How the probe of your scope can fool you (what the manuals do not say)

Hi all, probably a lot of you already know this, but may be it can help others…
The story starts during some tests on an Armega16 with an Arduino bootloader.
I wanted generate a clock output for others devices, so I made a simple sketch to generate it using Timer2 (see attached file Atmega16_test_1.ino) and output a 4MHz clock.

Well, I decided to see that clock with my scope:

TEST1.jpg (95.6 KiB) Viewed 824 times

attached file Atmega16_test_1.ino

(continued on next post)

In this book is clearly explained how the common standard 10:1 passive probe acts as an LCR circuit when used with “fast” digital signals and resonates due the input capacitance and the ground wire inductance at the high frequency components of the digital signal spectrum.
So the Atmega16 was enough fast to do that, and the standard 10:1 passive probe was the worst thing to use for “fast” digital signals measurement…

So I did an other measure using a more short ground wire to see any improvement:

TEST2.jpg (91.07 KiB) Viewed 823 times

Here is my homemade 21:1 low impedance probe:

PROBE1.jpg (75.74 KiB) Viewed 822 times

Well, now the question is… can I trust that..?

The only way is to use an active probe. A current active probe costs from about 1.500$ plus a new scope that supports it (> 3000$)….. a little bit expensive….

So after some research a found that it was possible to buy on ebay an 80’s active probe like the Tektronix P6202.
In particular the P6202 is a 500Mhz fet active probe with a provided power supply and can be used with any scope (note that the P6202A doesn’t have the power supply because it was made for Tektronix scopes that can power it).

So I bought one from ebay for about 100€:

PROBE2.jpg (79.49 KiB) Viewed 821 times

Not sure if I have linked this here before…
http://www.dfad.com.au/links/THE%20SECR … 0OCt09.pdf
… but it is well worth reading.

The best probe out of your collection above is the passive probe with 1k resistor.
With low L resistor and good 50ohm cable it works flat till 2GHz usually.
I doubt you will get better results with an active 500MHz probe for that specific purpose..
PS: When building the 1k probe – mind the most “o’scope probe cables” are not 50ohm, their inner wire is a “resistor” (around 200ohm).
Messing with above “1k’ probe and atmega (simulations, pictures, results):
http://forum.arduino.cc/index.php?topic=220942.0

Thanks for the links…

Pito wrote:
PS: When building the 1k probe – mind the most “o’scope probe cables” are not 50ohm, their inner wire is a “resistor” (around 200ohm).

To add to this interesting topic – the output impedance of the logic outputs on most ucontrollers (5V or 3.3V Vcc) is something like 150-250ohm (depending on the output driver settings when applicable). The 1Meg or 10Meg probes are not the best one for such measurements therefore..
PS: BTW: an other earthquake this early morning… this thing is becoming a little bit nasty this year…
Africa is pushing itself hard to Europe’s kitchen..

Pito wrote:To add to this interesting topic – the output impedance of the logic outputs on most ucontrollers (5V or 3.3V Vcc) is something like 150-250ohm (depending on the output driver settings when applicable). The 1Meg or 10Meg probes are not the best one for such measurements therefore..
PS: BTW: an other earthquake this early morning… this thing is becoming a little bit nasty this year…
Africa is pushing itself hard to Europe’s kitchen..

It is not about billion years, but rather a few million years..
The Alps were ~8-10km tall few mil. years back (as the Africa pushed it up), the most countryside around Alps (ie Bavaria, Austria, Italy, Swiss) is made of material eroded from the Alps heights, somewhere 2-3km deep layer. First the Gibraltar will close itself, Mediterranean Sea evaporates (as it was a desert only few million years ago)..

I found a strange spring in a set of probes. Accident? – I don’t think!

IMG_0066.JPG (103.34 KiB) Viewed 203 times

That short spring will decrease the inductance of the flying ground wire (the ringing will be smaller), but anyhow the 10:1 probe (or similar high impedance probe) is not well suited for high speed/broadbanded measurements (ie fast rising/falling edges).