LTC2977

Hi  Tom,

Yes, the 2977 is able to monitor and control eight power supplies. The chip only monitors DC voltages.  Maybe you can provide more info on the AC/DC converter, or send a diagram.

The 'control' aspect is enabling and disabling the individual supplies. The supplies can be enabled/disabled either with a CONTROL pin, with a PMbus command, or it will just respond to Vin.  The enable signal is a logic level signal.  There's also a sequencer built into the 2977 that allows you to set the time when enables fires to each supply. 

The 'monitor' aspect takes a bit more explanation, especially when high voltage is involved. The 2977 has eight voltage sensing pins, and the max voltage is limited to 6V.  If you need to sense a 40V rail, you'll need to divide down 40V to something less than 6V to satisfy the 2977 ratings. This can be accomplished with a simple voltage divider, or with an op-amp that drives a single-ended voltage into the Vsense pin.  The advantage of the op-amp is its low impedance output that drives Vsense. The resistor approach is fine but you need to select the values carefully, so as to avoid errors introduced due to Vsense pin current being injected into the divider. The demo board DC2518 covers some of these issues.

When discussing HV power management, we will shortly introduce a 2-ch PSM that supports voltages as high as +/-60V.  The part number is LTC2971. It controls and monitors 60V supplies and also measures output current.  You would need four 2971 devices to monitor/measure 8 power supply rails.  It also measures/monitors the input supply's voltage and current.

Whether you use a single 2977 or multiple 2971, the supplies can be sequenced based on the timing settings that you set. A timing signal called SHARE_CLK is tied across all PSM devices and is used to synchronize the on/off delay times.

I hope this gives better insight into the workings of a PSM device.

Mike

Thanks MPETERS for the reply, I have some follow-up questions:

Since the control portion of the LTC2977 is use to control the ON/OFF function of the power supply device then most likely I must select to use a DC/DC converter instead of a AC/DC converter because most AC/DC converter / regulator does not have an ON/OFF pin, ...do you agree. (maybe I'm missing something)

My system power distribution consist of 8 power supply: +/- 5v, +/- 15v, +/-20v, +40v and another master +5v power supply, They are all AC/DC switching converter from Acopian. Since they do not have an ON/OFF input, I probably have to switch them all to DC/DC converter instead if I were to perform power Sequencing. 

In this case, my power topology would change to include a high power AC/DC then the output would drive my eight DC/DC.

Originally, I thought the LTC2297 controls the distribution of all eight power by connecting the power supply voltages to the LTC2297 device input pin then performing the ON/OFF function internally and then sending the voltage out of the  LTC2297 device pin then to the "User Load". This is NOT the case of the ON/OFF function of the LTC2977,... correct ?

As far as monitoring a Negative power supply voltages, does the LTC2297 support negative voltage, example -5v, -15,....

Lastly, Can the LTC2977 monitor input and output current? If it can not then are there any current LTC / ADI device that can perform this?

I appreciate your prompt attention to this matter.

Regard.

Tom

Hi Tom, I am not familiar with Acopian power supplies but I looked a few of their products. It looks like a few have an INHIBIT input that could be used with the VOUT_EN of the 2977.  Not sure about the polarity of INHIBIT. 

The bigger issue, and you point it out, is that you need to decide on the power topology. It would seem that one AC/DC converter would be a better choice and select your 7 or 8 switching regulators to provide DC power. Maybe buy an AC/DC converter that drives +48V which would be your input supply for the 7 switchers.

The 2977 can measure negative supply voltages.  We show a simple way to sense negative voltage in Fig 29 on pg 85.

Finally, the 2977 is a power manager, does not provide power itself.  If you need to measure input supply current and output load current, I suggest you look at the 2975 (4ch) or 2972 (2ch).

Regards

Mike

Hi MPETER,

The LTC2977 has a separate VIN and VOUT: sensing, monitoring nomenclature.  Function wise, shouldn't they be the same. Why are they designated separately. sensing should be the same regardless if it come from VIN or VOUT of the DC/DC converter unless the function if different, and if they are different then can you explain,

Regard,

Tom

On the VIN side, you power the chip typically with the VPWR pin and use VIN_SNS to sense the input power supply voltage.  The VIN_ON and VIN_OFF are user defined thresholds that the chip reacts to start many of the management functions, such as sequencing and fault detection.  The VIN_SNS pin is one of the inputs to the mux'd ADC.  Both VPWR and VIN_SNS pins are allowed to go as high as 15V.

For the Vout side, the 2977 manages eight power supplies.  Each supply is enabled, monitored, supervised, and servo'd.  The VSENSE pins are allowed to directly touch output voltages up to 6V.  If higher voltages are present, you may use a voltage divider to keep the pin below 6V.  The mux'd ADC does a round robin measurement of these eight VSENSE pins.  Each pin also has a high speed supervisor that uses a comparator that trips if the OV or UV output voltages are exceeded.

Typically you would tie VIN_SNS to the system's input power supply which powers the eight 'output' supplies.  The VIN_SNS pin does not have a supervisor to quickly fault off the eight channels. Instead it uses the ADC result to determine if there's an OV on Vin.  It is common for Vin to be +12V, and this supply powers a number of buck regulators that may be at various 'low' voltages, such as 3.3V, 2.5V, 1.8V, etc.

I hope this gives you more context.

Mike

Thanks for reply, MPETER.

After diving into the LTC2977 datasheet, there are two things that is not clear.

1) On the Topic of trimming and servoing, it mentioned that the ADC and DAC function within the device can provides a close loop of trimming the output voltage of the DC/DC converter. This allows the feedback pin on the Converter to adjust the voltage. Does this represent the Converter that is used, must have this control input pin call Feedback signal. Most converter I use do not have this pin. Can you recommend a DC/DC Converter Part# that has this feature. To use this feature requires a device with this control pin?

2) secondly, the data sheet mentioned about Sequencing function and using the "TON_RISE" to control the power on rise time of the output voltage. In this case what kind of DC/DC device signal control pin is needed. Does the LTC2977 actually control the rise time of the DC/DC Converter output? Also, can you provide a Part number of a device that has this control pin.   It also talks about "Soft Connect" during this control function, what is this referring to and how does this affect the Converter. Do you have a schematic connection between the LTC2977 and the DC/DC.

Regard,

Tom

Trim/servo is the fine tuning of an output voltage. The 2977 uses the ADC to accurately measure Vout, and compares the result with the VOUT_COMMAND value that you set.  This is what I would call the 'target' voltage.  The 2977's servo/DAC then is instructed to move up/down.  The DAC pin is tied to a resistor that adds/subtracts current into the feedback node of the regulator, causing Vout to move to the target voltage.  Our PSM devices all have servos that use a DAC to connect to a feedback node or other control node.  The DAC has two important configuration bits, a high/low range bit and a polarity bit. With these two settings, the DAC can be configured/wired to the vast majority of switching regulators and LDOs.

The TON_RISE config register on the LTC297x PSM managers does not control the ramp. Our PSM controllers (LTC388x) control the ramp with the TON_RISE register.  It's a bit confusing but the 297x devices borrowed this PMbus command to use it for the soft-connect feature.  The 297x DAC is high impedance during the powerup ramp ... that is, the regulator defines its own ramp, typically with a soft start cap (SS pin).  The soft-connect feature uses TON_RISE to time when the DAC attempts to connect to the feedback node. The DAC's internal voltage will be very close to the feedback node voltage prior to connecting, so that when it actually connects, the Vout voltage will not be disturbed, no glitch.

For circuit examples, see pages 80 and 81 of the 2977 datasheet. We have other examples in the DC2518 demo manual.  The important aspect of servo is that you have a KCL at play. There are 3 current branches that must satisfy the KCL equation.  The bottom fb resistor has a constant current due to the fb node being at a constant voltage. The top fb resistor current changes if there's any DAC resistor current.

Mike

Trim/servo is the fine tuning of an output voltage. The 2977 uses the ADC to accurately measure Vout, and compares the result with the VOUT_COMMAND value that you set.  This is what I would call the 'target' voltage.  The 2977's servo/DAC then is instructed to move up/down.  The DAC pin is tied to a resistor that adds/subtracts current into the feedback node of the regulator, causing Vout to move to the target voltage.  Our PSM devices all have servos that use a DAC to connect to a feedback node or other control node.  The DAC has two important configuration bits, a high/low range bit and a polarity bit. With these two settings, the DAC can be configured/wired to the vast majority of switching regulators and LDOs.

The TON_RISE config register on the LTC297x PSM managers does not control the ramp. Our PSM controllers (LTC388x) control the ramp with the TON_RISE register.  It's a bit confusing but the 297x devices borrowed this PMbus command to use it for the soft-connect feature.  The 297x DAC is high impedance during the powerup ramp ... that is, the regulator defines its own ramp, typically with a soft start cap (SS pin).  The soft-connect feature uses TON_RISE to time when the DAC attempts to connect to the feedback node. The DAC's internal voltage will be very close to the feedback node voltage prior to connecting, so that when it actually connects, the Vout voltage will not be disturbed, no glitch.

For circuit examples, see pages 80 and 81 of the 2977 datasheet. We have other examples in the DC2518 demo manual.  The important aspect of servo is that you have a KCL at play. There are 3 current branches that must satisfy the KCL equation.  The bottom fb resistor has a constant current due to the fb node being at a constant voltage. The top fb resistor current changes if there's any DAC resistor current.

Mike

Hi MPETER.

Appreciate your prompt response as always.  I have some follow up question

1)  LTC2977, datasheet, figure 23 and 24 describe the trimming function.. Both figures are used to perform the same function; Figure 23 is the setup when the DC/DC PS has a Feedback pin and Figure 24 is used when the DC/DC PS has the Trim pin. Therefore depending on the pinout of the DC/DC PS, the availability  of either a TRIM or FeedBack pin, then select the appropriate figure 23 or 24 to use this TRIM/MARGIN function. Would this be correct?

2)  Secondly, the way you describe the TON_Rise register, this is the delay time after power-on to initiate the sequence of the PS. Therefore a simple timing sequence could be as follow:  1)  power-on, 2) when TON_Delay expire, the Enable is set, 3) when TON_Rise expire, final trimming start?  I believe this is the case, unless I totally got this wrong.

Therefore based on how long the soft-start cap pin from the DC/DC PS is, the TON_RISE value should be at least longer than this soft-start time?

3)  Lastly, I believe the LTC2995 current monitoring feature is easier to implement than using the LTC2977 (odd channel only)? it this correct? LTC2975 didn't really cover on how to select the correct resistor value for the sensing resistor to measure/monitor current, possible I could have miss it, can you elaborate on this?

Many Thanks,

Tom

1) Figure 23 represents the more common fb node that you see on most switching regulators and LDOs. The TRIM pin shown in Fig 24 is not as common but can be found on aftermarket modules.  What regulator or power module do you have in mind?

2) TON_RISE time is measured from VOUT_EN high to the start of the soft-connect operation. This applies to each channel since they each have an enable.

When power is applied, there is a tINIT time that you need to account for. It is essentially the boot time of the chip, the time it takes to load the RAM from NVM and reset all logic. When tINIT expires, this is considered time=zero, or the start of all TON_DELAYs.  Each channel will be enabled relative to time 0.  Upon the channel's enable pin going high, the TON_RISE timer begins.  When the timer expires, soft-connect starts and the DAC begins to connect.  Consider this the start of the servo function.  Once the target voltage is reached, the servo goes into a mode that updates the DAC every ADC round trip (~90ms).

3) The LTC2975 has dedicated current sense functionality. It has hardware pins to support current measurement and it has PMbus registers that are intended to do the math and report a measured current value called READ_IOUT (in units of Amps).  By 'do the math', I mean that the chip senses a voltage and does a calculation to derive a current. It has registers the value of the Rsense or inductor DCR, which are programmed by the user.  This register is called IOUT_CAL_GAIN. The 'math' also includes sophisticated parameters that allows for temperature compensation and thermal time constants.  

If you use a 2977 (odd channel) for current measurement, it can be done but you need the host processor to do the math.

Hope this helps.

Mike

Hi MPETER,

I'll plan to use three LTC2975 to managed 9 DC/DC. I have one AC/DC driving 9 DC/DC. 

1) At this point, I would like to understand how to program the device. I believed one way is to use the DC1613A to program the device using the LTpower Play software. This might be the simplest way since the GUI  doesn't require any code programming. I assumed all the PMBus command is driven by the GUI and there is no need to learn any programming language. Once all the LTpower Play setting have be entered, I could upload this into my LTC2795 device into the EEPROM?

2) When power is re-apply, the LTC2975 will be configured and ill run autonomously?

3) In this case the DC1613A  will no longer be in the loop, at this point, how does the user communicate with the LTC2975? What is the approach the user has in term of interfacing it from the Window PC application. Is there some software driver that is available that allow the user to control and monitor the LTC2795 from the Linus or Window computer. This is where I'm not sure what the approach should be? 

4) I'm not a software engineer and need to understand how to communicate with the device from within a software appllication running from a PC

One idea is to connect the computer via USB to a FTDI USB controller and connect this to a FPGA. The FPGA would perform the USB to PMBus- I2C  function. then connect to the LTC2975. What practical alternative way do you recommend.

5) Lastly, is there a LTC2975 Demo board that is available for evaluation with the DC1613 and LTpower Play 

Thanks you,

Tom

Hi Tom, 

Programming the 2975 using LTpowerPlay is probably your easiest path. It's is a free download and runs on Windows. LTpp has a built-in Programming Utility that takes the configuration data you have saved and writes it to the chip's EEPROM. After power cycling, the chip automatically loads its RAM from the EEPROM and is ready to run autonomously.

To communicate with the 2975 without the DC1613 dongle, you could run telemetry or write commands to the chip via Linux or Linduino.  Or even run LTpowerplay from a Windows shell. We have an online FAQ that covers these options.  Also we recently completed a site on Github for our digital PSM. 

Github(DPSM)

Linduino

http://ltpowerplay.com/help/PSM-FAQ.pdf

We have a few demo boards that showcase the LTC2975.  The DC2022, DC2518, DC2428 are your options.  The DC2022 is a very small board that has low power switchers that have supplies from 1.5V to 3.3V.  The DC2518 is a board that has +/-12V and +/-48V supplies. The DC2428 is a two-board set that plugs together, one is the 2975 PSM board and the other is a LTM4644 uModule 4-ch power board.

Mike

Hi MPETER,

I scan through the GitHub and Linduino site and they contain many useful software tool support for programming the LTC2975 for our software engineer. They can use these information to design the programming application require to communicate with the PMBus device which is great.

1) But I need to provide the proper hardware interface to the computer in order to develop the software application. What is a practical hardware solution to interface with the PMBus on the LTC2975? One simple approach is to locate a USB to PMBus device but not sure if this device exist. 

A USB to I2C device might be easier to locate, therefore can the LTC2975 communicating with a I2C connection instead of a PMBus connection? I know they are similar and one is derive from the other but I want to make sure that this will work.

Any recommendation for a simple and basic "Hardware interface" between the Computer to the LTC2975 PMBus?

Thanks,

Tom

Hi Tom, we use an FTDI chip in the DC1613 USB dongle. The schematic for the DC1613 is on the website. There is a Windows driver for the FT245BL chip.

If you have an FPGA, why not run commands directly between the FPGA and 2975?  Doesn't the FPGA have tools to run USB/computer to FPGA?

Mike

Hi MPETER,

I believed FTDI chip in the DC1613 design is only the USB front end with a FIFO interface bus. The rest of the circuit within the DC1613 will be required to implement the I2C side. Thsi seen rather involved.

I could use the FPGA with a USB controller IP to interface with the Computer, but then the translation of USB stream into I2C is required within the FPGA. before connecting to the LTC2795. A reverse engineering is probably required to understand the USB byte format before translating it into PMBus stream.

Not knowing exactly what the translation from C programming content to USB byte format look like before-hand and also what the LTC2975 expect to see on the PMBus for every PMBus Command will make the FPGA  translation function very error prone. 

Since you mentioned about the FTDI device, I found a FTDI chip called FT201x. This is a USB I2C slave IC. This should fit perfectly between the computer USB and the LTC2975. Your opinion on this.

Thanks,

Tom