This guide is for electronics engineers and procurement specialists who are comparing components from different vendors—specifically, scenarios where you're evaluating an alternative like a Kemet capacitor against a competitor's equivalent (say, an NXP vs. a C300 series part). It's a 5-step checklist I use before signing off on any alternative source.
When to Use This Checklist
This isn't a 'general sourcing' guide. This checklist is for the specific moment when you've been given the green light to evaluate a second-source component to reduce cost or improve lead time, but the application is critical. You have an incumbent part (maybe a specific NXP microcontroller or a legacy C300 series relay), and you're looking to qualify a replacement, perhaps from Kemet's portfolio.
The single biggest mistake I see isn't choosing a bad part. It's skipping the verification steps that go beyond comparing datasheet headline figures. This checklist fixes that.
The Checklist: 5 Steps to a Legitimate Component Comparison
Step 1: Map the 'Specs That Matter' vs. the 'Specs That Are Listed'
The Trap: You compare the operating temperature range and the tolerance, and they match. Looks good, right? Maybe. But a datasheet threshold (e.g., -40°C to +85°C) doesn't tell you performance at the extremes.
The Action: Create a side-by-side list of your application's three most critical performance parameters. For a Kemet tantalum capacitor, it might be ESR stability at high frequency. For a relay (like the C300-line type it might replace), it could be contact resistance after 100,000 cycles. Compare those specs, not just the generic ones.
Checkpoint: I literally highlight the datasheet row for my critical parameter. If the vendor's datasheet doesn't include a graph for it, I flag that as a higher risk item. (Which, honestly, I've seen happen more often with newer entrants).
Step 2: Find the 'NXP vs. Kemet' Equivalency Story, Not Just the Part Number
The Trap: You ask for the 'equivalent' to a part, and the vendor sales engineer answers quickly. They often give a 'direct cross.' But what story is behind that cross-reference?
The Action: Request the specific application note or internal test report that justifies that cross-reference. If they say 'it's the same,' ask which internal testing protocol they used to verify it. If it's a competitor like NXP, did they do a direct pin-compatibility test? If it's a relay vs. the C300, did they test the mechanical footprint 1:1? (Circa 2023, I found a 'direct replacement' for a C300 relay that was physically 0.5mm taller—not mentioned in the cross guide.)
Checkpoint: Write down their justification. Then, ask yourself: did they give you a detailed explanation, or a generic 'industry standard' answer? The latter is a yellow flag.
Step 3: Run the 'First Article' as a Stress Test, Not a Functionality Test
The Trap: You get the sample. You plug it into the test board. It works. You sign off. That was the trap I fell into for years.
The Action: Your first article test must include a condition it will only see once in 100,000 cycles: high temperature + high ripple current (for caps), or a high inrush current (for relays). 'Working' under ideal bench conditions proves nothing.
Checkpoint: The specific test should be: measure the parameter when the part is at its max temperature and at 90% of its max rated voltage. I ran a blind test with our lab team comparing a Kemet foil capacitor and an NXP-class competitor. At 85°C and full ripple, the competitor's capacitance dropped by 18%. The Kemet one dropped by 6%. That difference wouldn't show up on a standard test.
Step 4: Verify the 'Hidden' Supply Chain Specs (The Part Nobody Checks)
The Trap: Everyone looks at price, lead time, and voltage rating. Almost no one looks at the packaging and moisture sensitivity level (MSL).
The Action: Check the MSL rating of the alternative part. A part rated MSL 2 vs. MSL 1 might need different storage and a different baking protocol before assembly. If you're running a high-volume line, that's not an abstract risk—it's a real, measurable cost in reduced throughput or rework. For Kemet parts from their Mexico facility, I've found their packaging specs to be very consistent—MSL ratings match the stated levels almost 100%. But we had a batch of generic alternatives (shipped from a different Fort Lauderdale distribution hub) where the stated MSL was wrong, causing a $2,200 rework.
Checkpoint: Get the packaging spec sheet. Check the MSL. Check the reel orientation. Check the lead finish (Sn vs. NiPdAu). These aren't 'minor' details. They are the ones that cause line stoppages.
Step 5: Do the 'One-Week-Later' Gut Check
The Trap: You finish the tests. All green lights. You approve the BOM change. You close the file. Then, three weeks later, the procurement team gets the part and it's a different date code or has a different internal marking.
The Action: Before you finalize a Yes, do this: Ask yourself, 'If this batch fails six months from now, and we have to revert to the original part, can I get that original part quickly without re-qualifying it?' The risk of using a new vendor isn't the parts being bad—it's that you lose the relationship history with your incumbent vendor. I did a blind assessment once: the cost savings was $1,200 per lot, but the risk of having to requalify the old part cost us $4,200 in engineering time. The expected value said 'go,' but the downside risk felt too high.
Checkpoint: Draft an 'escape plan.' If you reject this competitor part in three months, do you have a pre-negotiated price and lead time with your incumbent vendor? If not, your decision is reversible, but painful. I want to say that's rare, but don't quote me on that—it's happened to me twice.
Common Mistakes (That I Have Made)
- Skipping Step 2. I approved a 'direct cross' for a microcontroller (NXP vs. a clone) because the speeds matched. The new part had a different internal timing for the start-up sequence. It caused an entire batch of 1,000 units to fail because the bootloader didn't sync. That quality issue cost us a $22,000 redo and delayed launch by two weeks.
- Assuming 'Same Factory' = 'Same Part.' I once evaluated a relay from a company that used the same die as the C300. The mold compound was different. It changed the thermal expansion characteristics. The part worked fine in the lab, but failed after 80 cycles in our end-use product.
- Ignoring the 'Kemet Mexico' vs. 'Kemet Fort Lauderdale' Effect. This goes for any vendor, not just Kemet. A part from one manufacturing location can have slightly different internal tolerances due to specific process controls. You need to verify the specific manufacturing source, not just the brand. (As of our Q1 2024 audit, the Fort Lauderdale line had a slightly higher failure rate than the Mexico line for a specific small case size capacitor, so we started requesting the Mexico-line parts specifically.)
This checklist isn't about being paranoid. It's about being professional. When you present your final decision to your engineering team or manager, your goal is to say, 'I verified this, and here's the audit trail.' That one sentence saves more time than any price comparison ever will.
