The datasheet for the LT3015 negative regulator calls for the EP underneath the part to be connected to Vin. What happens if it is connected to ground? It seems to work but not at high temperature.
In this case I rely on my discussions on this topic with the design group:
As mentioned in the original post, the exposed pad of the LT3015 should be soldered to IN and not to GND. If the exposed pad of the LT3015 is inadvertently soldered to GND instead of IN, the device may work but there is a risk, device-to-device or over time, that an electrical connection through the device substrate and on the order of 50 ohms may develop between the LT3015 IN pins and the pcb GND. If that electrical connection occurs or develops, the LT3015 and possibly the remainder of the circuit will not function properly.
No operational issues will occur If the IN pins and the exposed pad remain isolated. In other words, the LT3015 will still function properly if the pcb does not make an electrical connection between the exposed pad and the IN pins. It is still important to solder the exposed pad to the pcb for proper thermal performance.
BACKGROUND: In manufacturing the exposed pad for the LT3015 is supposed to be electrically connected to the die and consequently also the LT3015 IN pins, but an inadvertent thin layer of oxide occurs on the die before the exposed pad gets glued down, so the exposed pad and the IN pins usually don’t ohm out as the same connection. Still, users should not rely on the thin layer of oxide as electrical isolation.
I am not sure why the circuit in the original post does not work at high temperatures. It seems like the exposed pad is properly soldered to the pcb for thermal purposes (but not for electrical purposes).
Why design the EP to connect to Vin if the pad's main purpose is heat removal? For most ICs of this type EP is ground which is conducted to the pcb housing through ground layers while the housing metalization is also ground. I'm baffled why it ever worked if it was strictly relying on an oxidation layer. I'm not sure this is a good part to use for that very reason.
The second layer of the pcb board shoud be Vin, not the GND for thermal purpose. You can use the standard demo board design as your reference.
The EPAD of the die should be connected to the lowest potential, which is “VIN” for LT3015. Connecting the EPAD to “GND” will definitely create issues with operation, since this will forward bias the diodes in the silicon. So for proper operation of LT3015, the EPAD must be connected to VIN.
I appreciate the feedback however tying the EP to VIN makes removing heat much more difficult which for my application is a major issue. We have another negative regulator, TI’s TPS73AA, following the LT3015 in our circuit which does tie its EP to ground. When we respin the board we’ll use it in place of the LT3015.
TSC, Phase IV Systems Operation
310 Voyager Way
Huntsville, Al 35806
I agree with Alustrisimo's Jan. 29 post that basically says the pcb should be designed so the exposed pad of the LT3015 connects to IN and not GND. The LT3015 datasheet says the same thing. Sometimes customers miss this point and layout their pcbs so the LT3015 exposed pad is connected to GND by mistake. If the customers find their mistake, they sometimes contact me and I say to them the same thing as in my Jan. 29 post above.
Thanks to SCA for the original post. I'm not happy he may use a competitor's part, but his post may help others avoid an incorrect pcb layout when using the LT3015.
Retrieving data ...