MPS Four-Channel Output Power Module EVM - Review

Table of contents

RoadTest: MPS Four-Channel Output Power Module EVM

Author: saadtiwana_int

Creation date:

Evaluation Type: Development Boards & Tools

Did you receive all parts the manufacturer stated would be included in the package?: True

What other parts do you consider comparable to this product?: Similar parts listed in Review, below.

What were the biggest problems encountered?: Biggest issues was installation of USB-I2C driver until I found out (later) that it was part of "Virtual Bench Pro". Small quirks in GUI. Relevant info pages on website not always linked.

Detailed Review:

Attend the Webinar: Techniques and Technologies for Better Power Management Design (Win a Power Module!)

How I got here:

I have a confession: I have been using single channel power converters in almost all of my designs for the past several years. Personally, the power section of the board is not what i enjoy designing most, nor do I consider myself an expert at it. Mostly, I just need the power section to work (well!) so i can focus more on the "higher level" functionality of the project. So, I will admit to cheating from reference designs or existing similar designs whenever possible to save time, and manage risk. However, SoC/FPGAs have been getting more and more complex, requiring multiple power rails and very specific power sequencing requirements, which has not made things any easier. Designs using discrete DC/DC converters takes significant portion of BOM count as well as board space. And there's decent design effort involved. For this reason, when I saw the MPM54304 boasting small size, multiple power rails and features like software-programmable power sequencing, it immediately caught my eyes. Is this fancy little component going to be my ticket to easier, better, smaller power stages? I was excited!


Shipped Package, Unboxing:

The package came decently-packed and contained everything as promised. The package came under DDP (Delivered-Duty-Paid) terms so I did not have to do anything in the process other than receiving the package at the gate. Since the package shipped from Element14, I'd say credit for this excellent execution goes to Randy et al image.


The package included the following parts:

- Evaluation board EVM54304-MN-01A

- USB-I2C device kit (EVKT-USBI2C-02) including:

   - USB to I2C communication device

   - USB cable

   - Ribbon and jumper cables


The kit did not come with a printed quick-start guide.


Comparable/Similar Parts:

As stated previously, I have been using discrete, single channel power converters in most of my designs, so after getting introduced to this shiny new component, I was curious to see what other similar(-ish) components are being offered by different manufacturers. Turns out, the right keyword to unlock good search results is "PMIC". The following are not apples-to-apples comparable components, but broadly are in the same multi-rail PMIC category:

- Microchip - MCP16501 (3 Buck regulators@1A, 1 LDO@0.3A, VIN: 2.7-5.5V)

- Microchip - MCP16502 (4 Buck regulators@1A, 2 LDO@0.3A, VIN: 2.7-5.5V)

- Maxim - MAX77714 (4 Buck regulators, 9 LDO, VIN: 2.6-5.5V, GPIO, RTC)

- TI - TPS650861 (3 DC/DC converters, 4 LDO, 3 load switches)

- TI - TPS6521815 (6 DC/DC converters, 1 LDO, 3 load switches)

- TI - TPS652170 (3 DC/DC converters, 4 LDO, battery charger, etc)

- TI - LP87524-Q1 (4 Buck converters, switches, smaller VIN range)


Note that this is NOT an exhaustive list. Also, the MPM54304 is NOT the only PMIC being offered by MPS; infact the MPS offerings are growing rapidly.


As can be noted, the different PMICs vary a lot in features (and prices). Depending on the SoC/FPGA/processor that you're using, and also depending on what all other power requirements you have in your design, some of them will be better choice than others. There's certainly no one-PMIC-to-rule-them-all! Also, note that different PMICs above have different requirements for external components; for example some of them require external inductors while others (such as MPM54304) have those built-in. Same goes for other external components (resistors, capacitors, etc).



The MPM54304 IC:

Let's first talk about the MPM54304, the power module featured on the EVM54304 Evaluation board. The official page has plenty of information on the device, however, let me start by sharing an overview video which gives a fair idea in under 2 minutes:


For more detailed information, I highly encourage referring to the datasheet which is packed with 45 fun-filled pages of detailed information covering everything one expects of a good datasheet.

