XTAL Identification

Scope: Identification, characterisation and pass/fail screening of surface-mount clock sources with 4 or 6 pads, without altering or damaging the device under test (DUT).

Out of scope: through-hole DIP-14/DIP-8 oscillators, 2-pad crystals, MEMS programmable parts requiring I²C/OTP programming, radiation or life testing, decapsulation or cross-section.

The whole job is: work out which one it is, work out what each pad does, then check it works.

The golden rule: find out what the pads are before you connect power. Guessing a power pad is the number one way to kill a good part.

1.1 The two parts in stock to practise on

Do pk 588 first. You already know the answer, so it tells you whether your technique is right.

2. What you need

From InvenTree

Not in InvenTree — get these

• A solderless breadboard. Any half-size board is fine. There isn't one in the system, so add it when you buy it.
• Male Dupont / header pins, 4 to 6 of them.
• Very thin bare copper wire, 0.1–0.3 mm. Easiest source: take one jumper wire from pk 323, strip 30 mm, and pull out a single strand. That single strand is exactly the right thickness and it's bare copper, so there's no enamel to scrape off.

3. Safety limits

Short list. Do not go past these.

4. The flow

flowchart TD
    A([Part in hand]) --> B[STEP 1<br/>Caliper: length, width, height]
    B --> C[STEP 2<br/>Microscope: copy the text<br/>EXACTLY as laid out]
    C --> D[STEP 3<br/>Count pads, find the pin-1 dot,<br/>number them P1..Pn]
 
    D --> E[STEP 4 - THE KEY TEST<br/>Beep each pad against<br/>the metal case]
    E --> F{How many pads<br/>beep to the case?}
 
    F -->|TWO| G[TYPE C - CRYSTAL<br/>Those two are the GND pads.<br/>The other two are the electrodes]
    F -->|ONE| H[OSCILLATOR<br/>That pad is GND]
    F -->|NONE| I[Lid is not connected.<br/>Use the fallback in 5.4]
    I --> H
 
    H --> J{Pad count?}
    J -->|4| K[TYPE A]
    J -->|6| L[TYPE B]
 
    G --> G1[STEP 5C<br/>Build the breadboard jig]
    G1 --> G2[STEP 6C<br/>Pierce test with 74HC00<br/>Does it oscillate? What frequency?]
    G2 --> G3[STEP 7C<br/>Identify the exact part<br/>from size + text + frequency]
    G3 --> Z
 
    K --> M[STEP 5<br/>VDD is the pad DIAGONALLY<br/>OPPOSITE the GND pad.<br/>Confirm with the meter]
    L --> M
    M --> N[STEP 6<br/>Build the breadboard jig:<br/>4 pins in a rectangle,<br/>part in the middle, pads up]
    N --> O[STEP 7<br/>Beep every pin to its pad<br/>BEFORE connecting power]
    O --> P{All four<br/>connect?}
    P -->|No| Q[Re-bend the wires.<br/>Do not power a jig<br/>that does not beep]
    Q --> O
    P -->|Yes| R[STEP 8<br/>Power up: 1.8 V, 20 mA limit<br/>Watch the current]
 
    R --> S{Current<br/>sensible?}
    S -->|Limit hit| T[POWER OFF<br/>Pads are wrong.<br/>Go back to STEP 4]
    S -->|Yes| U[STEP 9<br/>Raise to 3.3 V, 50 mA limit]
    U --> V[STEP 10<br/>Scope BOTH remaining pads<br/>at the same time]
 
    V --> W{Which pads<br/>have a signal?}
    W -->|One| X[That pad is OUT.<br/>The last one is the control pad]
    W -->|Adjacent pair| Y2[Differential output pair]
    W -->|None| Y3[FAIL - no output]
 
    X --> AA[STEP 11<br/>Control pad test<br/>through a 10k resistor]
    Y2 --> AA
    AA --> AB[STEP 12<br/>Frequency, duty cycle, current]
    AB --> AC[STEP 13<br/>Identify the exact part]
    AC --> Z
 
    Z[STEP 14<br/>Write it all into InvenTree]
    Z --> END([Done - part back in its bag])
    Y3 --> END
    T --> END

5. Steps 1–4: Identify the pads

5.1 Step 1 — Measure it

Use the caliper. Close it fully and press ZERO before you start.

