Dear Analog/LT team, I'm in dire need of a little help.
I've read the data sheet of the LTM4653 over and over again and I am still not sure, why this is happening exactly:
Under some strange conditions the LTM just bursts in flames. My input voltage is 53V and there is no load (besides some caps) in the moment of ignition. I tried to reproduce the error with the dev board and it worked (as you can see in the picture). I modified the dev board to match my own schematic, so EXTV is supplied by a separate power source and the enable pin is controlled externally as well:
So when the external 5V are missing (I think this was the cause) and the RUN has low level the LTM dies in an spectacular explosion which is kind of impractical for my use case. The data sheet states that the LTM will derive its internal voltage if there is no external supplied and switches over automatically.
"EXTVCC (F3): External Bias, Auxiliary Input to the INTVCC Regulator. When EXTVCC exceeds 3.2V and SVIN exceeds 5V, the INTVCC LDO derives power from EXTVCC bias instead of the SVIN path."
But there seems to be either something wrong or I'm doing something not exactly right (which is more likely in my opinion). I've attached some more photos of the broken µModule and the PCB underneath if that helps someone determine the cause. It's hard to measure things because every time the LTM dies it costs me 15€.... So I would really appreciate it if an applications engineer could help me finding the error...
Ok, got a reply from the customer support:
The issue here is the generation of INTVcc.INTVcc is important, because it supplies a number of internal circuits.1. If RUN>1.2V , then INTVcc LDO is ON…
Hello- It would help to get back the damaged modules and demo boards. We will replace them at no charge. Is there a particular action that triggers the failure? What is the sequence between EXTVCC and RUN? Is the failure repeatable? I would add 20 Ohms in series with the 5V EXTVCC source, with the 1uF directly on the EXTVCC pin. Ship returns to:
ADI attn- David Burgoon
1530 Buckeye Dr
Milpitas, CA 95035
Do you need the modules for further research? Otherwise I would throw them away because shipping and handling to the USA is nearly as expensive as the modules itself and I doubt they will pass customs without trouble :D
The issue here is the generation of INTVcc.INTVcc is important, because it supplies a number of internal circuits.1. If RUN>1.2V , then INTVcc LDO is ON and converts SVin to INTVcc.2. If RUN<1.2V, then INTVcc LDO is OFF (INTVcc is unregulated).
Therefore if the RUN-pin is low (internal LDO turned off) and EXTVcc has no voltage (or ramps up too slow) then the internal circuitry is not properly biased. If the input voltage is high enough you have a nice little camp fire on your desk ;)
Hi i'm also currently in the process of designing in multiple LTM4653's (6 pieces) at once in a control-board of our company where we need to have a configurabe DC output voltage from 48VDC in.
I came across this post literally by accident while doing some google searching.
As the LTM4653s in my application are operating at a relatively high input voltage of 48V (which can be 52V worst case) i also thought about using the EXTVCC pin of the µModule to save nearly 1W of power dissipation per module.
But if I do so a really strong requirement is, that there is no synchronisation between RUN-pin state and EXTVCC-state necessary. Reading the posts above it seems thats not the case.
Here are my requirements in detail. It would be perfekt, if an applications engineer could have a look on it and tell me if I could run into some problems with what I'm trying to do.
The 3,3V EXTVCC voltage will be present the whole time and not switched off at any time. Altough there is no power sequencing intended. So at startup the LTM4653 should survive a condition where the VIN is already present, but the EXTVCC is not and the RUN-Pin is held low. If this is a problem would it be a solution to tie the RUN-pin to Vin with a high value resistor and pull it down with an open-drain output? I understand that there would possibly a series diode needed in the EXTVCC supply path, so that when the EXTVCC-rail is still not up, the internal LDO does not get overloaded trying to backsupply the EXTVCC-rail.
Your help would be highly appreciated,
Dilbert from his cubical!
In my experience it depends on the output voltage. I never had any issues with a 54V to 12V converter. Tough with higher output voltages (54V to 48V in my case) don't **ever** let the run pin be pulled down without EXTVCC applied. I realized this by gating my enable signal with the PGD signal of the EXTVCC regulator. Since then I didn't have any further issues with the modules.