The Xiegu G90 SDR transceiver woes, but maybe fixes?

[ThaDood]

       Buying a used Xiegu G90 SDRtransceiver, used for $369.00, from MTC Radio, was a deal that I couldn't ignor. However, with this bare-bones, 20W, SDR HF transceiver being out now for about 6 years, problems are developing with them, including the RF cut-out problem that I had. How did I fix mine? The following is a Copy & Paste e-mail that I had with Tom, WB8ICT, (Whom also has a Xiegu G90.), where I explained to him what I did with my rig. So, contemplating that this INFO might be useful to other Xiegu G90 owners, here's what I'd come across:

"Hey Tom???
Just to give ya a run-down on what I did on my Xiegu G90. Again, Rick, W8ZT, nailed-it with the Si heatsink grease being dried out, under the Voltage Regulators and all three Mitsubishi diver and finals MOSFET pair. Yep... It just flaked off, and after only 4 years. The hardest part is the around of disassembly that I had to do to reach those. BTW, the MOSFET final, closest towards the front, has that thermo resistive sensor over it. Looks like a black dot with two legs, hovering loose over that final's body, and close to that 5W, 330 Ohm, resistor. I suspect that therm resistor gets more heat from that. I did also resolder what looked like a cold joint to one of those 330 Ohm, 5W, resistors. Other bad solder joints found were the 13.8VDC main lead-ins to the board and that filter torroid, just 1 inch up further. I've also replaced those chintzy antenna and GND leads to the SO-239 connector. In fact, the hex nut that locks the GND connection screw was free spinning. (Good QA inspection there, eh?.) I've swapped that out for a high temp, Teflon-based, 100% shielded, RG-174-like, coax. (Leftovers from my days with RF plasma generators.) Then, reassembled to about 85% together and did long key-down tests for over 10min at a time on AM, FM, and SSB at the 20W level. Granted, 1.1:1 VSWR in a dummyload. I let the rig burn on RX overnight for 24hrs. Then, did a auto-tuner test using a 12VDC trailer light bulb. It, did not like it, and at best, got the match to 2:1. Ran that for like 10min, to also warmer than I could touch. The rig didn't cut-out, and heat was much more uniformed throughout the rig's chassis, and not just by the finals and REG's. So, the big test was Saturday evening. Apparently, it passed that QSO'ing. BTW... The set screws, that locked-in the REG's and finals were not tight to begin with either. And, no mica insulators required.
So, thar' ya is, if you need to work on your Xiegu G90. Doable, but definitely room for improvements. And, found nothing on-line explaining any of this."

Hopefully, this INFO will help. BTW, I already have firmware ver 1.75 in this rig, so the on-line fixes to upgrade from ver 1.73, was mal-en-void, for me. 73!!!!

[RobRich]

A thermal compound issue does not surprise me. Many compounds, from cheap to even rather expensive exotics, tend to dry out under heavy cycling and/or high-temp loading. Pump-out is another concern as well.

I often use Shin-Etsu G751 (4.5W/mk), X23-7762 (6.0W/mk), or similar for standard thermal interfaces. Usually better than the cheapest stuff. Tends to hold up for years under normal usage. Shin-Etsu is popular in the OEM computer industry for a reason.

I do use "better" pastes for certain projects. Case in point, and IIRC, I used Thermal Grizzly Hydronaut (11.8W/mk) when last upgrading the processors in my Xeon 96-core / 192-thread server. It is formulated for long-term stability instead of outright thermal conductivity. Seems to be working fine so far.

For a direct comparison from the same brand, Thermal Grizzly Kryonaut (12.5W/mk) is (was?) popular among overclocking enthusiasts. It is also one of those formulas that tends to dry out quicker under heavy thermal loading. It is not worth the potential hassle for that particular system IMO.