Measure length, width and height. Take each one three times and write down all three — if they disagree by more than 0.03 mm you are squeezing too hard or catching an edge.

Look up the size:

Height helps too. A crystal is thinner than an oscillator in the same footprint, because the oscillator has a chip inside as well as the quartz. A 5032 crystal is around 0.9 mm; a 7050 oscillator is around 1.3 mm.

5.2 Step 2 — Copy the text exactly

Put it under the microscope, text the right way up.

Copy the layout, not just the letters. Manufacturers hide the model code in which line things sit on. If you flatten it to one line you throw that away.

Write it like this:

Line 1:  [TXC]  25.0
Line 2:  UR03
Extras:  small dot in the top-left corner
Lines:   2

• Put [ ] around a logo.
• Copy 25.0 as 25.0, not 25. The number of digits is part of the code.
• If you see a voltage like 3.3V, that's a strong hint it's an oscillator.
• Photograph the top and the bottom.

5.3 Step 3 — Number the pads

Turn it pads-up. Find the pin-1 mark: a dot, a cut-off corner, a notch, or one pad shaped differently. Starting there, number the pads P1, P2, P3… going clockwise.

No mark at all? Pick a corner, call it P1, and draw it on your photo so you can find it again. Just write down that the numbering is your own.

5.4 Step 4 — Find GND using the metal case ⭐

This is the test that decides everything, and it's the easiest one in the procedure.

The top of the part is a metal lid. Inside, that lid is welded to the ground connection. So any pad that is ground will be electrically connected to the lid — and you can just beep it out.

Do this:

1. Set the Brymen to continuity (the ♪)) symbol). Touch the two probe tips together — it should beep. That's your check that the meter is set right.
2. Lay the part pads-up on the bench.
3. Hold one probe tip on the metal lid. Press on the shiny welded seam right at the edge of the lid — that's where the metal is cleanest.
4. Touch the other probe to each pad in turn. Note down which ones beep.
   Do not press hard, and do not scratch or sand the lid. If it won't beep, move along the seam and try a different spot.

Now read the result:

That single reading has just told you the device type and found your ground pad, with no power and no risk.

A note on why this works, and when it doesn't. Nearly all these parts have the lid bonded to ground because the lid doubles as a shield. But it isn't a law. A few designs leave the lid floating, and some lids have a thin oxide or lacquer film that stops the probe making contact. So beeping is proof; not beeping is not proof of the opposite. That's why there's a fallback below.

5.5 Fallback — if nothing beeps to the lid

Use the meter's resistance range and measure between every pair of pads, both ways round (swap the probes and measure again). You're looking for one pair that behaves differently depending on which way round you measure.

• A pair reading a few hundred Ω to a few kΩ one way, and open the other way → that's the power pair. The pad that had the red probe on it in the low-reading direction is GND; the other is VDD.
• A pair reading near zero Ω both ways, with the other two pads open → those are the crystal case pads. Treat it as a crystal.
• Everything open both ways → it may be a crystal whose case pads aren't connected to anything. Skip to §7 and try the Pierce test on each diagonal pair. If neither oscillates, stop and quarantine it. Do not start guessing power pads on a part that has shown you no power pair.

6. Build the breadboard jig

Everything from here on runs on a breadboard. You are not soldering to the part — you're making four thin wires press onto its four pads.

