DIY AC current probe to measure quartz crystal drive level

[mickab]

I'm sharing with you the realization of a very small probe inspired by the CT-6 tektronix in terms of form factor.
The probe was designed to measure crystal drive levels but BTW we can measure anything else like mosfet gate drive current or any type of measurement requiring a really small probe in the <100mA range.
The main goal is to measure the drive level using a current probe like the AN-830 explained Figure 1 :
https://www.renesas.com/en/document/apn/830-quartz-crystal-drive-level?srsltid=AfmBOop9F-hQOzeEjaKozB6p6m6hfijKkWmu828jqtUFeVQD4wPCHuMj

Similar way at Murata :
https://www.murata.com/en-global/products/timingdevice/crystalu/overview/basic/drivelevel

I used a B64290P0036X830 core 5 turns of AWG26, a 49.9R and an SMA. Everything else was made in-house.
The mini_probe_skeleton.zip contain only the CAD file in Solidworks format (.sldprt) of the plastic part. Only an SLA or SLS printer can be used for such a small part like that one. The printer used was an Halot Mage Pro with a layer height of 30um just for information.
The plastic part itself measure 5.6mmx10mm with 0.4mm wall.

Assembly of the probe :


Poting with the same resin as the used one for the printer :


Final assembly :


The probe measures with +/-3% accuracy up to 100MHz and has a bandwidth of ~700MHz (-3dB) with a sensitivity of 4mV / mA.
VNA result:


Doing a rapid test on an EVK.
In this example the crystal was a LFXTAL011301 from IQD FREQUENCY PRODUCTS.
As we can see in the Datasheet the max Drive Level is 500uW and the ESR is 40Ω.
According to the formula give by Murata :
Drive level = I² ･R1 -> (3.521*10-3)² * 40 = 495uW



Some measurements with a scope and a test jig :
Test jig :



C1 : Injected signal to the two 50ohm load (one in termination for the scope and the other one on the test jig itself)
C3 : The raw value (in V) measured by the scope from the current probe
F1 : The C3 value rescale in mA to match the 4mV/mA

[Reminx123]

Hi Mickab,

Did you noticed this video: https://youtu.be/DQ1Gsfw5nk0?si=gG5DOwvwjGlzndx_ ?

Ciao Re

[bson]

Drive level is voltage based though, not current - isn't it?  More specifically, don't you want to check the oscillator input voltage waveform - the undriven side - to make sure it's not flattening at the peaks, which would indicate the crystal is overdriven?  Usually the oscillator datasheet (or MCU etc IC datasheet section) will specify what the input amplitude should be; if too low it's underdriven. I usually do this with a 0.5pF FET probe and a scope.

I'd be curious to know what your probe (or any other small-signal high-bandwidth current probe) adds in terms of trace impedance or loss since it obviously has to EM couple.  The tip capacitance of a FET probe is a well-known quantity, so the amplitude droop it causes can be corrected for, but in general it's small compared to the load + pin capacitance.

[RoGeorge]

I didn't know the current through a quartz crystal can be in the tens of mA range! 

Thanks for posting the measured current screenshot.  Never plugged any numbers to find out the current, and I would have wrongly expect it to be in the range of uA, or maybe less, nA. 

Even for a 32kHz tuning fork (Pmax=1uW / ESR=50k\$\Omega\$), the peak current is in the range of 10uA.

Code: [Select]

$ qalc
> sqrt(1uW/50kohm)*sqrt(2)
  sqrt((1 microwatt) / (50 kiloohms)) × sqrt(2) ≈ 6.324555320 μA
>

Though, some 32kHz resonators datasheets show not 1uW, but 0.1uW and 70k\$\Omega\$ (expected Ipp in the range of ~5uA).

[David Hess]

Jim Williams did the same thing at a much lower power level for a 32kHz crystal.  He needed a preamplifier and bandpass filter to get enough sensitivity.

https://www.analog.com/media/en/technical-documentation/tech-articles/lt-journal-article/LTMag-V17N02-15-SubMicroAmpCurrentMeasurement-Williams.pdf

[peter-h]

Yes; tens of uA, not tens of mA.

For a watch xtal, much less still since the whole RTC chip can be found at 200nA Icc (DS1302) or < 50nA (Artasie chip).

A most impressive current probe design! I remember the 4 digit priced Tek ones I used in the late 1970s.

