The linked post argues that Texas Instruments' current NE5532 is no longer the same workhorse audio op-amp engineers have used for decades, despite carrying the same part number. Commenters highlight three concrete changes called out in Dave Jones' EEVblog video and in the post itself: the input stage appears to have changed from NPN to PNP, the slew rate dropped from 9 V/µs to 5 V/µs, and the maximum supply rating fell from ±22 V to ±18 V. That is not a cosmetic respin. It is a different analog part wearing a legacy label.
If a commodity analog part number no longer guarantees drop-in behavior, then procurement, compliance, and field reliability all get riskier in ways that small hardware teams usually do not have the process to catch.
Strongly negative toward Texas Instruments. People saw this as a breach of trust around a commodity part number, with real risk for interchangeability, repairs, compliance, and latent failures. A few commenters noted that process migrations and loose part-number conventions are common in analog, but even they largely agreed the NE5532 changes were too large to hide behind the same SKU.
01 The real failure mode is supply-chain contamination, not just a bad datasheet update.
A jellybean part like the 5532 is supposed to be sourceable by generic number across vendors and even through co-mingled distributor inventory. Once TI's version no longer matches the de facto standard, a purchasing team can break a product without ever making a visible part substitution. That turns second sourcing from a resilience tactic into a hidden liability.
For commodity analog parts, the part number itself is part of the interface. Break that, and ordinary purchasing becomes an engineering risk.
02 The most plausible root cause is fab migration, not random negligence.
TI is reportedly moving legacy products onto newer processes and 300 mm wafer lines, and analog circuits often do not survive that move unchanged. The hard choice is to obsolete the old part or ship a behaviorally different replacement under the old code. Commenters saw TI as choosing continuity for large customers over true compatibility for everyone else.
Modernizing manufacturing can collide directly with backward compatibility. Analog parts are especially vulnerable because process details are the product.
03 Regulated hardware makes part renaming expensive, but silent equivalence is worse.
One commenter noted that changing a part number can trigger thousands of dollars in regulatory paperwork for small companies. Another pointed out the obvious catch: regulators are likely to care far more if the same approved part number now hides materially different behavior. The paperwork cost is real, but it does not justify a vendor smuggling in a new part under an old identity.
Compliance regimes reward stable identifiers, but only when those identifiers still mean something. Fake continuity creates a nastier problem than explicit change.
04 This exposed a gap between digital instincts and analog reality.
Several commenters said semiconductor part numbers have never guaranteed exact sameness the way many software-minded engineers assume, and analog designers are taught to leave margin because process spread, aging, and circuit stability all matter. But they also stressed that even benign-looking 'improvements' can destabilize real circuits, raise EMI, or surface timing bugs. In other words, analog tolerates variation right up until it does not, which is why large spec shifts under the same number are so dangerous.
Analog engineers expect variation, not identity. That is exactly why they react so strongly when a vendor changes first-order behavior, not just manufacturing detail.
05 The backlash is also about accumulated distrust, not just one op-amp.
Commenters brought up TI's earlier attempts to remove archived datasheets, reported die changes in other products, and the loss of responsive support. That history makes this look less like an isolated judgment call and more like a company that no longer treats long-lived documentation and part identity as sacred.
Vendor reputation compounds technical risk. Once engineers stop trusting your documentation discipline, every change looks more dangerous.
01 Blind reuse of old bins and old assumptions is part of the problem.
One commenter argued that engineers should prototype with newly sourced parts, read current datasheets, and stop trusting the 'magic parts box' or institutional memory. That does not excuse TI, but it does point to a real failure mode inside small teams that assume old subcircuits stay valid forever.
Good component hygiene still matters. Fresh stock and current datasheets catch some of the damage before it ships.
02 The practical answer may be to stop caring about the 5532 at all.
One commenter argued that newer parts like the OPA1612 already outperform the NE5532 on paper for high-end audio, so engineers should migrate instead of defending a decades-old default. That sidesteps the naming controversy by treating the part as effectively obsolete for new designs.
For new products, replacement may be cheaper than policing legacy equivalence.
03 Part numbers are not universal promises of every observed behavior.
A few commenters pushed back on the outrage by saying that if a design depends on behavior outside the published spec, the fault lies with the designer, not the vendor. That argument is weaker here because commenters say the changed parameters were first-order specs, but it is still a useful reminder that drop-in assumptions often exceed what suppliers actually guarantee.
Do not confuse market convention with a formal contract. Some of the industry's pain comes from relying on conventions as if they were standards.