Author Topic: Any clues why this buck-boost IC's internal 5V Vreg never starts? [Solved]  (Read 982 times)

0 Members and 1 Guest are viewing this topic.

Online PsiTopic starter

  • Super Contributor
  • ***
  • Posts: 9924
  • Country: nz
I'm trying to design a buck-boost using the MAX25431 IC but am having an odd issue with the first RnD prototype.

The chip doesn't want to start up its internal 5V Vreg  (VCC pin).
There is nothing on the PCB loading the VCC output, (I removed IC and fed in 5V myself just to be sure, no current draw).
The cap on the line is 4.7uF as specified by datasheet, and I checked it, value is good.
But VCC always rises to 100mV or so and just sits at that voltage.
It should be 4.9-5.1V and nothing is going to startup inside the chip until its 4V, as there's a 4V under-volt lockout on VCC.

Schematic and PCB images below.
I have removed U113 and C195 from the current sense system as well as Q4 to keep things simple for now, none of that should interfere with starting.
I've also tried connecting enable line to both 5V and to 14V input and it doesn't help. (datasheet wasn't clear if you can tie enable to VCC or have to supply external power but neither worked)
Have also tried a few different MAX25431 chips incase i blew it up somehow, they all do the same thing.
VCC internal Vreg refuses to start so nothing else starts.
Chips were from digikey, so i'm sure they are genuine.

Any ideas?


Datasheet
https://www.analog.com/media/en/technical-documentation/data-sheets/MAX25431.pdf


« Last Edit: September 29, 2023, 11:42:26 am by Psi »
Greek letter 'Psi' (not Pounds per Square Inch)
 

Online PsiTopic starter

  • Super Contributor
  • ***
  • Posts: 9924
  • Country: nz
Re: Any clues why this buck-boost IC's internal 5V Vreg never starts?
« Reply #1 on: September 28, 2023, 01:22:55 am »
I'm tempted to feed external 5V into VCC myself from the PSU and see if the chip starts.
But that is probably a bad idea, it's not starting for a reason.
Greek letter 'Psi' (not Pounds per Square Inch)
 

Offline jwet

  • Frequent Contributor
  • **
  • Posts: 452
  • Country: us
Re: Any clues why this buck-boost IC's internal 5V Vreg never starts?
« Reply #2 on: September 28, 2023, 02:04:33 am »
The little outboard Isense amp, PFET Q4 and Q159 all look a little suspect.  Can you disable this circuit temporarily, I think its getting in the way of startup.
 

Online PsiTopic starter

  • Super Contributor
  • ***
  • Posts: 9924
  • Country: nz
Re: Any clues why this buck-boost IC's internal 5V Vreg never starts?
« Reply #3 on: September 28, 2023, 02:15:01 am »
The little outboard Isense amp, PFET Q4 and Q159 all look a little suspect.  Can you disable this circuit temporarily, I think its getting in the way of startup.

yeah, that was my first thought too so I removed it.
Q4 was removed as was U113 (the IsenAmp) and C195 (its cap to GND)
So that has removed it all from the VCC rail circuit and shouldn't interfere with VCC or feedback if the max chip does decide to start up.

I also tried feeding 5V into the VCC externally at power on, just to see what happened.
I had psu set to 5V and 180mA current limit.
It imminently went to 1.1V at 180mA and the IC started to heat up.

So I have no idea why the IC cant handle 5V on its 5V output.  Doing that may have popped this IC, but was worth testing just to see what happened.
« Last Edit: September 28, 2023, 02:36:51 am by Psi »
Greek letter 'Psi' (not Pounds per Square Inch)
 

Offline slugrustle

  • Frequent Contributor
  • **
  • Posts: 278
  • Country: us
Re: Any clues why this buck-boost IC's internal 5V Vreg never starts?
« Reply #4 on: September 28, 2023, 02:38:24 am »
I think you're meant to tie EN to VIN rather than to VCC, given that EN is rated for the full VIN voltage.

I didn't read the datasheet closely enough to find out (and maybe it doesn't even say), but I'll bet the VIN UVLO has to clear and the bandgap reference has to start up before the LDO that supplies VCC is enabled. But, it's not going to do this if EN is held low...