6.1 The idea

        breadboard holes (2.54 mm apart)
                                                    
     o     o     o     o     o     o     o
                                                    
     o     P1----o-----o----P4     o     o     <- header pins pushed in here
     o     |                 |     o     o
     o     |   +---------+   |     o     o
     o     |   |  PART   |   |     o     o        part sits PADS UP
     o     |   | pads up |   |     o     o        in the middle
     o     |   +---------+   |     o     o
     o     |                 |     o     o
     o     P2----o-----o----P3     o     o
                                                    
     o     o     o     o     o     o     o
 
  Each header pin gets one thin copper wire.
  The wire arches over and presses DOWN onto its pad.

Four header pins in a rectangle. Part in the middle, pads facing up. One hair-thin wire from each pin, bent over so its tip rests on one pad. The springiness of the wire holds it down.

6.2 Building it

1. Make the wires. Take a jumper wire (pk 323), strip about 30 mm, and pull out a single strand. Cut four pieces about 25 mm long. Bare copper, nothing to scrape.
2. Fit the pins. Push four male header pins into the breadboard in a rectangle roughly the size of your part. On a 2.54 mm breadboard, two holes apart is 5.08 mm and one hole is 2.54 mm — close enough for a 7050 or 5032. It won't match exactly, and it doesn't need to; you'll bend the wire tips in or out to land on the pads.
3. Attach a wire to each pin. Wrap two or three turns tightly around the top of the pin. Tug it gently — it should not come off.
4. Place the part. A pea-sized piece of poster tack or double-sided tape under it stops it sliding. Pads up.
5. Bend each wire down so its tip sits on the pad you want. Aim for the middle of the pad. A slight downward curve gives you springy contact.
6. Label the rows. Write on masking tape which breadboard row is which pad. You will mix them up otherwise.
7. Add the decoupling capacitor. For oscillators only: a 100 nF (pk 523) between the VDD row and the GND row, as close to the part as it will go. This is not optional — without it a fast oscillator can misbehave or refuse to start.

6.3 Check the jig before powering anything

Beep from the top of each header pin to its pad, one at a time.

All four must beep. If one doesn't, re-bend that wire and check again. Never apply power to a jig that hasn't passed this check — an intermittent contact on a power pad is exactly how parts get damaged.

Also beep between neighbouring pins. They must not beep to each other. If they do, two wires are touching.

6.4 What a breadboard costs you

Breadboards add a few pF between rows and the wires add inductance. At 48 MHz your waveform will look rounded and messy, and there may be ringing. That's the breadboard, not the part.

So: trust the breadboard for is it working, what frequency, what voltage levels and how much current. Do not trust it for rise time, fall time or overshoot. Those need a proper board.

7. Type C — testing a crystal

A crystal makes nothing on its own, so you have to build a circuit around it. Never connect it to a power supply directly.

7.1 The Pierce test circuit

One gate of a 74HC00 with both its inputs tied together works as an inverter. Built on the same breadboard.

              1 MOhm
          +---/\/\/---+
          |           |
          |  gate A   |          10 kOhm
   +------+--[>o------+---------/\/\/-----+
   |                                      |
 pad 1                                  pad 2
 (crystal)                            (crystal)
   |                                      |
 22 pF                                  22 pF
   |                                      |
  GND                                    GND
 
 gate B: inputs tied to gate A's output --> scope CH1 (10:1 probe)
 Chip power: 3.0 V, 20 mA limit, 100 nF across pins 7 and 14

Parts: 74HC00 (pk 258); 1 MΩ and 10 kΩ (pk 521); two 22 pF and one 100 nF (pk 523).

7.2 Running it

1. Build the circuit with the crystal held in your jig. Keep the wires to the crystal short — under 20 mm.
2. Set the supply to 3.0 V, 20 mA limit. Switch on.
3. Probe gate B's output, not the crystal. Touching a probe onto the crystal itself loads it and can stop it oscillating — you'd conclude the part is dead when it isn't.
4. Read the frequency from the scope's counter.

For a 32.768 kHz crystal, swap the 10 kΩ for 330 kΩ and be patient; it can take several seconds to start.

