Z-One Digital IBA Forums

DPM doesn't not answer to its address

Steph — Thu Aug 21, 2008 5:57 pm

Author: Steph

Posted: Thu Aug 21, 2008 5:57 pm

Hello,
we've got 2 DPM mapped at address 0x50 and 0x52, and one RTC on the same I2C bus and with a different address (of course...).
by software construction we insure that when we communicate with the RTC, the I2C bus is free.
We experienced that sometimes, after X commands successfull, the both two DPM doesn't answer (Ack) to their address and are freezed in this state.

Any ideas to debug that pb ??

Thanks for advance,

Steph.

Test post from user "JJ"

JJ — Fri Aug 08, 2008 6:20 pm

Author: JJ

Posted: Fri Aug 08, 2008 6:20 pm

Hi there!

I2C Clock Stretching

Anonymous — Wed Jul 02, 2008 6:37 pm

Author: Anonymous

Posted: Wed Jul 02, 2008 6:37 pm

Hi Steph,

As the DPM is a real time controller, many of its timing parameters are specified in its data sheet, but unfortunately not the I2C clock stretching timing. The DPM firmware has to put the priority on the management of the devices it is controlling, while I2C communication is done as a lower priority task. So, the I2C clock stretching depends on the processor loading of the DPM, which depends on the task queue and I/O stimulus. There is not a firmware upgrade to make the I2C communication priority higher as this would surely impact the other performance parameters of the DPM.

Brad

ZM7300 DPM Voltages and Resistances

Anonymous — Wed Jun 18, 2008 6:28 pm

Author: Anonymous

Posted: Wed Jun 18, 2008 6:28 pm

The attached Excel table can be used to troubleshoot the DPM (for example, to find solder bridges, opens, pins damaged by electrical overstress.)

If a DPM issue is suspected, it is recommended to go through and measure the resistance to gnd and voltage on each pin and compare it to the attached table (or to a known good board).

POL IDENTIFICATION REGISTER (Address=0x28-47)

Anonymous — Wed Jun 18, 2008 4:58 pm

Author: Anonymous

Posted: Wed Jun 18, 2008 4:58 pm

Hi Steph,

The POL IDENTIFICATION registers are not automatically updated by the DPM at start up. The values that you are reading of all 0x08 (ZY7115) are typical default settings that you would see with a "blank" DPM.

Since the POLs all use the same controller IC which does not have constant memory or any way of self identification, the DPM has to be programmed with the correct information.

When using the GUI, it takes care of setting up the POL IDENTIFICATION registers (and all other registers). But in the case of manual programming over the I2C bus, all registers need to be manually set up.

Avoiding "Chicken and Egg" Situations

Anonymous — Thu Jun 12, 2008 12:05 am

Author: Anonymous

Posted: Thu Jun 12, 2008 12:05 am

Avoiding "Chicken and Egg" Situations

Connecting Z-One signals to other devices in the system (e.g. to an FPGA for supplemental monitoring and/or control purposes) can create a "chicken and egg" situation. If the device in question is not powered or not programmed, it can load down the Z-One signal making it impossible for the Z-One power system to start up, be programmed, and/or communicate.

� Avoid connecting the Z-One signals to devices that are powered by the Z-One system itself.
� Avoid connecting the Z-One signals to devices that are un-powered.
� Avoid connecting the Z-One signals to devices that are un-programmed.
� Avoid connecting Z-One signal pull-ups to Z-One controlled power converters.
� If there "needs" to be such a connection, provide zero-ohm jumpers so the Z-One signal can be isolated in case a problem arises.

AREF = 0V Issue

Anonymous — Fri May 09, 2008 4:49 pm

Author: Anonymous

Posted: Fri May 09, 2008 4:49 pm

AREF = 0V Issue

Issue:
In automated test, the AREF pin measures as 0V where as on a working board it is 2.56V. The test station considers this a failure and does not proceed to program the DPM.

Solution:
This is specific to parts configured for JTAG programming, i.e. ZM7304G-65505, ZM7308G-65506, ZM7316G-65507, ZMY316G-65508. The AREF voltage should be set as a �don�t care� to the test station because the voltage on the ZM7300�s AREF pin varies depending the DPM�s configuration and on whether it�s undergone JTAG programming.

The ZM7300 has a feature to allow a choice between either an internal or external reference. To avoid the possible condition of having 2 references fighting each other, the JTAG programmable DPM�s are initially set to use the external reference. This way, the AREF pin is an input and is a high impedance.

If the intention is to use the internal reference, simply program the DPM with the appropriate configuration file and the AREF pin will be connected to the internal 2.56V reference.

Keywords: ZM7304, ZM7308, ZM7316, ZM7332, ZM7304G, ZM7308G, ZM7316G, ZM7332G

DPM : Always answer DeviceCode+Read bit ???

Steph — Thu Apr 10, 2008 7:06 pm

Author: Steph

Posted: Thu Apr 10, 2008 7:06 pm

Hi,
It finally works...
Here are the debug session, for user that encounter same pb
1) Try to read DPM ID register at start up
2) Read FF FF ==> I thought the problem was my I2C interface from STR9 not support clock stretching
3) I implement a pause (in fact a breakpoint in debug mode) during the DPM manage the command, then I read the DeviceCode+Read bit
4) After reflexion, documentation and various trial, I finally
- Add a wait between the STR9 reset complete and the first command to DPM
- Let the STR9 manage the clock stretching : I insert the re-start function call as soon as last data byte is sent : the STR9 wait for the SCL line free to transmit the start signal.
==> And it works!!

In Conclusion :
Pb = Answer FF ==> Wait for DPM reset complete
Pb = Answer DeviceCode ==> Don't add any wait before ReStart.

Thanks for the support,

Steph.

�Check for updates� is not working

Anonymous — Mon Mar 24, 2008 4:25 pm

Author: Anonymous

Posted: Mon Mar 24, 2008 4:25 pm

�Check for updates� is not working

Question:
When I try to update the GUI using the �Check for updates� button, I get a timeout error. How can I fix this?

Answer:
Your firewall needs to have port 22 open for sftp (secure file transfer protocol). Have your IT department check it for you.

Is Repeated Start without Stop Required?

Anonymous — Tue Feb 19, 2008 6:20 pm

Author: Anonymous

Posted: Tue Feb 19, 2008 6:20 pm

Is Repeated Start without Stop Required?

Question (via email):
In Section 8 (I2C Interface) of the ZM7300 DPM Programming manual, a Repeated Start is specified to be sent between the I2C command and response. Is the Repeated Start required, or is it okay to send a Stop to the DPM before sending a Start and reading the result from the DPM?

Answer:
The Repeated Start without a Stop is required. Sending a Stop frees the i2c bus and therefore is not allowed on devices like the DPM that need to do Clock Stretching (clock extension). Clock Stretching allows the DPM to do the required processing before sending the response. See [1] and section 3.9 of [2] for more information.

[1] http://www.z-alliance.org/forums/viewtopic.php?t=62
[2] http://www.standardics.nxp.com/literature/books/i2c/pdf/i2c.bus.specification.pdf

Using POLs on Daughter Cards

Anonymous — Wed Jan 16, 2008 8:03 pm

Author: Anonymous

Posted: Wed Jan 16, 2008 8:03 pm

Using POLs on Daughter Cards

Question via email:

I want to have some POLs on a daughter card that does not have its own DPM.

How can I make the POLs get programmed when the daughter card is plugged in? And how can I turn them on?

Answer:

The DPM has a feature called �Group Reprogramming Trigger� that was designed for this purpose.

This is described in section 10.5.4.2 and Figure 19 of the DPM data sheet [1].

[1] http://www.power-one.com/resources/products/datasheet/zm7300.pdf

(Portions of [1] are copied below for reference and to support forum text searches.)

Quote:

10.5.4.2 Group Reprogramming Trigger
An interrupt that is programmed as a group reprogramming trigger always acts only on one group of POL converters. Interrupt 0 acts on Group A, Interrupt 1 acts on Group B and so on. The assignment is fixed and cannot be changed by the user.

When the interrupt is pulled low, the DPM will program the group of POL converters. Upon completion of the programming, the DPM will turn-on the POL converters, if the Auto Turn-On is enabled. When the interrupt input is released, the DPM will pull the appropriate OK line low. The POL converters in the group will execute regular turn-off ramping their output voltages down according to the turn-off delay and falling slew rate settings. In addition, the DPM will clear all bits in the POL Programming Status registers.

The group reprogramming trigger is mostly used to support hot swap of boards and daughter cards that do not have a DPM installed on them as shown in Figure 19.

<Figure 19>

In this configuration the Interrupt 0 (INT0_N) is configured as the group A reprogramming trigger. The DPM is installed on a mother board or a backplane. A daughter card with a group of POL converters is being inserted in the system during normal operation. At first, the long pins carrying power and the OK_A line signal make contact. Then the short pins carrying the SD and interrupt signals make contact. Once the interrupt senses low input voltage, it will command the DPM to program all POL converters in the group A. Upon completion of the programming, the DPM will turn-on the POL converters, if the Auto Turn-On is enabled.

When the daughter card is being removed, the interrupt input is released as soon as the short pins break the contact. The DPM will immediately pull the OK_A line low turning off all POL converters in the group A according to the turn-off delay and falling slew rate settings.

ZM7300 DPM Write Protection

Anonymous — Wed Jan 16, 2008 5:50 pm

Author: Anonymous

Posted: Wed Jan 16, 2008 5:50 pm

ZM7300 DPM Question via email:
Once the write protection is enabled, is the configuration of the DPM permanently locked?

Answer:

No. The write protection register is RAM. It defaults to �write protected� upon power up, and can be cleared by I2C command. There is also a hardware lock pin. For more information, see the DPM data sheet [1] (Some pertinent sections are copied below for reference.)

[1] http://www.power-one.com/resources/products/datasheet/zm7300.pdf

10.1 DPM Memory
... The LCK_N pin and the write protection register (WP) limit the write access to the memory blocks in the DPM and POL converters. The WP register content is defaulted to write protect upon powering up the DPM.

10.1.1 Write Protection
There are hardware-based and software-based memory write protections. The hardware protection takes precedence over the software protection.

10.1.1.1 Hardware Protection
The LCK_N pin enables the hardware memory write protection. If the pin is pulled low, the hardware lock is active and the memory blocks are then read-only. I2C write commands to the DPM return an error code (0x00). The write commands to the POL converters bypassing the DPM are also disabled. If the pin is left floating, the hardware lock is disabled and the software write protection is active.

10.1.1.2 Software Protection
The software write protection allows users to protect the various memory blocks from being overwritten through the I2C bus. At the power-up the WP register is defaulted to write protect. The software write protection can be disabled by checking appropriate boxes in the Write Protections window shown in Figure 5 or via the I2C bus by writing directly into the register. The write protections are automatically restored when the DPM�s input power is recycled.

Stability with Capacitors Removed

Anonymous — Tue Jan 15, 2008 6:59 pm

Author: Anonymous

Posted: Tue Jan 15, 2008 6:59 pm

Question by email:
We are seeing an over-voltage condition on one of our Z-POLs at turn-on. We found that it is happening in a condition when not all of the daughter cards are plugged in and therefore not all capacitors are present on the Z-POL�s output. Then we found that the problem goes away if we just put the capacitors on the Z-POL, so it seems to be a stability problem. How can we make the system stable with capacitors removed?

Answer:
The Z-POLs are auto-compensated for stability in the GUI at design-time based on the user specified output capacitors and ESR. It definitely can be a bad thing to remove capacitors after the POL was compensated based on assuming those capacitors would be there. This is because removing capacitors makes the loop gain higher and generally less stable.

If you need to have varying amounts of capacitance, it is better to design the loop compensation for the minimum capacitance and then add capacitors. Simulations should then be run in the GUI over the expected range of capacitors and ERS to verify stability.

Brad

Z-One POL Bandwidth

Anonymous — Tue Jan 15, 2008 6:05 pm

Author: Anonymous

Posted: Tue Jan 15, 2008 6:05 pm

Question by email:
I noticed the auto compensate adjusts for a bandwidth around 30 KHz. If there are no step loads to worry about, should the bandwidth be reduced or is this a don't care? I've adjusted all of the converter's bandwidth to 6 KHz. My concern is will noise be a problem for the converters?

Alan

Answer:
Hi Alan,

The normal bandwidth range for Z-One POLs is 10 to 40 KHz. We have done all of our testing and data sheet characterization with the bandwidth in that range. I do not see it as a noise problem.

I have seen that the POLs may not turn on properly with the bandwidth that low (6 KHz). They need to be able to track the reference as it ramps up at the programmed rate (0.1 to 8.3 V/ms) otherwise you can get an under-voltage fault. The turn-on can be checked in the Transient Simulation window.

Also, if the bandwidth is lower than the output filter's LC resonance frequency of about 10 KHz, the LC will be un-damped and you will see ringing at that frequency.

That said, as long as you are able to turn them on and like what you are seeing, you can go with whatever bandwidth works for you.

Brad

The GUI In�s and Out�s of DPM User Memory

Anonymous — Fri Jan 11, 2008 6:51 pm

Author: Anonymous

Posted: Fri Jan 11, 2008 6:51 pm

The GUI In�s and Out�s of DPM User Memory

The DPM has 1024 bytes of FLASH memory referred to as �User Memory�. The basic features are as follows:

� Can be programmed/read �on the fly� via the GUI or I2C commands
� Is saved into the configuration file
� Is programmed into the DPM when parts are bought against a configuration file

However, in the GUI, the User Memory is programmed and read separately from the rest of the DPM�s configuration memory. If this is not understood, it can lead to some confusing results. The following describes the GUI�s normal behavior regarding the User Memory:

System Configuration Window Buttons
� Read Config� � User Memory is NOT read from the DPM
� Program Config� � User Memory is NOT programmed into the DPM
� Edit User Mem� � Leads to the User Memory window

User Memory Window Buttons
� OK � Saves the current text into the GUI internal configuration and returns to the System Configuration window. The internal configuration can be saved to a p1c file in the System Configuration window.
� Load from DPM � Loads the user memory from the DPM into the user interface for viewing and editing.
� Save to DPM � Saves the current text into the DPM and returns to the System Configuration window.
� Cancel � Cancels the operation without saving the data and returns to the System Configuration window.
� Help � Opens the help sheet with more detailed information. (Recommended)