Author Topic: Scope rising time  (Read 29953 times)

0 Members and 1 Guest are viewing this topic.

Offline Technojunk

  • Contributor
  • Posts: 15
Scope rising time
« on: May 24, 2010, 09:54:57 pm »
Hi folks,

I was watching Dave's blog about the rigol 100MHz hack, chekking the rise time of the signal.
Now I was wondering, what kind of signal do I have to put in, to check if my scope is really 100MHz?

0.35 / Trise = bandwith, but if I put an signal in of 100Khz the bandwith will be different with a 20Mhz signal in!

Thanks allot.
 

alm

  • Guest
Re: Scope rising time
« Reply #1 on: May 24, 2010, 10:21:53 pm »
The rise time should be much faster that the rise time of the scope, say a 1ns or less (for a 3.5ns scope) step of known shape. You can make do with a known risetime in the same range (if you feed it a 3.5ns signal, the measured 10-90 rise time should be about 5ns, since they add as the square root of the sum of squares). The rise time is independent of the frequency (although it will always be lower than half the period), you can have a 1kHz signal with a really fast rise time. Note that the rise time and aberrations of the probe (if you use it) are added to those of the signal and the scope, so you should really use a 50ohm BNC-BNC coax cable and a 50ohm feed-through terminator right on the scope BNC connector (assuming you use a 50ohm source).
 

Offline chscholz

  • Regular Contributor
  • *
  • Posts: 85
Re: Scope rising time
« Reply #2 on: May 25, 2010, 02:27:46 am »
What Alm said...

You need to be a bit careful: the BW*t_r product of modern digital storage oscilloscope can significantly deviate from 0.35. As a general rule, the higher the bandwidth the higher the BW*t_r product tends to be.
It is always a good idea to check the manufacturer's spec sheet.

Chris
 

Offline saturation

  • Super Contributor
  • ***
  • Posts: 4788
  • Country: us
  • Doveryai, no proveryai
    • NIST
Re: Scope rising time
« Reply #3 on: May 25, 2010, 10:32:47 am »
The rise time of a system is a constant, practically, regardless of frequency. But you need a a square wave to do the test.  Any frequency will do and you simply stretch the time base to find the rise time of the step.  Since you can discern the 7th harmonic sine wave component of sq wave, to see the effect of 100MHz easily try ~ 100/7 = 12 MHz square will start to push 100MHz limits on the scope.

Chris is right, but these calculations are only estimates.  

http://www.tek.com/Measurement/scopes/selection/performance/vertical_sys.html

Aside, be aware of the limitations of your frequency generator and your scope probes when making the measurements, what you see as its bandwidth maybe the limitations of these items.



Hi folks,

I was watching Dave's blog about the rigol 100MHz hack, chekking the rise time of the signal.
Now I was wondering, what kind of signal do I have to put in, to check if my scope is really 100MHz?

0.35 / Trise = bandwith, but if I put an signal in of 100Khz the bandwith will be different with a 20Mhz signal in!

Thanks allot.
« Last Edit: May 25, 2010, 10:35:27 am by saturation »
Best Wishes,

 Saturation
 

alm

  • Guest
Re: Scope rising time
« Reply #4 on: May 25, 2010, 05:42:34 pm »
It's true that any ideal square wave would do, but I'd be very careful to estimate the rise time purely based on repetition frequency. Depending on the quality of the generator, a square wave near the top of the frequency range may be almost a sine, so you'd be lucky to even find a third harmonic of any significance. Most generators will specify the rise time, which is indeed usually independent of the output frequency.
 

Offline Technojunk

  • Contributor
  • Posts: 15
Re: Scope rising time
« Reply #5 on: May 26, 2010, 10:23:24 am »
Well, Here a measurement with a signal of 10Vtt, @ 100Khz.
                   

If i do the math: 0.35 / 3.5E-7 = 1 000 000 , so its says my scope is 1Mhz?
 

Offline saturation

  • Super Contributor
  • ***
  • Posts: 4788
  • Country: us
  • Doveryai, no proveryai
    • NIST
Re: Scope rising time
« Reply #6 on: May 26, 2010, 11:53:18 am »
Hi, thanks for posting the image, it says a lot.

Math is right, but your measurement is off.  It looks like this scope can do this measurement with ease.
If those settings are right, it says its 50nS per division horizontal sweep and 2V p-p.

You do not have it zeroed.  Your waveform is 1.5 graticules below the zero mark.  Make sure the p-p reach from 0 to 100% on this scope.

Those vertical bars, seem to be movable.  The rise time is between the 10% and 90% marks; align one vertical bar at the 10 and 90% mark.

Your trace is either out of focus or its near the limits of its frequency response: its a bit fat or its jittering, thus it looks fat.

I eyeballed that waveform, and with your equation, the rise time gives a frequency response ~ 2 MHz.  The roll off near the top of the wave suggests its near the limits of its response.

Now, that is the rise time of the worst case element in your test set up, it could be the probes, the signal generator, or the scope. 




Enjoy.



Well, Here a measurement with a signal of 10Vtt, @ 100Khz.
                   