Here, I will just share a snip of the datasheet's first page, which gives a good summary of main features:





The Evaluation Board

The main page for the Evaluation board is linked from the MP54304 page directly

The Evaluation board comes with it's own 12 page datasheet which has useful information including schematic, BOM, some test results and PCB layout.


One thing I want to point out is that the EVM pictures in most places show the top side of the board, with the MPM54304 and 5 capacitors next to it. However, just be aware that there are components on the back side also, which can add to the required PCB space depending on your particular configuration. The official main page DOES show both sides, though, if you scroll through the pictures. Below are two pictures next to a millimeter scale for both top and bottom sides, for scale.


The evaluation board does not come with on-board load and USB/UART/similar interface as is the case for some other evaluation kits, but I think this is not a bad thing since the EVK can be cheap and external interface device can be reused across different EVKs. More about this later.


The USB-I2C Communication Interface

The USB-I2C communication interface is based on a Silicon Labs CP210x chip. According to the main page, it is designed to work with MPS I2C and PMBus products, Virtual Bench Pro, and I2C GUI tools. Basically it allows easy and quick evaluation of the relevant MPS products using the software tools provided by MPS.

One issue I had was that despite an updated Windows-10 (Pro), it did not get installed by itself. The main page gives a link to Silicon Labs download page, however that requires you to first create an account, and secondly there is no straight-forward option to download just the driver.


Moreover, the first option is a 194MB download and installs the full SDK. I found that the second option "USBXpress Dev Kit" was a smaller 7MB download. However, this still installed the "Silicon Labs USBXpress Development Kit" before proceeding to install what I really only needed, the "USBXpress Driver".  Anyway, once this was done, the device appeared correctly installed as a "USBXpress Device" in the Device Manager.


As per the quick start guide for the EVK-USBI2C-02, The next step was to install the "I2C GUI Software". A search on the MPS website seemed to imply that the I2C GUI software does not exist for this particular IC, and so i moved on to downloading the Virtual Bench Pro software: 

Thankfully, the MPS website did not need an account for downloading the software. Once I installed it, I noticed that in the end it DID install the Silicon Labs USB device driver. Had I installed the Virtual Bench Pro BEFORE plugging-in the USB-I2C device, it would have been a much smoother process. However, I had been merely trying to follow the sequence in the quick start guide. Perhaps something to improve upon in the quick start guide, or the EVK pages.


Getting the hardware connected together was straight forward. I ended up using the 3-wire jumper cable to connect the GND, and SCL, SDA signals. Although, I did notice that you *could* (with careful placement) also use the supplied Ribbon cable to connect the Evaluation board to the USB-I2C device, since the relative sequence of the 3 connections is the same on the board as well as the USB-I2C device.


Using the EVK with Virtual Bench Pro:

After starting Virtual Bench Pro, the "Auto Scan" option did not work and I had to resort to manual connection, for which and address had to be selected. The evaluation board datasheet did not mention the address, however, I was lucky to get the first option in the manual address selection dialog correct (0x68). A good connection is indicated by a green circle once you press OK.


If, for any reason, you have a device configured to a different address, there is also a very convenient I2C search function located in top-right of screen, that will scan and return any addresses that responded.


After this, evaluation of the EVK/MPM54304 features was very easy, as the GUI lets you conveniently change parameters all parameters on System level as well as for each individual Buck Converter. The interface is very detailed and nicely done.

I did come across a few quirks that can be improved:

1. The OTP only has two write cycles, and the button to write to OTP is right next to the button to write to Volatile memory (RAM). A user might accidently use up the OTP cycles. (I would put the OTP write button in RED!)


2. The GUI seems to monitor what values are changed since last write to RAM, and only send the values when GUI values are changed. Normally this is not an issue as long as the IC registers are in-sync with the GUI. However, if you power-cycle the IC (without reconnecting/reopening the GUI) and try sending the current settings in GUI to the IC again without change, then no changes are sent to the IC as the GUI thinks the IC already has the same settings. The GUI also shows no error in this case. The work-around I found was to change any 1 setting, send to RAM, and then change back the setting to the previous setting followed by another Send-to-RAM. It's a bit annoying, but at-least it doesn't force you to re-do ALL settings in the GUI after powercycle of EVK.


One thing i really liked is the info/hint popups if you leave your mouse on the "i" icon next to any button. Saves a lot of time and prevents having to go back to datasheet frequently.


