Part 1: A 20 Vrms Rating Whose Curve Stops at Half of It

The Pro-Wave 250ST160 is rated for 20 Vrms continuous drive. On the next page, the drive-voltage curve is drawn in Vp-p and stops at 30 - which is 10.6 Vrms.

Both numbers are correct. They describe the same quantity in different units, and they never meet. The rating table says 20 Vrms; the chart beside it plots sound pressure against a horizontal axis labelled Vp-p, running 0 to 30.

Convert them and the gap opens up: 20 Vrms equals 56.6 Vp-p, and the curve ends at 30. The published characterisation covers 53% of the rated drive range. The upper half of the rating has no curve behind it.

It gets more interesting when you look at the sibling part. The 250ST/R180 carries the same chart, with the same title and the same 0-to-30 range - but its axis is labelled Vrms. Same manufacturer, same family, same numbers along the axis, units differing by a factor of 2.83.

Put those two curves side by side and the 18 mm part appears to deliver the same sound pressure for a third of the drive. It does not.

What it costs: an engineer sizing a driver stage from the curve would size it for a third of the voltage the device is actually rated for.

Read the other way, the same mistake makes a 10 Vrms operating point look like the top of the range when it sits at roughly half.

Part 2: An AL Value That Turned Out to Be Permeability

The Cosmo Ferrites CF196T1303 has an AL of 800 nH per turn squared. The number 2000, which circulates alongside it, is the material's initial permeability.

Two quantities, 2.5× apart, both belonging to the same material, both printed in the same manufacturer catalogue - in different tables.

You can tell which is which, and it is worth showing the arithmetic because it settles the question rather than deferring to it. The relationship is:

AL = µ₀ · µi · Ae / Le

For the T1303 core the same catalogue gives Le = 29.5 mm and Ae = 9.3 mm². Put those in with µi = 2000 and you get 792 nH against the catalogue's printed 800 - a 1.0% difference.

Run that across all sixteen toroid sizes and every row reproduces within a few percent. That single test proves the AL column is not the permeability - and it delivers a second result for free: one material, sixteen cores, AL ranging from 800 nH to 2750 nH.

There is a third layer. The catalogue's material table carries a footnote stating its values "are obtained with Toroid T 2512" - a much larger core whose AL is 2400 nH. Cross-reference that table into a T1303 design and you have silently swapped the geometry.

What it costs: winding for 1 mH at a mistaken AL of 2000 gives 22 turns instead of 36. At the real AL of 800, those 22 turns produce 400 µH - 40% of the target.

A 400 µH choke in a 1 mH design does not read as a small error. It reads as ripple, saturation, or a converter that will not start.

Part 3: Two Datasheets, Two Frequencies, One Family

MACOM's CGH40006P is characterised at 2.0 GHz. Its sibling, the CGH40006S, is characterised at 5.8 GHz. Its larger sibling, the CGH40010P, is characterised at 3.7 GHz.

All three run on the same 28 V rail over the same DC–6 GHz band. Their gain and output figures sit in adjacent columns of a comparison table, and none of them were measured at the same place.

Read across and the 6 W parts look like a straight downgrade of the 10 W part. They are not comparable at all - 13 dB against 11.8 dB is not a 1.2 dB gap, it is two measurements taken 3.8 GHz apart.

The S variant is more than a package change, too. Its junction rating is 50 °C lower than the pill version's.

The thermal figure carries its own trap. The 8.0 °C/W on the 10 W part is footnoted as measured on the CGH40010F, the screw-down flange variant, at 14 W dissipation. The 6 W part's 9.5 °C/W was measured on the P variant at 8 W.

What it costs: a gain budget assembled from three datasheets, each quoted at its own frequency, produces a number that describes no operating point on any of them.

Part 4: A Suffix Letter Everyone Reads Wrong

The Thinking TVR20561KSW and TVR20561KSY differ by a factor of two in surge current. The obvious explanation - that W and Y are two surge grades - is wrong.

The two parts, both Φ20 mm varistors at 560 V:

Eight rows out of eight that are not surge-related are identical, including the disc geometry. The doubling is not a bigger part.

Now the letters. In the manufacturer's own part-number decoder, the eleventh position carries these definitions:

W raises the surge rating. Y declares RoHS. They are not two grades of the same thing, and reading them as a like-for-like swap produces the wrong answer.

