About a year ago, First Solar announced that it was making a fundamental change in the architecture of the big solar power plants it builds and operates by moving from a conventional string architecture to a medium-voltage DC (MVDC) scheme.

GTM has now learned from several sources that First Solar has indefinitely postponed that transition, and First Solar CTO Raffi Garabedian has confirmed.

Garabedian told GTM, "We’ve postponed MVDC indefinitely given the increased competition and margin compression we’re seeing in central and string inverters. We still believe that the MVDC architecture has a substantial system-level cost entitlement, but we’re waiting for conventional inverter prices to stabilize before we restart. On the standard inverter front, there are a number of new (or “new to us”) players [Editor's note: Huawei] in the market who are competing aggressively with high-quality products. We think this situation will remain the norm in the power conversion business for some time."


At First Solar's analyst day in April of last year, Garabedian called MVDC "a completely new, radically game-changing power plant architecture for solar."

He said at the time, "If you look at a conventional architecture, you've got modules in the field that are connected in what's called strings. Multiple strings of modules are combined in combiner boxes where all of that energy is aggregated in parallel, and fuse-protected, and then sent through feeder lines to power conversion stations. That transmission of energy is done at 1,500 volts today with our 1,500-volt inverter system. More commonly, it's done at 800 or 1,000 volts in the industry," he said.

"Each power conversion station is a combination of an inverter and a transformer that steps up the voltage roughly to 35 kilovolts, which goes to a PV combiner switch gear and then onto the substation and onto the grid connection. That's how power plants are built today," Garabedian continued.

MVDC would change "the number of elements in this power plant, reducing the number of parts and improving the capacity factor of the energy generation of the plant. By replacing the combiner boxes here with DC-to-DC converters that step up the voltage from the strings to roughly 10x the string voltage of 1,500 volts, we can dramatically improve the cost of wiring and the resistive losses in the wiring, and by using a very large-scale DC-to-AC converter, which is commonly available in the utility industry (used for grid ties globally), and by leveraging that pre-existing technology, we can achieve a very good cost structure from the MVDC point forward to the grid connection," according to the CTO.

MVDC could "also bring to bear grid control and support functions that are already available in these large-scale DC-to-AC converters, which are not available in standard PV inverter systems."

"What are the advantages? We can reduce the number of components in the field. We can reduce the amount of labor in the field, and we can reduce the energy losses in the power conversion systems." According to Garabedian, "It doesn't stop there. There's one more benefit. This system is storage-ready. What that means is that we are architecting the system to accept battery storage systems integration into the DC side of the plant."

But those benefits will have to wait until inverter prices stabilize.

MVDC "exponentially more ambitious than 1,500 volts"

MJ Shiao, GTM's director of solar research, summed up the MVDC approach as a "super-centralized" model with a very large medium-voltage DC-to-AC converter creating "a 100-megawatt power conversion station."

Shiao identifies all the benefits of raising DC voltages: "Less wiring, less trenching, eliminating combiner boxes (although you replace that with the DC-DC optimizer), fixed voltage output that reduces the DC-AC conversion complexity, reduced DC losses, reduced site grading/civil work, etc."

But he also had some concerns: "MVDC means a whole new ball game for construction and O&M. There's not much solar-specific equipment beyond the DC-DC optimizers in this case, but anything that's not off-the-shelf needs to be certified by UL, etc., and that could take a while."

Shiao pointed out that the architecture isn't totally brand new. Alencon, with funding from the DOE and backing from Stephens Capital Partners, has been working on this type of architecture, albeit with a 2,500 VDC bus. Shiao suggests that SunPower is working on something along these lines (based on the DragonFly acquisition), while Ampt has also been pushing string-level DC optimization in the utility space for the past few years.  

Scott Moskowitz, GTM Research solar analyst, notes, "This is exponentially more ambitious than [the move to] 1,500 volts," with "far more significant hurdles." Shifting to 1,500 volts only required small changes in supply chains, some R&D to reconfigure and test products, incremental changes to standards for PV components, and there was precedent from the earlier evolution to 1,000 volts. Medium-voltage DC would require significantly more technology development and would require borrowing conversion technologies from other energy industries."

Someday: Greater use of DC bus and AC conversion at higher voltages

In 2015, former First Solar CEO Jim Hughes said, "The large arrays that are built today are essentially a scaled-up rooftop system -- because it's what the code authorities in ancient years [understood]. If you were building a large-scale array from scratch on a clean sheet of paper with no prior contamination, what you would build would look different than what we build today."

"I think over the next five years, you'll see the industry transition to a fundamentally different architecture with greater use of DC bus [and] use of AC conversion at higher voltages with fewer inverters. [...] I also think we need tighter integration with the [way we] use electricity." He said there is an opportunity to do things on the DC side, as well. "If you can co-locate or locate within in a reasonable distance, you can put DC right into the data center," resulting in "a 15 percent cost savings from avoiding the whole AC conversion side of the equation," Hughes said.

Again, those benefits will have to wait until inverter prices stabilize.

Garabedian notes, "A similar situation exists in the structure realm, where we’ve seen new suppliers entering the market with well-engineered and locally cost-optimized solutions. Rather than try to be the best at everything everywhere, we’re taking advantage of this dynamic through an ecosystem initiative whereby we’re working closely with trusted structure suppliers to assure low-cost compatible and reliable 'First Solar-ready' structures."