Anyway, once all this was sorted out, I started playing around with the various settings on the MPM54304 for quite some time, before moving on to some lab testing. The GUI  is detailed and seems to map the internal registers very well and so using it was a pleasure.



For lab tests, I was lucky to get permission to use some nice equipment at my workplace-lab (for which I am very grateful!).

I used the following equipment in my tests:

- TTI CPX200 bench power supply

- Kikusui PLX334W Electronic Loads

- Fluke 52 II Thermometer

- Agilent DSO-X 3024A oscilloscope with 10:1 probes

- A concoction of cables and clips

- A multimeter


Unfortunately, the number Electronic loads and oscillocope probes available to me were limited to 2 each, so I decided to do the tests with Channel s1&2 in parallel, and Channels 3&4 in parallel. This reduced my outputs to 2 voltage rails, but at higher power sourcing capacities.


Here's a picture of my test setup:


Power up/down andSequencing Tests:

Most FPGA/SoC that I have worked with, and even many specialized ICs have specific power sequencing needs. Usually the designer has to play around with a combination of IC "enables", "power good" signals and RC circuits for delays, for power sequencing.

The MPM54304 handles all of that internally and configurable through I2C. I wanted to try this out myself.


Power-on sequence test (NO LOAD):
- Ch1&2 in parallel, set to 2.5V, No load on output (scope ch1 - green)

- Ch3&4 in parallel, set to 5V, No load on output (scope ch 2: yellow)

- Power up delay of 1ms between outputs.

(Everything else was left at EVK defaults)



Power down sequence test (same settings, NO LOAD):

The MPM54304 had Vout Discharge resistors (45Ohm) enabled, which i believe were resposible for the RC-shaped curves obtained.



In both tests, the sequencing worked as configured. Note that Delays are configurable from 0-4ms in 1ms increments. The soft start time (ramp-up slope) is also configurable.


Power-on test under heavy load:

I wanted to see how well the device powers up under heavy load. In the past, I have seen some buck converters going into perpetual short circuit protection or hiccup recovery when loaded at power up (still within the specs).

- Ch1&2 in parallel, set to 2.5V, constant-current 5A load on output (scope ch1 - green)

- Ch3&4 in parallel, set to 5V, constant-current 3A load on output (scope ch 2: yellow)

- Power up delay of 1ms between outputs.


Even under such heavy load, the power up happened almost same as unloaded. Quite impressive.


Lastly, here's a power down under same heavy load, and same settings. This time the outputs fell rapidly due to the external load, as expected. Power-down sequence timing is still spot-on.




Voltage Ripple Performance under Loaded conditions:

For this test, I wanted to see the ripple on the outputs under load, as well as the ripple introduced on the input power rail by the MPM54304.


From earlier tests, I knew that my particular test setup was drawing ~2.6-2.7A from the power supply at 12V input voltage.

To establish the ripple/noise baseline, I connected the Power supply output to electronic load set to draw 2.6A in constant-current mode and probed the loaded 12V output with oscilloscope. MPM54304 was NOT connected. I used AC coupling on the oscilloscope to observe these.


This gives us the basline noise/ripple on our 12V input as <20mV pk-pk


Next, I removed the Electronic load and connected the 12V power to MPM54304's input. MPM54304 was configured as following:

- Ch1&2 in parallel, set to 2.5V, NO LOAD on output

- Ch3&4 in parallel, set to 5V, NO LOAD on output

- Outputs Enabled(but no load)

- Switching frequency of 800khz (EVK default)


As can be seen, even without load the MPM54304 puts some noise (~60mV pk-pk) on the input rail.


Next, I loaded the outputs to near-max for the configuration:

- Ch1&2 in parallel, set to 2.5V, constant-current 5A load on output

- Ch3&4 in parallel, set to 5V, constant-current 3A load on output


Ripple on Input (12V) supply: (~210mV pk-pk)



Ripple on Channel 1&2 output: (~25mV pk-pk)



Ripple on Channel 3&4: (~50mV pk-pk)




- The MPM54304 does add noise to the input power supply. In my tests the magnitude varied between 20-210mV pk-pk depending on load. So when designing with MPM54304, depending on what else is on the 12V input rail, and loads on the outputs, some additional filtering might be needed on the input of MPM54304 to keep these ripples isolated.

