First reply would why these bypass caps are needed..... Following several reference books on power amps (Bob Cordell, Douglas Self), the caps are merely to account for resonance frequency cancellation of the big reservoir caps beyond the 500 kHz region, and less for RFI/EMI HF decoupling as often mentioned. As such 1uf to 2.2 uF is often more than appropriate, and you can keep using polypropylene capacitors as Mr Yamaha did. Polyester caps are not distortion free, and starting with a distortion source at the very beginning of the power supply chain.... Secondly, the whole resonance topic of the large reservoir caps in the higher kHz region is sensitive (worsened) by inductive elements, and as such soldering them as close as possible to the pins of the reservoir caps is key. Soldering those caps on the VFET boards is adding a lot of inductance in the 'system' due to the wiring and PCB track between reservoir caps and the bypass caps..... not good, potentially making things worse than not having them at all. As for HF decoupling, on the contrary.... decouple as close as possible where needed, i.e. on the VFET boards themselves, using SMALL value bypass caps in the order of 47nf to 100nF (max, like for the +/-B2 rails on the driver board); or else you may start coupling instead of decoupling..... So it's a dual job to do. PS: the B+/B- power rails on the VFET boards continue to the driver boards via de connecting PCB below the heatsink, and have no further HF decoupling on the driver boards (which is a shame), so adding those 47~100nF bypass caps on the VFET board incoming power rails (to E) is no luxury at all in the era of wireless communication.