Power supply design

[elsalahy]

Summary:

Power supply design discussions and decisions.

edit by @azerimaker:

Step-down regulator choices:

• TPS62840DLCR

  • Iq -120 nA (quiescent current)
  • Input range 1.8 to 6.5V
  • Price 0.62$ from TI (1K+)
  • Features: PFM/PWM, RUN/STOP modes (high-level power saving modes)

• BD70522GUL

  • Iq -180 nA (quiescent current)
  • Input range 2.5 to 5.5V
  • Price 1.029$ from DigiKey (1K+)
  • Features: PWRGOOD output, 100% ON mode

[elsalahy]

@azerimaker can you please add relevant design documents and relevant design decisions here. We should also target asking specific questions here.

[azerimaker]

After many discussions and past mistakes with the GN, we decided to go with the ### 2x AA battery solution. By doing so, will provide us with a working input voltage range of 1.8-3.6V which aligns perfectly with the useful operating voltage of the STM32WL and other on-board devices. Would love to get @johanstokking's opinion on this choice.

[azerimaker]

If we go the 2x AA way, we can use a step-down converter or a low power linear power regulator (LDO), by operating at a lower voltage, say @2V we can attain a much more stable device, with longer battery life, wider voltage input range (1.8 - 5.5V). Having wider input range also enables us to use many other types of battery chemistries within the range. I've narrowed down our regulator choice to TPS62840 from TI, which is highly-integrated, extremely low-power buck converter with only 120 nA of quiescent current and up to 6.5V input range.

tps62840.pdf

[johanstokking]

I've got too little experience to say something more meaningful about this than what you describe above.

What speaks for 2x AA is the ability to easily replace the batteries anywhere. If there are other very commonly used batteries that provide better performance and are smaller, for example, then we need to consider that.

One of the mistakes made with GN was, I believe, that there were some tweaks needed to power the ATECC608A with 2.0 V (?)

[azerimaker]

There are other batteries, of course, (smaller, higher energy density), but they all come with added cost. Cheapest option is the AA alkaline.

@johanstokking if we believe the datasheets (which we shouldn't) operating voltage starts from 2V. The buck converter has a selectable output ranging from 1.8-3.3V, if needed we can adjust the output voltage with a quick solder bridge.

[elsalahy]

Design decision: The power supply design will be greatly affected by which transmission power the device must support?

One idea is to target 14 dBm only design? Pros:

• Better battery design
• Better power efficiency
• Reduced cost

Cons:

• Reduced range in countries utilising +20 dBm like the US

Am I missing other cons? then why is the LW standard mentioning +20 as a recommendation for the US

@azerimaker @wienke @johanstokking what are your thought on this?

[johanstokking]

Can you give some figures about better battery design, better power efficiency and reduced costs, when going for a design that supports only 14 dBm?

TX power is driven by Adaptive Data Rate and/or the application running on the end device. We wouldn't always make use of 20 dBm in US915, but it would be a possibility. It's 4 times as much power which can become.

[elsalahy]

Regarding better power efficiency:

• +20 dBm requires a minimum input voltage of 2.7V in comparison to 2V for 14 dBm
• The increase in voltage will increase leakage current and cause an estimated 30% increase in the current consumption compared to a 2.0V input
• 20+ dBm Tx current is 5x higher than 14 dBm for the same input voltage Example: Simple sensor data fetch done with 5mA at 2.0V will cost 7mA at 2.7V for x time for the same operation

@azerimaker can you provide some estimation and figures regarding the better battery design and the reduced cost that will be gained if we limit only to 14dBm?

[azerimaker]

@johanstokking @elsalahy @wienke, the buck converter we want to use has selectable outputs which can be set during the assembly. We can also read this jumper status and program the firmware to operate accordingly.

If we set the operating voltage to 2V for EU and 2.7V for US with a penalty of slightly more current consumption, we can have singe BOM to cover both regions.

[johanstokking]

with a penalty of slightly more current consumption

What would be the estimate?

@elsalahy estimates 30%, which is not "slightly"

we can have singe BOM to cover both regions.

This sounds great

[azerimaker]

@johanstokking increasing transmit power (by increasing the voltage) will always end up in higher energy consumption. With the help of ADR US customers can still save a lot of energy. We can also let them to switch the operating voltage down to 2V if longer battery life is what they want.

