Scope and Probes - Measuring Crystal

[WanaGo]

Hi,

Bit of a noob question, but I am wanting to measure/visualise the oscillations of a number of crystals on some products we have, to try and tune the optimal capacitors for those crystals. Things were designed some years ago, and over the years the suppliers have changed, but the loading caps have never changed, and I am wanting to check to ensure we are using the right caps for the crystals we now use.

I have a Hantek DSO-1152S scope, 150Mhz (hacked to be a DSO-1202S thanks to this forum).
I have the standard 1x/10x probes, and when looking at the output side of the xtal on the 10x setting, it seems to load the crystal somewhat, and instead of showing 12Mhz crystal, it's much lower. On some boards, it's around 460Hz and some are 1.2Khz.
I am guessing this is from the capacitance of the probes?
Will a 100x probe solve this ?
Is there such a thing as a cheap 100x probe? I don't need it for high voltage etc, its purely for low power things like this.

Thanks

[tautech]

I am guessing this is from the capacitance of the probes?

Yep.

Will a 100x probe solve this ?

Partially as there still will be some tip capacitance just less so.

Is there such a thing as a cheap 100x probe? I don't need it for high voltage etc, its purely for low power things like this.

I've got some Pintek 100x 300MHz probes for $75 + GST
http://www.pintek.com.tw/files/pintek/CP-3308R-spec.pdf

Send PM if interested.

[WanaGo]

I am guessing this is from the capacitance of the probes?

Yep.

Will a 100x probe solve this ?

Partially as there still will be some tip capacitance just less so.

Is there such a thing as a cheap 100x probe? I don't need it for high voltage etc, its purely for low power things like this.

I've got some Pintek 100x 300MHz probes for $75 + GST
http://www.pintek.com.tw/files/pintek/CP-3308R-spec.pdf

Send PM if interested.

Thank you.
$75 + GST is in NZD I assume, you are local to me it seems - nice

So if 100:1 is better but still not perfect, what is the ideal solution for this sort of measurement?

Thanks

[KaneTW]

Intuitively, I'd impedance match for the crystal frequency so that the complex impedance component cancels out.

[bdunham7]

Hi,

Bit of a noob question, but I am wanting to measure/visualise the oscillations of a number of crystals on some products we have, to try and tune the optimal capacitors for those crystals. Things were designed some years ago, and over the years the suppliers have changed, but the loading caps have never changed, and I am wanting to check to ensure we are using the right caps for the crystals we now use.

I have a Hantek DSO-1152S scope, 150Mhz (hacked to be a DSO-1202S thanks to this forum).
I have the standard 1x/10x probes, and when looking at the output side of the xtal on the 10x setting, it seems to load the crystal somewhat, and instead of showing 12Mhz crystal, it's much lower. On some boards, it's around 460Hz and some are 1.2Khz.
I am guessing this is from the capacitance of the probes?
Will a 100x probe solve this ?
Is there such a thing as a cheap 100x probe? I don't need it for high voltage etc, its purely for low power things like this.

Thanks

A while back I was fixing something with an intermittent starting crystal oscillator just under 5 MHz.  Granted, my oscillator was weak to begin with, but I couldn't scope it with anything without killing it, even after I had repaired it to work reliably.  Finally, I tried my 100X 100M (not all 100X are 100M...) 4pf probe from eBay--which is surprisingly good-- and it worked, or at least it didn't kill the oscillator.  However, that's nowhere near the goal of not significantly loading the circuit.  I was just trying to tell the difference between working or not. If you want an accurate picture, even just connecting the ground lead will significantly change the circuit, let alone the probe.  To optimize and adjust, you will need a very low capacitance, as in <1pf, active differential probe like the Tek P6247. 

[WanaGo]

Finally, I tried my 100X 100M (not all 100X are 100M...) 4pf probe from eBay--which is surprisingly good-- and it worked, or at least it didn't kill the oscillator.  However, that's nowhere near the goal of not significantly loading the circuit.  I was just trying to tell the difference between working or not. If you want an accurate picture, even just connecting the ground lead will significantly change the circuit, let alone the probe.  To optimize and adjust, you will need a very low capacitance, as in <1pf, active differential probe like the Tek P6247.

Thanks. Yeah, seems it is all down to the capacitance of the probe. The one Tautech offered above looks to be 5.5pF (http://www.pintek.com.tw/files/pintek/CP-3308R-spec.pdf), which is lower than some of the other ones I have seen locally which were all well over 10pF. I cant justify spending big money on an Active probe though, and that Tek P6247 looks like crazy money.

Thanks for your help

[WanaGo]

Intuitively, I'd impedance match for the crystal frequency so that the complex impedance component cancels out.

