Scope measurement question

Go To Last Post
31 posts / 0 new
Author
Message
#1
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I have a Tektronix TDS 2012B (100MHz, 1.0GS/s) and its not clear to me how frequency is calculated in various readout, and the manual isn't much help.

I am measuring what should be an 8Mhz clock signal. On the scope, the trigger frequency displays a rock solid 8.00012 Mhz (measured at 50% on either raising or falling edges). That's great and it seems to indicate that I have the load caps correct, etc.

But when I hit the "MEASURE" button on my scope and watch the frequency of the channel, it fluctuates from 7.991 to 8.009 and everywhere in between.

My guess is that the trigger display is an average. So I changed the channel acquire mode from "sample" to "average 128". After that frequency is more stable, but still moves occasionally between 7.996 and 8.000.

Questions:
1. Do you think my guess about the trigger being an average is correct?
2. Seems odd the the trigger average is above 8.0 while the channel measure average is below 8.0
3. Any clever way to determine if the changing values are due to the measurement equipment or if they are really happening?

-Brad

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Brad -

Don't know about that specific scope so some conjecture.

I would guess that the value provided as "Trigger Frequency" is the average of many samples. Its probably difficult to get that resolution, otherwise.

Part of the sample-sample variation you are seeing is just trigger level uncertainty (aka "noise"). A variation in the sense level (the noise) translates into variation in the time interval. This variation will increase as the slope decreases.

My humble opinion as a long time scope user is that they excel at visualizing time relationships. But, accurate time measurements are less optimum; thats what counters are for.

Jim

Jim Wagner Oregon Research Electronics, Consulting Div. Tangent, OR, USA http://www.orelectronics.net

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I have the 60mhz version of your scope. Jim is correct that the reading can be quite inaccurate, but I have had pretty good results. Set your trigger to auto, and bring the threshhold into an area that gives the most stable result. In the measure area, keep in mind you can use the cursors to get a good idea where you are at.

Keep in Mind Jim is 100% correct that nothing beatsa good Counter

The 'other' Jim

I would rather attempt something great and fail, than attempt nothing and succeed - Fortune Cookie

 

"The critical shortage here is not stuff, but time." - Johan Ekdahl

 

"Step N is required before you can do step N+1!" - ka7ehk

 

"If you want a career with a known path - become an undertaker. Dead people don't sue!" - Kartman

"Why is there a "Highway to Hell" and only a "Stairway to Heaven"? A prediction of the expected traffic load?"  - Lee "theusch"

 

Speak sweetly. It makes your words easier to digest when at a later date you have to eat them ;-)  - Source Unknown

Please Read: Code-of-Conduct

Atmel Studio6.2/AS7, DipTrace, Quartus, MPLAB, RSLogix user

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Hmm what is a counter? Is that something like a logic analyzer?

-Brad

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I also have a TDS2012 Oscilloscope.

When I get home from work tomorrow evening, I'll look at an 8MHz signal on my STK500 and see how stable it is, and post back.

My observation though, is that when looking at 16MHz & 20MHz frequencies, the frequency readout of the TDS2012 oscilloscope is quite stable - much more then I had expected.

You can avoid reality, for a while.  But you can't avoid the consequences of reality! - C.W. Livingston

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

A frequency meter works by either timing the difference between successive edges of the signal - subject to jitter on the triggering point and limited by the resolution of the counter clock - or by counting the number of transitions in a fixed time - usually a multiple or submultiple of a second.

This second version is much more accurate but is slower in its response to changes in the frequency - it might take one or ten seconds to display the current frequency.

Both mechanisms rely on very accurate clocks.

It's instructive to see just how accurate a standard crystal can be, even without special tricks; I have a clock using a 4.096MHz crystal which is accurate to a second a day (one part in 86,400) - the normal sixty-parts-in-a-million would suggest around five seconds a day. If you put it in a temperature controlled oven, it can be *lots* better.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

The triggerfrequency measurement is propably the average of many period timings. The Scop averaging does average amplitude and this way gives only marginal improvements for timing one period.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Is a counter a feature of higher ends scopes or is it a totally different device? I'm pretty sure mine doesn't have that. Although perhaps if the trigger frequency is an average over a long period it is similar. I have noticed that it takes a noticeable amount of time (1s or so) to resolve.

