suitable scope to use for RF work

[jason695]

hi anyone,

i hope this doesnt seem to broad or vague a question.

can anyone answer my question about a suitable dso if any.

i am trying to do some RF measurement. it will all be done on unmodulated sine waves. i was looking ultimately to measure a 100 mhz sine wave. however, most of my work will be done at about 30mhz.

the only problem is significant parts of my study is being done on single transistor oscillators designed and constructed by me. these are simple sine wave oscillators e.g. colpitts, hartley etc. im not using premade square wave clocks etc.

in this situation obviously there'll be more significant harmonics than from signal generators or premade i.c. but, more importantly i think, there'll might be distortion as in "squashed sine waves" to some extent.

obviously the DS1052e is only 50mhz banwidth but, it is 1ghz sample rate so although attenuated it can theoretical measure 100mhz sinewaves. i know about nyquist etc but, would such signals be heavily aliased & what about limited bandwidth of 50mhz.

can anyone tell me if the ds1052e would be suitable with all these effects present.

thanks everyone
jason

[alm]

Sounds like you need a spectrum analyzer, not a scope. A scope is only useful to establish the presence of an RF signal. It will lack the bandwidth to show much of the harmonics, and FFT is a lousy replacement for a real spectrum analyzer due to very limited vertical resolution.

[gregariz]

I regularly use both specans and scopes for this type of work. The scope is a useful tool to look at things like clipping and waveform shapes that give one an idea of circuit function and design problems. I would suggest you find a 100 or 150MHz scope if your intention is to look at the higher frequency. Regardless of which scope you buy it will have a higher frequency rolloff that will result in any higher frequency components either being missed or distorted.

[alm]

For 30MHz signals a 100-150MHz scope may be OK. For 100MHz, I would recommend a 500MHz scope if you want to measure the shape as the signal, and not use it as a clumsy frequency counter / power meter. Note that the scope bandwidth is the point at which signals will be 30% down in amplitude. It will have a hard time displaying anything but an attenuated sine at its rated bandwidth.  Any signal except perfectly sinusoidal signals will have a larger bandwidth than just the fundamental frequency.

[vk6zgo]

Obviously,a Spectrum Analyser would be nice,but it may be outside your budget.
 
As far as Oscilloscopes go,I would suggest you look around for an analog 'scope,as you may be able to get a 400MHz one reasonably cheaply.
Unless your sinewave is severely mangled,the level of any harmonics above the 3rd, probably won't be too high so 400MHz BW would be fairly useful.
If you already have a DS1052e,then try it & see what happens.

Other than a Spectrum Analyser,there are other & sometimes more useful ways than looking at waveform distortion to check for harmonics of a sine wave,such as:-

(1)
Absorption Wavemeter ( basically a tuneable detector--you tune it for the suspected harmonic).,
These are very old technology now,you might find one on EBay,or you could make one!
Commercially made ones have a calibration chart,so you can set the fundamental as 0dB,& read off the amplitude of harmonics in dBs down.

(2) VHF/UHF receiver---obviously,these won't give you an accurate measurement,due to AGC,etc,but they will indicate the presence of a harmonic.

This is all presupposing a limited budget.If you can spend around $1600 you can pick up a brand new Chinese Spectrum Analyser which should make things quite easy.

VK6ZGO
.

[Mechatrommer]

my way. not how it supposed to be... Observing 433MHz RF signal using 100MHz Scope (DS1102E)

[vk6zgo]

my way. not how it supposed to be... Observing 433MHz RF signal using 100MHz Scope (DS1102E)

Where are you looking at?

Simple Amplitude Modulation gives an output which includes:-

 fc =the RF carrier frequency---433MHz in this case
 
 fm=the modulating frequency ----837kHz?

 fc-fm or 432.163 MHz

 fc+fm or 433.837MHz

All these frequencies will be present at the output of the modulator or modulated stage.

It is necessary to restrict the bandwidth after the modulator using a bandpass filter (BPF),or high pass filter (HPF)to remove the unwanted modulating signal  fm.

From the look of the waveforms presented,it seems that the probe is placed prior to any filtering,so that the modulating frequency is the major component of the signal.

