In 90% failed cases, the issue wasn’t “whether it can sensitively detect CO,” but rather that “time, environment, batch, and aging” weren’t considered core variables during the design phase. A performance-stable carbon monoxide alarm can consistently make correct decisions and responses according to the regulatory “time-concentration model” throughout its 10-year lifespan.
If you want to design a performance-stable CO alarm, you must solve the following problems simultaneously:
CO Sensor Stability
CO sensor materials are key components in alarm products; the success or failure of an alarm depends on a stable design that begins with material selection.
1. Choosing the right sensor technology, not just the lowest price.
| Sensor Technology | Performance stability | key issues | manufacturers |
| Semiconductors | No | large temperature drift and rapid aging | Figaro Engineering, New Cosmos |
| Electrochemical | Mainstream | Aging and consistency issues need to be addressed | AMETEK, KIDDE, CFS |
| NDIR | Very stable | high cost and large size | Senseair, Sensirion |
The standards UL2034 & EN50291 of household CO alarms use an electrochemical sensor. CFS produces 1,000,000 pcs of CO electrochemical sensors per year, and all CO alarms use CFS’s own electrochemical sensors.
2. Three key points in sensor stability design
(1) CO Sensor Sorting (many factories don’t do)
- There are initial sensitivity varies and the zero-point offset varies for CO sensors of the same model.
- CO sensors must be sorted into categories, not mixed before assembly.
CFS assembles CO sensors and does performance tests, then ages them for 14 days in-house, and then their performance and parallel impedance are tested again. We sort the current in the range of 1.2nA to 2.5nA, then make bin codes for every 0.3nA.
(2) The sensitivity working range is in the middle, not at the extremes
Many companies make the sensor operate near its full range to appear “very sensitive.” CFS focuses on fewer false alarms in a 10-year lifespan and allows the CO sensor to operate in the linear middle section of its response curve for extended periods. We can make our CO sensor have minimal temperature drift and a predictable aging slope.
(3) Designing for “aging” from the outset
CO stable sensitivity isn’t about preventing aging, but rather the design must make the inevitable aging controllable. CFS uses sensor aging models, software compensation provisions, and life-span calibration window design to keep stable performance.
Signal Circuit Design: 90% of False Alarms Originate Here
The Analog Front-End (AFE) is important for CO detection. Some alarm companies use the high-gain, high-noise generic amplifier circuits, while others use a custom amplifier for low-noise amplification and a stable reference source. CFS engineers add a guard ring.
In weak signal amplification, especially in the amplification of weak currents (our CO sensor produces weak signals in the nA range), a guard ring (or shielding ring) needs to be added to the input of the operational amplifier. The purpose is to suppress the influence of leakage current on the operational amplifier’s input.
Leakage current is also a form of current, and its formation is influenced by factors such as PCB materials, poor PCB trace layers, and PCB contamination.
PCB contamination includes oil stains on the circuit board surface, air humidity, flux, and circuit board cleaning agents. The presence of these substances can create leakage current paths. For example, if the voltage between the 5V power supply and the input pin is 5V, and due to circuit board contamination, the leakage resistance between them is 100M ohms, then the resulting leakage current is 50nA. This value may be tens of times larger than the bias current of a high-precision operational amplifier, which is why it is necessary to suppress leakage current.
Algorithm and Threshold Stability
Algorithm Design is the “soul” of stable performance. There are 3 items impacting stability.
1. CO alarm ≠ Threshold alarm
This is the biggest misconception. The regulatory logic is: Concentration × Time = Risk. Therefore, the algorithm must include time integration, segmented judgment, alarm window control and non-alarm range protection.
2. Comply with UL2034 or EN50291
A stable performance CO sensor always falls within the middle of the regulatory UL2034 or EN50291 sensitivity window, not at the edges.
| Concentration | UL2034 Standard | EN50291 Standard | ||
| No alarm allowed | Alarm required | No alarm allowed | Alarm required | |
| 30PPM | ≥ 30 days | Not allowed | ≥ 120 mins | —— |
| 50PPM | —— | —— | < 60 mins | 60~90 mins |
| 70PPM | < 60 mins | 60~240 mins | —— | —— |
| 100PPM | —— | —— | < 10 mins | 10~40 mins |
| 150PPM | < 10 mins | 10~50 mins | —— | —— |
| 300PPM | —— | —— | —— | < 3 mins |
| 400PPM | < 4 mins | 4~ 15 mins | —— | —— |
‘——’ Means Not apply.
3. The software algorithm should be “less sensitive,” not “more sensitive.”
Algorithms closer to the boundaries are less stable in the long run, based on 20 years of engineering experience. CFS engineers put hysteresis, buffering, noise suppression, and outlier removal in our software design.
Changeable environmental stability
There are 4 variables in the Real World (Not a Laboratory), they are as follows:
CFS team puts temperature and humidity compensation matrix, dynamic baseline tracking, long-term drift detection, and abnormal environment handling in the firmware.
Mass production batch stability
Batch stability is the “watershed” for a factory. You may see some interesting things for many products. It is perfect for lab samples, but the products are terrible and appalling in the market.
The reason is very simple: “mass production statistics” were not considered during design. The CFS team established three mechanisms during the design of the hardware and firmware.
- Parametric convergence design
- Not dependent on a “single calibration point.”
- Tolerates device variability
- SPC-driven design
- Not post-hoc verification
- Design-to-fit SPC
- Cross-batch consistency verification
Our team uses the same algorithm to verify Batch A, Batch B, and Batch C or D from different environments, different product lines, and different people.
Lifecycle Stability
Lifecycle stability is the real difference between brands. When the CFS team starts to design a new fire alarm, we usually ask our members the following question:
Our products employ low-power design to avoid voltage noise, model battery life curves, and separately design end-of-life logic.
Summary
A performance-stable CO alarm is not designed to “detect CO well one time”, but to “make the right alert for 10 years”.
CFS provide one stop ODM/OEM solution for your brand.