I believe I have a dedicated (powered) oscillator sample that is supposed to be accurate. Maybe I'll stick that on a breadboard alone and see how it looks in my scope.

-Brad

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Brad -

Typically, a "counter" has a precision time base (clock) which sets a count window or "gate time". It then just counts the total number of events in that window. In doing that, it is averaging. Depending on the frequency and the desired resolution, the gate time can be 0.1 sec, 1 sec, 10 sec, etc.

The other way to get frequency is to measure the period of one cycle and compute the reciprocal, which is frequency, This is a non-averaging process but gets the result much faster.

I suspect that this is actually what is going on in the scope. In one mode, it probably is doing the standard counter-like event counting in a gate interval, In the other, it is probably measuring period and displaying the reciprocal.

A dedicated counter lets you vary things like clock rate, gate time, and such so that you can trade response speed, resolution, and so forth. The scope probably does all this under software control, trying for some predetermined "optimum" over which you have no control. On the other hand, most counters have a low impedance (usually 50 ohm) input and you cannot connect them arbitrarily to a circuit and, especially not a microprocessor oscillator.

Hope this helps,
Jim

Jim Wagner Oregon Research Electronics, Consulting Div. Tangent, OR, USA http://www.orelectronics.net

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Thanks for the info Jim.

microcarl wrote:
When I get home from work tomorrow evening, I'll look at an 8MHz signal on my STK500 and see how stable it is, and post back.

What did you find Carl?

-Brad

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

If the scope measurement is after the digitized point and not done on an analog measurement, then you have the normal quantizing error of the scope, the lack of synch of the timebase and the sampling frequency, etc..

I suspect that those could make the measurement jitter as well.

Harvey

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

schickb wrote:
What did you find Carl?

I just looked at a Mega32 on my STK600 running with an external 14.7456MHz crystal. My TDS2012 displays a "Rock Solid " 14.7503MHz oscillator signal at XTAL1 & XTAL2. There isn't even a one least significant digit bobble.

Hope this helps...

You can avoid reality, for a while.  But you can't avoid the consequences of reality! - C.W. Livingston

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

A note on Barnacle's statement about "counting the number of transitions in a fixed time" counters.

--begin longwinded description--

They are actually the worst of the bunch. If the signal is 8MHz, they can give you 8000000 counts per second, or 0.125 ppm resolution with a one second gate time. To do ten times better you have to wait ten times longer. If the signal is 10Hz, you need to wait 100000 seconds to get 1 ppm resolution. Not something you want to do often.

The reciprocal counter, a reduced version of a time interval counter, works the other way around and timestamps edges on the signal. With a 10MHz reference in a cheap counter the resolution is 0.1 ppm for a 1 second gate time, regardless of wether the input is 8MHz or 10Hz.

A TIC+reference like the one I have on my desk uses a time interpolator to timestamp the edges of the signal to ~50ps, and has 0.00005 ppm raw resolution for a 1 second gate time. With the regression function enabled on an 8MHz signal it's 100 times better, or 0.5 in 10^12.
My reference isn't better than 10^-10 though.

--end longwinded description--

A TIC is the tool of choice if you want to measure frequency, stability, jitter, propagation delay, skew, and similar things. Time machines.
Scopes, counters, and cocaine: you don't miss them if you've never tried them.

microcarl: That's 320 ppm off, more than you should be able to get with a terrible crystal and a poor timebase in the scope, which hints that it's showing more digits than it has resolution for.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

@KKP
Assume we had a AVR clocked at 20MHz. The AVR clock
is locked to a reference received on Longwave
(DCF77 or TDF or BBC or similar). Now
we would use the capture-unit to get
timestamps of a (jitter-free) square-wave
with an approx. frequency of f=1 kHz (the
unknown frequency) .

What accuracy can we expect when we have 1
second time. Has anyone built such a TIC
using an AVR ?

