U1 vs U3 microSD: speed classes and bulk purchase requirements

U1 and U3 differ in their minimum sustained write class: U1 is 10 MB/s and U3 is 30 MB/s, under the conditions specified by the relevant standard. This does not mean every U3 card is three times faster in every device, or that read performance, endurance and compatibility improve together. For a bulk order, qualify an identified SKU in the intended host rather than buying an isolated symbol [1][3].

Image: AI-generated production-scene illustration. It is not an actual factory photograph, a daily inspection record or evidence of speed-class compliance.

Why does a quote stating 100 MB/s not replace a U3 requirement?

If the quoted value is a maximum read speed, it does not answer the sustained-writing question for loop recording. Even a peak write result needs its test capacity, data volume, reader, host interface and duration. A short desktop benchmark may never reach the conditions of a continuous recording workload.

Keep three separate purchase fields: declared speed class, the target host's sustained-writing requirement, and the read-speed objective for exporting files. These can lead to different configuration choices and acceptance procedures. A single column labelled “speed” hides the difference.

Can U3, V30 and UHS-I replace one another?

U3 and V30 both specify a 30 MB/s write floor, but belong to different rating systems. The number alone does not establish identical access conditions or host support. UHS-I describes a bus interface; it is not another way of writing U1. Follow the device manual or approved compatibility list [1].

The table compares numerical floors, not interchangeable certifications. The required host class and bus mode still apply; V60 and V90 require UHS-II or a supported faster mode [1]. Multiple markings on one card do not multiply its performance.

For purchasing, record what each field can and cannot establish:

What does a bitrate calculation actually prove?

Assume a recorder has a total average bitrate of 60 Mbit/s:

60 Mbit/s / 8 = 7.5 MB/s

This is only the average encoded-data rate. It excludes some overhead and does not describe bursts, host pauses or synchronization behaviour. “7.5 is below 10” is therefore not sufficient evidence that a U1 card meets that recorder's requirements.

Supply the device model, firmware, channel count, average and peak bitrate, continuous recording duration, temperature and retention period. Validate candidates under the same conditions. If peak data is unknown, mark it for measurement rather than inventing a universal margin percentage. The microSD selection guide helps separate use cases.

A desktop result above 30 MB/s does not authorize a buyer to apply U3 or V30 markings. Request the declaration for the exact product and keep target-device results as a separate record [1].

How do A1 and A2 fit into the decision?

Application Performance Class addresses application workloads, including random access [2].

The random-access figures use specified 4 KB operations under standard conditions. A2 performance requires supporting host and card functions. Its marking is not a promise of peak sequential speed, loop-recording endurance or compatibility with every USB reader.

For recording, check the host's U/V requirement and the real workload. For applications, also assess random access and host support. If a device mixes video, databases and logs, evaluate that combined workload instead of assuming any one symbol covers it.

Speed is not endurance

No speed marking establishes a service life for continuous writing. A V30 A2 card can still be unsuitable for a specific recording duty cycle. Require part-specific endurance evidence and agreed operating conditions; do not substitute a class label for those questions. See high-endurance versus consumer microSD.

How should a purchase requirement be written?

Do not stop at “128GB U3.” A useful requirement identifies a 128GB candidate microSD, the host model and firmware, the device-required speed class, and agreed acceptance for sustained writing, loop overwrite and recovery. Record artwork, configuration-change rules and lot identity separately.

Kalstor can help identify candidate hardware configurations for a microSD project. In the RFQ, provide the device requirements, quantity per capacity and read/write objectives. Available configurations, declared classes and lead time require written confirmation.

For repeat orders, connect the approved sample and specification to the delivered lot using the incoming-inspection checklist. A passing sample is evidence for its recorded conditions, not a guarantee that all future lots are identical.