Push-Pull start up issue?

[Warpspeed]

The problem is not core saturation, the core is not going to be damaged if it saturates.

But the switching devices may easily fail if saturation causes the peak current into an uncontrolled upward spike.
So a very fast peak current limit is what is really needed, and current mode control provides that feature.

 

[treez]

So a very fast peak current limit is what is really needed, and current mode control provides that feature.

yes, I agree and I am sure you would agree that current mode involves "leading edge blanking" time...this is time where the current mode "thing" is not working at all...it is OFF......and in this incidious, but very necessary time, the inductor current can staircase upwards in situations where the error amplifier gets railed high such as overloaded output or non-soft re-start.

I am afraid that during the repetitive leading edge blanking time, you do not have any current limiting at all. None whatsoever.
This is easily shown on simulation, as you will know, and on the bench.

 

[Warpspeed]

There are two possibilities here, the first is that the control loop is operating well within its small signal control range, and the current peaks are well controlled.

There may be some extreme initial condition during start up, where the closed loop completely saturates and locks up, and stays out of its control range for a time.

But there are tools to get around that problem, such as soft start, and a separate direct acting fast current limit pin that overrides the control loop.

 

[treez]

separate direct acting fast current limit pin that overrides the control loop.

..I suppose in the ideal world we would have that, but in most controllers, when you trip the second current limit, you go into a temporary shutdown where the soft start pin gets discharged, and you have to wait for it to kind of restart.

Soft start is great if you have it, but of course, as you know, it takes a devilish amount of extra circuitry to ensure that you always get soft "re-starts"....I reckon i'd rather just gap the core...some cores come with a 0.2mm (or less) gapped version off the shelf. I wouldn't mind betting that's for the reasons described here.....I reckon its even worth having a cheap mini gap by adding shims....that gaps the outer legs too, bad news, but if its a tiny gap, who really cares.

I'd say shim it, scrub all the extra circuitry, and sleep easier that you are further away from saturation than those who don't gap.

 

[Warpspeed]

..I suppose in the ideal world we would have that, but in most controllers, when you trip the second current limit, you go into a temporary shutdown where the soft start pin gets discharged, and you have to wait for it to kind of restart.

That surely has to be better than letting the smoke out ?

If the thing always stutters into action, there is a serious design problem that needs to be investigated.
Designing reliable gear means first ensuring bad things can NEVER happen.
But when they invariably do, have multiple layers of protection.
Finally take steps to ensure that there can never be a cascading type of failure where one small fault can lead to total destruction of almost every component on the board.

Don't laugh, when supply rails go way beyond anything reasonable the result can be pretty major.

 

[treez]

Yes I agree those ‘trip and re-try circuits’ definitely have their useage…but I am sure you would agree that sometimes you want to ride through an overload, only tripping out if the overload goes on for too long…….but if you don’t gap, you can’t ride through a short term overload…for example, class d audio amp SMPS’s often go in and out of overload, that’s how a class d works, -you could just make a bigger smps but cost means that’s not viable…so gap and ride through the overload.

Designing reliable gear means first ensuring bad things can NEVER happen

I am sure you would agree that cost means that you often can't put in a protection circuit to prevent every single mishap....so making it more robust by gapping is a nice half way house.

I think there are 2 points at this stage
1....we agree that a tiny gap does no harm
2....we agree that a tiny gap makes it more robust than no gap.

 

[Warpspeed]

From what I have seen, the gap solution is usually a panic band aid solution after customer fury morphs into management fury, and the (new replacement) engineer is tasked with fixing his predecessors screw up with minimal drama.

- - - Updated - - -

*edit*

I am sure you would agree that cost means that you often can't put in a protection circuit to prevent every single mishap...

That is the difference between a young engineer and an old engineer.

At a planning meeting, management INSIST you save ten cents by not fitting a fuse. The young engineer says "yes sir".

The old engineer gets mean and onery at planning meetings, narrows his eyes like Clint Eastwood, pulls out his gun and says "punk, we gonna fit a fuse to this mother" ten cents be damned.