If i do the math: 0.35 / 3.5E-7 = 1 000 000 , so its says my scope is 1Mhz?
Best Wishes,

 Saturation
 

Offline tecman

  • Frequent Contributor
  • **
  • Posts: 434
  • Country: us
Re: Scope rising time
« Reply #7 on: May 26, 2010, 02:05:11 pm »
I still think your source is the issue.  Without a good pulse generator, into the correct impedance, you are likely to not have a good reference source.  Even HCT logic gates at 5 volts are in the 20 ns time range.  ECL logic offers the best rise/fall times (< 5 ns) but harder to find and apply.

Early pulse generators used a special mercury relay to generate pulses.  Other schemes use spark gaps.  You could estimate using a 100 mHz sine generator, if it has a leveled output, but I suspect that the source you are measuring is just a lot slower than your scope.

Paul
 

alm

  • Guest
Re: Scope rising time
« Reply #8 on: May 26, 2010, 04:34:19 pm »
It's actually the rise time of your total system, sqrt(t_r(source)^2+t_r(probe)^2+t_r(scope)^2). In this case, it's likely dominated by the source. You don't mention the source, if it's a function generator, the rise time is usually specified. I wouldn't count on a <50MHz or so function generator to have a fast rise time, you might have better luck with a dedicated pulse generator, or you can built something yourself. The avalanche pulser from Jim Williams is a bit hard to use if you don't know the real output amplitude (hard to measure without a really fast scope), but I think there's a version of that design with a charge line which delivers a longer pulse. Something like 1ns might just be achievable with fast logic gates or something discrete. Even then, it's hard to verify a system with a signal generator of unknown specifications.

A leveled signal source with a bandwidth from 50kHz or so to >100MHz is also a good way, you use it to measure the -3dB point directly (increase the frequency until it's just 1/sqrt(2) times the amplitude at some low frequency like 50kHz).
 

Offline Technojunk

  • Contributor
  • Posts: 15
Re: Scope rising time
« Reply #9 on: May 26, 2010, 05:20:52 pm »
I'm using a home-build frequency generator, based on an XR2206 article here: www.technojunk.nl/files/84111NL.PDF [14MB!]
@saturation:
The fatness is becouse the delayed timebase, and may not focused right.

@alm:
Using a piece of coax with propper BNC connectors, so no probes.


EDIT:
With my probe on the CAL signal on the scope gives the same result: ~350nS risetime!
« Last Edit: May 26, 2010, 05:49:03 pm by Technojunk »
 

Offline tecman

  • Frequent Contributor
  • **
  • Posts: 434
  • Country: us
Re: Scope rising time
« Reply #10 on: May 26, 2010, 07:21:14 pm »
Specs on the XR 2206 chip for a square wave output is 250 ns.  If you have the supporting circuity to support that sort of bandwidth, the best you will get is 250 ns.

As I mentioned before, the generator is your problem.

Paul
 

Offline Technojunk

  • Contributor
  • Posts: 15
Re: Scope rising time
« Reply #11 on: May 26, 2010, 07:26:20 pm »
Thanks Paul for the quik research!
What could I bult, to do a propper measure? something with a square signal, with 500pS ?
 

Offline Matt

  • Contributor
  • Posts: 19
Re: Scope rising time
« Reply #12 on: May 26, 2010, 07:36:46 pm »
Seen this referenced a couple places:  http://www.i9t.net/fast-pulse/fast-pulse.html

Haven't really looked at it, i just bookmarked it to look into it at a later date.

-Matt
 

Offline Technojunk

  • Contributor
  • Posts: 15
Re: Scope rising time
« Reply #13 on: May 26, 2010, 08:17:10 pm »
Thanks!
What is the best way to connect this generator to my scope?

BNC output from generator to scope, 50R resistor from signal to ground on both sides?
[not really in to terminators etc)
 

alm

  • Guest
Re: Scope rising time
« Reply #14 on: May 26, 2010, 09:23:15 pm »
That was the Jim Williams design that I mentioned. The issue with that design is that if the rise/fall time is faster than the system rise time of your scope, you can't determine the actual amplitude, so you can't determine the 10%-90% points (rise times are usually specified between 10% of the amplitude and 90% of the amplitude). He did publish an improved version with a charge line, but this one is more complex. For both designs, you're flying blind without a sampling scope to tune it (which he did), especially for those 'adjust for best pulse shape' components.

The source should have a 50ohm output amplitude (I think the Jim Williams design does, and most commercial sources do too), and you should terminate it with 50ohm on the scope end. If the scope has a 50 ohm setting, that's the best option (probably not for a 100MHz scope), second best is a feed-through terminator (basically a BNC male-female coupler with a termination resistor in the middle), and third best is a BNC T-adapter with a standard terminator on one side. Make sure to use good quality BNC hardware, not the cheap crap designed for 10base2. The Jim Williams design has harmonics up to 1GHz or so, so you're dealing with pretty high frequency stuff. I don't think you'd get an acceptable performance with soldering a resistor to a coax cable.
 

