Tweet

Posted on 05 November 2019

Trade-offs When Selecting Automotive Dc/Dc Converters

Free Bodo's Power Magazines!

 

 

 

In automotive designs using dc/dc converters, strict conducted and radiated emissions have to be met. Emissions in the Radio LW, MW, SW and FM bands are more tightly controlled to avoid interfering with the vehicle’s radio reception, Figure 1.

By Dan Tooth, Texas Instruments

The most tightly controlled being the FM band, which can have an average conducted emissions limit that is almost three-times lower than in the MW or SW bands. (Note – the MW band is sometimes referred to as the AM band).

Typical Conducted Automotive Emissions Requirements, highlighting the Radio Bands. Also the positions of a Fundamental Switching Frequency, Fsw and its low order Harmonics

Also, the ambient temperature in automotive applications is expected to reach 85°C. When the temperature rise due to power losses within a dc/dc converter IC are added to the ambient temperature, junction temperatures can reach 125°C. These two different yet related requirements are often a focus for the automotive electronics design engineer.

Dc/dc converters produce switching frequency noise at their switching frequency fundamental (Fsw) plus harmonics of Fsw. In addition they produce much higher frequency noise which can fall within the FM band, due to parasitic resonances excited in the dc/dc converter circuit. To avoid producing switching frequency noise in the MW band, you can select a switching frequency that lies just above it, at 2MHz ± 10%, where the emissions are permitted to be higher. This is straightforward when the input voltage (Vin) is low, e.g. 5V or 3V3, because the duty cycle and hence converter on-time (Ton) remain long enough to stay above the minimum on-time (Ton(min)) of the dc/dc converter IC. Ton = Vout / Vin x 1 / Fsw. However, when the dc/dc converter in question is connected to the vehicle battery, Ton becomes very short. For automotive applications, the maximum Vin the converter is required to support varies, but can be 18V for continuous operation, 24V for 2 x battery and up to 32V for load-dump transients. (As the input voltage increases, Ton of the step-down dc/dc converter decreases. If the Ton is required to be less than the IC’s Ton(min) then this leads to pulse-skipping [1] behavior, Figure 2, and increased RF emissions in unpredictable lower frequency bands. To avoid this behaviour, the switching frequency must be decreased, which results in Ton increasing for a given duty cycle D = Ton x Fsw).

Increasing Vin leads to Decreasing Ton, Until Ton(min) of the IC is reached and pulse-skipping behavior is Encountered

Another approach is to set the switching frequency of the converter connected to the battery below the MW band. The four or five-times reduction in switching frequency compared with switching at a frequency above the MW band makes Ton(min) of the IC no longer an issue. With a longer Ton, the dV/dt of the switch-node waveform can also be reduced, exciting less ringing of the resonant parasitic tank circuits and less emissions in the FM band. Figure 1 shows that the fundamental, Fsw, is located at a frequency where the emissions are permitted to be higher. The 2nd and 3rd harmonics of Fsw do fall into the MW band, but they are lower in amplitude compared with the fundamental. The input filtering is manageable – Texas Instruments designed a two-stage input filter comprising an inductor and a bead inductor, that passed the automotive test limits when tested according to CISPR25 at an independent test facility [2].This was for TPS54360- Q1 with a 3.5A load.

The high ambient temperature which automotive electronics have to be designed for (>85°C) can create issues with ICs getting too hot. Synchronous dc/dc converters (two integrated MOSFETs in the IC) have all the losses in one IC pkg. Non-synchronous dc/dc converters (one integrated MOSFET plus an external Schottky diode) spread the losses between two packages, with a corresponding bigger thermal area. Furthermore, when computing MOSFET power losses it is important to use their on-resistance value, Rds(on), at a realistic junction temperature of 125°C, not room temperature. Calculating or measuring conduction losses at room temperature is not realistic and could easily lead to disappointment when the ambient is moved up to 85°C. At high temperature then MOSFET Rds(on) is significantly higher, (by 0,5%/°C). Conversely, the forward voltage of a Schottky diode decreases with temperature, reducing losses. The conduction losses in a MOSFET are proportional to the rms current squared, whereas in a diode they are proportional to the average current. Switching power loss, Psw, in the top MOSFET is approximately Psw = Vin x Iout x Fsw x Tr where Tr is the rise time of the voltage on the switching node. (The switching loss in the bottom MOSFET is small and ignored in this discussion). From that equation, and all other things being equal, switching above the MW band compared to below it gives a four or five times increase in switching losses, simply because the switching frequency is four or five times higher. To calculate the total power dissipation in the IC, then switching losses have to be added to conduction losses.

In summary, a synchronous buck converter has two sets of MOSFET conduction losses plus one of switching losses dissipated in one IC package. A non-synchronous buck converter has one set of conduction and switching losses in the IC and one set of conduction losses plus a small switching loss in the external diode.

Two dc/dc Converters with Clock Synchronization at Different Frequencies via a Divide by Five. TPS54360-Q1 is switching below the MW band and the PMIC above it

In an automotive power topology and given the tradeoffs discussed, then a low input voltage power management IC (PMIC) can be used, switching at 2.2MHz(typ), plus an upstream converter, TPS54360-Q1, as shown in Figure 3. TPS54360-Q1 was operating off the car battery and is switching just below the MW band. The switching frequencies of the converters can be phase-locked by dividing-down the 2.2MHz PMIC clock by a divide by five circuit [3]. The output of the TPS54360- Q1 can be set to either 3V3 or 5V, but the current is lower for the same output power when 5V is used and so it is preferred.

The TPS54360-Q1 is part of a new family of non-synchronous (MOSFET + external diode) automotive dc/dc converters with a max Vin of 60V, or 42V for TPS54340-Q1 and are rated at 3.5A. TPS54540-Q1 and TPS54560-Q1 are higher current 5A versions. (The term “nonsynchronous dc/dc converter” only refers to the fact it has one MOSFET and a diode and is not referring to their ability (or not) to accept an external switching frequency clock synchronization signal.) The ICs have a low drop-out voltage, which means that they retain Vout voltage regulation when the Vin has decreased so that it is only a few hundreds of milli-Volts above Vout and the duty cycle is high, D = Vout / Vin. Their SOIC plus power-pad package means the pin-pin and pinpad metal-metal spacing is large at 0.8mm, to increase reliability under tough high humidity environmental conditions. As expected, their cost is very competitive as they are non-synchronous buck converters, which are lower cost than synchronous buck converters. These new ICs can all be synchronized to an external clock.

Conclusion

Synchronous buck dc/dc converters switching above the MW band are not a panacea. Non-synchronous dc/dc buck converters like TPS54360-Q1 family, switching below the MW band have many desirable qualities in automotive applications. Thermally, the losses are spread between two packages; the IC and the external diode, plus the switching losses are lower. The conducted EMI in the MW band can be managed by locating the switching frequency just below it and in the FM band by managing the switch node dv/dt. A simple two stage input filter can meet the conducted emissions requirements.

References
1. Frank Dehmelt, Texas Instruments. “Pulse-skipping: reasons, effects and relevance for automotive applications” Part 1 and Part 2. August 2013. EETimes Europe, Power Management section.
2. TPS54360-Q1 filter design and testing performed by Anthony Fagnani of the SWIFTTM Power group of Texas Instruments.
3. Dan Tooth, Texas Instruments “Divide by N for Synchronizing Dc/Dc Converter Clocks.”

 

VN:F [1.9.17_1161]
Rating: 0.0/6 (0 votes cast)

This post was written by:

- who has written 791 posts on PowerGuru - Power Electronics Information Portal.


Contact the author

Leave a Response

You must be logged in to post a comment.