So then there was this 10-20 gallon aquarium pump.
It does have an inlet which generates a vacuum, but the difficulty of connecting to that inlet led to another use as a compressor & hope the container didn't explode. It doesn't need a 1 way valve to manetain a seal. It's just a simple 4k ohm solenoid which oscillates a ferrite magnet. It burns 2.5W at a .95 power factor.
The great challenge was turning it on.
This thing is based on the BTA41-600B Triac & MOC3021 optocoupler. The 1st problem was it not turning off. Removing the 10 ohm snubber resistor was the only way it would turn off. It could be the 2.5W load wasn't enough to turn it off.
The other problem was a lack of desire to have a live load when it's off. As originally labeled, the load was always live & the neutral was switched off. It actually works if live & neutral are reversed. To make this arrangement work, the line has to plug in to the triac side. The line provides the voltage to turn on the gate. This might not work for inductive loads which require the snubber circuit.
Enclosing the new mess of pumps & electronicals will require a 3D printed enclosure. As hoped, the pump is totally silent. Unfortunately, the 10 gallon aquarium pump was nowhere close to as powerful as the inkjet pump so daytime dewpoints ran away. The most efficient drying happens when the inside dewpoint is as low as possible during the highest temperatures. There shouldn't be any water hanging around above the ambient dewpoint.
There are delusions of rewinding the solenoid to run on lower voltage & higher current & higher frequency. The only real capacity change would come from a bigger pump. The volume of air is limited by the mechanics. Bigger aquarium pumps take more space, cost more & are louder.
2 days showed it pulling more water out than the passive valves but not as much as the adafruit pump. At least it was silent.
A 40 gallon pump quickly replaced the 10-20 gallon one. It was just as silent but had comparable thrust to the inkjet pump. It burned 3.5W.
Interior dew points were pulled right down to exterior dew points. Unfortunately, it still requires 120V. After 9 days, the triac failed on again & it started pumping in water instead of pumping it out. By this time, the lion kingdom was leaning on a defective triac. There could be 1 more attempt at a pulldown resistor between the gate & neutral but the mane idea was to buy a bag of triacs & a relay.
The next problem is the CPU fan as a circulation fan is dying. There's no way to lubricate it & it's not up to the conditions in the dehydrator. CPU fans are intended to blow cold air at a hot CPU, not circulate hot air. A more serviceable blower is needed.