That's ok, the reason for the SWR meter looking good, is because it's in line BEFORE the amp, and the RF power The Amp sends out the door - so it will always look good to the operator, - but I am glad you see this effort on the EXTERNAL meter you put in, it tells a different story.
The amp board is absorbing - like a Roll of Toilet Paper - all the SWR issues, so that Amp will run hot - and fail quite quickly - if this issue isn't tamed or somehow turned down. The Reflected side of that power not going to the antenna is making the amp work harder as dissipating the heat it already is producing.
So since the SWR internal meter seems to be a little too "dense" to Sense the problem, I'd keep an eye on this condition as well as the power amperage the supply will need to push into the radio to make that Amp work.
Seen several MOSFET types that kill their power supplies because the Gate Voltage and their mA settings are a little too much ON and not enough "OFF" - which too can set a spurious problem or two.
The Reason for the killing power supply, or a good symptom of it, is the "Wink Blink" or just Cut out - go Dark until you turn it off - those moments, the MOSFET or Bipolar, either way, they are on their way out - and will need to be replaced before it kills the supply.
Then the Gate voltages are getting set a bit too high.
The Gates need to have their CURRENT side, that jumper method used, set to 50mA
Voltages are ok, to "set" - but Current is their "Suck point" (SEE: Vacuum Cleaner) Not the Voltage, which is the "turn on" threshold for that gate, but set too high in voltage, they will stay on too long during their cycle.
The benefit of MOSFET, is also their Demise. They have very low-ohmic impedance when they are turned on, it's not like the Bipolar which has a working "Knee" or linear operation. The MOSFET is a nearly perfect switch, very fast on-to off to on times, Bipolar have a Hfe (Gain) values that is Frequency Dependent - MOSFET On Junction Resistance is measured in Siemens or Ohms of Impedance per their AC character - pretty low making them perfect for RF, but too much Gate Voltage, they'll stay on too long and make the power supply fail trying to keep up with the current demands these things, when latched, will pull from the supply rails...
As a side note, many people think of the MOSFET's Rise Times and Fall Times as their switching rates, yes, you can determine the working range of the part this way, But they are not taking into account - the effect the INDUCTANCE (those coils in the strip) as another part of this equation. These coils filter out the bad, but in turn also affect the way the Bipolar used to work, (think Miller Effect or Flywheel - Required Capacitance) MOSFET's see the Inductance and work differently - because of their "Field" effect the Gates have, the Inductance in the Load side of the line determines their actual frequency response.
Too much on time will force the Gate on longer because of the lag behind effect the loading the inductance will impose to force the field to becomes a pulse against the Gate - keeping the Gate threshold "on" for longer periods of time. You don't want that.
if you use the 520 or 13N10's - use 50mA OR LESS to drive these parts before the RF goes into the amp - simply because they will heat up and the Bias to their gates won't track the best to follow along, instead, it will cause them to "Latch" on and stay on longer as they heat up and since they are connected directly to the AM Regulator and the Power Positive Rail, they can suck all the power out of the power supply and literally dim the lights in the Radio and even pop a breaker - from the Power supply trying to catch up as a DC voltage across the MOSFET Drain and Source.