For pk 588 (25 MHz): this jig is right at the limit of what a 74HC00 will start. Plan on the no-capacitor retry. A no-start here is inconclusive, not a failed part.

7.3 Identifying the exact part

You can call it confirmed only when all four of these agree with one datasheet: measured size, pad count and layout, the marking, and the measured frequency. Three out of four is a candidate — say so.

1. Turn length × width into a package code.
2. Search the manufacturer's marking guide for the series that uses that package and that text layout.
3. Narrow it down using what §5.4 told you — crystal or oscillator. Manufacturers often use neighbouring codes for the crystal and the oscillator in the same package, so this is the step that separates them.
4. Check the frequency matches.
5. Check the datasheet's height against your caliper reading. Height is the thing marking-only identifications always miss.
   Worked example, pk 588. 5.0 × 3.2 mm → 5032. Two lines of text, logo and frequency on line 1. §5.4 found two pads beeping to the case → crystal. TXC's only 4-pad 5032 crystal family is 7B (7A is the 2-pad version; 9B is a big through-hole can). Height measured 0.94 mm against 7B's 0.9 mm nominal — the alternatives are 1.2 mm and up, so it isn't close. Settled: 7B-25.000M___-T. The three blanks are tolerance, stability and load capacitance, and they are not on the case — UR03 is a batch code. Those need an LCR meter the lab doesn't have, so they stay blank rather than being guessed.

8. Types A and B — testing an oscillator

8.1 Step 5 — Work out VDD

You have GND from §5.4. On both standard pinouts, VDD is the pad diagonally opposite GND.

Confirm it before trusting it. Put the meter on resistance and measure between GND and that diagonal pad, both ways round. You should get a reading of a few hundred Ω to a few kΩ with the red probe on GND, and open the other way. That's the internal circuitry showing itself.

If the diagonal pad shows nothing either way, or one of the other pads shows the effect instead, stop. Something doesn't match the standard layout and it's worth a second opinion before you power it.

For reference, the standard layouts (numbering counter-clockwise from the pin-1 mark, viewed from the top):

4-pad: 1 = enable or not-connected · 2 = GND · 3 = OUT · 4 = VDD
6-pad: 1 = enable/NC · 2 = NC or voltage control · 3 = GND · 4 = OUT− · 5 = OUT+ · 6 = VDD

8.2 Steps 6–7 — Jig and check

Build the jig per §6, wire the decoupling capacitor, and run the continuity check in §6.3. Don't skip it.

8.3 Step 8 — First power-up, gently

1. Set the supply to 1.8 V with the current limit at 20 mA, output still OFF. Check the limit reads back correctly before you connect anything.
2. Connect only VDD and GND. Leave the other pads unconnected.
3. Switch the output on and watch the current.
   Switch off straight away if: the supply goes into current limit (the CC light comes on), or the current keeps climbing instead of settling, or it jumps to tens of mA immediately.

If that happens your pad identification is wrong. Go back to §5.4 — don't just try the other polarity and hope.

A good result is a small, steady current. Many parts will sit silent at 1.8 V, and that's fine — this step is only checking you haven't got it backwards.

8.4 Step 9 — Raise the voltage ⚠

Important, and a correction to earlier revisions of this document. 20 mA is fine for the polarity check but too low to actually run a fast oscillator. A 7050 CMOS oscillator can draw 30 mA or more, and more still at high frequencies. If you leave the limit at 20 mA the supply will fold back on a perfectly healthy part and you'll record it as a failure.

Once §8.3 has passed cleanly:

1. Raise the current limit to 50 mA.
2. Step the voltage: 1.8 V → 2.5 V → 3.3 V, pausing at each.
3. Write down the current at each step. That number helps pin down the exact model later.
   Never go above 3.6 V.

8.5 Step 10 — Find the output

Don't probe one pad at a time — you'll miss things and it takes four times as long.

4-pad part: put scope CH1 and CH2 on the two remaining pads at once, both probes on 10:1. One will show a clock. That's OUT. The other is the control pad.

6-pad part: clip the UT-M15 logic pod (pk 302) onto all four remaining pads and take one capture. It shows you immediately which pads are switching. Then move the scope's analogue channels onto those for the detail.

If two neighbouring pads both switch, it's a differential output. Check with the scope's A − B maths function: a real pair gives a clean doubled signal. Then read the middle voltage of the pair to work out the type:

Work out which it is before connecting any termination resistors. The wrong termination gives you a wrong answer and stresses the part.

8.6 Step 11 — The control pad

Connect the last pad to GND through a 10 kΩ resistor — never directly.

For that last one, drive the pad from channel 2 of the UDP3305S and step it from 0 V up to VDD in 0.2 V steps, reading the frequency each time.

Then repeat the whole test with the 10 kΩ going to VDD instead, to confirm it behaves the opposite way.

8.7 Step 12 — Measure it

Remember §6.4: on a breadboard, rise and fall times are meaningless. Note them as "breadboard — not valid" rather than writing down a number that isn't real.

Also watch the scope's own limit: the MSO2204 handles 200 MHz, so at 48 MHz your frequency and voltage readings are solid, but edge shape is already being softened by the instrument as well as the breadboard.

8.8 Step 13 — Identify the exact part

Same rule as §7.3, with two extra clues you now have: which supply voltage actually worked, and what the control pad does. Those two often separate model numbers that are otherwise identical.

9. Step 14 — Write it into InvenTree

A test nobody records has to be done again. For each part:

1. Put the confirmed type, series, supply voltage, output type and pinout into the description.
2. Delete provisional keywords like type-unconfirmed.
3. Fix the category if the test contradicts it.
4. Attach your text transcription, both photos, and the measured frequency.
5. Write confirmed or candidate in plain words. An unlabelled guess turns into an accepted fact within a month.

10. Record sheet

PART: __________  DATE: __________  TESTED BY: __________
INVENTREE PK: __________
 
STEP 1  SIZE      L ______  W ______  H ______ mm    Package: ________
STEP 2  TEXT      Line 1: ____________________________
                  Line 2: ____________________________
                  Extras: ____________________________
                  Photos taken:  top [ ]  bottom [ ]
STEP 3  PADS: ____   Pin-1 mark: dot / corner / notch / none
 
STEP 4  CASE CONTINUITY TEST
        Pads that beeped to the lid: ______________
        -> TWO = crystal   ONE = oscillator, that pad is GND   NONE = use fallback
        GND is pad: ____
        Fallback used?  Y / N
 
TYPE:   A (4-pad osc)  /  B (6-pad osc)  /  C (crystal)
 
JIG BUILT AND CONTINUITY-CHECKED:  Y / N
Decoupling cap fitted (oscillators):  Y / N
 
--- CRYSTAL ---
Pierce test oscillated:  Y / N        Frequency: ____________ Hz
22 pF caps fitted / removed: ____________
 
--- OSCILLATOR ---
1.8 V polarity check passed:  Y / N     Current: ______ mA
Working voltage: ______ V               Current: ______ mA
PINOUT:  1:______ 2:______ 3:______ 4:______ 5:______ 6:______
Output type: ______________   Control pad: NC / OE / standby / VC
Frequency: ____________ Hz   Duty: ______ %
High: ______ V   Low: ______ V
(rise/fall not measured - breadboard)
 
IDENTIFICATION
Maker: __________  Series: __________  Part number: ________________
Datasheet: ____________________________
Agreement:  size [ ]  text [ ]  type/pinout [ ]  frequency [ ]
Status:  CONFIRMED / CANDIDATE
InvenTree updated:  Y / N
 
RESULT: PASS / FAIL      SIGNED: ______________

11. If something goes wrong

12. What to buy to do this properly