My language was misleading,
...so filling in a location in your profile might be useful. Not only mains voltages differ between countries/continents...
I have two transformers (30va) with 2*120:2*18.
Also, i am testing without a load on the output, could this be the issue?
As previously said, dual 18VAC secondaries wasn't called for and will be a heat generator.
AC mains voltage isn't a constant and might vary by +/-10%, depending on location or time of day.
18VAC is the secondary voltage at rated load, that will be higher by maybe 10% unloaded (look up the datasheet of your transformer for the voltage regulation % number of your mains transformer).
For a high mains and transformer unloaded scenario, this 18VAC will give a raw DC voltage in front of the 78S12 voltage regulator of 18VAC * 1,1 for 10% high mains * 1,1 for no-load * SQR(2) for fullwave bridge rectification = 30.8VDC.
Are your -from schematic 25VDC rated- caps C16,C17,C18 at least rated for this 30.8 DC voltage or are they already damaged.
The voltage differential between raw VDC-in and VDC-out times connected load (light and tube heaters, about 350mA) is transfered to heat by the voltage regulator (78S12). For a 15VAC secondary, heavy heatsinking was already called for. For your 18VAC secondary transformer, generated heat for 30.8VDC-in - 12VDC-out = 18.8VDC no load differential (or 16VDC differential at rated load) times connected load, you'd need a river to cool it down.
Your parallel connected 120VAC primary /18VAC secondary windings transformer is a factor 6.666 step down ratio. For the HT rail, this 18VAC winding connected to a 2nd backwards connected same type transformer will give you a 13.333 step up ratio with the primary windings connected in series for about 240VAC between the outer winding ends on paper. This 240VAC then is fullwave bridge rectified to about 340V raw DC.
In your shoes, I'd disconnect the mains transformers from pcb and measure these seperatly for their real world no-load primary/secondary ratio or voltage.