Two Infineon PDM microphone datasheets list supply currents from 980 µA at a 3.072 MHz clock down to 25 µA in standby and 1 µA with the clock off. The microphone is one line in a pendant's budget. Measure MCU, VAD, BLE and regulator current on your own board.
A voice pendant that waits for speech and records on demand spends most of its life in the listening path. The microphone, the voice activity detection (VAD) code, the processor's sleep state and the radio all draw from the same small battery. This guide lists the microphone currents we verified, a measurement checklist for everything else, and questions to answer before you choose a design. It contains no measured result from our own boards and no Duobay price.
What do the microphone datasheets say?
These are typical values from two Infineon PDM microphone datasheets, read on October 31, 2026. The IM69D130 current values are stated with no load on the data line.
| Part | Mode | Supply current | Source |
|---|---|---|---|
| IM69D130 | Clock 3.072 MHz | 980 µA | Infineon IM69D130 datasheet, as of October 31, 2026 |
| IM69D130 | Clock 2.4 MHz | 800 µA | Infineon IM69D130 datasheet, as of October 31, 2026 |
| IM69D130 | Clock 1.536 MHz | 620 µA | Infineon IM69D130 datasheet, as of October 31, 2026 |
| IM69D130 | Clock 768 kHz | 300 µA | Infineon IM69D130 datasheet, as of October 31, 2026 |
| IM69D130 | Standby mode | 25 µA typical (50 µA max) | Infineon IM69D130 datasheet, as of October 31, 2026 |
| IM69D130 | Clock-off mode | 1 µA, clock pulled low | Infineon IM69D130 datasheet, as of October 31, 2026 |
| IM69D128S | High-performance mode | 520 µA | Infineon IM69D128S datasheet, as of October 31, 2026 |
| IM69D128S | Low-power mode, clock 768 kHz | 180 µA typical (250 µA max) | Infineon IM69D128S datasheet, as of October 31, 2026 |
Two things in this table matter for design. First, the IM69D128S lists a low-power mode at 180 µA typical against 520 µA in high-performance mode, so the mode you run changes the mic's share of the budget. Second, the IM69D130 datasheet lists a mode switch time of 20 ms for +/-0.5 dB sensitivity accuracy, which limits how aggressively the mic can be duty-cycled.
The standby and clock-off rows come from the IM69D130 datasheet only. We did not verify a standby current for the IM69D128S, so check its datasheet if you plan to duty-cycle that part.
What else is in the listening path?
The mic is one line in the budget. MCU sleep, VAD compute, BLE advertising and regulator quiescent current each draw too, and so does a recording indicator if you add one. The mic figures above apply at the mic supply pin. A regulator sits between that pin and the battery. We publish no figures for the other blocks, because the vendor does not publish them for your algorithm, advertising settings or regulator. Take them from the official datasheet of each part you choose, then confirm them with a capture on your own board.
What should you measure on your own board?
Use a power profiler or a current meter with enough range for both sleep and active currents. For each item, record what the table lists so a second engineer can repeat the capture.
| What to measure | How to instrument or capture | Raw data to record |
|---|---|---|
| Mic current in each mode you plan to use | Put a current meter or power profiler on the mic supply, with the clock at the frequency your firmware uses. Capture each mode on its own. | Raw current trace, clock frequency, supply voltage, temperature, mode name, instrument settings. |
| Mode switch behavior if you duty-cycle the mic | Trigger a mode change from firmware and capture the supply current across the switch. Mark the firmware event with a GPIO pin so the trace lines up with the code. | Time from the switch command to usable audio, current during the switch, and the datasheet row you compared against. |
| VAD compute cost | Run your real algorithm on the real MCU with the mic path live. Toggle a GPIO at the start and end of each frame so compute shows apart from sleep. | Current during compute, current during sleep between frames, frame length, MCU clock setting, algorithm version. |
| BLE advertising | Capture a window of current at your real advertising interval and payload. | Interval, payload size, transmit power setting, capture window, and the trace itself. |
| Regulator overhead | Measure at the battery terminal and again at the load side while the device sits in the listening state. | Both currents, battery voltage, load voltage, regulator part number and settings. |
| Whole listening path in a normal session | Capture from the battery with the device in its normal listening state, including wake events and recording starts. | Battery-side current trace, event markers, number of wakes in the window. |
How does this fit a recording product?
Listening for a trigger and recording are different states. A product that records on demand should make recording visible, with an indicator that shows whenever audio is being saved, and should handle consent for the people being recorded. The indicator draws current too, so give it its own row in the budget. Never design for hidden recording. This is an engineering summary. It does not give legal advice.
What should you ask before you choose the listening path?
Take this list to your team, and to any supplier who proposes a microphone, MCU or VAD design.
- Which exact mic part and which mode does the firmware use while listening, and which datasheet row states its current?
- Does our VAD need valid audio the moment the mic wakes, and does the mode switch time on the datasheet allow that?
- What clock frequency do we run the mic at, and does the datasheet publish a current for that exact frequency?
- Which MCU sleep state do we use between frames, and where is the official datasheet line for its current?
- What does our VAD algorithm cost on the chosen MCU, and has anyone measured it on a board?
- What is the average current of our BLE advertising at the interval and payload we plan to ship?
- What is the regulator quiescent current in the listening state, and was it measured at the battery terminal?
- Is there a visible indicator that shows whenever the device records, and does its current belong in the budget?
- How is the user asked for consent, and what does the device do between the VAD trigger and the start of a recording?
- Which numbers in our budget are datasheet values, and which are still assumptions?
Questions about always-on listening power
How much current does the microphone alone draw?
Can I duty-cycle the microphone?
The IM69D130 datasheet lists a mode switch time of 20 ms for +/-0.5 dB sensitivity accuracy. That limits how aggressively you can duty-cycle it. Measure the behavior on your board.
Where do I get MCU sleep and active current numbers?
From the official datasheet of the exact part and mode you choose, and from your own capture. We publish no MCU figures here.
Does an always-listening pendant record all the time?
A pendant can listen for a trigger and record only on demand. Design recording to be visible, with an indicator that is on during recording, and ask for consent from the people being recorded. This article covers power. It is an engineering summary. It does not give legal advice.
What does a Gate 0 Feasibility study cover?
A Gate 0 Feasibility study covers chip and module selection, a system architecture document and a costed BOM. We do not publish prices. Tell us about your device and we will explain how it works.
If you want help choosing the listening path for your pendant, tell us about your device. You can also read about our AI recording wearable development.
Sources
All pages accessed October 31, 2026. Datasheets are revised from time to time, so re-read them before you decide.
- Infineon, IM69D130 datasheet: https://www.infineon.com/dgdl/Infineon-IM69D130-DS-v01_00-EN.pdf (accessed 2026-10-31)
- Infineon, IM69D128S datasheet: https://www.infineon.com/dgdl/Infineon-IM69D128S-DataSheet-v01_00-EN.pdf (accessed 2026-10-31)
Planning an AI recording wearable? Tell us what it needs to do. A Duobay engineer replies within 1 business day. Send a project brief →