Electronics Weekly Electronics Design & Components Tech News
Updated below to include driving requirements
Called IPAM01210C10, “the power module was designed specifically to enable SSCBs to deliver very low conduction losses, while operating orders of magnitude faster than conventional electromechanical breakers”, said Ideal Power CEO Dan Brdar, adding that he sees them paralleled in utility, industrial, railway and military applications. “We look forward to our first commercial sales later in 2023.”
The bipolar technology (central part of module circuit left) has been branded ‘B-Tran’, and its low forward on-state voltage drop, well under a volt, is its key parameter.
This is the normally-on variant of the device, hence the module has cascoded low-voltage high-current mosfets to turn it off.
“The normally-off version of B-Tran has much higher Vce(on) than normally-on version,” the company told Electronics Weekly. “We are continuing engineering both versions.”
It claims 600mV drop for its normally-on chip, compared with 2.75V from back-to-back IGBTs with anti-parallel diodes for an example 1.2kV bidirectional application, and goes on to say that losses in a 12kV 50A switch would be 360W compared with 1.5kW for an IGBT version.
Being fully bidirectional, only nine are needed in a three-phase to three-phase matrix converters, for example, compared with 18 IGBTs plus 18 diodes.
However, unlike IGBTs, the bases of the bipolar structure (one possible B-Tran structure right) require current-based driving, and the two bases each have to be driven relative to their own emitter.
“For the power module, under 120A emitter current, base drive current is 24A [β=5],” Brdar told Electronics Weekly. “For 60A emitter current, base drive current is 8.5A [β=7].”
A series of choreographed steps on the four control inputs are required to cleanly turn the module on and off. “The mosfets are also used to configure power module in right operation mode,” said Brdar.
- Turn-on sequence (below left achieves 30A 600mVce(on)):
Turn on cascode mosfet gate
Turn on base injection - Turn-on sequence (below right configures as a 1.2kV reverse biased diode):
Turn off cascode mosfet gate
Turn off base injection
Disconnect E2
Connect to B2
“They are low voltage mosfets with under 2mΩ resistance,” continued Brdar. “B-Tran is the bi-directional high-voltage device, not these mosfets. For a B-Tran chip with 0.6V Von at 30A, with a cascode mosfet this will be under 0.61V.”
According to the data brief, the typical end-to-end saturation figure for the module is 560mV (800mVmax) at 100A with the appropriate base driven to 1.5V. Where is that going to come from?
“For the 100A 0.56V typical on-state, total driving current is 20A when driving 1.5V,” said Brdar. We have a “bi-directional floating driver with a buck converter that converts a 24V power supply to ~20A 1.5V. We are further improving B-Tran chip design to improve beta, so to reduce driving to 10A.”
In a 20A application, he sees on-state module losses as 18W (600mV 20A module + 1.5V 4A drive) compared with his estimated 55W for an IGBT (2.75V 20A module + not much drive)
The 44 x 54 x 5.6mm module makes use of the double side structure of the high-voltage die (example left) and offers double-sided cooling, and built-in temperature sensing. “Heat is dissipated from both the top and bottom surface without wire bonding,” according to the company.
Dissipation is rated at 1.6kW at 25°C (400W 100°C). Zero power operation is across -40 to +125°C.
Speed-wise, the module is recommended for switching at 30kHz.
At 600V 100A with 1.5V drive, data brief timings are: 250ns turn-on delay, 100ns rise, 450ns turn-off delay and 200ns fall. Energies are 2mJ turn-on and 7.2mJ turn-off.
The device can also be used in the bi-directional position within T-type inverters, according to Ideal Power, which is based in Austin Texas
I am fascinated. This is not my area, though. Those base currents look high to me and I can see why they want to get the beta up. I would like to see how the efficiency compares for a typical system vs other solutions, rather than on the individual device level. Is anyone more familiar with the area able to do a back-of-the-envelope assessment?
Morning DB
I have asked for more information on base driving, particularly cascode base driving.
20A 1.5V does not seem awful if it is saving more than a volt of drop at 100A through the device, but this seems to need a base drive that can see which way the polarity is across the device – or it will have to drive both bases – consuming 60W.