Sorry can you say that again in layman's terms for me Thanks
Im not quite sure what to do with that information, but thank you

[tautech]

I am guessing this is from the capacitance of the probes?

Yep.

Will a 100x probe solve this ?

Partially as there still will be some tip capacitance just less so.

Is there such a thing as a cheap 100x probe? I don't need it for high voltage etc, its purely for low power things like this.

I've got some Pintek 100x 300MHz probes for $75 + GST
http://www.pintek.com.tw/files/pintek/CP-3308R-spec.pdf

Send PM if interested.

Thank you.
$75 + GST is in NZD I assume, you are local to me it seems - nice

So if 100:1 is better but still not perfect, what is the ideal solution for this sort of measurement?

Thanks

Yep NZD, replied via your email.

Best solution is a active probe......but cost $ $ %

[David Hess]

Is there such a thing as a cheap 100x probe? I don't need it for high voltage etc, its purely for low power things like this.

There is something even better, an RF JFET like a 2N4416.  Design it into the circuit as a source follower to buffer the signal to the oscilloscope probe.  Or build a low capacitance active probe like that shown below.

But better is to measure the signal at the output of the amplifier where a typical probe will have no effect.  Besides frequency, the startup time after power is applied will give some idea of the operating margin.

[vk6zgo]

I've found with discrete crystal oscillators that just resting the probe tip on the outside of the crystal is enough for quite a good display.
This should have very little loading effect ------whether it works with packaged oscillators, I don't know.

In analog stuff, there is usually a "buffer" stage following the crystal, & with digital circuitry, there will sutely be places where the clock, (or some derivative of it) is isolated from the oscillator, & which can be probed without affecting it.

[WanaGo]

These circuits are simple SMT 4 pad 12Mhz crystals with currently 2x 22pF caps to Gnd, and crystal directly into a microcontroller of sorts, that run between 40Mhz and 80Mhz, depending on the product they are in.

Based on the reading I have done, the loading caps (currently 22pF) should be selected based on a formula similar to
Caps = 2x CL - Cstray
Where Cstray is somewhere between 2-5pF, and CL comes from the Crystal datasheet.
I then found variations on this, where its 2x CL - 2x Cstray
And then another variation which ends up much the same as the second one, which is CL = ((Cap x Cap) / (Cap + Cap)) + Cstray

Either way, the current values we have been using I am imagining are on the fridge of being correct now, and are not ideal at all. Assuming those formulas are anything like being correct.

22pF was selected way back in 2010 and the supplier of the crystals has changed a few times, and staff also have changed, so the spec has somewhat slipped.

Looking on the scope, one product the scope loads down to about 400Hz instead of 12Mhz, but the sine wave is beautiful. Another one also loads it down about the same, but the Sine wave is not very Sine wavey at all. Another product it loads it down to 1.2Khz and is also beautiful. So I really just want to see what its like when its running at 12Mhz if I can, to hopefully then be able to judge which is a 'better' cap, if we have a few variants side by side and view them all.

Money is an issue, and so is time, so I would prefer to be able to just get a 100x probe off the shelf and quickly see if we are in the right zone. Active probe is cost-prohibitive, as is building a circuit I think (although this would be neat to try next if the probe doesnt do what I want). Because its all low voltage I deal with, many of these probes seem specific to HV stuff, which I guess is where the $$ comes in.

[bdunham7]

You are saying your 12MHz oscillator is stable at 400Hz with the scope probe connected???   

[vk6zgo]

Looking on the scope, one product the scope loads down to about 400Hz instead of 12Mhz, but the sine wave is beautiful. Another one also loads it down about the same, but the Sine wave is not very Sine wavey at all. Another product it loads it down to 1.2Khz and is also beautiful. So I really just want to see what its like when its running at 12Mhz if I can, to hopefully then be able to judge which is a 'better' cap, if we have a few variants side by side and view them all.

Please tell me that what you wrote there is a typo, & what you meant was " the scope loads down by "400Hz" rather than "to 400Hz"!

And again, " by 1.2kHz", rather than " to 1.2kHz"!

A frequency shift of 400Hz or even 1.2kHz at 12MHz is possible, but if you are seeing an actual signal at 400Hz or 1.2kHz, either the xtal isn't oscillating, & you are seeing some other random signals, or you have major aliasing problems with a DSO.

[WanaGo]

Yep wasnt a typo. Auto freq was showing 400hz and 1.2Khz.
I need to test this again as you are right, that is rubbish.
These are on the standard probes, of which I have no idea what the capacitance is, likely 20-40pF

[WanaGo]

Yeah not sure what happened there, its 12Mhz now on the 10x probe.
I found an older 100x probe which I will test out tomorrow and see how it goes

[MosherIV]

Hi. As vk6zgo said, it is better to measure the buffered output of the xtal oscillator.
Some micros have a dedicated gpio that can be set to output the crystal frequency.