- Same goes for the ripples on the outputs. Additional filtering may need to be added if design requires less noise/ripple than what was seen here (EVK has 2 x 22uF capacitors on each output).


Note that changing the switching frequency should also alter the noise figures.


Thermal Performance:

Since the MPM54304 packs such high power conversion/sourcing capabilities in such a small package (7x7mm) and with internal Inductors, I was very curious to see how quickly it heats up under heavy loads, close to it's advertised maximum.


For my setup, I used Kapton tape to secure a thermocouple on the top surface of the IC.



I made a very rookie mistake initially when I attached the electronic load just to the VOUT1 and VOUT4 terminals ("because they're connected in parallel internally as per my configuration"...i thought). But I noticed that I was getting Over-Temperature warning very quickly when running them with high load. I then realized my mistake: When you use channels in parallel mode, the internal inductor for each channel output is still handling it's own half of the total for the two channels combined. So you still need to connect them in parallel externally. On this realization, I added extra wires to connect ch1&2 in together and ch3&4 together. After this, things started worked as expected.


Two Electronic loads were used in constant current mode. One connected to the parallelized Ch1&2, and other to parallelized Ch3&4.



Thermal Test - Ch1&2:

- Input Power: 12V, 3A limit

- Ch1&2 in parallel, set to 2.5V, constant-current 6A load on output

- Ch3&4 in parallel, set to 5V, NO LOAD

- Ambient temperature: 23C, no fans in room


Result: Reached stable temperature of ~70C and did not indicate Over temperature warning (5mins running)



Thermal Test - Ch3&4:

- Input Power: 12V, 3A limit

- Ch1&2 in parallel, set to 2.5V, NO LOAD

- Ch3&4 in parallel, set to 5V, constant-current 4A load on output

- Ambient temperature: 23C, no fans in room


Result: Reached stable temperature of ~70C and did not indicate Over temperature warning (5mins running)



Thermal Test - All Channels:

Ok, for this test, initially i loaded Ch1&2 with 6A load, and Ch3&4 with 5A load at the same time. Despite operating within limits (as per my understanding), the device immediately started going into over-current protection mode (Current Limit was set to "5A Valley/7A Output" option on both sets of channels). Over-temperature warnings were not triggered, so it wasn't overheating. I couldn't figure it out the issue at the time (and it was getting late in the night) so I decided to just slightly lower the load currents for this test:


- Input Power: 12V, 3A limit

- Ch1&2 in parallel, set to 2.5V, constant-current 5A load on output

- Ch3&4 in parallel, set to 5V, constant-current 3A load on output

- Ambient temperature: 23C, no fans in room


Result: Power supply drawing 2.7A@12V (32.4W) Reached stable temperature of ~81C in 7mins running. Showed Over Temperature warning once surface temperature crossed 80C. From the values, I calculated the operating efficiency at ~85% under these conditions.



- Overall, I was really impressed with how much current this small IC was able to handle without any heat-sink on top, (and with internal inductors!).

- The IC did get quite hot when fully loaded close to it's limits. However, this is to be expected when a 7x7mm component is handling so much power. The thermal dissipation/design should be given special consideration when designing with the MPM54304.


Note that the MPM54304 has both Over-temperature warning reporting mechanisms as well as auto-thermal shutdown if temperature goes too high. In any case the component should not over-heat and burn itself out.


Availability of Design Resources:

The datasheet for the MPM54304 is detailed and quite well-written from what I could see.


My main interest in evaluation this PMIC was for possible use in future FPGA designs, especially for small form-factor "SoM" boards. To that end, I prefer when the manufacturer provides a "ready-made" design for popular FPGA/SOCs. In my case, I had Xilinx's Zynq-7000 in mind. On the main page for MPM54304 I did not find any link to reference designs. However, a quick google search for keywords "MPM54304 zynq" gave me the correct link in 3rd place. ( ).


As per the reference design, for Zynq-7000, one would need the MPM54304 with two additional ICs. When looking for the other two ICs (MP20073, MP20043) I ran into the same issue as the MPM54304: No stock anywhere on the popular distributors (as per Octopart).