What do you see with the RF turned off?

What do you see with the modulation turned off?

VK6ZGO

[joelby]

Yeah, you're really not going to see any of the 433 MHz signal on a 100 MHz scope using a 150 (?) MHz CRO probe. Anything you do see is probably noise.

If you wanted to observe the RF signal, you could mix 433 MHz down to a low IF (e.g. 10.7 MHz) using a 422.3 MHz local oscillator.

[Mechatrommer]

Where are you looking at?

err (biting my nail) 
What do you see with the RF turned off? = 10kbps pwm signal (feed to radio)
What do you see with the modulation turned off? = 433MHz signal (i suspect) from the spec. i'm not able to make the stable signal, its premade and only modulate when pwm goes hi->lo

It is necessary to restrict the bandwidth after the modulator using a bandpass filter (BPF),or high pass filter (HPF)to remove the unwanted modulating signal  fm.
From the look of the waveforms presented,it seems that the probe is placed prior to any filtering,so that the modulating frequency is the major component of the signal.

you are right, there's no filter at all, except DC filter. i'm using premade rf module, use as is, no filter.

my point is not filtering, i know i can see some spike/noise there but, i only care the carrier signal (thick line below the "distorted pwm" shape)

[Mechatrommer]

Yeah, you're really not going to see any of the 433 MHz signal on a 100 MHz scope using a 150 (?) MHz CRO probe. Anything you do see is probably noise.

yea maybe you are right. but in the link i've provided the noise level (no signal) and the ~433MHz "noise" when i punch in the signal and calculating/estimating the "noise" freq.

If you wanted to observe the RF signal, you could mix 433 MHz down to a low IF (e.g. 10.7 MHz) using a 422.3 MHz local oscillator.

i guess you and vk6 are refering to FM radio?

[Mechatrommer]

as i said... not how it supposed to be. buy a spectrum analyzer! that how it supposed to be.

[joelby]

yea maybe you are right. but in the link i've provided the noise level (no signal) and the ~433MHz "noise" when i punch in the signal and calculating/estimating the "noise" freq.

It's still effectively noise, since you don't know if it's really a 433 MHz carrier, part of the modulation, aliasing, outside RF being picked up by your test setup, etc. If you have a signal generator, you can try observing the output of that at 433 MHz on your oscilloscope to provide a comparison of a clean, unmodulated signal.

i guess you and vk6 are refering to FM radio?

Nothing that sophisticated - if you mix 433 MHz with 422.3 MHz, it will be down-converted to 10.7 MHz (plus a bunch of image frequencies, but 10.7 is a convenient frequency that you can see on a 100 MHz scope and filter with a cheap filter). I suppose the question to ask is exactly what you're trying to observe here.

[amspire]

It didn't make sense that the Rigol couldn't show a 433MHz waveform, unless the trigger didn't work at 433Mhz.

My first CRO was 5MHz analog and I used it all the time to look at 20MHz+ waveforms.  OK the amplitude was low, and it was useless at waveforms,but it displayed the signal fine.

I just tried it on a DS1052E that was upgraded to 100MHz. The best source I had at hand was a 433MHz transmitter, and the signal was pretty messy and there was a lot of background noise, but the trigger was rock solid at 433MHz and the waveform is fine.

Here it is:



I used the 25GS/a Equivalent Sampling, and as long as you can trigger off a repetitive signal, it should show a real waveform at only 433MHz. That A/D's gate time is well below 1nS so it can capture 433MHz easily.

What I found is that the FFT works fine (it was working better earlier, but this evening, the background noise has gone right up). If you stop the trace on the oscilloscope, it only shows a single shot capture (rather then Equivalent Sampling), but if you press the Stop in the control box on Ultrascope, it freezes the Equivalent waveform.  I don't have the Rigol probes on hand, and I used some really crappy 100MHz probes instead. I will dig up the Rigol ones tomorrow. I might also try a 500MHz probe and see if it make a difference.

Richard

[amspire]

A bit clearer with a  500MHz probe. Still masses of noise as I can't get a decent ground near the antenna.

The 433Mhz signal is definitely there. Be much easier with a decent signal of known amplitude to test, but I think the highest I have from a generator is 150MHz.



