Oscilloscope vs. probe frequency ratings

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

After a lucky impulse buy of a 100MHz* scope last night, I may as well take advantage of combined shipping and grab some probes of which I am short. I need at least 2 more for my 250MHz TDS485. One would think that 250MHz probes would do the trick, but I don't know what the advertised probe frequency actually means. Does it mean the probe will add no measurable error to 250MHz measurements, or is it a classic -3dB (cutoff) frequency? And, at what input capacitance is this frequency set to? Surely, after adjusting the probe to match the 'scope, this value will change?

For example, if f_probe_cutoff is measured at a standardized 10 pF, but I then add 5 pF to match the scope, does it degrade?

Anyone have some experience in the area? ('Suppose that's most regulars...)

*it can be increased from 50MHz to 100MHz through software.

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

Actually, I worked for Tek, even on probes. But, that was "the olden days" and there was no probe bandwidth spec.

I SUSPECT that this is what is happening. "Good" scope probes have a resistive center conductor in the cable to dampen transmission line reflections. So, that adds a low-pass RC to the probe frequency response.

As a result of this, I SUSPECT that if you put a 100MHz probe on a 500MHz scope, the system bandwidth will be 100MHz (though it might be more complex than this, maybe with a pole, a zero, then the scope bandwidth pole).

And, I SUSPECT that a probe rated at 100MHz on a scope rated at 100MHz will give a system bandwidth of 100MHz but it MIGHT be 100MHz/sqrt(2) (the latter is what you get with a cascade of two equal bandwidths).

This probably does not help a whole lot with all of these "I suspect ...". Tek OUGHT to have something on its website about probe bandwidths. I'd trust that a lot more than me.

Jim

 

Until Black Lives Matter, we do not have "All Lives Matter"!

 

 

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

"Probing High-Speed Digital Designs":
http://www.signalintegrity.com/Pubs/straight/probes.htm
You'll need greater than 250MHz probes if you want a 250MHz bandwidth instrument (scope + probe).
Greatly less expensive and much less capacitive load is a 1K ohm probe for most digital.
http://emcesd.com/1ghzprob.htm
For that URL, I'd replace the 4 200-ohm resistors with 4 one pF (probably) capacitors (an alternate frequency compensation method) since usually don't need to have a 50 ohm source (probe output) impedance for this.
An alternate probe source:
http://www.ezprobe.com/
If you don't need the DC level:
http://www.auburntec.com/
A good read:
http://www.evaluationengineering.com/index.php/solutions/instrumentation/eight-hints-for-better-scope-probing.html

"Dare to be naïve." - Buckminster Fuller

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

I'm not nearly as qualified as the previous 2 responders, but that's never stopped me before. It's easier since I essentially agree.

A 250 MHz probe will be down 3 db @ 250, same goes for the scope. In my simplistic world view, these attenuations add up. So, you won't have the BW that you want.

The tricks mentioned by gchapman probably will work well. Many years ago, when I was young, I used a 'sampling scope' made by Tektronix. It could display repetitive signals into the GHz -- not an easy trick in those days. I think that it had 500 ohm probes. ( It also had other quirks, but it was still pretty amazing).

Jim, I think that they never used to bother with probe BW, because it was well in excess of the scopes at the time. ( At least of my time :) )

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

If You want a higher bandwidth probe place a terminating resistance at the input BNC connector.

Use a transmission line with matched characteristic impedance for that better response.

Typically a 50 ohm termination and 50 ohm coax.

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

Thats where the 500 ohm "probes" mentioned by ford2go come from. 50 ohm terminated system, 450 ohm series resistor at the signal end. Makes 10:1 with a 500 ohm input impedance. They actually work pretty well if you connect it to a circuit sanely.

But, the question at hand is HiZ probe bandwidth spec. The consensus seems to be that the probe bandwidth cascades with the scope bandwidth. The over-all effect is that a cascade of two single-pole rise times makes a composite rise time that is square-root of the sum of the squares of the individual rise times. For a single pole system, 3db bandwidth is related to rise time by risetime * bandwidth = 0.35. Its not a big error to assume the same for two single poles in cascade.

Jim

 

Until Black Lives Matter, we do not have "All Lives Matter"!

