.mvp/id/5250

 

Ohh and by the way, it is possible to repair these damage chips. The process is extremely expensive and typically used on the initial tape-out at the fab. First prototype of a new microprocessor for example. Ironically the art is becoming more refined, yet, at the same time simulation tools are alleviating the need. More and more first-run chips are working as intended. And when they don't, it's usually a gross overlooked logic flaw not covered in simulation. Most complex microprocessors can have some flaws programmed out via uploadable microcode, too. Or pathways rerouted, at the expense of clock-speed.

 

---

 

So back to the VCS and videogames and vintage electronics! While semi-randomly chosen component swaps based on symptoms can work sometimes. The real solution is going to come from a thorough analysis. It will come from knowing what parts do what, how they do it, and how they interact with other parts in any given circuit. Good luck to all in current and future repairs!



#57 Nebulon OFFLINE  

Nebulon

    Stargunner

  • 1,588 posts

Posted Fri Mar 17, 2017 2:41 PM

I'm definitely not well-versed in electronics. So I'll ask the potentially stupid question:

 

Is there some kind of alternative to the typical types of capacitors that are in the retro machines we know and love?

 

In other words, is there something better that we could be re-capping the caps with to ensure longer system life?



#58 Keatah ONLINE  

Keatah

    Quadrunner

  • 17,310 posts

Posted Fri Mar 17, 2017 5:12 PM

Sure. There's gel and solid-state capacitors. Most are surface mount, so you'd need to rig up a way of securing the part to the board, and then jumpering it in place. Caps like from PC motherboards and graphic cards. Some of those are real high quality made to deal with loads in excess of what's in the VCS.






0 user(s) are browsing this forum

0 members, 0 guests, 0 anonymous users