Push Pull Transformer Driver design

[mag_therm]

Yes, but since that time in 2020  I have made a few more hobby push pull pwm inverters.

I am using the IXGH32N60C IGBT (without antiparallel diode). Current range 12 Amp DC, up to 24 Amp peak/side.
Take a look at Fig 4 of its datasheet  which shows a negative  Vce tempco in that current range.

I am using no airgap in the ferrite transformers. The inverters have perfect current balance, as measured.

I do remember the olden days. With 50 Hz UPS fundamental scr inverters in 30 ~ 100kVA range,
we used a low grade lamination (like motor grade) as part of keeping the currents even.
As the inverter heated up on full load, the current at end of conduction would start to curl up on one side.
Further indication of an offset in the core was to hang a steel wrench (spanner) above the yoke.
With a little offset the wrench would start to swing like a compass in the yoke flux. Great Fun?

[T3sl4co1l]

The MOSFET is used as the main switching transistor of the push-pull converter, which can have the characteristic of automatic "imbalance correction" during its operation. In practical applications, the on-time of two transistors in a push-pull converter cannot be guaranteed to be 100% completely symmetrical, and the slight deviation of the turn-on time Ton still causes the volt-second products of the push-pull transformer unable to be exactly equal, resulting in an imbalance of magnetic flux. This imbalance process leads to an increase in the operating current of its corresponding circuit, and the corresponding MOSFET generates additional losses, causes its temperature to rise. Due to the positive temperature coefficient of MOSFET's R(on), the voltage drop during MOSFET's conduction increases accordingly, this eventually causes the voltage amplitude Vp in the turn-on circuitry - obtained via voltage division from the primary winding of the transformer - to decrease, hence achieving automatic "imbalance correction".

Interesting, suggests they have output sensing. Or maybe it's just because the die is heated locally which affects the nearby drive circuitry, but that seems like it would be VERY minor.

Thanks, have been curious about equivalents, SN6505 is, either obsolete or poorly available.

Tim

[niconiconi]

After further investigation, this paragraph actually came from TI SN6507 datasheet, or another SN-series chip datasheet from TI (unsurprisingly, the same company also cloned other transformer driver chips in the same category).

https://www.ti.com/lit/gpn/sn6507

Here's the uncorrupted original text:

Fortunately, due to the positive temperature coefficient of a MOSFET’s on-resistance, the output FETs of the SN6507 have a self-correcting effect on V-t imbalance. In the case of a slightly longer on-time, the prolonged current flow through a FET gradually heats the transistor which leads to an increase in Rds(on). The higher resistance then causes the drain-source voltage, VDS, to rise. Because the voltage at the primary is the difference between the constant input voltage, VIN, and the voltage drop across the MOSFET, VP = VIN – VDS, VP is gradually reduced and V-t balance restored.

[Zoli]

...
Interesting, suggests they have output sensing. Or maybe it's just because the die is heated locally which affects the nearby drive circuitry, but that seems like it would be VERY minor.

Thanks, have been curious about equivalents, SN6505 is, either obsolete or poorly available.

Tim

Take a look on LCSC for VPSC:
https://www.lcsc.com/brand-detail/12739.html
They have a family of 6 in this category. One warning: all the products are MSL-3.

[niconiconi]

Thanks, have been curious about equivalents, SN6505 is, either obsolete or poorly available.

Chip shortage. SN6501 is out of stock  , SN6505 is out of stock  , SN6507 is out of stock  , MAX258 is out of stock  , MAX13253 is out of stock  , LT3439 is available but it's a premium chip...  Mouser says a new batch of SN6505 will arrive in 2024. I guess SN6505 was already a low-demand but still in-production chip before the shortage and the supply got a hard hit. The SN6507 looks better, as it's a new chip and a batch should arrive in January 2022.

On the meantime, the clones are the solution.