The confusion has a locatable source: in the standard-series decoder, Y appears twice - once as an appearance code for a kink-lead body, and again as the optional suffix meaning RoHS compliant. One letter occupying two fields is enough to make the two part numbers look comparable.

Same first ten characters, same tolerance code, same lead form. The surge class lives entirely in the eleventh.

What it costs: specifying the standard part for a high-energy transient environment, because the two part numbers look like a suffix variation rather than two different series.

Why the Condition Keeps Going Missing

It is a documentation-structure problem, not an authoring mistake. Nobody at any of these four manufacturers set out to publish a misleading number. The condition drifted away from the value for structural reasons, and the same four reasons recur across the industry.

Separate documents per series. The Thinking TVR standard series and the TVR-D high-surge series are two datasheets. Each is internally complete and internally consistent. Neither contains the other's surge figures, so a customer holding one has no way to see that the other exists.

Separate documents per variant. MACOM publishes the CGH40006P and the CGH40006S as two datasheets with two revision numbers, two package drawings and two RF tables quoted at two different frequencies. A designer who downloads one has no signal that the other chose a different reference point.

Footnotes that carry real weight. The Cosmo Ferrites material table's footnote about the T 2512 reference core is one line at the bottom of a page. It changes what every number above it means.

The Pro-Wave family page footnote defining the 3.9 kΩ receiver load is the only place that condition appears. It is not in the datasheet at all.

Conventions that differ between documents. Vrms and Vp-p are both standard. Choosing one in a rating table and the other on a curve is entirely legitimate, and it is also how a 2.83× error gets built into a design without anyone making a mistake.

The practical consequence is that the condition is usually one document or one footnote away from the number, and the only reliable defence is to go and find it.

When a supplier or a distributor quotes a figure, the question worth asking is not whether the number is right. It usually is. The question is which document the condition lives in.

Four Questions to Ask Before You Use a Number

None of the four problems above were hidden. Every number was printed, correct, and in the manufacturer's own document. What was missing was the connective tissue. Here is the check that finds it.

1. What unit is the number in - and is the curve next to it using the same one? Vrms against Vp-p differs by 2.83×. A rating table and a characterisation chart can disagree about units without either being wrong.
2. Which variant was it measured on? A thermal resistance measured on the flange version does not describe the pill version. A junction limit quoted on one package does not carry to the other.
3. What was the measurement taken on - this part, or a stand-in? A material table measured on a larger reference core is a material property, not a specification for the core you are buying.
4. Do the numbers you are comparing come from the same test conditions? Gain quoted at 2.0 GHz, 3.7 GHz and 5.8 GHz cannot be put in one column, no matter how similar the part numbers look.

When all four answers line up, you have a specification. When any one is missing, you have a plausible number and a design decision resting on it.

Frequently Asked Questions

Q1: Isn't this just normal datasheet reading?

A: It is normal datasheet reading, and it is also where most sourcing errors come from. The four cases above are not obscure. Every number is printed correctly in the manufacturer's own document; the condition is simply stored somewhere else.

Q2: Which of the four is the most dangerous?

A: The suffix letter, because it looks like a naming detail rather than a spec. The TVR20561KSW and KSY share their first ten characters and differ by a factor of two in surge current. Nothing in the part number signals the size of the gap.

Q3: How would I have caught the AL value?

A: By arithmetic rather than by reading. AL = µ₀ · µi · Ae / Le reproduces the catalogue column within a few percent across sixteen cores. A formula check costs a minute and settles the question.

Q4: Can I compare gain figures across MACOM's 6 W and 10 W GaN HEMTs?

A: Not directly. The CGH40006P is characterised at 2.0 GHz, the CGH40006S at 5.8 GHz and the CGH40010P at 3.7 GHz. Pull each part's figure at the frequency you actually operate at.

Q5: If a curve stops short of the rated limit, is the part still usable up to the rating?

A: The rating is the rating; the curve is characterisation. The Pro-Wave 250ST160 is rated to 20 Vrms continuous, but its published drive-voltage curve stops at 10.6 Vrms. Above that point you are outside the documented range.

Q6: Does the same problem show up in the parts you stock?

A: It shows up in every part family we handle, which is why we quote with the condition attached. If you send us the operating point you are designing to, we will tell you which of the published figures actually covers it.