We are using the LTC3789 alonside the LTC4000 in an arrangment which is identical to the DC1721A-A Demo board (schematic and layout) for a LiFe battery charger and portable radio PSU application. Unfortunately we have had half a dozen failures of the LTC3789, causing product failure in the field. The 3789 device fails without warning and ends up running very hot, The VCCINT regulator output has been seen at 9V+ on some failed boards, so it looks like the internal 5V5 regulators are cooked. We don't know if the failure coincides with a user operation, e.g. connecting/disconnecting Vin or the load. The input voltage is 12V nominal, and we have plentry of extra protection (surge and polarity) on the board ahead of the LTC3789. The 12V is supplied from a good quality AC-DC adapter block spec'd for an 11.5A output. Our maximum load is approx 6A at 12V, and we sometimes run the load with the 12V input supply connected to the board, regardless of battery charge state.
if anyone can give me some pointers as to what could be causing the failures than it would be much appreciated.
Looks like this is a loop filter issue. Using just a series C and R between the CC and ITH pins on the 4000 charger and removing the 3-component filter between the ITH pin and GND on the 3789 switcher…
Were you able to resolve this? I'm having a similar issue.
Hi Jeff, apologies for the slow reply. I haven't logged in for a few months but have been looking at the problem again recently. What I have found is that when the charge termination timer of our LTC4000 expires, at the moment the PMOS between the charger and the battery is switched off, the LTC3789 output spikes briefly up to 20V. The spike rises almost instantaneously then decays back to our target Vout of 15V within approx 40ms. We had the 5V6 zener in series back to the EXTVCC pin, so I'm now suspecting that this pin was seeing at least 15V, albeit briefly, which might be the cause of the failures since this pin has a max input of 14V. We now no longer fit the zener or use the EXTVCC pin, and haven't had any more failures but it's early days.
Looks like this is a loop filter issue. Using just a series C and R between the CC and ITH pins on the 4000 charger and removing the 3-component filter between the ITH pin and GND on the 3789 switcher improves things. It improves start up and reduces the overshoot when the PMOS to the battery is disconnected, which can happen when the charge termination timer expires and on application of external DC if there is a fault condition (e.g. NTC). I'm now seeing an overshoot of about 2V above target Vout. Previously I saw 7V, which was bad news for EXTVCC even with the 5V6 zener. The data sheet for the 4000 charger discusses selection of Cc and Rc but not the additional 3 components of the switcher filter, so I assume that including both filters in this configuration is an error.