Group Actions
Engagement
  • Author Author: Jan Cumps
  • Date Created: Date Created
  • Views 4235 views
  • Likes 6 likes
  • Comments 25 comments
Related
Recommended

Low Voltage Step-Down Converter TPS54A20 - First Check

Jan Cumps

I received an evaluation board for the TPS54A20 DC/DC converter from TI.

This switcher is specific for low voltage designs. The output range is 0.5 - 2 V.

That's a very narrow range. In that range it can deliver 10 A, with a typical input of 12 V.

Efficiency is in the lower-to-mid 80%.

With a switch frequency of 2 MHz (more on that later) it allows for small passive components and a condensed footprint of the whole stepdown module.

 

image

The reason that I want to review this particular converter is because it's a two-phase type.

TI specifies it's in essence a 4 MHz design because of this. I'll validate if that's sales lingo or truth.

 

 

There's a new regulation in the US that stipulates that I have to disclose that I received this board from Texas Instruments. I don't like that. It's none of your business, and anyways, it's not a product review or endorsement. It's the comments and explanation of an interested person.

I don't like to be told by a gouvernment what legal clauses I should put in an amateur blog.

I don't put this disclosure here because I have the intention to comply.

Only to avoid that TI gets sanctioned.

 

Basic Specs

 

The full specs of the converter are available on the TPS54A20 landing page

The parameters are all focused on getting a small-footprint (both pcb real-estate and height) and reasonable efficient conversion from 12 V in to maximum 2 V out.

 

  • Output Voltage Range: 0.5 - 2 V
  • Output Current: 10 A
  • Typical switching frequency: 2 MHz (per phase)
  • Input Voltage Range: 8 - 14V
  • Max. efficiency is just above 90% in the IC datasheet. The evaluation board has a maximum efficiency of 84.7%

 

Because the passives required by the converter are really small, the converter and it's surrounding components can easily be mounted on the underside of a PCB.

On the evaluation board, the tallest components are the two coils. They are MLA-FY12NR22N-M3-RU. They have a height of 1.2 mm.

 

Two-Phase? Series Capacitor?

 

This is the reason why I'm interested in this converter. It's a particular design that splits into two converters that are phase shifted.

 

image

Image from the product landing page

 

The name TI gives to this design is "two-phase, synchronous series capacitor buck converter".

In this design you have one inductor per phase. You can find an explanation of the concept here.

The image below is taken from that link. Check the topology section of the datasheet for a per-phase breakdown of the circuit.

image

 

I can't explain it better than what Paul Pickering tells us in that link.

In essence, the capacitor Ct (the Series Capacitor"), because of where it's positioned in the circuit, has half VIN over it in steady state.

So one component deals with stepping down the input voltage by half, with theoretically no energy loss.

What remains abstracts to two out-of-phase buck converters.

Leave your better explanation of the design in the comments.

 

Evaluation Board Configuration

 

The board is configured for 1.2 V output and 2 MHz switching frequency.

Input voltage is between 9.2 and 14 V, although 9.4 V is required initially to prime the converter.

Below is the schematic of the switching core.

I've left out the output filters and the transient load circuit (this merits a separate blog).

 

image

 

This is virtually the same as the typical application from the datasheet.

The frequency is set by the resistor R1 connected to SS/F SEL. See this table to find out how to translate frequency, soft start time and hiccup (related to overcurrent recovery) time against resistor value.

On the evaluation moule, R1 is not populated, so these are the parameters applicable for that setup:

RSS/FSEL (kΩ)FOSC (MHz)FSW (MHz)Soft Start Time (µs)Hiccup Time (ms)
Open4251232.8

 

 

This is a light touch on the subject. In the next posts I'll go deeper into the topology and the design choices made on the EVM board.

And maybe the first measurements...

 

 

Related Blog
Low Voltage Step-Down Converter TPS54A20 - First Check
Low Voltage Step-Down Converter TPS54A20 - Series Capacitor
Parents

  • This is fun. Couldn't resist looking at the compensation on the feedback.

     

    Here's what it looks like in a simulator.

     

    image

     

    image

     

    The response [of the converter] is rolled off from a couple of 100kHz upwards. Might be interesting to hit the output with a heavy load at a few hundred kHz and see what happens (as that's the area where they think the compensation is becoming necessary but it's only just starting to have effect). With a power FET, load resistor and a waveform generator you could do a slowish sweep from 100kHz up to 2MHz and quickly see if there were any problem frequencies where the output was getting a bit skittish.

     

    Sorry if I'm hijacking your thread a bit. Tell me to go away and do blogs of my own, if you want.

  • in reply to jc2048

    ,  that's no hijack. That's all cool things. Thank you!

     

    I'll have to purchase a new chip though. Even though I tried to be as careful as possible, and used a magnifying glass while probing, I damaged it. Given the package's pin layout, with relative large ground areas underneath, it's not going to be simple to rework.

     

    in the mean time, someone just wrote an in-depth explanation of how the serial capacitor mechanism works on TI's E2E pages.

  • in reply to Jan Cumps

    If all is ok, the replacement should arrive tomorrow...

  • in reply to jc2048

    I've pulled the board out again and am prepping it for tests.

     

    I'll solder some short pieces of enamel wire on the SMD components that I want to measure, and connect scope or multimeter probes on those.

    I'm not planning to buy yet another board.

     

    Because I don't have a differential probe, I'll use channel A and B of my scope to measure both sides of the series capacitor and use my scope's math functions to get at the voltage across it.

    It's not ideal but it'll have to do.

     

    According to this article (same as I linked in a previous comment), the voltage over the cap swings around half of the input voltage

    (its nominal value is approximately the half of that input voltage - just a little above it in that article's measurement).

    Ignore the Step 4 mention in the caption ofn the drawing below. The trace is showing a few full switching cycles in steady state.

    image

    source: Step by step: How the series capacitor buck converter works

  • in reply to Jan Cumps

    I've added probe points for the series capacitor. You won't see me probing around this board directly  anymore image

     

    image

     

    I've checked if all is still ok on the board: no shortcuts created or connections broken. And that's fine.

    I'll now power up the board. If you don't hear from me in the next 24 hours, I'll probably be in a pub crying.

  • in reply to Jan Cumps

    The outcome is as expected.

    I've put my scope probes on probe points A and B - left and right side of the series capacitor.

     

    image

     

    My input signal is 10V.

    The capture below is:

     

    image

     

    Yellow: probe point A; Bounces between 5 and 10 V.

    Blue: probe point B: Bounces between 0 and 5V.

    Purple: math A-B: hoovers around Vin/2. Because my load is very low at the moment (*needs electronic load*), it is almost flat except for ripple.

    With a significant load, I should be able to capture an outspoken charge/discharge curve over the series capacitor that folds around Vin/2.

     

    edit: output current is 414mA

     

    More measurements, and collation of the cap signal versus several other interesting probe points are for an upcoming blog.

Comment
  • in reply to Jan Cumps

    The outcome is as expected.

    I've put my scope probes on probe points A and B - left and right side of the series capacitor.

     

    image

     

    My input signal is 10V.

    The capture below is:

     

    image

     

    Yellow: probe point A; Bounces between 5 and 10 V.

    Blue: probe point B: Bounces between 0 and 5V.

    Purple: math A-B: hoovers around Vin/2. Because my load is very low at the moment (*needs electronic load*), it is almost flat except for ripple.

    With a significant load, I should be able to capture an outspoken charge/discharge curve over the series capacitor that folds around Vin/2.

     

    edit: output current is 414mA

     

    More measurements, and collation of the cap signal versus several other interesting probe points are for an upcoming blog.

Children
No Data