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

KKP wrote:
microcarl: That's 320 ppm off, more than you should be able to get with a terrible crystal and a poor timebase in the scope, which hints that it's showing more digits than it has resolution for.

I thought the OP was concerned about the changes from sample to sample, of which, I have none that I can visibly observe.

The fact that I stated one frequency for the crystal and yet another from the actual scope reading, leads me to believe that the crystal frequency is off a bit. When I measure the calibration oscillator on the scope, it's right exactly where it should be, - 1.00000KHz. I have a pretty decent frequency counter that measures that same test signal to be 999.99987Hz, and bobbling +/-0.00001Hz.

I would not be as concerned about any error in frequency from that stated on the crystal. My Oscilloscope reads 14.74503MHz and my Frequency counter reads 14.7450159MHz.

So, where's the problem? There is about 14Hz difference between the two, which between them,, that's a 1PPM difference. So, I think I can say with at least a little confidence, that it is the crystal that is off frequency, and not the Oscilloscope.

I would however, be much more concerned if my frequency counter was stable, and my scope was not - which isn't the case. My scope is "Rock Solid !", as is my frequency counter.

But the jumping of the frequency all over the place of the frequency readings, from sample to sample, that is probably not normal, or acceptable. Again, my Oscilloscope is very stable.

Now, exactly what were you saying???

You can avoid reality, for a while.  But you can't avoid the consequences of reality! - C.W. Livingston

Last Edited: Sun. Apr 13, 2008 - 01:31 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

ossi wrote:
@KKP
Assume we had a AVR clocked at 20MHz. The AVR clock
is locked to a reference received on Longwave
(DCF77 or TDF or BBC or similar). Now
we would use the capture-unit to get
timestamps of a (jitter-free) square-wave
with an approx. frequency of f=1 kHz (the
unknown frequency) .

What accuracy can we expect when we have 1
second time. Has anyone built such a TIC
using an AVR ?

With a time accumulating frequency measurement (or any time based measurement) the accuracy of measurement will depend on the accuracy of the timebase. I.E. the crystal establishing the time reference.

If you want 10PPM, use at least a 10PPM crystal.

If you want 1PPM accuracy, then use a crystal with at least a 1PPM accuracy.

If you need temperature stability, use a crystal that has a temperature controlled environment.

If you need verification of the timebase accuracy, use the highest quality (read most accurate, at least 5PPM) components that you can get, and if you provide a means for making very small adjustments of about +/-10PPM to the oscillator, then simply send the thing out to a calibration shop and have it certified to the National Bureau of Standards (NBS) and get a certificate stating so.

You can avoid reality, for a while.  But you can't avoid the consequences of reality! - C.W. Livingston

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

ossi:
This will be another long one since yes, I have built that:

One AVR is running off a 20MHz oscillator: It controls an oven heater to keep temperature constant, listens to the phase of the DCF77 transmitter, and trims the crystal using a varactor diode. On http://n1.taur.dk/dcf/ you can see the typical daily phase wander of DCF77 as received here, about 4us peak to peak.
For a 24 hour interval that would translate to 4.6*10^-11. We won't get that though.
The limit for such a long interval is the oscillator itself: A cheap AT cut crystal has rather large aging, around 4*10^-9 per day. It is however constant and predictable over a few days, so 4*10^-10 is achievable.
My reference can also run with GPS as input, in which case it does a lot better, more on that if you find it relevant.

Counters:

Your counter would capture 1000 edges with a resolution of 50 ns. So, just taking the first and last timestamp would give you 5*10^-8, 0.05 ppm, 50 ppb. The 50ns is two times the quantization error, and is constant no matter the measurement time.

Implementing a cheap counter with this approach is dead simple; You can use another timer as a prescaler, too.

This would be perfectly adequate for a crystal-referenced frequency meter but you specified something better. So let's improve it a little, there are things we can get for free.

Since the input can be considered a continuously advancing value, you can do regression on it: Do a straight line fit on the 1000 (edgenumber,time) values. This brings down the noise by approximately the square root of the number of samples, 30 times for 1kHz. If the input doesn't cause a sub-50ns beat with the reference, the counter will now have 2*10^-9, 2 ppb, resolutionon a 1kHz input. But if you try to compare one very high performance reference with another you will get such a beat, and resolution will degrade to the original 50 ppb (the readout will start flickering by 50ppb). It would still be a worthwhile use of CPU time since it filters the noise from what you're measuring, will always give some improvement, and requires no additional components.

That design beats all the build-a-counter designs I've found on the net so far.

To fix this quantization beat problem, we need to get rid of the 50 ns quantization error. This either means multi-GHz clocks, or using the A/D converter.
Consider the 50 ns time window where the input signal arrives at the AVR: There is a flipflop inside the AVR, and this flipflop delays the signal by a varying amount, depending on the the instantaneous phase of the AVR clock. We would like to know this amount.

This needs external hardware, unfortunately. Namely a time interpolator circuit.
Using a pair of flip-flops and a few diodes this time difference can be converted to an analog voltage, which we can measure, and use as a compensation value. Even with slow HC logic such a system has better than 1ns performance, bringing it into the system noise, and giving 0.03 ppb, 3*10^-11 performance at 1 second measurement. 11-12 good digits on the display.
Look at http://n1.taur.dk/permanent/frequencymeasurement.pdf page2, figure 2, for the basic interpolator circuit. The document covers my homebuilt, AVR based counter. It has about 25 ps noise, and 12 stable digits.

Last Edited: Sun. Apr 13, 2008 - 02:58 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

microcarl wrote:

Now, exactly what were you saying???

You stated a "Rock Solid " 14.7503MHz oscillator signal at XTAL1 & XTAL2. There isn't even a one least significant digit bobble
That's a rather large deviation from 14.7456, so I found it suspicious, and would have put the counter on it. You did

Quote:
I would not be as concerned about any error in frequency from that stated on the crystal. My Oscilloscope reads 14.74503MHz and my Frequency counter reads 14.7450159MHz.

So I'm saying either your scope's reference is flaky, or you dropped a '4' from the first figure you gave. Probably the latter. That's pretty good for a scope.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

KKP wrote:
You stated a "Rock Solid " 14.7503MHz oscillator signal at XTAL1 & XTAL2. There isn't even a one least significant digit bobble
That's a rather large deviation from 14.7456, so I found it suspicious, and would have put the counter on it. You did

So I'm saying either your scope's reference is flaky, or you dropped a '4' from the first figure you gave. Probably the latter. That's pretty good for a scope.

To the point, the OP stated:

schickb wrote:
But when I hit the "MEASURE" button on my scope and watch the frequency of the channel, it fluctuates from 7.991 to 8.009 and everywhere in between.

So I changed the channel acquire mode from "sample" to "average 128". After that frequency is more stable, but still moves occasionally between 7.996 and 8.000.


I'm not seeing this on my scope. So any argument that you make about resolution, bobble, inaccuracies, etc... is moot!

I have a TDS2012, but the color version, and it's rock solid, where the OP says that his is not! That is the point!

And considering that in the days of old, Oscilloscope measurements were +/-3%, at best, what I have is a far cry better then say, the wonderful TexTronix 465 Oscilloscope I use to use back in the mid to late 1970s.

And as the frequency counter built into the TekTronix TDS2012 is really a "Freebie " I don't think there should be any complaining. But as with the cost of development tool prices dropping to next to nothing, there's continual complaining about today's cost or the lack of functionality that is expected from a $20,000. development system of 20 years ago. So too, the ignorance of the day will complain about even the "Freebies " that they are given.

Case in point... Atmel literally gave us a new and really neat device called the Raven. And what have many of the ignorant individuals visiting this forum done? Well, they bitched because it lacked some pending software, or that it doesn't do exactly what they want it to.

If the TekTronix TDS2012 only has 5 least significant decimal places, Too bad! It was a "Freebie " just as the FFT functions in the TekTronix TDS2012 are free!

I say, just live with it.

And, if someone will settle for an old antiquated Oscilloscope from the 1960s & 1970s, what reason would they have to bitch about the lack of resolution on a "Freebie " frequency counter that is incorporated into an $1,800.00 technological wonder? By 1970s standards, the Tektronix TDS2012 is like comparing a Volkswagen to a Rolls Royce!