Richard

[Mechatrommer]

Still masses of noise as I can't get a decent ground near the antenna.

i dont think its the noise. i highly believe its because the equivalent sampling method, or sampling jitter or such, maybe. your FFT is clean, and wait... you turn ON the FFT? max is 500MS/s, violating Nyquist rule. just an assumption. since you have astable 433MHz source, why dont you try my setup... fft off,ch1 only, real time sampling. try both sinc(x) on and off. i'll be thankful if you can show the picture here.

[amspire]

It is noise - lower frequency . I know - I had to set the trigger as high a possible to trigger on the peak of the LF noise.

Nyquist has not been violated as equivalent sampling is real sampling for a repetitive signal.

So the sampling rate is 25GHz. It is not 1GHz . There are over 50 measurements per cycle evenly spread across the cycle. Until recently, sampling was the only way to view microwave waveforms.

Back in the 60's, you could get a 30GHz sampling scope, even though the scope mainframe could not cope with more then a few MHz of bandwidth.

Richard.

[Mechatrommer]

It is noise - lower frequency

the first time i've heard. yes, that was my suspicion, your 433MHz is not a clean/pure sine wave.

[amspire]

The noise is there with the transmitter unpowered. As far as I can tell, the waveform I'd clean.

You still seem to believe that the Rigol cannot see a 433MHz waveform.

It definitely can. It is definitely not an aliased fake waveform.

I have another transmitter at 310 MHz and if I connect it I see 310MHz.

Are you suggesting that the Equivalent Sampling on the Rigol does not work?

Richard

[Mechatrommer]

You still seem to believe that the Rigol cannot see a 433MHz waveform.

how can you make this conclusion? rewind to reply #5 in page 1. i'm with you

Are you suggesting that the Equivalent Sampling on the Rigol does not work?

no. i'm suggesting, your transmitter signal is not clean. which is you've already admitted.

i've used equ. sampling and it can better show higher frequency signal, ie near or higher than the scope BW. its just we have to grasp the experience of what the jumbling shape is. its your signal mixed with other freq components, as you said, low freq one. you can call it jitter(ing) or interference, but not noise. noise is alot higher freq than your signal and is random and alot low magnitude. will be difficult to distiguish between noise and harmonics (or freq components) in eqv. samp. mode. but i've seen the picture similar to yours while i was observing non-clean 75-100MHz signal in eqv. mode. if its noise, then your picture should be alot cleaner and less jumbling. but if your signal is near noise level, then it will be alot jumbling and probably you wont see anything meaningfull.

[alm]

So what does it tell you about the signal? Does it show the amplitude (minus how many dB)? Can you determine the frequency with any accuracy? Can you tell distortion? Harmonics? How is this better than a frequency counter or RF detector probe which will easily go to 500MHz for less money that even a 50MHz scope? At least these are specified for those frequencies.

Sure, if you feed it a ton of signal, some will make it through. It's hardly a measurement anymore, though.

[Mechatrommer]

So what does it tell you about the signal? Does it show the amplitude (minus how many dB)? Can you determine the frequency with any accuracy? Can you tell distortion? Harmonics? How is this better than a frequency counter or RF detector probe which will easily go to 500MHz for less money that even a 50MHz scope? At least these are specified for those frequencies.
Sure, if you feed it a ton of signal, some will make it through. It's hardly a measurement anymore, though.

as in my link in reply #5, we cannot measure it accurately, quantitatively speaking, but at least we see something persistent. if i want to build a high quality 400MHz signal, then i need to see if it has 4GHz component which is far from impossible using 1GS/s scope. but let say we are working on 40MHz signal, then we may have an idea what the noise/harmonics of the signal up to its 10th harmonics, but as i said, its amplitude is not accurate since i dont have the scope profile. if i'm one of rigol staff, then i may want to steal the profile from the engineer and do the diy compensation.

i never heard freq counter that can show many freq. or i'm not familiar enough with it. showing only one frequency will not be helpful in (mixed) signal analysis. if hi freq RF is thy main stuff, thy feed and thy shit with it, then thy should get a spectrum analyser, or maybe VNA? not sure.

