I am experiencing a challenge with the design of a DCDC Boost converter that is intended for gasoline direct injection control.

This DCDC converter is configured to generate 65V from a 12V supply, utilizing the NXP PT2001 controller (the MC33816 is quite similar, as the PT2001 manual is currently under NDA).

The controller is set up with a voltage control hysteresis ranging from 65V to 64V. The current flowing through the inductor is maintained between 2-5A (the MOSFET gate toggles on and off) when the voltage drops below 64V. Once the voltage reaches 65V, the control loop pauses until the voltage falls below 64V again.

While the DCDC converter operates effectively under low load conditions, a problem arises under high load scenarios (such as high RPM in the engine, which increases fuel injection within the same timeframe). During these instances, it appears that the MOSFET begins to close very slowly. I have included a picture of the circuit, along with a graph showing the gradual decrease in current as measured at the shunt resistor (10mOhm), and the gate voltage at that moment, measured between the gate resistor (2.2 ohm) and the gate.

Any insights or explanations regarding this phenomenon would be greatly appreciated!


Schematic of circuit (both of the MOSFET has been tested separately with same result)


Voltage at shunt resistor during slow closing/decrease of current


Gate voltage during slow close of MOSFET


Voltage at shunt resistor during normal/correct operation

 

Seems pretty normal.
When the coil is discharged to low impedance (aka high load) the discharging skew is very slow.
The ripple of inductor current is very low and the converter is deep in continuous conduction mode (CCM).

It’s a matter of basic (poor sophisticated) chip that doesn’t decrease the switching frequency during this state.

 

Thank you for your response!

The system maintains a steady voltage of 65V for approximately 10 seconds under high load conditions. However, it then experiences a rapid drop in boost voltage from 65V to 12V, resulting in a slower regulation response, as illustrated in the image from the initial post. Can you provide any theoretical insights into why the converter is unable to sustain the high voltage anymore?

Since the ripple current through the inductor is controllable via software, I have experimented with various settings. By increasing both the low and high current limits in equal increments, I was able to delay the voltage drop by a few seconds. However, when I increased the current span between the on and off states of the MOSFET beyond the NXP default of approximately 3A (specifically to 2A and 5A), the performance actually decreased.

 

Probably the big output cap (1100uF) is able to hold this overload state for certain period of time.

You can try to reduce it whether the output voltage falls quicker.

In high loads the inductor has to be charged to higher current to deliver enough energy to output.

 

What You are trying to accomplish is no easy task.
You need to build a very stout Power-Supply and then create
a PWM Current-Limiter-Circuit.

Below are the designs that I settled upon.
They have not been tested in the real World, only on simulation software.

Other concerns are the very precise Timing required for both opening and closing events.
I've worked this out using a Mega-Squirt-Fuel-Injection-Computer and
a custom designed Comp-Cams 4-Lobe High-Pressure-Fuel-Pump-Cam-Lobe.

Holley makes a special add-on-Box that will drive High-Pressure-Fuel-Injectors.
.
.
.

.

 

Is this a new design that is having problems, or is this an existing system that is suddenly not working correctly?? Or is this a development project still in process?? The diagnostic processes for the three different possibilities are quite different. So please clarify which it is, because for some reason it is not clear to me which is the situation.

 

This is a new design closely following a reference model provided by NXP.

Upon reflection, the earlier response suggested that the observed behavior resembles typical discharge behavior through the coil at high load. However, since the current shunt is positioned on the source side of the N-MOSFET, I would expect to see the current decrease rapidly during all conditions. Does this not imply that the MOSFET is not turning off as quickly as anticipated?

Application note AN4849 from NXP describes the solution I am trying to implement.

NXP PT2001 (Chip used in this design to control both injectors and DCDC)
NXP MC33816 (previous revision of the PT2001, but with some more info)

 

See precisely where the shunt is connected. The output ground and input ground are not the same as usual. The output ground is a shunt voltage higher.

Another words, both charging and discharging current flows through shunt.

 

In the one link I see a circuit with two shunt resistors, and the whole load current passes thru the shunt, in addition to the source current. That is also what I see in post #1.
In the simulation, are you able to watch the source current directly? THAT will show what is actually happening.