[HN Gopher] The Solid-State Shift: Reinventing the Transformer f...
       ___________________________________________________________________
        
       The Solid-State Shift: Reinventing the Transformer for Modern Grids
        
       Author : JumpCrisscross
       Score  : 48 points
       Date   : 2025-04-12 18:23 UTC (4 hours ago)
        
 (HTM) web link (www.powermag.com)
 (TXT) w3m dump (www.powermag.com)
        
       | Calwestjobs wrote:
       | figure no. 1 is missing green color on HVAC transmission line. i
       | do not have ExTwitter and i do not know other way to let them
       | know.
       | 
       | it is interesting to think about human made object in terms of
       | how much materials they use, for example big old transformer
       | contains 2 tons of iron, new solid-state transformer with same
       | capacity uses only 300 kg of silicon (Si), 120kg kg of plastics
       | and 50 kg of copper.
        
         | JumpCrisscross wrote:
         | E-mail the author spatel@powermag.com [1].
         | 
         | [1] https://www.powermag.com/contact-us/
        
         | mppm wrote:
         | Why do solid-state transformers need so much silicon?
        
           | LargoLasskhyfv wrote:
           | Because they contain long cascades of these
           | https://en.wikipedia.org/wiki/Insulated-
           | gate_bipolar_transis... to be arranged into these
           | https://en.wikipedia.org/wiki/Power_electronics
        
             | mppm wrote:
             | As a sibling comment implies, high-power IGBTs are still
             | built from small and thin silicon dies with most of the
             | module being heat spreader, insulator, interconnect and
             | packaging. Even MW-class modules probably contain a gram or
             | so of Si.
        
         | wbraun wrote:
         | Where are your weight figures coming from, they are not in the
         | linked article.
         | 
         | Packaged semiconductors are going to be more metal interconnect
         | / plastic encapsulation / ceramic insulation than silicon by
         | weight.
         | 
         | These systems will also have a significant weight fraction in
         | magnetic materials, either ferrite ceramics or amorphous
         | metals.
         | 
         | Still a huge weight savings, but the weight fractions you are
         | giving see off and are missing some important materials.
        
       | hristov wrote:
       | This is very interesting and cannot happen fast enough,
       | considering the current worldwide transformer shortage.
       | 
       | I have a question for people more familiar with these. What
       | exactly happens at the isolation stage. They say it includes a
       | high frequency transformer (HFT). But its input and out put is
       | DC. And classic transformers operate on AC. So in order to get
       | the transformer working, one would have to chop up the incoming
       | dc power into a square wave or a sine wave. But what transistors
       | can you use to do this, considering you are dealing both with
       | very high power and very high frequencies?
        
         | cyberax wrote:
         | You need to use nuclear-powered transistors!
         | 
         | No, I'm not joking. For these kinds of voltages, you need to
         | use highly homogenous doped silicon, and the only way to
         | produce it is to irradiate silicon with neutrons. It transmutes
         | some of the silicon atoms into phosphorus:
         | https://nrl.mit.edu/facilities/ntds/
        
           | yjftsjthsd-h wrote:
           | I assume that if people are going to the trouble to literally
           | irradiate the material in order to get what they need, they
           | can't get the results by just mixing in phosphorus. Could
           | somebody who actually understands this tell me why that is?
        
             | cyberax wrote:
             | Because there is no other way to produce homogenous enough
             | thick wafers of doped silicon. Other methods rely on
             | diffusion from the surface, which is not enough for this
             | case.
             | 
             | And doping during crystal growth doesn't produce homogenous
             | enough silicon.
        
             | CorrectHorseBat wrote:
             | https://www.waferworld.com/post/the-complete-guide-to-
             | doping...
             | 
             | It's apparently possible for boron doping.
             | 
             | I think because phosphorus is bigger than silicium you'll
             | get too many defects in the crystal, while with the smaller
             | boron it is not an issue.
        
         | itcrowd wrote:
         | The keyword you are looking for is IGBT (insulated-gate bipolar
         | transistor) -- the type of transistor used in such DC-DC
         | converters
        
           | hristov wrote:
           | I don't think IGBTs can handle frequencies up to "several
           | megahertz" as the article says.
        
         | bacon_waffle wrote:
         | There are techniques for "stack"ing transistors so that the
         | individual swtiching devices see potentials that are within
         | spec and much lower than the voltage switched by the overall
         | circuit.
        
         | Gibbon1 wrote:
         | Power transistors can be had that can switch up to maybe 3000V
         | at the most. But maybe 1200-1500V is more common.
         | 
         | You can stack power transistors to switch higher voltages on
         | the primary side. On the secondary side you just need an
         | H-bridge. Which can be made up of transistors in parallel.
         | 
         | We've had high power high voltage transistors for about 40
         | years now. A lot of this isn't technical but rather economic.
         | As the price falls the applications where they are cheaper
         | increases. It's notable for instance Toyota started work on
         | their hybrid drive in the mid nineties when inverters for
         | 10-100 HP motors became cheap enough.
        
       | Animats wrote:
       | Then they'll make them "cloud-enabled" and they will be hacked.
       | 
       | It's scary to see protective relays for power systems with
       | embedded web servers. "IEEE C37.118 synchrophasor measurement,
       | DNP3 Outstation, Modbus TCP/RTU, Telnet, FTP, Simple Network Time
       | Protocol (SNTP), built-in web server, and IEC 61850" [1]
       | 
       | [1] https://selinc.com/products/351/#
        
         | fc417fc802 wrote:
         | It's fine provided that the link from the equipment that feeds
         | it data is optically isolated to only go in one direction.
         | 
         | A public internet connected web server that enables remote
         | equipment control is indeed scary.
        
           | Animats wrote:
           | > It's fine provided that the link from the equipment that
           | feeds it data is optically isolated to only go in one
           | direction.
           | 
           | Then people get two of them, one for each direction.[1] Can
           | someone explain why this is supposed to be secure? It's
           | apparently a real product.
           | 
           | [1] https://owlcyberdefense.com/product/recon-2u/
        
         | sightbroke wrote:
         | > Then they'll make them "cloud-enabled" and they will be
         | hacked.
         | 
         | It's worse:
         | 
         | https://arstechnica.com/security/2025/01/could-hackers-use-n...
        
       | algo_trader wrote:
       | Ex-tesla Drew Baglino is starting a new company Heron for solid
       | state transformers
       | 
       | Of topic - dealing with medium voltage e.g. 6kv-10kv - are there
       | existing be-spoke solid-state solutions?
       | 
       | e.g. PV/battery 400V converted to 6kv without
       | inverter/transformer
        
       ___________________________________________________________________
       (page generated 2025-04-12 23:01 UTC)