LM34910: no switching when system voltages are sequenced 'on'

Katelyn,

I show two lists, one with Vin applied and below that another without Vin.  The difference at Feed back is rather insignificant both sitting around 1.15 v

When I apply only the Vin, yes this part starts normally.

I have several regulators, all using the LM34910c,  and most often 3 out of 5 do not start when sequencing.  The 'sequence' is to apply an external 3.2V to my MCU and associated circuits. In-turn the MCU decides when to turn on the Vin.  the two regulators that does turn on consistantly is shown as the +16V  & +12v regulators ( 12 is actually now operating at about 10V ).

Well I tried to insert the schematic but not sure about the result.

They are closing the doors now so I will look forward to your reply Monday.SCHEMATIC1 _ PWR_A.pdf

Schematic power section.docx

Katelyn,

More data and a threshold found.

Regarding the LM34910C failure to start.

As mentioned before the regulator will start if only Vin is applied first. Thus eleminateing the presence of leakage voltages from the system.

Here are some interesting measurements taken from the +5V regulator circuit when system voltages are present and varied for test.

Conditions:  applied various voltages to the system 3.3V rail, then increasing the Vin to about 8 or 9 volts to observe 'start' or 'no star't of the +5V regulator to thereby define a threshold/boundary.

summary; 

at 2.97V system voltage, followed with application of Vin,  the +5V regulator starts switching and regulation as expected.

at 3.05V system voltage, followed with application of Vin, the +5V regulator does not start.

Below are two sets of voltages measured at the pins of LM34910C. The first set, using system voltage: 2.97  and the second set using: 3.05 volts.

system voltage = 2.97V

Pin    volts

1     2.095

2     1.714

3     0

4     0

5     0

6     1.049

7     0

8     1.185

9     0.625

10   1.611

system voltage = 3.05V

pin     voltage

1     2.218

2     1.879

3     0

4     0

5     0

6     1.111

7     0

8     1.227

9     0.737

10   1.732

Without details ( transistor level schematic )  of the internal circuitry of the LM34910  all I can surmise is the possibility that an SCR type of function may be occurring that causes a latch-up. No excessive currents are drawn and the regulators seem to stay at ambient temperature, not the slightest warm to the touch.

Jaye

I will need a day to setup and capture the startup sequence.

As for the 'ideal' circuit design, please consider the following: 

           For economy and production matters I wish to have the inductor the same value in each regulator circuit. 

 The startup loading is quite light.  3V  is between  0.00  and  1.5A  [  the 3V regulator, as seen, is diode blocked and is wire-or'ed with a  3V, 1A source from the main power supply.  When the loading causes the main supply to droop the LM34910C regulator then sources current to minimize the droop. ]   In some product sub-models the 3V regulator will essentially never source current while another model will constantly source current.

  +5V  does not share any loading and provides about 0.1 A

  +/- 12V  which is now actually set for +/-10V ,  provides about 0.04A

I was at the point of re-running the Web Bench model  when I discovered you ( T.I. ) had disabled Web bench for the LM34910.  As an aside, the person/s who contacted me in response to my request for assistance  with Web Bench simply stated it had been disabled for LM34910.  They have never provided anything in the way of when it will be made available for the LM34910, nor any hint of why it was disabled.

I chose the LM34910C  because I have used it in another product have never had a failure since the product started shipping  4 years ago.  My goals are quite simple: have a solid design which will last no less than 6 years running 24/7 in very harsh environments with excursions to  80 deg. ambient.  

Thank you.

Jaye

Katelyn,

A discovery:

As you mentioned earlier, there is a relation between parasitic voltage present on the SW (output) pin before Vin is applied.

It also relates to the BST (boost) pin.  Apparently the parasitic voltage passes through the SW MosFet and biases the internal circuit resulting in a locked condition. When Vin is then applied, no switching results. After Vin is applied, with no switching, I used a 30 Ohm resistor with a test lead connecting one side of the resistor to ground and then momentarily and individually touched the resistor to the SW and BST pins . This loading effectively pulled the respective pins down through the 30 Ohms.  Both pins, individually, responded with the regulator commencing  operation. The BST pin appears the better possibility to create and apply an external pulse to.  I would use the rising event of Vin above the typical turn-on, conditioned with having no switching present, to result in a periodic pulse to BST.  The expectation would be 'within one perhaps two pulses the switcher starts'.   Well if it makes my system work it's not unreasonable.

Still does not explain precisely what is locking up, but having a possible means to unlock it is good. Potential problem is degrading the BST circuit by hanging anything onto it.

Perhaps you can find a better solution.

Thank you for pursuing the Web Bench team to have the LM34910  re-included.

Jaye

Katelyn,

Knowing the exact problem inside the LM34910 is key to knowing if I have fixed it.

However;

I have a functional solution to the problem where LM34910C fails to start if system leakage voltage appears on the SW pin before Vin is applied. Which is my case, due to having an off-board smps supplying 3.3V for the 'sleep' or 'keep alive' voltage while the other voltages are turned off in my system.

A combination of diode blocking the D.C. output from the regulator circuit and raising the feedback impedance allows the LM34910C to start.  However I need to make empirical measurements to see where the operational margin is ( that point where it may or may not start ). 

For the SW pin, if the leakage voltage present at SW  is less than 2V then  LM34910 will start.  If SW is above 2V it will not start when Vin is applied.  Exact voltages vary 10's of millivolts from board to board.

For the FB pin a low impedance and leakage voltage present influences the failure to start.

With Diode blocking the Filtered regulator output:   having low feedback impedance, ~ 1K Ohm, coupled with the system leakage voltage I observed an occasional failure to start. I raised the impedance to ~10K Ohm and have not observed a failure to start.

So while I appear to have a workable solution I still would like to know the mechanism of the LM34910C  which causes the failure to start When >2V is applied to the SW before Vin is applied. And how the FB path can influence the failure.

Thank you.

Jaye

Hi Jaye,

I see the same performance on the EVM.  If there is about 2V on the output, the device does not start up, but if the output sees less than about 2V, the device does start up.

The failure to start up is related to the voltage at the FB pin.  The FB voltage and the SS start voltage are compared through the regulation comparator in the device.  If there is a high voltage present on the FB pin, this will cause the SS voltage never to go high.  The SS voltage must go high in order for the device to start up.

The FB path is critical to fix this failure.  You will want to add a method to block the voltage from the feedback pin in order to ensure that none of the devices will fail to start up.

Best Regards,

Katelyn