It's plain ludicrous...

You can avoid reality, for a while.  But you can't avoid the consequences of reality! - C.W. Livingston

Last Edited: Sun. Apr 13, 2008 - 03:34 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

microcarl wrote:

My TDS2012 displays a "Rock Solid " 14.7503MHz

microcarl wrote:

My Oscilloscope reads 14.74503MHz and my Frequency counter reads 14.7450159MHz.

Carl, you missed 4 in the first post to OP, and KKP commented that your crystal frequency is too much off. After your second post everything is clear.

Pop

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

microcarl wrote:

My TDS2012 displays a "Rock Solid " 14.7503MHz

microcarl wrote:

My Oscilloscope reads 14.74503MHz and my Frequency counter reads 14.7450159MHz.

Carl, you missed 4 in the first post to OP, and KKP commented that your crystal frequency is too much off. After your second post everything is clear.

Pop

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I wonder if the OP set the scope to show only one or two periods of the measured signal or many periods.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

@Kasper
Thanks for the answer. It covered all aspects I had in
mind. I currently operate a 20MHz VCO that I can
lock to various Longwave transmitters. Its all
very simple. On my HP3585A spectrum-Analyser
I clearly can see the phase-noise if I go to
10Hz resolution BW. Unfortunately most LW transmitters
currently do some phase-modulation, sometimes at
rather low frequency. So the loop-bandwidth
probably must be very small to get it filtered.

On the other side I have (at home) no possibility
to really measure the parameters with enough precision.

I never used time-to-voltage or voltage-to-time
translators up to now. I first have seen them
used in the following instrument:

http://www.thinksrs.com/products...

I got two of them (defect) together with schematics.
Very interesting analog+digital electronics.
I think I only understand the simplest of the tricks.
The better ones are probably still buried.

Perhaps a advanced AVR frequency-meter would
really be an interesting project.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

@Carl

Why so grumpy today ??

Like your excellent math explanations , I liked Kaspers explanation. And I do believe I learned a bit here.

/Bingo

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Bingo600 wrote:
Carl, Why so grumpy today??

Well, you've heard of and probably seen "Grumpy Old Men "? The testosterone levels must be low... That, and having a headache for about the fifth day in a row.

But I wasn't trying to act grumpy. We all have our rants.

While I do think the OP either has some problem with his TDS2012, or some misconception about the thing's operation, there is nothing wrong with my Oscilloscope, it's resolution, accuracy, or otherwise. That is it's design limitations, and that doesn't reduce it's quality.

My Oscilloscope is stable, and the OP's isn't. So all of the technical jargon about display resolution, LSD bobble, and the like is only, that... Technical jargon. The complaint was instability of the frequencies being measured and if other TDS2012 Oscilloscopes exhibit the same symptoms. So, while I can't speak for the other 1 or 2 hundred thousand TDS2012 Oscilloscopes out there, I can speak for mine.

And, as for the 14.7456MHz crystal I'm using, and it's being so far off the specified frequency, that particular crystal is 14.74558MHZ in my workhorse STK500. So, all possibilities regarding frequency error seem to point to my new STK600 as having the potential for the real oscillator issue. But even that is moot right now, as I don't need any real frequency accuracy for what I'm doing at the moment. I'm simply working on an data entry method using a fairly expensive 32 position rotary optical encoder made by Gray-hill. The code is a modified version of that which I've posted on another thread somewhere over in the AVR8 forum, I think.

Anyway, if I sounded pissed off, I'm not. It's the Yankee emerging from within me. It's just so hard to suppress at times...

I suppose that if I used those "Smiley Face " thingies, it might lighten things up a bit. But I just don't always think about using them.

You can avoid reality, for a while.  But you can't avoid the consequences of reality! - C.W. Livingston

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

For what it's worth, the BBC's colour subcarrier frequency *used* to be the best freely available reference around since they were all phase and frequency locked to the master rubidium oscillator in Television Centre.

