exxos blog - random goings on

[exxos]

Reset circuit back together and all seems to be working fine now

IMG_3940.JPG (352.54 KiB) Viewed 206 times

@JezC 030 CPU & FPU was one that was only 40mhz capable. I've swapped both for 50MHz. Now running @dml FPU test programs..

Will let it all warm up, then I will have to go back to testing the clock patch stuff again.. Now that the reset side quest is completed.

IMG_3941.JPG (382.7 KiB) Viewed 206 times

IMG_3942.JPG (213.39 KiB) Viewed 206 times

IMG_3943.JPG (175.49 KiB) Viewed 206 times

EDIT:

IMG_3944.JPG (219.25 KiB) Viewed 203 times

Now to leave it running GB6 to make sure its stable in its current config.

[JezC]

Do all the FPUs fail at some point in the dml test @exxos ?

Only one failure in all those tests isn't bad, just wondering how reliable the FPUs are in general (given how few (Atari) systems have them as standard)?

[exxos]

Do all the FPUs fail at some point in the dml test @exxos ?

Only one failure in all those tests isn't bad, just wondering how reliable the FPUs are in general (given how few (Atari) systems have them as standard)?

Oh I hadn't noticed that one pass fail. I will reseat and retest when I get chance. I mean they were all tested in my own Falcon fine

But yeah, 50MHz is overclocked, even though I test them as much ass possible, I can't guarantee the outcome indefinitely.

I think part of the problem is the whole board heats up over time. It's not "out of spec" by far. But the heat from the CPU via the socket, and PLD warming things up, the whole PCB can get pretty warm even in "sunny old UK". and that's without being enclosed in the case as well. It's why I suggest some air movement over it all if shut up in the case just to keep things running cooler.

[JezC]

I have both large and small heatsinks from the store to fit plus a fan (Mega ST replacement, again from the store) destined for this setup once I get it back so hopefully it will run cool enough to avoid problems!

Glad it's helping the investigation even if it is slow going!

[exxos]

I have both large and small heatsinks from the store to fit plus a fan

It is possible the metalwork could even contact the top of the CPU with some compound or some conductive carbon sheet. But is not something I have had time to look into. I did put a small fan in my metalwork back in the 90s.

[exxos]

Seems I'm getting much more expected results now DFB1X is running stable now.

A few days ago I had a series resistor with the SDMA clock on the V3 (62R) where it did not work with this resistor. It is now running fine.. But need to leave it on another hour to make doubly sure..

I'm also testing on various voltages on the clock patch driving the buffer from around 4.50V to 5.85V to see if voltage makes any odds.

When I was doing tests before, and things were unstable, or very intermittent failing, it must have been down to the reset circuit as I am not experiencing anywhere near as many problems as I was originally.

I'm just guessing but maybe the reset circuit could have been pulsing fast enough to crash the machine but not long enough to cause a total system reset.

I think the changes I was making were mostly red herrings. As it seemed like it was more stable when the machine was cold and the longer it was on the more unstable it become. But this I think is down to whatever is was shorting out on the reset circuit. As it warmed up the short got worse I guess.

I'm still perplexed about the whole thing! I mean I have poked around and prodded everything around that reset circuit. Even heated the chips up to try and remove them as well so if there was a bad solder joint it would have reflowed anyway.

Plus the oddity where it was a consistent 150R all the time which is weird. Maybe runtime wasn't a factor after all.. Or maybe it could have been a high resistance originally before I started measuring it.. Hard to tell now..

In any case it has not tripped up once all day..yet...

Won't have time to do anything over the weekend but will probably continue on Monday doing more clock patch tests.

[Steve]

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.

[exxos]

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.

That vaguely rings the bell now, you mention it..

I asked Grok what it thought, but I think half the time it just agrees with whatever I say.

. Yes, soldering flux residues can become conductive over time, particularly after decades like 30 years, due to factors such as moisture absorption (hygroscopic properties), corrosion, dust accumulation containing conductive particles, or chemical changes in the residue itself. This is more likely with older fluxes from the 1980s and 1990s, which were often rosin-based (like RMA or RA types) or even acid-based in some applications, as these could leave behind residues that weren't always fully cleaned or neutralized. In electronics from that era, improper cleaning could lead to gradual conductivity increases, potentially causing shorts, leakage currents, or failures in high-impedance circuits. Even "no-clean" fluxes, which became more common in the 1990s, aren't always truly residue-free and can pose similar risks under humid or contaminated conditions over long periods.

[exxos]

Asked grok more info. I've not fact checked it all...


Understanding Conductive Flux Residues Over Time in Older Electronics

Why Flux Residues Become Conductive Over Time

I've been troubleshooting an issue on an old PCB from the 1980s-1990s era with a signal line showing a malfunction—measuring about 150 ohms to ground despite a 1k pullup to 5V. Cleaning the area with isopropyl alcohol (IPA) resolved it, pointing to old soldering flux residues as the culprit. Below, I merge and explain the key details on why this happens, including mechanisms, experimental evidence, and specifics for older leaded solder systems. This is based on technical studies and standards—hope it helps if you're dealing with vintage electronics!

• Moisture Absorption and Electrolyte Formation: Many fluxes (e.g., rosin-based like RMA or RA types common back then) contain ionic activators, weak organic acids (WOAs like succinic or glutaric), or halides. Over time, these residues absorb atmospheric moisture, forming a thin water film or electrolyte on the PCB surface. This enables ionic mobility, reducing surface insulation resistance (SIR) and creating leakage paths—think partial shorts like your 150-ohm reading to ground.
• Electrochemical Migration (ECM) and Dendrite Growth: Under circuit bias (e.g., your 5V pullup), metal ions (from tin, lead, or copper) migrate, forming needle-like dendrites that bridge conductors. This is amplified by flux acids or halides in humid conditions (e.g., 85-90% RH at 40-85°C), leading to conductive paths and failures in high-impedance circuits.
• Aging and Contamination: Older fluxes weren't always fully cleaned (e.g., using CFC solvents phased out post-1987), leaving higher ionic contamination. Over 30+ years, dust with conductive particles accumulates, or residues hydrolyze, increasing conductivity. Even "no-clean" fluxes from the 1990s can degrade under humidity. Industry stats show over 50% of PCB failures stem from such corrosion or shorts.

IPA cleaning works because it dissolves organic residues, removing the ionic contaminants and breaking the path. Use 99% pure IPA, brush/agitate, and dry thoroughly to avoid recurrence. For long-term fix, consider conformal coating in humid setups.

Experimental Evidence Proving Conductivity Over Time

Multiple studies use accelerated aging (high humidity/temperature) to simulate decades of exposure. Here's a summary of key experiments:

• Transformation of Reflow Solder Flux Residue Under Humid Conditions (2021): Tested ROL0 pastes with WOAs on PCBs. Under low humidity (<60% RH), no changes; but high RH caused residues to "open up," absorb moisture, and increase conductivity via acidic evolution. Measured via extract conductivity, acidity gels, and microscopy—proves time-dependent degradation.
• Effect of Flux Activator on Climatic Reliability (2023): Evaluated pastes on SIR patterns under cycling (25–55°C/98% RH) and constant humidity. SIR chronoamperometry showed leakage currents >10 µA, EIS impedance drops after 7–14 days, and dendrite growth. Conductivity rose (e.g., to 75 µS/cm) due to hydrolysis—worse with halogens.
• Corrosion Failure Due to Flux Residues (2010): In wind turbine switches, potentiodynamic tests in flux solutions showed increased corrosion/ECM vs. water alone, leading to hydrophilic surfaces and progressive conductivity.
• Additional Studies: 2009 wave solder tests reduced time-to-failure via ECM; 2015 no-clean fluxes showed higher leakage under humidity; 2019 reflow spatter increased conductivity in humid setups.

These align with standards like IPC-TM-650 for SIR testing.

Specifics for Older Leaded Solder (Pre-RoHS Era)

Older leaded solders (e.g., Sn-Pb 63/37, melting at ~183°C) were paired with milder rosin fluxes, less aggressive than modern lead-free (e.g., SAC305 at 217°C, needing halogen fluxes). This means initial residues are less corrosive, but over 30 years, they can still degrade similarly via moisture and ECM.

• Leaded joints have better mechanical aging but flux residues can cause dendrites under bias.
• Less whisker growth than lead-free, but uncleaned residues lead to conductivity issues in humid environments.
• Lead-free traps more residues due to higher tension, worsening risks—but older leaded boards often lacked proper cleaning, amplifying long-term problems.

In summary, for vintage PCBs, flux is the main issue, not the solder alloy. Clean thoroughly or coat for protection.

If anyone has similar experiences or questions, chime in!

[exxos]

This video, fleetingly mentioned short circuits due to flux.





dendrites are tiny, needle-like metallic structures that form on a printed circuit board (PCB) due to electrochemical migration (ECM). They grow between conductive paths (e.g., traces, solder joints) under the influence of an electric field, moisture, and ionic contaminants, such as those from aged flux residues. These structures can create unintended conductive paths, leading to partial shorts or low-resistance issues.


What Are Dendrites and How Do They Form?

In the context of my issue with an old 1980s-1990s PCB showing a 150-ohm path to ground (despite a 1k pullup to 5V), dendrites are tiny, needle-like metallic structures that form on a printed circuit board (PCB) due to electrochemical migration (ECM). They grow between conductive paths (e.g., traces, solder joints) under the influence of an electric field, moisture, and ionic contaminants from aged flux residues. These structures can create unintended conductive paths, leading to partial shorts or low-resistance issues like I observed, which was fixed by cleaning with isopropyl alcohol (IPA).

• Definition: Dendrites are crystalline, tree-like growths (from the Greek "dendron," meaning tree) of metal (e.g., tin, lead, copper, or silver) that form on PCB surfaces or solder joints. They typically grow from a cathode (negative) to an anode (positive) under a voltage bias, such as my 5V pullup circuit.
• Formation Process:
  1. Ionic Contaminants: Flux residues, especially from older rosin-based or no-clean fluxes used in 1980s-1990s leaded soldering, contain ionic activators (e.g., halides, weak organic acids like succinic or glutaric). Over time, these residues absorb moisture (hygroscopic behavior) in humid environments, forming an electrolyte film.
  2. Electrochemical Migration: When a voltage (e.g., 5V) is applied across closely spaced conductors, metal ions (e.g., Sn²⁺ or Pb²⁺ from leaded solder) dissolve at the anode, migrate through the electrolyte, and deposit at the cathode, forming dendritic growths.
  3. Growth: Dendrites extend as thin, conductive filaments, bridging gaps between traces or pads. They can create paths with resistances like my 150 ohms, causing signal malfunctions or intermittent failures in high-impedance circuits.
• Conditions Favoring Growth:
  • Humidity: High relative humidity (e.g., >60% RH, especially 85-90% RH) accelerates electrolyte formation.
  • Voltage Bias: Even low voltages (e.g., 5V) can drive ECM, especially in dense PCBs.
  • Contaminants: Uncleaned flux residues, dust, or salts (e.g., from handling) provide ions for migration.
  • Time: Over decades, as in my 30-year-old PCB, residues degrade, increasing ionic activity.