2
\$\begingroup\$

I have a Tektronix P6245 active probe (allows only DC input) and wish to use it to see noise on various power supply rails. I've done this already on a +3.3V rail, and can get sufficient DC offset on the scope (Tek TDS744a) to see millivolts.

However, it gets tricky with higher voltages (within the probe limits). For example, a +12V DC rail is tough. Bringing the vertical resolution to a level that will provide meaningful data is simply not possible.

I'm considering a dc blocking capacitor between the probe and the rail under test. Is this a reasonable option? Any suggestions for best way to do this?

\$\endgroup\$
  • 1
    \$\begingroup\$ I see no reason why you cannot use series cap and R to gnd to monitor with probe. Layout and ground path may compromise its 5GHz bandwidth but no problem here. Suggest NPO ceramic or plastic cap. But also any 10:1 probe with direct connection to tip/ring of probe should be ok with pulse calibration. \$\endgroup\$ – Sunnyskyguy EE75 Mar 15 '17 at 16:39
  • 1
    \$\begingroup\$ What are the relative magnitudes and frequencies of the noise you are interested in probing? \$\endgroup\$ – user49628 Mar 15 '17 at 16:47
  • \$\begingroup\$ The noise I'm interested in might be from 50Hz to 1.5MHz or so. \$\endgroup\$ – Mark Richards Mar 17 '17 at 0:31
2
\$\begingroup\$

Yes, go ahead and use a DC blocking cap. Active probes do have high input impedance, so peculiarities of capacitor actual R-L-C do not matter much. If you need more accuracy along the spectrum, you can always calibrate your tip.

I usually use the standard Tektronix tip (plug), shave the sharp edge flat (with Dremel), and solder a 0603 ceramic capacitor as a new tip. A 0.1uF 25V MLCC cap should give you a bandwidth from about 1.6Hz and up.

CAUTION: However, be aware that the single-ended probe must have good grounding, so be careful not to pick up noise on the ground lead over the air, or from improper grounding point. Instead of long lead, use a pogo-pin (spring-loaded) plug into ground receptacle of the P6245 probe, and have an open ground surface near your point of measurement. You will need to properly align both tips to make good connection.

\$\endgroup\$
  • \$\begingroup\$ I think this is a very reasonable suggestion. The reason I want to use the P6245 is simply because it's virtually noise-free. Perhaps this is the result of its 50 ohm termination. Whatever. Will give this suggestion a try. Cheaper than a differential probe or made-for-purpose system! \$\endgroup\$ – Mark Richards Mar 17 '17 at 0:37
2
\$\begingroup\$

Options:

  1. Money no object? R&S has a new probe specifically for this application, with separate AC and DC signal paths so that large offsets can be accomodated while measuring small AC signals. The cost of a new R&S scope to use it with may be prohibitive, though. I'd be surprised if Tek, Keysight, LeCroy, etc., don't come up with similar probes in the near future.

  2. Use a different probe that allows for ac-coupling.

  3. Forget the fancy probe and just contact the power rail through a 50-ohm resistor with the center conductor of a chopped off piece of coax (and contact the o.c. to ground nearby), and ac-couple the other end to your 'scope.

\$\endgroup\$
  • 1
    \$\begingroup\$ +1 but please terminate your coax ;) \$\endgroup\$ – peufeu Mar 15 '17 at 16:49
1
\$\begingroup\$

Your probe seems to have a gain of x1, therefore it provides only a dubious benefit in this case.

Since a power supply has low impedance (hopefully it has decoupling capacitors!) the easiest (and also best) way to measure its noise is to solder a 50 ohms resistor to your test point, and connect this to the scope using a piece of 50 ohms coax with a BNC at the end.

Set scope to AC coupling. You don't need to enable internal 50R termination, since you have 50R source termination.

Soldering the coax frees up your hand to perform other tests, if you want to measure your supply noise with your microcontroller/board/DSP/whatever doing various things, for example.

I like MELF resistors for this purpose, because they are small, have good HF behavior, and unlike SMD chip resistors, won't break when the cable pulls slightly.

This will give you the best measurement possible.

The place you solder the coax ground to is also important.

\$\endgroup\$
  • 1
    \$\begingroup\$ A false suggestion! Only the termination at the scope end prevents the reflection. You may think "Who cares, it will be killed at the source termination!" True, but it still exists for the scope and it's frequency dependent. \$\endgroup\$ – user287001 Mar 15 '17 at 16:42
  • \$\begingroup\$ Yes, there will be reflection at the scope. Not really a problem, though. If you want a little more noise and this keeps you up at night, be my guest and enable scope 50R termination... and correct results for attenuation of course. \$\endgroup\$ – peufeu Mar 15 '17 at 16:47
  • 1
    \$\begingroup\$ nothing special , only -3dB at 60MHz, -6dB at 100MHz, when there's an ideal 50Ohm coax Td=5ns and scope input is usual 1MOhm 50pF. Eat well! \$\endgroup\$ – user287001 Mar 15 '17 at 17:16
  • \$\begingroup\$ I'd worry measuring a +12V rail with a 50 ohm termination. Bye bye scope? \$\endgroup\$ – Mark Richards Mar 17 '17 at 0:34
0
\$\begingroup\$

So the best way would be to just use a normal probe lower bandwidth probe (if you don't have one maybe you should get one) and run the probe in AC mode at a higher gain than 1x and take out the hassle of adding capacitors.

I also find it hard to believe that you can't set the scope to AC coupling.

\$\endgroup\$
  • \$\begingroup\$ You're adding series capacitance, not shunting capacitance. You're sure that's going to affect the bandwidth the way you're saying? \$\endgroup\$ – The Photon Mar 15 '17 at 16:23
  • \$\begingroup\$ No I'm not sure, faulty logic on my part. After looking at it it only messes up the bandwidth around the 1GHz point by creating a resonance point. \$\endgroup\$ – Voltage Spike Mar 15 '17 at 16:40
  • \$\begingroup\$ Active probes (as P6245) are DC amplifiers with limited (+-8V) input DC range relative to ground. They are terminated with 50-Ohm load at the scope end, and AC coupling cannot help. The P6245, however, has 10V offset range, so measuring 12V rail will end with 2V DC, and the amplification of only 500mV/div will bring you out of dynamic range, which might be not enough to see noise details. \$\endgroup\$ – Ale..chenski Mar 16 '17 at 5:23

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.