[azerimaker]

after revising the power supply section we now have two design options if we want to achieve global operation using single BOM. Option 1 - using LDO + Boost Converter Pros:

• default operation at low voltage (2V )and Dynamic Voltage Scaling up to 3.3V for high power (+22dBm) transmissions
• extended Battery lifetime due to the low operating voltage

Cons:

• Dynamic Voltage scaling might introduce oscillations, needs to be carefully tested
• Extra regulator (LDO + Switching) is necessary (0.3 $ added cost)

[azerimaker]

Option 2 - using Buck/Boost Converter

Pros:

• Using only one regulator
• Single BOM for all regions

Cons:

• HW jumper switch for region selection is necessary (EU @2.5V, US @3.3V)
• More current consumption in IDLE (Always ON switching regulator, slightly higher Iq)

[elsalahy]

Before I provide my recommendation, @johanstokking you mentioned a new TNGLORA (V3) in development, is it the same as the previous version HW wise?

[johanstokking]

Yes it's a ATECC608B version with the same specifications otherwise

[azerimaker]

@elsalahy, the Linear regulator I've picked has selectable output with a resistive voltage divider, we can set whatever operating voltage we want between 1.8-3.3V. If the secure element behaves erratically at 2V, we can operate at 2.1 or 2.2 instead. We have that flexibility.

[elsalahy]

@azerimaker Ok great, one tip is to analyse the latest old GN schematics and ensure we use similar capacitors values and setup.

[azerimaker]

@wienke , @elsalahy , @johanstokking , I'm adding these battery energy calculations (in Watt-hours instead of Amp-hours) so we can have better understanding when we choose which AA or AAA combo we want to use. I'm using Energizer L91 and L92 batteries as a reference:

• 1x AA , 1.5V, 3000mAh = 4.5 Wh
• 2x AA in series, 3.0V, 3000mAh = 9 Wh
• 2x AAA in series, 3.0V, 1200mAh = 3.6 Wh
• 3x AAA in series, 4.5V, 1200mAh = 5.4 Wh

The takeaway is that, 2x AA has 66% percent more capacity than 3x AAA.

[azerimaker]

I think we can all agree that going with 2xAA route would be the best option, at least for the first prototype.

[azerimaker]

We now have one last crucial decision to make. After all the back and forth discussions, it's clear that we have two options to proceed for our power supply approach, which are:

1. We design two slightly different and region optimized BOMs for the and the US. EU one would be optimized to work at max. +14dBm output, while operating at lover voltage (2V), and the US one will have higher operating voltage to accommodate high RF power output.

Advantages:

• [x] Huge power saving for EU device (~20% increase)
• [x] Better RF performance due to the simplified RF front-end
• [x] ~10% lower overall BOM cost due to the simplified RF front-end

Disadvantages:

• [x] Risks involved with managing two different BOMs.
• [x] Might be problematic for Lacuna in EU (they will be forced to use the US version in EU)

[azerimaker]

2. We design single BOM with dual-band RF front-end. We again have two sub-options here:

2.1 Have a fixed middle voltage (say ~2.7V) with max. RF power of +20 dBm.

Advantages:

• [x] Simplified BOM management (although greater number of components)
• [x] Interoperability with Lacuna

Disadvantages:

• [x] Less power efficient due to the increased operating voltage
• [x] ~5% higher BOM cost due to the dual-band RF front-end.

• [x] added RF switches might reduce RF performance and limit the output power, we may not achieve full +20 dBm.

  2.2 Have a dynamically scaling power source (default +2V, can jump up to +3.3V) Advantages:

• [x] Huge power saving also for the US region if +22 dBm isn't necessary all the time
• [x] Simplified BOM management (although greater number of components)
• [x] Interoperability with Lacuna

Disadvantages:

• [x] a bit more complex power supply circuitry (boosting w/ bypass mode)
• [x] ~10% higher BOM cost due to the dual-band RF front-end and booster circuitry
• [x] added RF switches might reduce RF performance and limit the output power, we may not achieve full +22 dBm.

[azerimaker]

@wienke, @johanstokking, @elsalahy, I know it's a bit overwhelming to decide which route to take, I've been scratching my head for a month now.

If you ask me, if the added 5% component cost and slightly compromised battery life isn't an issue, I would pick option 2.1 at least for the first prototypes to be on the safe side, but if you say we should aim for lower BOM cost from the beginning, then we should go with option 1.

Option 2.2 could be an experimental one, once we have a working prototype of option 2.1. It can be an upgrade.

[elsalahy]

I support 2.1 as the best choice. I support 1 as the second best choice.

[johanstokking]

Let's go for 2.1 for now indeed, thanks for the considerations.

[wienke]

Same