As for the current draw on VCC in backfeed, a few hypotheses, perhaps none of which are correct:
  • LDOs tend to have a body diode in the pass transistor.  Maybe you were backfeeding a bunch of stuff through that.
  • You've got those bootstrap diodes.  Who knows what they hook up to in the chip.
  • This chip may intentionally discharge VCC when it is disabled so startup is clean the next time and VCC UVLO doesn't just immediately pass on next startup.
  • To hand wave a bit, it's not surprising the chip is biased funny in this state.

EDIT: The schematic in the datasheet for the eval kit has a jumper that connects EN to VIN to start the IC and that connects EN to GND to disable the IC.  Connecting EN to VIN is at least worth trying.
« Last Edit: September 28, 2023, 02:48:45 am by slugrustle »
 

Online PsiTopic starter

  • Super Contributor
  • ***
  • Posts: 9924
  • Country: nz
Re: Any clues why this buck-boost IC's internal 5V Vreg never starts?
« Reply #5 on: September 28, 2023, 02:44:03 am »
I think you're meant to tie EN to VIN rather than to VCC, given that EN is rated for the full VIN voltage.

I didn't read the datasheet closely enough to find out (and maybe it doesn't even say), but I'll bet the VIN UVLO has to clear and the bandgap reference has to start up before the LDO that supplies VCC is enabled. But, it's not going to do this if EN is held low...

Yeah, I noticed that this morning so cut the track and added a bodge wire to feed VIN into EN instead of VCC. I think it threw me off a bit because the EN pin is talked about in the 'logic level' section of the datasheet so I connected it to VCC thinking it was logic level.
But fixing this didn't solve the problem though.
However, maybe I'll just try a fresh MAX chip with EN=VIN, Just in case the old chip got damaged by having EN connected to VCC somehow. Maybe it oscillated and damaged the regulator somehow.
 

As for the current draw on VCC in backfeed, a few hypotheses, perhaps none of which are correct:
  • LDOs tend to have a body diode in the pass transistor.  Maybe you were backfeeding a bunch of stuff through that.
  • You've got those bootstrap diodes.  Who knows what they hook up to in the chip.
  • This chip may intentionally discharge VCC when it is disabled so startup is clean the next time and VCC UVLO doesn't just immediately pass on next startup.
  • To hand wave a bit, it's not surprising the chip is biased funny in this state.

Yeah, i'm not too surprised about the back-feeding causing problems.
This chip has been giving me grief all day and I just felt like doing it  :-DD
I didnt really expect it to prove anything. Was just curious
« Last Edit: September 28, 2023, 02:48:01 am by Psi »
Greek letter 'Psi' (not Pounds per Square Inch)
 

Offline slugrustle

  • Frequent Contributor
  • **
  • Posts: 278
  • Country: us
Re: Any clues why this buck-boost IC's internal 5V Vreg never starts?
« Reply #6 on: September 28, 2023, 02:50:42 am »
It's very common for regulators or controllers with enable pins to have logic level thresholds (so they can be driven from a microcontroller or other logic) and to also tolerate full input voltage so users who simply want the chip to turn on when power is applied can hook enable to the input voltage.  I see this a lot.
 

Online PsiTopic starter

  • Super Contributor
  • ***
  • Posts: 9924
  • Country: nz
Re: Any clues why this buck-boost IC's internal 5V Vreg never starts?
« Reply #7 on: September 28, 2023, 03:28:40 am »
I put a new chip on and this time it blew up.  Which I guess is somewhat good news,
When it blew up i had the scope on VCC and i saw a much higher voltage than ever before.
So I guess it is progress, even if it did blew up, it had VCC running this time.

The magic smoke took out a PCB trace to the chip (VIN), so I'm currently moving parts on a fresh PCB.
Strangely none of the fets blew, just the IC. So i'm not really sure why that happened or how to prevent it next time other than lower voltage and current limit. Which doesn't always work on switchmodes.
« Last Edit: September 28, 2023, 03:30:39 am by Psi »
Greek letter 'Psi' (not Pounds per Square Inch)
 

Online PsiTopic starter

  • Super Contributor
  • ***
  • Posts: 9924
  • Country: nz
Re: Any clues why this buck-boost IC's internal 5V Vreg never starts?
« Reply #8 on: September 28, 2023, 08:01:41 am »
So I had working, barely, for a short time, 10V in 17V output. However, what's odd is that it was clicking every second but the input current was ok. (no output load)
but it just randomly died after a power cycle and now the VCC is stuck at 1V.
So it looks like the weakest link in these chips is the internal 5v regulator. 