Offline tecman

  • Frequent Contributor
  • **
  • Posts: 434
  • Country: us
Re: Scope rising time
« Reply #15 on: May 26, 2010, 10:40:58 pm »
Thanks!
What is the best way to connect this generator to my scope?

BNC output from generator to scope, 50R resistor from signal to ground on both sides?
[not really in to terminators etc)

Source has the 50 ohm.  50 ohm coax cable to the scope with a 50 ohm terminator (it should be a BNC coaxial terminator).  The 50 ohms at the scope is critical.  If you don't use a proper terminator you may see ringing.

paul
 

Offline Technojunk

  • Contributor
  • Posts: 15
Re: Scope rising time
« Reply #16 on: June 18, 2010, 09:35:09 pm »
So, i build this: http://www.i9t.net/fast-pulse/fast-pulse.html
Only the down part, 2N2369 with 3 resistors and cap.

When i put 90V on the input (1M resistor) i'm seeing a kind of saw signal of 200mV?
Is this good, or do i have to build te upper part to?
 

Offline saturation

  • Super Contributor
  • ***
  • Posts: 4788
  • Country: us
  • Doveryai, no proveryai
    • NIST
Re: Scope rising time
« Reply #17 on: June 18, 2010, 11:09:44 pm »
How are you generating the 90V?  From a PSU?  Chances are you are seeing the ripple, or its oscillating.  Send a pic if you can.
You'd best make the 'upper half' too, its AA battery powered for a reason, for cleanest power.




So, i build this: http://www.i9t.net/fast-pulse/fast-pulse.html
Only the down part, 2N2369 with 3 resistors and cap.

When i put 90V on the input (1M resistor) i'm seeing a kind of saw signal of 200mV?
Is this good, or do i have to build te upper part to?
« Last Edit: June 18, 2010, 11:11:15 pm by saturation »
Best Wishes,

 Saturation
 

Offline Technojunk

  • Contributor
  • Posts: 15
Re: Scope rising time
« Reply #18 on: June 18, 2010, 11:17:11 pm »
2 bench power supplies in series.
I think its better to to bult the upper part, but where do i grab the LT1073 off?
 

Offline saturation

  • Super Contributor
  • ***
  • Posts: 4788
  • Country: us
  • Doveryai, no proveryai
    • NIST
Re: Scope rising time
« Reply #19 on: June 18, 2010, 11:27:12 pm »
2 bench power supplies in series.
I think its better to to bult the upper part, but where do i grab the LT1073 off?

Digikey or if you are not in the USA, see findchips.com or octoparts.com

About $3-5, to make it worth your while in S&H buy all the parts from the same supplier.

http://search.digikey.com/scripts/DkSearch/dksus.dll?Detail&name=LT1073CN8%23PBF-ND

Best Wishes,

 Saturation
 

Offline Technojunk

  • Contributor
  • Posts: 15
Re: Scope rising time
« Reply #20 on: June 19, 2010, 12:53:32 pm »
Here a pic of what I build:


On my scope:
 

alm

  • Guest
Re: Scope rising time
« Reply #21 on: June 19, 2010, 02:32:09 pm »
Looks like ringing, not surprising since you use a probe and long ground lead. I think it's designed for a 50 ohm load, so you should terminate it with 50 ohms. The best way would be to solder a 50 ohm BNC connector right to the transistor lead, use a 50 ohm cable, and terminate at the scope. Terminating at the source and use a probe with a really short ground lead (BNC connector + probe-tip-to-BNC adapter or ground spring) might also work. The 50-ohms resistors should have a low inductance (standard metal film might be too much), since there are frequency components up to 1GHz or so in this signal.

As I mentioned before, you need to know the real amplitude to calculate the actual rise time.
 

Offline Technojunk

  • Contributor
  • Posts: 15
Re: Scope rising time
« Reply #22 on: June 19, 2010, 02:54:38 pm »
Or the design is crap!
I mean, why put 90V on a transistor that is made for MAX. 40V?

Now, with terminators and 50R cable:
     
 

Offline tecman

  • Frequent Contributor
  • **
  • Posts: 434
  • Country: us
Re: Scope rising time
« Reply #23 on: June 19, 2010, 05:21:34 pm »
The circuit uses the transistor breakdown to generate the fast pulse.  The high supply impedance keeps the transistor from being destroyed.

It looks like you have grounding/termination issues by the amount of ringing you are showing.

Paul
 
 

alm

  • Guest
Re: Scope rising time
« Reply #24 on: June 19, 2010, 05:49:33 pm »
Not sure what that is, it's really low level, a 167kHz radio station being picked up? Or from a switching power supply somewhere? Does not look like a signal from the avalanche transistor to me, or it's just not breaking down and this is the capacitor charging? I would just solder a female BNC connector to it, they're cheap. And keep wiring outside the coax as short as possible. That's probably why the i9t guy used a shielded Pomona box with integrated BNC connector ($$$). The original appnote (I hate how people keep linking to the website from one guy who built it instead of the Jim Williams appnotes) might contain more details and pictures.
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf