Paddle testing · bench data · OEM / ODM
We test pickleball paddles for OEM and private label orders
Six numbers from one locked sample decide which build goes into production. Since 1 October 2026, one of the six is pass or fail, because USAP now counts spin directly and caps it at 2,100 RPM.
The six
Six readings
Each number decides one line of a purchase order. It decides nothing else. The table shows how we read each number, what it decides, and what it cannot decide.
| Reading | How we read it | What it decides | What it cannot decide |
|---|---|---|---|
| Exit velocity | 90 km/h ball, five impacts, ball speed off the face averaged | which of two builds hits the ball faster at the same weight | how the paddle feels in your hand |
| Swing weight | measured about a 4 in pivot, not on a scale | how fast the head comes through at the net | how far a hit off-center still flies straight |
| Spin rate | dual-camera count across the same 100 impacts used for wear | whether the model passes the 2,100 RPM cap | how long the face keeps that number |
| Twist weight | kg·cm² about the long axis, read on the bare frame | how much the head twists on a hit off-center, and how often a widebody comes back as a return | power, which comes from mass and face |
| Consistency | 20 impacts across the hit zone, spread of ball speed | whether a repeat order still matches the sample you approved | whether the design is any good |
| Balance point | millimeters from the butt cap, paddle on a level edge | player profile, head-heavy or quick-hand | total weight, which is set separately |
Thickness is set by a professional pickleball paddle manufacturer before the sample is made, so it never appears as a test reading. Every number above is measured on that sample. If you change the thickness later, you need a new guard mold, a different core density and a new quote.
Method
How we read the six
The cannon fires 100 times at the same spot, and we count spin every ten hits, so one run gives a curve, not a single number. We only compare two faces on one core, at one thickness and one weight. A conventional textured face lost about 100 RPM over the same run, and that is where 2.1× comes from. Both faces are ours, so the number says nothing about another factory.
Both faces ran on the same core and thickness, at the same weight.
Handfeel
Swing, twist and balance
Three numbers describe how a paddle swings, and none of them appears on a sheet that lists grams and millimeters. All three are read on the bare frame, before the guard and the grip go on.
Swing weight
A 216 g paddle with the weight in the head swings heavier than a 225 g paddle with the weight in the handle. We read both at a 4 in pivot so the two numbers can be compared.
Twist weight
Our 216 g game frame reads 44.2 kg·cm². Ten grams at three and nine o'clock adds about 2.0, and that is the cheapest way to stop the head twisting.
Balance point
Eight millimeters more weight toward the head adds 11.2% swing weight. One gram on the face adds about 0.32% ball speed.
The third number is the one buyers underrate. The player feels it on the first off-center hit, and what 10 g at three and nine does to a paddle never appears on a spec sheet.
Limits
One sample, one date
A bench reading belongs to one sample and one date, and it does not carry over to a production batch by itself. A brand that writes it into the spec will be arguing about numbers by month four.
What these numbers cannot do
- A new core lot, a new mold or a new face roll resets the reading, even when nothing on the drawing changed.
- Two cannons cannot be compared. Ball type, humidity and where the ball hits change the number more than most design changes do.
- Our readings cannot go into an approval file. USAP tests the model at its own facility, and your own bench data does not count there.
- The 2,100 RPM cap is a ceiling, not a target. A face built to land right on 2,100 will fail some batches.
- A good reading does not make a good paddle. It only rules out one way a batch can fail.
A bench is for finding what changed when something goes wrong, not for ranking one paddle against another. The approval route lives in a different file, and how a model reaches the approved paddle list is where those papers go.
Paperwork
Five contract lines
Five lines make a reading something you can hold a supplier to. They cost nothing when you sign, and they are the only part of a test report that still works after the staff changes.
| Line to write | What goes in it | Why it works |
|---|---|---|
| Retained sample | one sealed paddle per model, kept by both sides | a comparison at month seven needs the same object |
| Batch binding | core lot and face roll recorded on the carton label | a reading with no lot number cannot be traced back |
| Re-test point | first production batch, then every fifth lot | one pass on the sample says nothing about the next lot |
| Tolerance | a window agreed in writing, for weight and for spin retention | a single number with no window becomes an argument |
| Failure action | hold the batch, re-test, then decide who carries the cost | it is cheaper to decide before the container leaves |
Fit
Which orders need a test
A reading saves money on some orders and changes nothing on others. The table shows where a number changes what you do, and where it only adds a page to a file.
| Your order | What a reading buys you | Our advice |
|---|---|---|
| First order for a new brand | picks one of two directions before tooling, and saves a sample round | Test both. It is cheaper than the second round. |
| Matching a model you already sell | confirms the new build has not gone softer or heavier | Test the reference build first, then the copy. |
| Volume model at a mainstream price | keeps batches identical, and nothing more | Skip the performance run. Test consistency only. |
| A certification you intend to file | tells you whether sending the model is worth the fee | Pre-test here, then send it to the body. |
| A 14 mm version of a 16 mm line | shows what the thinner core does to weight and spin | Test the thin build at its own target weight. |
Evidence
Factory record
We keep six records for every batch we ship. When a buyer says a batch is wrong, these six are what we open, and they let both sides argue with numbers instead of adjectives.
Ball cannon
100 impacts at 90 km/h, spin counted every ten, curve filed with the batch.
X-ray
Layer order and core bond read on a cut sample from the run.
Compression
Core set measured after a load cycle, before the batch is packed.
PBCoR pre-check
An in-house reading against the 0.44 and 0.43 thresholds.
Batch QC report
Weight, balance, thickness, guard bond and print on one sheet.
Retained paddle
One sealed sample per model, held for twelve months.
After the test
FAQ
Do you test every batch or only the approved sample?
Both, and they do different jobs. The sample run picks one of two directions. Every fifth production lot is tested again to see whether that direction is still holding.
Can I use your readings in a USAP application?
Those readings cannot go into an application. USAP tests the model at its own facility, and our numbers are a pre-check that tells you whether sending the model is worth the fee.
What does the 2,100 RPM cap change for my face choice?
It makes spin a pass or fail number rather than a selling point. A face that reads above the cap on a new paddle can fall under it after break-in.
Why do USAP and UPA-A read the same face differently?
Both bodies set the same 2,100 RPM cap. USAP reads a new paddle and UPA-A reads one after break-in, so one face can land on opposite sides of the same line.
What is the difference between swing weight and static weight?
Static weight is what a scale reports. Swing weight is what a hand feels once the paddle moves. Two paddles can match on the scale and swing differently, because the weight is in different places.
Can you match a reading from another brand's paddle?
We can build to a target reading on our own bench and show you the curve. We do not copy another brand's graphics or its trade dress, and a number from a different machine will not transfer.
How much time does testing add to a sample order?
Nothing on the first sample, because the test runs inside the seven to fifteen day sample window. A separate comparison of two directions adds about a week.
What happens if a batch misses the reading?
The batch is held and re-tested before packing. If the second run shows the same shift, we trace it to the core lot, the face roll or the press recipe. We settle who pays before the container leaves.
The numbers above describe one face over one run. The measurement itself is set out in how spin readings are taken on a textured face. The paperwork around a sample round is set out in how a sample test is run step by step.
Next step
Send the spec
Give us the shape, the thickness, the target weight and the channel you sell into. We will tell you which of the six readings will decide the build, and which of them your order does not need.
Reply within 24 hours · lixifirm@outlook.com · MOQ 200 pcs