[amspire]

So what does it tell you about the signal? Does it show the amplitude (minus how many dB)? Can you determine the frequency with any accuracy? Can you tell distortion? Harmonics? How is this better than a frequency counter or RF detector probe which will easily go to 500MHz for less money that even a 50MHz scope? At least these are specified for those frequencies.

Sure, if you feed it a ton of signal, some will make it through. It's hardly a measurement anymore, though.

It says a lot. Yes you can use a rf probe and it gives you a different set of information. The frequency attenuation will be something that can be measured and then you can compensate for it,so yes you can get amplitude.

Definitely you can spend a lot of money buying better gear, but it is good to learn the limits of equipment too. The Rigol provides very useful information above 100MHz.

I would hate to suggest to someone that they couldn't do rf work because they only had a Rigol scope.

Richard.

[alm]

The frequency attenuation will be something that can be measured and then you can compensate for it,so yes you can get amplitude.

You may be able to get a ballpark estimate. I wouldn't expect the attenuation to be very linear over the vertical sensitivity range or maybe even other factors like vertical position or individual channel. You're characterizing something that even the manufacturer didn't bother to spec. You're also presenting a fairly low-impedance load to the signal source with a 10Mohm passive probe (about 30 ohm), something that will load down the signal.

How would you calibrate it? You'd need access to a fairly clean signal source of known flatness that spans the 50-500MHz range. Or something else to measure the power/amplitude accurately.

Definitely you can spend a lot of money buying better gear, but it is good to learn the limits of equipment too. The Rigol provides very useful information above 100MHz.

It provides some information. Basically that there is a signal that includes a frequency component around 433MHz. It may also include other frequencies. It may even be a 433MHz square wave. Granted, it's infinitely better than no information. But it's not very detailed or reliable information.

I would hate to suggest to someone that they couldn't do rf work because they only had a Rigol scope.

I wouldn't suggest that either, you can even do RF without a scope (it just gets a lot harder), but suggesting that it's a good buy to observe signals near or above its rated bandwidth isn't correct either. It may be usable in a pinch.

[amspire]

Alm,

Basically I think we probably agree. Absolutely if you can have 4GS/s oscilloscope and a wideband spectrum analyser - fantastic. From where I am coming from, I just hate the idea of not trying something because you don't have the right equipment. When you have to do something and you don't have the right specialized piece of equipment - you have to be creative and that's fun.

I bet the person who designed the first spectrum analyzer would have loved to have a  .... spectrum analyzer.

Richard.

[Mechatrommer]

From where I am coming from, I just hate the idea of not trying something because you don't have the right equipment. When you have to do something and you don't have the right specialized piece of equipment.

vote with both hand i know i need to have a SA, but i know i cannot afford it, and maybe, this is the only time i'm playing with RF, so is it worth it? dont i have the right to play around?
i dont thing so.

[vk6zgo]

Alm,

Basically I think we probably agree. Absolutely if you can have 4GS/s oscilloscope and a wideband spectrum analyser - fantastic. From where I am coming from, I just hate the idea of not trying something because you don't have the right equipment. When you have to do something and you don't have the right specialized piece of equipment - you have to be creative and that's fun.

I bet the person who designed the first spectrum analyzer would have loved to have a  .... spectrum analyzer.

Richard.

He probably has a Selective Millivoltmeter!
Seriously,though,Selective Milli (& micro)voltmeters were widely used at up to UHF frequencies,& you may be able to pick one up.
VK6ZGO

[amspire]

Seriously,though,Selective Milli (& micro)voltmeters were widely used at up to UHF frequencies,& you may be able to pick one up.
VK6ZGO

You have convinced me. I want one.

As long as you don't tell me the good ones weigh about 50 KGs and use up half the bench.

Richard.

[joelby]

Well, here you go!

Unfortunately I don't quite have all the bits available to do this properly, but I did the following:

HP 8922S in CW (signal generator) mode.

Rigol DS1102E with 10x probe and ground spring, stuck in the 8922S IF output port. Equivalent time was used, though there seems to be a bug in my Rigol where stopping acquisition (to save the waveform image) reverts to normal acquisition mode. This completely messed up the 433 MHz reading, so I took a photograph instead.