Another thing I noticed is that although the Zynq reference design page boasts availability of Test Report, Layout, Schematics and BOM, the actual files there are for the EVM54304-MN-01A evaluation board, and NOT the reference design shown on the page.


Side Note: I have been thinking of testing the MPM54304 with an existing FPGA board by disconnecting the onboard power supplies and supplying power from the EVK using jumper wires. I did do some work towards this goal, however, in the end, I could not manage to do this as my board is still under use and I could not risk damaging the board in current circumstances. Perhaps a test for future.



Evaluation Board Pricing: I like the fact that MPS has made the Evaluation board available at a very reasonable price of 25$. This is very good because at this price this is affordable for anyone interested in evaluation, even for a hobbyist. In contrast, my usual gripe with most other Evaluation boards is that they cost several times more, which can be can be an issue, especially when you want to evaluate multiple ICs for a project. For me, a cheap board that has just the necessary external circuitry, and lets me quickly evaluate a board (albeit with some external test instruments) without building a PCB is a good evaluation board in many cases. So for that part, good job MPS!

Note that you still need to buy the USB-I2C separately (Another 25$) for evaluation, but if you'e evaluating multiple modules from MPS it's a one time expense.


PMIC pricing: The MPM54304 itself feels priced slightly on the higher side compared to using discrete ICs for the 4 rails. However, some of this difference gets covered when you consider the lesser number of external components, and savings related to maintaining a smaller BOM over a product's lifetime.


Component Availability:

At the time of this writing (Aug 2021), one of the biggest issues I see is lack of component availability. For MPM54304, octopart is showing zero stock with all major distributors. Restock times are also long. My assumption is that this is due to the current (2021) global chip shortage and not unique to this component. I am not sure if there's much that can be done about it at this point. I hope this situation improves soon. Personally, I find it hard to put a component into a new design if I don't see it available in decent numbers with the major distributors. So for my own designs, at the moment I will wait for availability situation to improve.



Summary of my thought/conclusions:

- Over-all a nice, feature-packed, solid PMIC which should cover many use cases, especially for compact SoC/FPGA boards.

- I am thoroughly impressed by the amount of power this small IC can handle, with much grace!

- The Evaluation Board is well-priced! USB-I2C module can be reused across EVKs from MPS.

- The price of the IC feels a bit higher than equivalent discrete ICs. However, the smaller external BOM, redued design effort and easier management of smaller BOMs in the long term (procurement, storage, obsolescence tracking, etc) also counts for something!

- MPS has made available decent amount of design resources available. Datasheet for MPM54304 is very detailed.

- Overall, the Virtual Bench Pro is a very nice piece of software for evaluation, despite a few quirks mentioned previously. Points for modern, clean design, generous info/hints and well-organized presentation of information.

- MPM54304 uses a 33 pin LGA package. In terms of one-off hand-soldering for testing on your custom board, i still consider it manageable (albeit with some reflow experience) compared to the BGA packages for some other PMICs that I came across.




Wish-list/Suggestions for Sponsor:

- Shipping the Evaluation Board with a quick-start guide would be nicer. Even if brief.

- Would be nicer if the USB-I2C device either installs by itself, or if "just" the drivers can be downloaded from the page directly without having to create accounts and installing SDK/Dev Kits. UPDATE (06/09/2021): Virtual Bench Pro seems to have these included. Perhaps the documentation can just explicitly mention installing the Virtual Bench Pro before plugging in the USB-I2C device.

- Reference designs should ideally be linked directly from the pages of component.

- EVK-USBI2C-02 mentions the Virtual Bench Pro and I2C gui tools, but does not provide links. Direct links would be nicer.

- EVK datasheet does not mention default I2C address for the device on evaluation board.

- Can the device availability be improved? Right now Octopart shows 0 stock everywhere for MPM54304.


Bottom Line:

The bottom line for me is that I would definitely consider using the MPM54304 for my next design, once I see the devices back in stock with the popular distributors. The size is small, performance is decent and features are plenty. Checks most of the boxes for my usual needs.



I am very grateful to the Sponsor (MPS) as well as Element14 for sending me this nice piece of kit to try out. I got introduced to something that I had not used before (muti-rail PMICs), and I learned a LOT while trying to read-up on MPM54304 and while road-testing the kit. Thanks a lot!