[MathWizard]

Thanks for posting about this, I thought it was way harder to sense low currents like this. I was looking up DIY current probes for a while, and thought about trying a 'Rogwoski'? coil type, but gave up on that since that's usually for much higher currents. Although I also wanted a current probe for AC power measurements of my computer PSU.

But so these don't work near DC is that it ? But yeah I'll have to order some small ferrite core, and make 1. I didn't realize such a simple circuit for these, gave such useful results.

So why again to big brandname current probes cost so much, even for low BW like upto 100kHz ? Maybe some of them do DC sensing too.

I was thinking of getting a Micsig current probe, but I think the BW was so low, I went off looking into the DIY probes.

I'll have to try this sometime soon.

[SteveThackery]

For a watch xtal, much less still since the whole RTC chip can be found at 200nA Icc (DS1302) or < 50nA (Artasie chip).

It's worth pointing out that 32,768Hz crystals specifically made for watches are different from the ones that intended for connection to a microcontroller. The ones for watches run at 1.5V or less, and their frequency goes way out of spec if driven with 3.3V or 5V.

[voltsandjolts]

It's worth pointing out that 32,768Hz crystals specifically made for watches are different from the ones that intended for connection to a microcontroller. The ones for watches run at 1.5V or less, and their frequency goes way out of spec if driven with 3.3V or 5V.

I'm aware of uW power level limitations to avoid damage to watch/rtc crystals, but haven't seen voltage limits.
Can you point to some datasheets that make this voltage limit clear?
thanks

[RoGeorge]

I don't know if the 32768Hz resonators for MCUs are different (I would guess they are the same).

Fun fact, the 32kHz quartz is, in fact, shaped like a tuning fork tuned on 32768Hz, not a (round) solid block.  You can see one extracted from its enclosure at minute 6:00 in this vodeo by Steve Mould https://youtu.be/_2By2ane2I4?t=360 ).  A good ultrasound microphone may hear a quartz clock (by its 32kHz sound), particularly by conduction.  This trick is sometimes used to fine-tune the exact running frequency for fancy quartz wristwatches (at factory/service) .

Another interesting fact, the tuning fork can have 2 oscillation modes (my guess is this might correspond with the series and the parallel resonance?  ):

In-phase mode of a quartz tuning fork
Castellanos-Gomez Lab


Anti-phase oscillation of a quartz tuning fork
Castellanos-Gomez Lab

[SteveThackery]

It's worth pointing out that 32,768Hz crystals specifically made for watches are different from the ones that intended for connection to a microcontroller. The ones for watches run at 1.5V or less, and their frequency goes way out of spec if driven with 3.3V or 5V.

I'm aware of uW power level limitations to avoid damage to watch/rtc crystals, but haven't seen voltage limits.
Can you point to some datasheets that make this voltage limit clear?
thanks

Actually I cannot. I was repeating something an electronics developer told me - he has made products for use in the horological trade. I've been to RS and checked out their crystals. The watch crystals have a maximum drive level of 1 to 1.5 uW.  A random look at other crystals suggests a maximum drive level of around 1 mW, so a thousand times higher.

I admit that I don't know how the drive power level is related to the operating voltage. I don't know how an oscillator is designed to drive its crystal with a particular power.

I've got a feeling, though, that plugging a 1 uW crystal into an oscillator designed for a 1 mW crystal might end badly.

Clearly I've got a lot to learn. Maybe a new thread is warranted.

[SteveThackery]

A good ultrasound microphone may hear a quartz clock (by its 32kHz sound), particularly by conduction.  This trick is sometimes used to fine-tune the exact running frequency for fancy quartz wristwatches (at factory/service) .

The crystal is easy to hear. For the past few decades watch crystals running slightly faster than 32768Hz have been used, and an electronic circuit drops one of the pulses at preprogrammed intervals. The drop rate is set by cutting a particular combination of tracks on the PCB, which tells the pulse-skipper how often to skip a pulse. The cutting is done by laser.

That's why you can't use the frequency of the quartz crystal as a measure of the watch's timekeeping. You need to know the skip rate as well. The skipping is called "inhibition", and you need to know whether a watch uses it or not when measuring its accuracy.

EDIT: Swatch pioneered a different system. The crystals were laser-trimmed to the desired frequency; the laser actually vapourised minute bits of the tuning fork.