TPS7370 Regulator problems

Hello all
I'm having some trouble with a TPS7370 regulator. I'm using the DCQ packaging. It's being used to produce 3.3 volts for 4 different chips: a microcontroller, MAX232, M41T94 and a VCH 373 buffer. There are 2 more TPS7370's on the board which operate without any trouble. The problematic regulator works fine until it drops its voltage to 0.8. I've already replaced it twice. The chip is used on a 2 layered PCB. Sadly, there are no power planes on the PCB. To somewhat solve this problem I've used pretty wide tracks for GND and main power. Any help would be appreciated.
Thanks

Hi,

1) what´s the load current?

2) datasheet says there is no need for a imput capacitor as long as the input impedance is low enough. But it also says it is good practice to add the capacitor.
Is your source low impedance enough? Mind that you don´t have planes for GND and supply. I expect some R and L in the source path.

3) the schematic and the part values are not clear form your picure.

Klaus

Clarify following doubts,

1. What is the input voltage you are feeding to it?
2. Is particular IC's are getting into problem repeatedly? If Yes, How many regulators giving problem out of 6?
3. Is all the ground lines of 6 6 regulators and input ground is connected together?

Hi KlausST
Thanks for the reply. The load current is below 0.3 amps. Approximately 0.2 A to be exact. I'll put the schematic for my input power and faulty 3.3 regulator at the end. The regulator uses a 1uF decoupling capacitor between its output voltage and GND. As for the input impedance, the input power uses a 200uF decoupling capacitor for noise reduction. For a 3.3 volts, the regulator uses a 12K resistor for R2 and a 6.8k resistor for R1. I used a bread board to test the output voltages prior to assembling the main board.
As you mentioned, not having a plane for GND is a huge mistake. The board uses the UART protocol to communicate with the PC. During one of my tests, when the boards power supply was unplugged, the UART socket was still connected; I noticed that the regulators have static charge. The wall sockets of the PC and the board power supply were different, causing a very unstable GND signal. By using the the wall socket for the board and the PC the problem seems to have been alleviated. I'm not sure though. The regulator has been working since the change. What are your thoughts? Could this have been the culprit?

- - - Updated - - -

Hey HobbyCKTS
Thanks for the reply. The input voltage for the circuit is 5 volts. As for your second question, the regulator responsible for generating 3.3 volts has burned out twice. It suddenly dropped its voltage to 0.8 volts during operation. The GND tracks are all connected. I think that the problem had to do with static charge due to an unstable GND. The wall sockets used for the PC, connected to the board via UART, and the board were different. By using the same wall socket, things seem to be working, but I'm not sure. Any suggestions would be appreciated.
Thanks

Hi,

there is no heatsink. Dissipated power should be below 0.5W. It will be hot, but no everheat either.

***

The wiring is a riddle. The output is on pin2. i see the trace. But there is no connection to the feedback resistors.
The feedback path should be short, and not influenced by load current. So i expected itto be all on the top side with short traces.
It seems it is wired across the bottom side. There you could improve.

***

If I´m correct, then you use two wall warts at the same time. Maybe there is a Ground loop problem, because many wall warts hve relatively high capacitive coupling between mains and secondary side.
If this current flows through your GND traces on your PCB, then this could make problems.



Klaus

Hi KlausST
Could track width be an issue? The width of the wires was set to 9 mil's. Regarding wiring, were these issues also present in the schematic? Could you please elaborate on this subject?
As for the wall warts, I'm using only one 5 volt adapter for my board. The UART is from the PC. The wall plugs were different though. My guess is the common GND signal between the wall plugs is not very stable, caused by bad building wiring. Thanks for the suggestions and insights.
CasualStranger

Track width is fine I guess. Want to know about the connection from the board to PC. As you said UART is used how exactly the connection between these two has been carried out. Are you using MAX IC's?

Hi,

track width: 9 mils are OK for signals without significant current. (feedback path)
But not for supply rails with unknown current, even worse with pulsed current.

Wiring is a riddle for me because I don´t see PCB bottom side.

Wiring:
* use a GND plane.
* Bulk capacitor should be placed near regulator.
* First and second feedback resistor should be placed near FB pin of regualtor.

feedback: decide one star point for power distribution. Often I use the bulk capacitor for this.
* From this point route an independent wire to the first feedback resistor.
* Wire from first feedback resistor to second feedback resistor. Extra wire from second feedback resistor to FB pin
* wire from second feedback resistor to GND plane

Klaus

Hey Hobbyckts
Yep, I'm using a max232 for level conversion between PC and the board.

- - - Updated - - -

Hey Klaus
During routing, I devised a hierarchical approach for the width of the tracks. The main power lines from the input, both GND and VDD, were 40 mi. The secondary power lines, the regulator outputs, were 20 and all the other wires were set to 9 mil. I will definitely use a GND plane for my PCB revision. As for the actual PCB, the picture was of low quality. My bad. I've used a 1 uf, as in the schematic, for the output. I've used this point to route the output to the feed back resistors as displayed in the datasheet.
You mentioned:"Extra wire from second feedback resistor to FB pin". By extra do you mean a wider track from the second resistor to the FB pin?
Thanks

Hi,

You mentioned:"Extra wire from second feedback resistor to FB pin". By extra do you mean a wider track from the second resistor to the FB pin?

Click to expand...

No, just a single trace from resistor to FB pin. Carrying no other current.

Klaus