4.4336187500MHz, plus or minus damn all (I forget the number now).

I saw 'was' because now you will get a feed - at best - from a local oscillator in a local production centre since everything is now synchronised digitally (or digitally produced to start with) and the need for the sync is much less. It should still be 4.43361875 +/- 0.025Hz, though.

Neil (don't quote me on this; I'm not allowed to know these things any more :mrgreen: )

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

And in my area (ad probably most part of Germany)
terrestrial TV transmitters went digital: DVBT,
so no longer a plain old TV-signal on the air.

But I think for my purposes DCF77 is surely precise
enough.

If I consider building a "precision AVR frequency meter"
I see two alternatives:

a) Lock the AVR to a reference (DCF77 or similar
transmitter) using a PLL with a (20MHz) VCXO.

b) Clock the AVR by a (eventually low-drift)
XTAL-Oscillator. Receive the reference from the
LW-transmitter (DCF) and use it to determine the
frequency of the XTAL with high accuracy. Use this
as factor to correct the readouts of the AVR
frequency measurements.

What method do you think is better ?

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

ossi:
They are equal, only b) will not provide a precise frequency reference for the other lab equipment and the ham gear. My 10MHz source drives the counters, generators, and the local NTP server.

there is also c) Lock to the PPS signal from a GPS receiver for everyone without a transmitter nearby. It's technically a lot simpler but costs the GPS receiver.

OP:
The trigger-readout is most likely the most correct value. The MEASURE function works with the waveform that is in memory, measures the period, and converts that to frequency. You have limited time resolution (1Gsps=1ns), helped by the slope interpolation wizardry by a factor of maybe 10, so maybe 100ps resolution. That figure would cause the readout to flicker between 124.9/125/125.1 ns, or 8.006 MHz, 8MHz, and 7.994MHz.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

KKP wrote:
OP:
The trigger-readout is most likely the most correct value. The MEASURE function works with the waveform that is in memory, measures the period, and converts that to frequency. You have limited time resolution (1Gsps=1ns), helped by the slope interpolation wizardry by a factor of maybe 10, so maybe 100ps resolution. That figure would cause the readout to flicker between 124.9/125/125.1 ns, or 8.006 MHz, 8MHz, and 7.994MHz.

Thanks, that does match what I see.

microcarl wrote:
I just looked at a Mega32 on my STK600 running with an external 14.7456MHz crystal. My TDS2012 displays a "Rock Solid " 14.7503MHz oscillator signal at XTAL1 & XTAL2. There isn't even a one least significant digit bobble.

Carl since we have similar scopes could you tell me how you get this value from yours? I'm a bit of newbie with scopes... and well electronics in general :) Also, as has been mentioned, I assume you meant 14.74503MHz

-Brad

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

jayjay1974 wrote:
I wonder if the OP set the scope to show only one or two periods of the measured signal or many periods.
I tried both. Didn't make much difference if I remember correctly, but I'll try it again just to be sure.

-Brad

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

schickb wrote:
KKP wrote:
OP:
The trigger-readout is most likely the most correct value. The MEASURE function works with the waveform that is in memory, measures the period, and converts that to frequency. You have limited time resolution (1Gsps=1ns), helped by the slope interpolation wizardry by a factor of maybe 10, so maybe 100ps resolution. That figure would cause the readout to flicker between 124.9/125/125.1 ns, or 8.006 MHz, 8MHz, and 7.994MHz.

Thanks, that does match what I see.

microcarl wrote:
I just looked at a Mega32 on my STK600 running with an external 14.7456MHz crystal. My TDS2012 displays a "Rock Solid " 14.7503MHz oscillator signal at XTAL1 & XTAL2. There isn't even a one least significant digit bobble.

Carl since we have similar scopes could you tell me how you get this value from yours? I'm a bit of newbie with scopes... and well electronics in general :) Also, as has been mentioned, I assume you meant 14.74503MHz

Unfortunately, I'm currently at work. I'll pick this up when I get home this afternoon, at about 3:30pm.

Hang in there...

You can avoid reality, for a while.  But you can't avoid the consequences of reality! - C.W. Livingston