 

 

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

A Tek document (one that gchapman linked) agrees with you fellas, in that the frequency rating of a probe (or a 'scope) is indeed its -3dB cutoff frequency. In combination, the rise or fall time is this: t_rise_combination = √(t_scope^2 + t_probe^2). The rise time of a 'scope or probe is approximately 0.35/f, as Jim mentioned. I could measure, with a 250MHz 'scope and probe,

t_rise = √((0.35/(250 MHz))^2+(0.35/(250 MHz))^2) =  1.9799 ns = √(2)⨉0.35/(250 MHz)
f_combined = 0.35 / t_rise_combined = 177 MHz = 250 MHz / √(2)

Mo' bettah:

1/f_combined^2 = 1/f_scope^2 + 1/f_probe^2
...
f_probe = 1/√(1/f_combination^2 - 1/f_scope^2)

If I want to probe a 200 MHz digital signal, I need at least a 333 MHz probe. I've found some pretty good deals on 300 MHz probes, which will give me a 192 MHz system. Good'nuf, and I'll have a basic idea of attenuation near and above that figure. Of course, the final verdict comes after calibration with the internal reference clock. As for noise floor... I'll figure that out when I need to! (Which, according to Murphy, is probably tomorrow.)

Thanks guys 'n gals. :)

About the 50-ohm stuff -- I'm going through alot of this in my microwave class, and see that this is relevant in the GHz range. In fact, this week we'll probe a 500 MHz - 10GHz circuit (we made some lame antenna), so I'll be sure to grill the prof about the 'scope setup, armed with your info.

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

The "500 ohm probe" is a bit marginal. It depends on perfect termination at the far end so that there will NOT be any reflections due to the source mismatch.

You CAN make a double-matched system with 10X attenuation and a probe impedance near 250 ohms. The solution to that one is left to the reader.

Jim

 

Until Black Lives Matter, we do not have "All Lives Matter"!

 

 

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

50 ohm stuff works becuase the input impedance of the oscilloscope is 1 Meg // x pF.

It is the pF that kills the input signal at the BNC connector due to effective source impedance.

BWD was an australian manufacturer of instruments among them scopes. I worked at BWD a lifetime ago when we still made things.

The way that the bandwidth of a scope was measured was simply to terminate the input BNC connector with a 50 ohm feed through termination and drive it with a leveled RF generator and look for a drop of 0.7etc in onscreen signal level.

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

Many of the "modern" high frequency scopes (above 100MHz, mostly) have an internal 50 ohm termination. This carries the 50 ohm environment all the way to the first stage of the vertical amplifier. Much better step response. You also need no feedthrough terminator. The terminator tolerance then also becomes part of the total scope specs.

The same high frequency scopes also pay a lot more attention to getting the input capacitance down. I remember when the standard input C on Tek scopes was 47pf!!! Now, a run of the mill, not so high frequency scope is around 10-12pf. Now, scope designers and users understand the consequences of input C in high bandwidth applications.

Jim

 

Until Black Lives Matter, we do not have "All Lives Matter"!

 

 

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

tlucas wrote:
A Tek document (one that gchapman linked) ...
I wasn't aware of a Tektronix link but here's one I found that has the rise-time formula:
"The Probe"
http://www2.tek.com/cmsreplive/tirep/16581/3GW_24992_0_LR_2010.01.12.16.00.19_16581_EN.pdf

"Dare to be naïve." - Buckminster Fuller

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

tlucas wrote:
About the 50-ohm stuff -- I'm going through alot of this in my microwave class, and see that this is relevant in the GHz range. In fact, this week we'll probe a 500 MHz - 10GHz circuit (we made some lame antenna), so I'll be sure to grill the prof about the 'scope setup, armed with your info.
GHz range - digital gets there fast when its rise-time is short; that's one of Dr. Johnson's driving points.
50-ohm - Tektronix makes two good value low capacitance probes: P6158 (3 GHz bandwidth, 1K ohm load) and P6150 (9 GHz, 500 ohm). One can usually find these used in good condition. Most digital circuits can handle a 1K ohm load.

"Dare to be naïve." - Buckminster Fuller

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

Don't I have to match the probe capacitance with the scope capacitance, regardless of how low I want it to be? Or can the calibration routine take care of that large of a range? For example, the TDS485 scope is 20pF, so I need to adjust the probe's compensation network to add 10 pF to the 10 pF LeCroy PP016, or the attenuation won't be 10X over the full range...