I did discover i forgot to connect the feedback line when I rebuilt it after it blew up before. So it was probably running flat out trying to get the voltage up and that may have produced a voltage so high it damage the chip.

Ok, new chip time. This is my last chip before I have to order more from Digikey.
Lets hope it was just the lack of feedback.
« Last Edit: September 28, 2023, 08:04:11 am by Psi »
Greek letter 'Psi' (not Pounds per Square Inch)
 

Online PsiTopic starter

  • Super Contributor
  • ***
  • Posts: 9924
  • Country: nz
Re: Any clues why this buck-boost IC's internal 5V Vreg never starts?
« Reply #9 on: September 28, 2023, 09:02:42 am »
I have it working 10V in 16V out.
The input current is practically nothing atm, a few 10's of mA and output is unloaded.

However the new problem is it randomly going unstable and drawing amps (unloaded).

Probably need to optimize frequency, comp, inductor value.
It's strange though, I used the values from the analog.com Esim calc for my in/out voltage/current application.
Or maybe something about my PCB layout is terrible for 1.4mhz operation.


« Last Edit: September 28, 2023, 09:10:48 am by Psi »
Greek letter 'Psi' (not Pounds per Square Inch)
 

Online deepfryed

  • Supporter
  • ****
  • Posts: 129
  • Country: au
Re: Any clues why this buck-boost IC's internal 5V Vreg never starts?
« Reply #10 on: September 29, 2023, 01:04:06 am »
noob question, should there be a ground pour under the inductor  between the 2 pads ? I always thought that should be kept clear.
 

Online PsiTopic starter

  • Super Contributor
  • ***
  • Posts: 9924
  • Country: nz
Re: Any clues why this buck-boost IC's internal 5V Vreg never starts?
« Reply #11 on: September 29, 2023, 02:06:08 am »
noob question, should there be a ground pour under the inductor  between the 2 pads ? I always thought that should be kept clear.

interesting point, I have no idea.
I guess it would make sense to prevent the magnetic field seeing nearby copper as a shorted turn and reducing efficiency.

Now that I have this thing running I think I will make a new thread dedicated to the instability issue, once i get some scope traces and such.
Probably doesn't make sense to keep going on a thread about a fixed issue.
« Last Edit: September 29, 2023, 02:07:46 am by Psi »
Greek letter 'Psi' (not Pounds per Square Inch)
 

Offline moffy

  • Super Contributor
  • ***
  • Posts: 1710
  • Country: au
Re: Any clues why this buck-boost IC's internal 5V Vreg never starts?
« Reply #12 on: September 29, 2023, 04:48:17 am »
I have it working 10V in 16V out.
The input current is practically nothing atm, a few 10's of mA and output is unloaded.

However the new problem is it randomly going unstable and drawing amps (unloaded).

Probably need to optimize frequency, comp, inductor value.
It's strange though, I used the values from the analog.com Esim calc for my in/out voltage/current application.
Or maybe something about my PCB layout is terrible for 1.4mhz operation.

Stability should improve with a minimum load, from a control standpoint a load helps reduce the loop gain and improves stability.
 

Offline SiliconWizard

  • Super Contributor
  • ***
  • Posts: 14431
  • Country: fr
Re: Any clues why this buck-boost IC's internal 5V Vreg never starts?
« Reply #13 on: September 29, 2023, 05:11:39 am »
noob question, should there be a ground pour under the inductor  between the 2 pads ? I always thought that should be kept clear.

That's a good question. Both can be seen in actual designs that work.
Note that firstly, it would depend on whether the inductor is shielded or not. If you use a shielded power inductor (which is a relatively common thing these days), then it's likely not going to make a lot of difference.
If it's unshielded, there may be some minor effect, although in all application notes I've seen, they were at least recommending a solid ground plane in an internal layer below *all* of the converter's layout without keepout for inductors or such, and that would be like just 0.2mm below the surface, or even less for thin PCBs. So...
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf