Author Topic: Couple of questions for DC/DC guru's to ponder for me?  (Read 572 times)

0 Members and 1 Guest are viewing this topic.

Offline max_torqueTopic starter

  • Super Contributor
  • ***
  • Posts: 1327
  • Country: gb
    • bitdynamics
Couple of questions for DC/DC guru's to ponder for me?
« on: April 07, 2020, 11:39:23 am »
Hello everybody!  I hope we are all keeping safe and sound in these, er, interesting times!  But anyway to my question:

I'm reverse engineering an automotive DC/DC converter, non isolated 12v to 48v, bi-directional, reasonable power (think it's about 1200W).  There's nothing too un-conventional about it, using off-the-shelf components to implement a 6 phase interleaved architecture so it can drop phases at light loads to help boost efficiency across a wider operating window (and of course to help minimise the fundamental ripple at high loads).

Question 1:

The mofset bridges each use a pair of mosfets (high and low) driven in parallel from a single bootstrapped N channel gate driver IC (ISL7840, with selectable deadtime, for those playing along at home):

965124-0

but, why have they laid out some of the power tracing to the devices as 3 parallel single traces, rather than say a block pour?  I've marked the 3 parallel traces in blue and red, blue is the ground Low Fet reference and red is the Vsupply High reference.   Any good reason anyone can think of? Thermal/expansion issues? Lower total plane area for lower EMC (seems unlikely as some of these traces aren't really in the switching path). Autorouted perhaps? or just quicker for a manual op to throw click in 3 parallel "follow me" traces than put in a and dimension a poured block etc?


Question 2:

They are using an OTS PWM controller, the ISL78220, which can drive 6 interleaved phases so all the fast and hardlifting work is being done in this chip, meaning the devices system controller (and STMicro uC) is just left to do the housekeeping as it were.  BUT: they have installed two parallel, multiplexed PWM controllers on the board?

965120-1

On the upper (in that pic) device labelled "U7" the PWM output comes out pin 14, i've  highlighted it in red, runs through an inner layer to a Texas Inst SN74LV4053 tripple dual channel multiplexer ("U16") arriving on Pin 1 2Y1.

The corresponding PWM pin on the lower of the PWM controller ("U8"), highlighted in green, runs up and into Pin2 2Y0 of the that multiplexer.  The output from that multiplexer channel, PIN15 2COM, which runs off to the "PWM input" pin of the mosfet gate driver for one of the bridge pairs, can therefore be in one of three states:

1) Isolated, neither controller gets to drive the gate driver, when the enable pin of the multiplexer is not asserted - shut down mode, allows the system controller to call time on proceedings at any moment independant of the actual PWM controllers actions

2) Connected to U7 the upper PWM controller

3) Connected to U8 the lower PWM controller

So why have they done this??   Possible reasons i can think of:

a) redundancy?  this is an high reliability automotive product after all, but it seems a strange choice

b) the controllers have different "settings" in their hardware/ in-circuit setup, settings that can't be easily dynamically controlled by the system controller, for example PWM frequency perhaps, or some other drive modulation that could need to be different between some operating mode requirement of the controller (ie different when current is going 12V-48V as opposed to 48V -> 12V)

c) Some other reason i haven't thought of   :-DD



I'd be interested to hear peoples thoughts on the matter    :-+



« Last Edit: April 07, 2020, 11:42:16 am by max_torque »
 

Offline max_torqueTopic starter

  • Super Contributor
  • ***
  • Posts: 1327
  • Country: gb
    • bitdynamics
Re: Couple of questions for DC/DC guru's to ponder for me?
« Reply #1 on: April 07, 2020, 12:09:39 pm »
I may have answered Q2 myself, as it looks like the voltage reference for the PWM controler PIN 4 FB, is connected to the 12V side on U7, and to the 48V side on U8.  I guess therefore somehow the system controller must switch between the two controllers depending effectively on which voltage it wants to control, ie if the DC/DC should source current from the 12v side to keep the 48v at 48v, or if it should source current from the 48V side to keep the 12v at 12V.  Kind of like a basic current controlled converter?
 

Offline digsys

  • Supporter
  • ****
  • Posts: 2209
  • Country: au
    • DIGSYS
Re: Couple of questions for DC/DC guru's to ponder for me?
« Reply #2 on: April 07, 2020, 01:11:21 pm »
I have worked with these "type" of converters before, but without knowing the make / model you refer to, it's difficult to even make any sort of "intelligent" comment.
I may "possibly" even have a service manual or similar, if it is one I know.
Hello <tap> <tap> .. is this thing on?
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf