Erroneus readings with function gen and accessories

[mcinque]

Hi to all,

this is my first post, I'm glad to be part of this forum.
Thank you Dave for giving me many fun and interesting videos!

My English is not so good, so please try to tolerate me  .

I'm quite new to istruments like function generators and DSOs, so I would like to receive some advices to understand better what's goin'on on my readings.

I've set my function generator (DG1022) to 1/3MHz 5Vpp Square Wave and start reading that signal with my DSO (DS1052E) with band limit disabled, using its probe (1X).

I noticed that :

if I read the signal directly on the output BNC of the function gen, I read a clean wave
if I read the signal at the end of an BNC cable (50Ohm) 1 meter long, I read a little overshoot
if I read the signal with the BNC cable and a Pomona grabber I read a high overshoot (and undershoot on 3MHz)

Are this readings "normal"? I mean, can a good and short BNC cable or a little grabber distort so much a square wave?

Most important: do you have any book to suggest about function generators and waveforms related measurements?

I appreciate any suggestion that make me understand what's goin'on! 

[staze]

What you're seeing is "normal". If you want to use just BNC into the scope, you'll need a 50ohm thru terminator. I would guess that if you zoomed in on that overshoot, you'd see "ringing".

That said, the DG1022 looks like it has the ability to set the input impedance of the "LUT". Looks like you do that under Mode/Functions (on the DG1022), then your channel you're outputting from. Set it to High-Z. That SHOULD resolve the ringing.

As a note, I'd set your probe to 10X and leave it there. There are very few times where you actually need to use 1X on probes. Just make sure to tell the DS1052 that the probe is 10X, and also make sure to set the compensation if you haven't already.

Good luck, and let us know!

[c4757p]

I would guess that if you zoomed in on that overshoot, you'd see "ringing".

It's quite visible in the third picture.

By the way, I like your diagram. Very nice, clear way to convey your setup leaving nothing to the imagination. If there's anything wrong with your English (there isn't), the picture more than compensates.

[staze]

By the way, I like your diagram. Very nice, clear way to convey your setup leaving nothing to the imagination. If there's anything wrong with your English (there isn't), the picture more than compensates.

Agreed!

[mcinque]

Thank you staze and c4757p for your replies.

The readings in the picture was effectively taken with 50Ohm impedence set on function generator, and since I've got the same results even with BNC cable plugged in directly into the input of the DSO, I guess this can be caused by the BNC cable and grabber capacitance.

By the way, when I'm injecting a signal into a circuit and take a reading,  I wonder how can I distinguish those artifacts from those caused by circuit issues!

The only solution that came to my mind is to take a snapshot of the reading without the circuit in between the gen and the dso, put it on the screen as reference and only after this take the reading, comparing the differences.

A sort of poor man's REL button available on most DMM 

What do you think about it?

p.s. Thank you for the suggestion on the 10X probe setting; if you have time to dedicate me, I'd like to understand in wich rare situations is required the 1X settings.

[c4757p]

The readings in the picture was effectively taken with 50Ohm impedence set on function generator, and since I've got the same results even with BNC cable plugged in directly into the input of the DSO, I guess this can be caused by the BNC cable and grabber capacitance.

The impedance must match - 50 ohm on the generator, 50 ohm terminator at the scope end of the cable - to correct the ringing. You are correct that it has to do with the cable capacitance (and inductance) - since you can't eliminate them completely, you have to work with them instead of against them at higher frequencies; that's what terminators are for. The capacitance and inductance of the cable form its characteristic impedance, and if you match that at both ends, the frequency response becomes (ideally) flat and the ringing goes away.

By the way, when I'm injecting a signal into a circuit and take a reading,  I wonder how can I distinguish those artifacts from those caused by circuit issues!

Intuition and careful probing. If you're injecting the signal into the circuit, try placing a 50 ohm resistor (47 will do) across the cable right where you connect to the circuit. And watch out - if the generator expects a 50 ohm terminator and doesn't get one, the signal amplitude will double, which could easily fry a digital circuit.

The probe itself can also ring - in fact, above a couple MHz, it will. This is caused by the length of the ground lead forming extra inductance. If you need to probe high-speed circuits, remove the ground lead. There's a ground ring on the probe right by the tip - you can attach a very small "spring" ground to this (make it with a piece of stiff wire if you must; keep it less than about 2cm long), and ground it to a point very close to where you are probing. At very high frequencies, it can be useful to build test points with convenient grounds into the circuit itself. At ridiculously high frequencies, you might even include a small RF connector on the board so you can use a coax cable directly without a probe... But if you're anywhere near that domain, you'll know it!

p.s. Thank you for the suggestion on the 10X probe setting; if you have time to dedicate me, I'd like to understand in wich rare situations is required the 1X settings.

Small signals, where the 10X attenuation makes the signal too small to see clearly on the screen.

If you didn't know, the probe's bandwidth becomes almost ridiculously low on 1X:

[staze]

The readings in the picture was effectively taken with 50Ohm impedence set on function generator, and since I've got the same results even with BNC cable plugged in directly into the input of the DSO, I guess this can be caused by the BNC cable and grabber capacitance.

By the way, when I'm injecting a signal into a circuit and take a reading,  I wonder how can I distinguish those artifacts from those caused by circuit issues!

You'll want to set your Function Gen to High-Z when you apply to a circuit unless you know the circuit has a 50ohm impedance (like c4757p said). Your scope, for example, has a 1M input impedance (higher end scopes, which I don't have, have this settable...)

By and large, my FG sits at the High-Z setting, but it's really going to depend on what you're doing with it.

Good luck!

[c4757p]

Your scope, for example, has a 1M input impedance (higher end scopes, which I don't have, have this settable...)

It's amazing what gets left out sometimes, though, especially on older equipment. I've got a very nice 400 MHz / 2GSa/s Tektronix DSO from the 90s that omits the internal 50 ohm setting (though my 150 MHz analog has it), and a nice HP synthesized (all-digital control, not direct digital synthesis) function generator from the 80s that omits the Hi-Z setting. The function generator one is a bit more pathetic, since that would be just a software option...

Come to think about it, sometime if I'm up for a bit of firmware hacking I might see if I can add such a feature to it. I haven't played with a ROM dump in a while....

[Mechatrommer]

http://www.ece.vt.edu/cel/docs/TekProbeCircuits.pdf
http://cds.linear.com/docs/en/application-note/an47fa.pdf

[alm]

Note that in all signal sources that I know, the hi-Z setting does nothing but adjust the signal amplitude to be correct for an unloaded output. If set to 1 V in 50 Ohm mode, it will output 2 V because half of it will be dropped across its 50 Ohm output impedance. Setting the generator to hi-Z does not set its output impedance to hi-Z (whatever that's supposed to mean) or back terminate the output, and does nothing to improve signal integrity.

A BNC cable directly connected to a hi-Z input (eg. scope input or probe) represents an unterminated transmission. High frequency signals will be reflected at the end of the cable and create artifacts (eg. standing waves). A square wave with a fast edge rate will contain high frequency components. I'm sure there are some nice videos that demonstrate the behavior of terminated and unterminated transmission line (the AT&T wave behavior one?). The longer this unterminated stub is, the stronger the artifacts. By terminating the cable in 50 ohms as close as possible to the scope or probe, you convert it to a well-behaved terminated transmission line with only a centimeter or so of unterminated stub (from  terminator to scope input). For calibration, Tektronix even made BNC to probe tip adapters with a 50 Ohm terminator built in, so the probe tip could be right at the termination resistor.

Scope probes employ black magic (eg. see Dave's video about scope probe cables that c4757p quoted) like resistive coax to reduce these effects. A scope probe, even a 1x probe, is very different from a piece of RG174 cable with a resistor at the end.

This is not so much related to function generators, but more to transmission lines in general.

[staze]

Note that in all signal sources that I know, the hi-Z setting does nothing but adjust the signal amplitude to be correct for an unloaded output. If set to 1 V in 50 Ohm mode, it will output 2 V because half of it will be dropped across its 50 Ohm output impedance. Setting the generator to hi-Z does not set its output impedance to hi-Z (whatever that's supposed to mean) or back terminate the output, and does nothing to improve signal integrity.

A BNC cable directly connected to a hi-Z input (eg. scope input or probe) represents an unterminated transmission. High frequency signals will be reflected at the end of the cable and create artifacts (eg. standing waves). A square wave with a fast edge rate will contain high frequency components. I'm sure there are some nice videos that demonstrate the behavior of terminated and unterminated transmission line (the AT&T wave behavior one?). The longer this unterminated stub is, the stronger the artifacts. By terminating the cable in 50 ohms as close as possible to the scope or probe, you convert it to a well-behaved terminated transmission line with only a centimeter or so of unterminated stub (from  terminator to scope input). For calibration, Tektronix even made BNC to probe tip adapters with a 50 Ohm terminator built in, so the probe tip could be right at the termination resistor.

Scope probes employ black magic (eg. see Dave's video about scope probe cables that c4757p quoted) like resistive coax to reduce these effects. A scope probe, even a 1x probe, is very different from a piece of RG174 cable with a resistor at the end.

This is not so much related to function generators, but more to transmission lines in general.

alm,

Great explanation. You really are better off with a thru terminator, or at least a Tee and a 50ohm end terminator if you're going to feed the scope directly (or a scope that has a 50ohm termination option)...

[mcinque]

Note that in all signal sources that I know, the hi-Z setting does nothing but adjust the signal amplitude to be correct for an unloaded output. If set to 1 V in 50 Ohm mode, it will output 2 V because half of it will be dropped across its 50 Ohm output impedance. Setting the generator to hi-Z does not set its output impedance to hi-Z (whatever that's supposed to mean) or back terminate the output, and does nothing to improve signal integrity.

A BNC cable directly connected to a hi-Z input (eg. scope input or probe) represents an unterminated transmission. High frequency signals will be reflected at the end of the cable and create artifacts (eg. standing waves).

Damn, you're right. How can I forget when I've made my first LAN about 20 years ago with BNC, RG58 and 50Ohm terminators? 

http://www.ece.vt.edu/cel/docs/TekProbeCircuits.pdf
http://cds.linear.com/docs/en/application-note/an47fa.pdf

Thank you so much!! 

Intuition and careful probing. If you're injecting the signal into the circuit, try placing a 50 ohm resistor (47 will do) across the cable right where you connect to the circuit. And watch out - if the generator expects a 50 ohm terminator and doesn't get one, the signal amplitude will double, which could easily fry a digital circuit.
If you didn't know, the probe's bandwidth becomes almost ridiculously low on 1X:

Thank you again!