Stiffler Scientific added a diagram for one of their solid state CRE circuits. In the past they only showed a crude relay version. This one looks nice, but they say a low current version, (whatever that means). Anyway the circuit looks simple enough and makes good sense from what they claim.
It's at www.stifflerscientific.com/cre.html
By looking at the schematic it's seems that they are using the technique of energy storage&switching that Bedini, Bearden, Naudin and others have tried on other experiments.
Some simplification - and improvement - in the design could be to use MOSFETs all around instead of bipolars;
use a P-channel MOSFET instead of Q2/Q3 pair (e.g. an IRF5305, ?1.21@Farnell/$0.80@Newark),
use a 2N7000 instead of the Q1 inverter - and use e.g. IRF3709 (?1.34/$1.33) for M1.
Even D1/D2 could be MOSFETs to minimize losses, but this require some additional thought to get them driven properly so this is something that should be tried after one gets results with diodes.
Cost of a CRE, I do some machine work for the SS group and they give me server time in return, but I also begged a 5 amp proto from them. Up until the MOSFet and the bipolar driver and pre-driver the cost of the transistors is $0.09 each and the resistors are $0.01 each. THe MOSFet in my proto is $0.89 on www.jameco.com and the bipolar driver is $1.25, but the one I have is over kill as it is rated at 25A.
When my cell is disconnected the steady state current is only 12mA. The circuit on their site has a problem that was corrected in mine by the addition of another transistor a PNP. There seems to have been a timing issue in that when one shut down the other one was coming on and the MOSFet would divert some of the charge current away from the capacitor. As seen in the circuit all current is used, except for the 12mA which gopes I guess into the 555 and what they show as the pre-drive. In mine with the PNP there is no loss due to the pre-drive (guess this is what you call it).
They have a 50A unit which is the same as an 80A straight connect cell. In addition the use a cylinder for the +plate and big carbon rod for the -electrode. But bottom line is they still have gain even though those diodes get hot and I'm told they already tried MOSFet's but they act strange as they like to have a real DC path, where the SS people call this a virtual energy path???