how to shut down PMBus function if i connect to I2C BUS
Hello,
Which version of the MPM3695 are you using? Depending on the one that you may using, the page numbers may vary on the subject.
In any case, all versions have an ADDR pin such that the PMBus address may be configured to a specific hexadecimal address. If you tie this pin to GND, the device will not respond to any PMBus commands, effectively disabling PMBus functionality.
Note that the module will use default standard settings after disabling PMBus, so refer to the datasheet of the version of this part you have to see current thresholds and how to manually set your output voltage.
Hopefully this answered your question.
Best,
Krishan
Hi， MPM3695-10 is which i used. if I just tie the addr to gnd, will it have effects on other I2C device because the all tied on one I2C bus.
the problem is that at begain there are four device on the I2C bus,and there is no problem when we R/W the I2C, after i changed the SCH design, we changed two power module to MPM3695-10 and add two I2C on the same bus, then when we read the reg(not on 3695),it always timeout.
when we use the i2c bus,we didn’t R/W the MPM3695-10,but it always timeout or i2c died. when timeout, the scl is hold low,and you need send another command ,then it will be normal. if the bus died(master die,master is FPGA), after scl hold low a few second, it pulled high,and it does’t have any response.
Setting the ADDR pin to GND shouldn’t affect the rest of your I2C rail or any of the other associated devices. You will just have to be sure that the MPM3695-10 that you have is disconnected in terms of all I2C connections. But from the problem that you are talking about, this does not seem to solve the root cause of your issue.
As for your timeout issue that you are dealing with, we will work on figuring this out and get back to you on this forum post with questions and a possible diagnosis.
Thank you,
Krishan
update this problem after what i have done. one 3695-25,two 3695-10 on the board.and tie thier’s PMBus on same I2C bus.
1.i tied the addr pin to GND, also the problem existent。
2.i delete the two MPM3695-10, it’s OK.
3.i delete one MPM3695-10, the problem also exsitent.
4.cut the PMBus trace at pin out, it’s OK
Hello,
What are you using to communicate with the parts? I recommend using the I2C EVKT with the GUI because I have used it to connect to the mpm3695-10 before without issues. Is there any way you can capture a waveform of the problem you are seeing?
best,
-Kerr
I used our own board I2C on FPGA.
i have uploaded the waveform below，first is timeout and 2nd is normal
Have you gotten a 3695-10 to work on its own? I am curious if you have already seen it work without any other devices in parallel so that we can narrow it down to an issue with that part or with sharing the line.
-Kerr
no,but i have test it work without 3695-10,it works well. we have test four boards, they are all the same
Looking more closely at your waveforms it looks like the “abnormal” screenshot shows the chip sending a NACK command to your master. The 9th clock pulse is an opportunity for the chip to pull the SDA low to ACK to let the master know that it has successfully received data. You can see the transition from when the chip starts pulling low in the “normal” waveform. This NACK can happen for multitude reasons, including sending to the wrong address, improper signal timing, sending a command that the chip does not recognize, among other reasons. Was there any difference in the circuit between the two screen captures? Knowing that could help me narrow down what could be causing the issue.
I also noticed that there is a quite long pause between the start condition and the start of sending data, this delay is not specified by the protocol, but each chip may have a different amount of time it waits before timing out. These chips also support fast mode, see if you experience the same issue at 400kHz. Also, focus on getting the 3695-10 to work on it’s own before trying it in parallel with other chips.
I recommend also double checking the PMBus interface section starting on page 28 of the datasheet to make sure that you are sending to the right address as specified on table 2.
Are you seeing the chip power on when given power and enable or is it completely unresponsive?
best,
-Kerr
Hi
there is no difference between the two test screens. we have not tested the 3695-10 alone,because we didn’t operate the 3695-10. the fault is tested when we send command to other device.
i cut the i2c bus at pin fan out and disable the chip with out power,it works well
To clarify, if you send commands to the 3695-25 when the 3695-10 is connected to the I2C bus AND powered on then it will fail. If the 3695-10 is either not powered on or disconnected from the i2c bus then the command will successfully make it to the 3695-25. Is this correct?
The interesting thing is that since the ACK signal can be pulled down by any device on the bus we know that the 3695-10 being present is causing the 3695-25 to not pull down the SDA for an ACK. Check the ALT# and PG lines to see if there are any changes there when the 3695-10 is added. Are there any other connections between the 3695-10 and 3695-25? Are their outputs in parallel or independent?
Sometimes adding more devices can increase the capacitance beyond the limits of the I2C. Your waveform looks clean, but where is this being taken from?
I still recommend trying to talk to the 3695-10 on it’s own because eventually you will need communication to work to that part too, otherwise there is no point in connecting it to the I2C line in the first place. Getting the 3695-10 to work on its own may help you find issues related to the problem you are currently seeing.
Best,
-Kerr
Hi，Kerr
thanks for your reply. yes if i either not powered on or dis connected form the i2c bus,it will works well.but not when we send command to 3695-25, there are another two devices on the i2c bus and it fails when talks to the another devices. it is true we just leave the 3695-25 on the i2c bus, it can works well,but even one 3695-10 is on, it will fail.
because our product need long time test, so we have cut the the route already , so i cann’t verify the 3695-10 alone on the same board.
Are there any connections between the 3695-10 and the 3695-25 besides the I2C lines?
Best,
-Kerr
no, they all work alone