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

tlucas wrote:
Don't I have to match the probe capacitance with the scope capacitance, regardless of how low I want it to be? Or can the calibration routine take care of that large of a range?
Yes and qualified yes (probe spec for its compensation will state the scope's required input capacitance range).
This article starts with general probe information but then becomes specific to 10x 10Meg ohm probes for 1Meg ohm input scopes:
"High-Speed Time-Domain Measurements—Practical Tips for Improvement"
http://www.analog.com/library/analogDialogue/archives/41-03/time_domain.html

"Dare to be naïve." - Buckminster Fuller

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

tlucas wrote:
Don't I have to match the probe capacitance with the scope capacitance, regardless of how low I want it to be? Or can the calibration routine take care of that large of a range? For example, the TDS485 scope is 20pF, so I need to adjust the probe's compensation network to add 10 pF to the 10 pF LeCroy PP016, or the attenuation won't be 10X over the full range...

The probe specifies what kind of scope capacitance range can be compensated by you, turning the variable capacitor screw (with preferably plastic screwdriver to avoid stray capacitance to metal screwdriver). After you have determined your probe and scope do match, then absolute values do not matter and you just turn the screw until square wave looks like square wave.

Nowadays the adjustable compensation capacitance is in parallel with the scope input (on BNC side of the cable) and there is a fixed feedforward capacitance in the probe.

For instance, my scope has 1Mohm +/- 2% with 20pF +/- 3pF in parallel. Not very modern scope.

Right now I have a borrowed 500MHz probe there, that says "Input impedance >2.2M // 12pF for 1M // 6-15pF", so basically the probe is not specified to compensate my 20pF input. However I still can adjust it fine.

So it seems the probe input has 2Mohm in series with input, and I imagine that is bypassed with about 13.33 pF feedforward capacitor. Then there is 286k to groud, creating 10:1 divider with the 1Mohm scope input (2M/9=222k). And the adjustable capacitor is in parallel with scope input capacitance, and the total input capacitance is adjusted to equal 9x13.33 pF = 120pF on scope input, so that both AC and DC see the 10x attenuation. Now if scope range is 6..15pF, the adjustment capacitor (perhaps with the help of a fixed capacitor) needs to do 105..114pF range to make total 120pF load in the scope end.

I must admit I never knew there were 2M2 10:1 input resistance probes, I have always been under the impression they all are with 10M input impedance. Does anyone know why is it like this? Better bandwidth? I see 10Mohm input impedance probes have less bandwidth on the catalogues?

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

Huh - a 500MHz, 10.8pF input, 5-30pF compensation, 10X probe for $65.00 [eBay link]. I am paranoid, as usual (the particularly bad chinglish doesn't help), but they have a decent website: http://www.texas.com.cn/probe04.... . Maybe I'll just have a few beer this afternoon to convince myself to get 2. ;)

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

Remember that a 10pF input capacitance at 500MHz means
an impedance of 32 Ohms. So it's no longer "high impedance". Therefore a simple 500 Ohm probe
with correct cable-terminations may perform better.

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

AFAIK this one http://www.emcesd.com/1ghzprob.htm is the most popular reference for building a DIY 1 GHz 1 k resistive probe. Note: Requires trimming.

Stealing Proteus doesn't make you an engineer.

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

tlucas wrote:
Huh - a 500MHz, 10.8pF input, 5-30pF compensation, 10X probe for $65.00 [eBay link]. I am paranoid, as usual (the particularly bad chinglish doesn't help), but they have a decent website: http://www.texas.com.cn/probe04.... . Maybe I'll just have a few beer this afternoon to convince myself to get 2. ;)
This one looks like (identical?) what you're considering and its from a local company thats a member of the Better Business Bureau:
http://www.tyrosys.com/mm5/merchant.mvc?Screen=PROD&Store_Code=&Product_Code=TEX500RA&Category_Code=PPRA

"Dare to be naïve." - Buckminster Fuller

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

gchapman wrote:
and its from a local company
Last time I checked a map Edmonton wasn't in Texas. It wasn't even in the USA. Maybe tlucas has more current information?

Stealing Proteus doesn't make you an engineer.

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

ArnoldB wrote:
gchapman wrote:
and its from a local company
Last time I checked a map Edmonton wasn't in Texas. It wasn't even in the USA. Maybe tlucas has more current information?
Please pardon me for I can be self-centered ;-)

"Dare to be naïve." - Buckminster Fuller

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

I find this a lil'bit funny: the eBay seller "tube_buyer" is in Hong Kong, the same city as the company that makes the probes, Texas. The 'local' company "tyrosys" is in Texas, importing the probes from Texas, Hong Kong. :D The BBB record does make a bit of a difference - thanks.