Signal generator set to -10 dBm at 100, 200, 300, and 433 MHz.

So yes! You should be able to see something, even with some pretty horrible mismatches, and as long as you use equivalent time sampling and your signal has a CW carrier.

[Mechatrommer]

thanx for the picture joelby. i wish i have that kind of generator.
but maybe a stupid question... why -10dBm? why not 0dBm?
assuming the generator is flat, you are showing around -3dB attenuation at each 100MHz interval.
and assuming the dso is speced as -3dB@100MHz, the signal coming out of the generator is somewhere 0.7Vpp.

[Zero999]

Obviously,a Spectrum Analyser would be nice,but it may be outside your budget.

He didn't say what his budget is.

I've found a cheap Chinese spectrum analyser on ebay. Whether it's any good oor not is another thing but it'll almost certainly be better than an oscilloscope of the same price. It covers 150kHz to 1050MHz. If you need to go below 150kHz then try the FFT on the Rigol or your PC sound card. If you need to go above 1GHz then you need to spend serious money.
http://www.ebay.co.uk/itm/ATTEN-AT5010-Spectrum-Analyzer-150-KHz-1G-Hz-/220847258416?pt=UK_BOI_Electrical_Test_Measurement_Equipment_ET&hash=item336b85bb30
 

As far as Oscilloscopes go,I would suggest you look around for an analog 'scope,as you may be able to get a 400MHz one reasonably cheaply.
Unless your sinewave is severely mangled,the level of any harmonics above the 3rd, probably won't be too high so 400MHz BW would be fairly useful.
If you already have a DS1052e,then try it & see what happens.

If he already has the DS1052E he should hack it to 100MHz.

[Mechatrommer]

i've been eyeing on that Atten SA. since 400MHz was(is) my interest, then a 4GHz is more suitable for the job. The cheapest, if i got money, i'll get back to this USB Signal Hound SA
(i know i know, its USB. i heard ya. one should get the more proper standalone with LCD SA which cost many thousands. and there's DIY SA link here if you search and powerful enough to build one)

[joelby]

thanx for the picture joelby. i wish i have that kind of generator.
but maybe a stupid question... why -10dBm? why not 0dBm?
assuming the generator is flat, you are showing around -3dB attenuation at each 100MHz interval.
and assuming the dso is speced as -3dB@100MHz, the signal coming out of the generator is somewhere 0.7Vpp.

I didn't want to run the signal generator at higher output powers without 50 Ohm termination and I don't have a feed-through terminator.

The Atten spectrum analysers don't have very great reviews. The Signal Hounds barely have any reviews at all! If you keep an eye on eBay you should be able to get something second hand and considerably better (3 GHz) for not much more.

I actually picked up the HP 8922S at a garage sale for $500. It's a 1GHz spectrum analyser, signal generator, and frequency counter, and a GSM base station.

[Mechatrommer]

If you keep an eye on eBay you should be able to get something second hand and considerably better (3 GHz) for not much more.

even if i get it for free. shipping will cost me the same as the new atten SA or more. even the "for parts" SA is not free.

[alm]

Is the output level and purity of your generator specced for high-impedance loads? Because most have to be terminated into their characteristic impedance to meet their specs. Reflections and standing waves can screw up these kinds of measurements. I would use feed-through terminators for these measurements. 433MHz is actually approaching the limit of these feed-through terminators, the capacitive stub formed by the scope input becomes a problem. Which is why 400MHz scopes tend to have real 50 ohm inputs.

[Mechatrommer]

agree with alm. a flat generator can be seen as non flat by the dso if unterminated properly. anything longer than 17cm is enough to make a perfect reflection on this 400MHz signal. but i think joelby's point is not about accuracy.

