That is indeed a serious problem being an OT and thinking (what else) that DEI meant Deliberative Engineering Inquiry. WE are not going to talk about the other currently popular meaning of DEI, but today's post is a trouble shooting problem.
A good friend and fellow podcast participant Dean, KK4DAS has a mysterious occurrence with one of his homebrew rigs. It is not a simple problem like R27 is the wrong value or it's a defective 2N3904, but one where you can see the problem, but the source and cure is hiding in the bushes.
So, how do you attack the problem? The old adage about time is healing applies here too. Step one is to back away from the problem for a short while and resist the immediacy of try this and try that where that only results in raising the frustration level.
A tool, I use, is to write things down on a piece of paper using a two-column approach. One column is marked Known and the other Unknown. My friend has some critical pieces of information in hand. The 1st is that the condition occurring in transmit depends on power level, meaning not so evident at 10 watts but really evident at 20 watts. This goes in the Known column.
In the Unknown may be a case of the Negroponte Switch Theorem (that which was sent by air is now sent by wire and that sent by wire is sent by air ~ Cable TV and Cell Phones). The Unknown is whether RF is floating around his Al Fresco layout causing the issue which is a shielding problem or RF is being fed into the power rail which is a decoupling problem. There is also the non-binary case where it is both!
Based on this "Unknown" entry we now can develop a plan and methodology to evaluate the two identified possibilities. Our tests should also look to uncouple the two issues. In the case of power being fed back into the power rails, using a 2nd supply to isolate the power supply condition then removes that as a cause. If the condition remains, then you can further explore the Al Fresco non shielded layout.
A short story about controlling the test conditions and page from when I was involved with Quality Processes during my aerospace years at McDonnell Douglas. We used this story as a training device. There was a manufacturer of cast/molded products that was having a high rejection rate. They called in an expert who observed three molding stations. Two of the stations had high rates of non-conformances while the 3rd station did not.
So, the expert set up a test where he observed the process and over and over again each station had total process control but two were always a bust. This was a mystery.
Purely by accident the expert came to work early one day and noted the worker who always was successful was cleaning up his mold but then noted the worker was rubbing the mold with a rag that was soaked with a bit of oil. He asked the worker what he was doing -- he said I always like to have my molds clean and shiny. This was not in the process but had not been observed because this was being done before work officially started.
The expert then asked the other two workers to do the same -- boom three perfect production pieces. This worker in his quest to have a clean mold was adding a release agent to the mold which then became the fix and part of the revised process. The hidden factory was at work!
My friend's project is actually a homebrew build of the SBitx and so care must be exercised when asking others "have you seen this"? They may be doing something slightly different (oil on the mold), and that issue does not come up in their build. All Information is important.
We may have the Unknown bits that we identified not resolve the matter but at least the testing would then exclude those as a root cause. The issue is relays in the rig are controlled by a RPi4 and the one set of relays being activated in a BPF indeed switch on command but the other relays in assembly also seem to partially conduct (chatter).
The control wiring to the banks of relays appears to be a cable bundle. Is shielded wiring needed? That goes into the Unknown column. Could a brute force solution be a totally shielded box with feedthrough caps on all of the power wiring? That is extreme but must be eliminated as the solution.
What we are advocating is DEI ~ Deliberative Engineering Inquiry.
TYGNYB!
73's
Pete N6QW