If it does not you just toggle a gpio pin either in code (you will need to work out the number of assembley instructions to work out number clock cycles between each toggle) or use a timer/counter peripheral being clocked by the xtal.

This will of course need some special test code.

[mjs]

Oscillator output is a good place to measure without loading the crystal. I've also used HP 3586C selective level meter to measure oscillators inside encapsulated devices. 32.768kHz RTC crystal can easily be received with a piece of wire or small coil.

[Ice-Tea]

Are you aware your measurements are mostly meaningless anyway as your scope has a Timebase Accuracy of ±50ppm?

In general your better of with a field probe so you can up the frequency without loading the circuit.

[fourfathom]

I use a spectrum analyzer and a little wire coil probe (a few turns about 1/16" diameter) soldered on a SMA connector, or at the end of the coax cable.  Just bring it in close to the xtal traces and I can measure frequency without significantly loading the circuit.

[CDaniel]

12MHz is not that high , I can measure this with no problems using a Siglent SDS1202X-E and 200MHz probes , clean signal on both pins ... That microcontroller oscillator is weak or the quartz needs other capacitors values . Some microcontrollers have  fuse settings for the oscillator proper operation ... Or you should check if your oscilloscope is to blame and the oscillator is actually still running. Even with an old 20MHz analog oscilloscope I don't remember having problems with microcontroller oscillators  These oscillators don't die so easily if you measure the frequency .

[vk6zgo]

Yep wasnt a typo. Auto freq was showing 400hz and 1.2Khz.
I need to test this again as you are right, that is rubbish.
These are on the standard probes, of which I have no idea what the capacitance is, likely 20-40pF

At the risk of sounding "snotty", based on your revealed knowledge, & the probable limitations of your DSO1152s, I don't think you should be contemplating making changes to equipment which is working OK as it is.

Everybody, except bdunham7, read your "400Hz & 1.2kHz" as variations of the nominal 12MHz frequency, because that was the obvious, logical conclusion.
Some still haven't caught on, & are posting good ideas to minimise any "loading" causing such error.

Honestly, you should have immediately realised that a few tens of pF loading could not cause a frequency change of four orders of magnitude, & that your DSO was lying to you!

In fact, nothing that you could do would cause a 12MHz oscillator of any kind to oscillate stably at 400Hz or 1.2KHz.

There are circuits which oscillate at audio & RF frequencies simultaneously, but they are designed that way.
It doesn't "just happen!"

[ogden]

If it is about "tuning" frequency of crystal oscillators, only way of probing is JFET probe. DIY is fine, just do not mess-up input tip capacitance. You *NEVER* shall measure "IN" pin of the oscillator, only "OUT". Use short as possible connection to nearest ground (ground spring clip) to avoid interference pickup by probe and "feedback" into oscillator.

If I would need to check oscillators, I would make special firmware which output XTAL frequency on other than oscillator, pin. If no access to firmware then I would make this probe design:

https://www.instructables.com/id/DIY-1GHz-Active-Probe-for-Under-20/

[vk6zgo]

One way to determine if your crystal oscillator has any discernible frequency change when you probe it with a normal probe is as follows :-

(1) Obtain a HF communications receiver.

(2) Attach a loose wire to the receiver's antenna input, & rest the other end fairly near the oscillator
Note:- this is just an antenna, & is not directly electrically connected to the oscillator wiring.

(3) As radios are highly sensitive, they don't need a lot of pickup, so your "antenna" will normally be far enough away from the oscillator circuitry that it doesn't add any significant loading.

(4) Assuming the OP's 12MHz frequency, tune the radio to that frequency with its mode set  to CW, USB or LSB.
As you approach the frequency, you will hear a high pitched tone, dropping in frequency as you get closer to the oscillator frequency, eventually being so low in frequency as to become inaudible--this is "zero beat"

(5) Now probe your oscillator--- any change in its frequency will be shown by the presence of an audible tone, as the radio will no longer be "zero beat".

[bdunham7]

I think we may be missing the point here.  I think the OP is concerned about the reliability of the oscillator, not its precise frequency.  The load capacitors will change many characteristics, including the voltage across the crystal and the startup time, or if they are off enough, they may prevent reliable startup at all.  I think you would select the capacitors by starting with the datasheets and a little math, then doing final testing with various values and testing both the crystal voltage to make sure it isn't too high and then the startup characteristics over the range of temperatures you expect the device to be subjected to.  If you want to measure and view those things, you need a probe such as the one I mentioned--and even that is going to have at least a small effect. 

This isn't tuning a transmitter.  How accurate do you thing the average internal MCU oscillator circuit is anyhow, even with a proper crystal? 

[Kean]

How about one of these probes?
https://www.ebay.com/itm/332393676082