[jason695]

thanks for your reply everyone, very interesting information thanks. i might be slightly late putting this on no but, going right back to the start of this blog .

on a suitable scope for 100mhz sinusoid LOL.

im not sure if what im about to say is wright but here goes anyway.

im just trying to observe the fundamental at 100mhz

if i had a 100mhz analogue scope, this isnt good enough to see all the harmonics on a 100mhz fundamental sinusoid.however, wouldnt its front end filters just act as a harmonic filter itself and id just be left with fundamental. obviously theres attenuation of 3db. isnt this ok with about 1 volt amplitudes.

can anyone please provide screen shots of a 100mhz scope observing a 100mhz fundamental sinusoid thats got harmonics on it.

oh yes lastly im trying to build these bjt single transistor oscillators. they all work at 100mhz easily. im trying to get one to work at 27mhz at its proving very hard. can anyone help me with a schematic for a 27 mhz bjt oscillator (preferablt colpitts type but, any will do).


regrads and thanks everyone.
jason

[alm]

Attached are some screenshots. The signal is a ~100MHz square wave (plenty of harmonics) with a specified rise time of <= 1 ns. The first screenshot is from a 500MHz/5GS/s scope, and shows a decent square wave with mainly odd harmonics, as expected for a square wave. The internal 50 ohm termination was used. The cursors indicate the ~100MHz and 1GHz point in the frequency domain.

The monochrome screenshots are from a 100MHz/1GS/s scope, with similar time base / vertical attenuation settings. An external feed-through terminator was connected to the scope's input. In the time domain, it looks pretty much like a sine to me. In the FFT, only a few harmonics are even present, and the third harmonic is at -30dB, as opposed to -10dB on the 500MHz scope. The fifth harmonic is near -20dB on the 500MHz scope (note that this is likely even a few dB down), but something like -55dB on the 100MHz scope, and only above the noise floor because I turned on averaging.

[vk6zgo]

Seriously,though,Selective Milli (& micro)voltmeters were widely used at up to UHF frequencies,& you may be able to pick one up.
VK6ZGO

You have convinced me. I want one.

As long as you don't tell me the good ones weigh about 50 KGs and use up half the bench.

Richard.

You've nailed it!
Yes they are very large !--but so was most of the test equipment of the time!

VK6ZGO

[jason695]

Alm, thanks for the screenshots.

so what do you think, on a 100mhz,1gs/s dso do you think id get the fundamental of 100mhz sine if it has harmonics it?

& similarly & probably better, dya think i'll get the 100mhz sine fundamental with 100mhz analogue scope then?

also anyone know anything on oscillators, i need a 27mhz.



lastly, to see the harmonics, cheap chinese spec analyser say ATTEN AT5010 Spectrum Analyzer 150 KHz to 1G Hz. what does anyone think to spec:-

http://www.ebay.com/itm/ATTEN-AT5010-Spectrum-Analyzer-1050MHz-/220751214997

thanks all
jason

[DaveW]

To have a point of comparison, I borrowed a 433MHz radio from shl876, and probed it using a 1.5GHz passive probe on a 4GS/s, 1GHz Agilent 7000 and on a Rigol DS1102E (not modified), both the screenshots are set to AC coupled mode, and I've set the cursors to check it's the actual signal

[amspire]

Dave,

Did you put the Rigol in Equivalent sampling mode?  If it is not, there is no way it can properly show a 433MHz waveform.

Richard.

[tinhead]

i've been eyeing on that Atten SA. since 400MHz was(is) my interest, then a 4GHz is more suitable for the job.

i consider buying one of the new ATTEN SAs, AT5010D or AT5030D.

They have 7" TFT screen and tracking generator, so it will be (well, not really) good replacement for my old HP 8590,
but it will at least not wake-up my neighborhoods nor drive power plant down while booting...

ATTEN have other nice RF equipment too, e.g. AT8010D which is actually a clone of R&S SMX/SMY?
Somethign like that, a guy from China told me this, no idea if this is truth or just misinterpretation of front panel design.

[Mechatrommer]

Did you put the Rigol in Equivalent sampling mode?  If it is not, there is no way it can properly show a 433MHz waveform.

yea, me curious. how can that signal looks nice on rigol? (even if aliasing is clearly seen, near nyquist sampling). maybe its sincx turned ON? my 433MHz is just triangle shape.
and wait. if i'm not wrong, rigol showing greater magnitude? 700mVpp @ 500mV/div vs Agilent 80mVpp @ 200mV/div? are you ok Dave?