Oscilloskope 50Ohm Input

[Apochrom]

Hello everyone,

I have a fundamental question about oscilloscopes. I have heard criticism several times when oscilloscopes do not have a 50 ohm input, or can not be switched to 50 ohms. For which specific case does one need the 50Ohm input?

Thanks all

[Moriambar]

Hello everyone,

I have a fundamental question about oscilloscopes. I have heard criticism several times when oscilloscopes do not have a 50 ohm input, or can not be switched to 50 ohms. For which specific case does one need the 50Ohm input?

Thanks all

I think it's to terminate 50 Ohm BNC cables

[dietert1]

When measuring above 100 or 200 MHz you need to use a preamplifier, usually called "active probe". Or you have a RF port on the "device under test" to connect a coaxial cable. Anyway the signal is routed to the scope using a coaxial cable, usually 50 Ohm, that needs to be terminated. The best location for that termination is at the end of the coaxial transmission line, which is somewhere inside the scope.

If a scope does not come with that built-in termination, the second best method is to use a feed through terminator on the input socket of the scope. Then only a very small piece of unterminated transmission line remains behind that, maybe some centimeters inside the scope.  So this method works well up to several hundred MHz.

Fast scopes above 2 GHz bandwidth often come with "50 Ohm only" inputs and with smaller input sockets. The usual passive probe can't be used with those scopes.

Regards, Dieter

[tggzzz]

Some outputs are defined to operate into a 50ohm load. If they don't have the specified load then:

• for low frequency signals the voltage will be higher than indicated
• the output could theoretically oscillate, but that shouldn't happen with any well designed circuit
• for high frequency signals (think RF), connecting cables act as transmission lines. If transmission lines are not terminated in the characteristic impedance, then the amplitude will vary with frequency. Search terms: VSWR, transmission line.

The last is the most important case.

Some scopes have an internal 50ohm termination. In most but not all low-end scopes, a 50ohm resistor is simply connected in parallel with the 1Mohm//20pF input. That is imperfect and the source of a specification "VSWR<1.3".

If there is no internal 50ohm termination, then it can be simulated by either a BNC t-piece plus 50ohm termination, or a through termination. That would still be in parallel with the 20pF, with VSWR consequences.

[Apochrom]

Thank you very much. That'll help me a lot. So for high frequencies from about 200Mhz you need an active sample and for that you need a BNC 50Ohm connector.
Many years ago I had difference active probes at the oscilloscopes in the training laboratory of Physics. However, we didn't only have them at high frequencies, but we could also measure with them safely without having to pay attention to the short-circuit problem.

Thanks, Jürgen

[tggzzz]

Thank you very much. That'll help me a lot. So for high frequencies from about 200Mhz you need an active sample and for that you need a BNC 50Ohm connector.
Many years ago I had difference active probes at the oscilloscopes in the training laboratory of Physics. However, we didn't only have them at high frequencies, but we could also measure with them safely without having to pay attention to the short-circuit problem.

Thanks, Jürgen

The 200MHz is not a hard and fast number. Understand VSWR and how it might affect your measurements. The scope and probe are all part of the circuit being tested

There are "low" impedance Z0 resistive divider probes that are purely passive, and are good to several GHz. At UHF and above their input impedance is higher than that of standard "high" impedance *10 probes.

[exe]

I strongly recommend watching this video: https://www.eevblog.com/forum/beginners/scope-input-vswr-figure-at-50-ohm-what-is-that/msg193640/#msg193640 . Then this one: .

I mean, you really, really should watch these

[Apochrom]

Thank you very much. This is really a fantastic video!
This guy makes a lot of very good videos, I think I will be busy for the next weeks... :-)

Regards, Jürgen

[rhb]

The attached screen shots show the difference.  The first is with a 50 ohm 36 ps rise time square wave input going in to the 1 M ohm input.  The 2nd is with a 50 ohm thru terminator.  The pulse generator is one of Leo Bodnar's.  There is an extensive thread on the forum:

https://www.eevblog.com/forum/projects/yet-another-fast-edge-pulse-generator/

That thru has no reflections up to about  2-3 GHz.  It increase the rise time from Leo's pulser from 55 ps to 65 ps using a 20 GHz, 14 ps rise time SD-26 sampling head on my 11801.  I think Leo used a 40 GHz SD-30 to make the step trace and rise time measurement that came with the unit.  A 50 GHz SD-32 might show it as a bit faster.  The spec for the Maxim 3949 laser diode driver is 21 ps.

The BNC is a pretty dubious connector for scopes that go over 1 GHz.  They can be made to work, but an N would be much more appropriate. 

My LeCroy DDA-125 has BNCs and 50 ohm inputs and goes to 1.5 GHz, but it's also got 20% overshoot to go with the <250 ps rise time.  If I ever get around to hacking it to get a single 3 GHz channel I'll install N-F connectors along with proper anti-alias filters. 4x 750 MHz, 2x 1.5 GHz, 1x 3 GHz and SMA relay switching.  It will merge all 4 channels to give an 8 MSa/s sample rate.

[dietert1]

At which frequencies a through terminator is useful depends on the scope.
As i wrote above, it depends how long is the unterminated part of the transmission line inside the scope from its input socket to the sampler. In your SD26 measurement this appears to be about 12 mm (lambda/4 at 3 GHz and half the speed of light). If that distance is 50 mm, then reflections will show up at 500 to 700 MHz.

Regards, Dieter

[rhb]

A thru is *required* when using a 50 ohm source to get accurate results independent of frequency.  For a 50 to 1 M ohm interface the reflection is |1e6-50|/(1e6+50)  which is very nearly 1.0.

The screen shots were from a 200 MHz Instek MSO2204EA.  The SD-26 sampling head is 50 ohms only, with a 3 V damage level.  It is not intended for use with probes.

The frequency where the reflection gets large is easily measured by setting cursors over a range of one cycle. In general, dispersion results in the reflections being spread out.

The strongest frequency is a function of the two way delay time between reflectors. So for reflectors 5 cm apart, the resonant frequency is 3e8/(2*0.5) = 3 GHz in air and 67% of that in typical cable.

137 shows the basic setup with the calibrator feeding the 13 ps rise time SD-26.  The calibrator is faster than the <35 ps specification. 138 shows the trace and the rise time more closely.

139 & 140 show the setup with an SMA tee feeding the head with the other side of the tee unterminated producing a reflection from an open circuit.

142 and 145 show the reflection with a 50 ohm terminator on the tee.  The tee is about 2 cm long, so the extra path to the terminator and back is 1 cm x 2 and accounting for a 67% velocity factor we get 100 ps delay between the incident and reflected wave arrivals at the head.  It's not a perfect match as the terminator reflection sees 25 ohms rather than 50 so there are 3 reflections instead of 1.

146 shows the reflection off the open end of  ~36 cm SMA-M to BNC-F jumper connected in place of the terminator.   The precursor shown in 147 is the 200 ps period reflection at the ends of the tee. 148 shows the bounce in the tee after the step.  It's ~300 ps as it's a round trip + 1/2.

149 shows the reflection from a ~36 cm  BNC-M to SMA-M with a SMA-F to SMA-F and the 50 ohm terminator.  150 shows the time of the approximate 2nd zero crossings giving a first order estimate of the BW of the BNC connectors of 3 GHz.  I've tried to explain this a bit better in a later post.

154 and 155 show the BNC thru reflection with the terminator removed.

158 shows the reflection step off the terminator.  159 shows the cumulative effect of the mismatches,

While it won't provide the resolution that the 20 GHz BW of the 11801 & SD-26 combination, the instek example shows that you can get very useful results if you have enough cable to get away from the ringing.  In response to my inquiry, Leo tells me he can supply a <40 ps BNC pulser with a 1 MHz clock in place of the 10 MHz clock for an extra 10 quid.  Thus one can easily determine the length and quality of a <45 m cable with a DSO and 1 MHz square wave generator to the BW limit of the DSO.  With all the other uses for Leo's pulser, that's a really good deal.

Edit:  fixed  blunders  (Thanks Dieter!), tried to improve the explanation and corrected autorotation artifacts. It would be nice if software were consistent and displayed things without showing one thing and doing something else.

[dietert1]

rhb, please check your calculation. I think the result of 3E8 m/sec divided by (2 times 0.05 m) is 3E9 /sec, which is 3 GHz.
In general traveling wave effects show up not when the delay is half the wavelength (maximum extinction) but already when it's a third or a quarter of the wavelength. That depends on how strict you are with your overshoot criteria.
When you go from half wavelength to quarter wavelength 3 GHz => 1.5 GHz, if you account for the speed factor 0,67 this becomes 1 GHz. My estimates were somewhat lower because for simplicity i used a speed factor 0.5.
Concerning your measurements: Did you disable the internal termination of the SD-26?

To come back to the thread: A low budget 350 or 400 MHz scope that comes without active probe ports and without internal terminators can well be used up to its full bandwidth with pass-through termination on its input sockets. Internal terminators are essential only at higher bandwidth.

And the missing active probe port isn't a real problem when measuring low impedance sources like differential digital signals. Often you don't need a preamplifier. The simplest high frequency probe in that case is a 450 Ohm resistor feeding a 50 Ohm coaxial cable. Together with the through terminator on the scope input socket this makes a reasonable 10:1 high frequency probe.

Regards, Dieter

[rhb]

Again, thanks Dieter for pointing out the blunders.

With respect to the BW.  The spacing of the zeros of a sinc(t) closest to T0 gives the BW which is 2/T.  If you pick the 2nd zeros crossing, then the BW is 1/T.  Out of habit I picked the 2nd zero crossings, but then explained it incorrectly.  So I've gone back and fixed that.  The waveform between the 2nd zeros is very similar to what is referred to as a Ricker wavelet in reflection seismology for which the BW is approximately 1/T.

The 11810 is a sampling scope.  It samples the input at 100 KSa/s.  Each sample is taken at j*dT after the trigger input.  They are not able to trigger on the input.  To do so would require a 50 ns delay line.  That would present so much capacitance that it would be impossible to get the fast rise time.

It's not like a DSO.  They are peculiar beasts designed to give very fine time resolution.  The minimum interval, dT, using 5120 samples per sweep is 200 fs (yes, 0.2 ps).  There is no internal termination.  The input circuit is a 50 ohm hybrid module.  Hence the 3 V damage level.

The thru is sitting in a 50 ohm line.  So the reflection shown in 149 consists of a BNC-F cable connector, the 50 ohm thru and the BNC-M cable connector in series.  As shown in 142 and 145, there are multiple reflections caused by the tee.  So the response of reflection response of the thru shown in 149 has been convolved with the reflection response of the tee to first order.  In fact, it's convolved with the reflections from all the interfaces which precede it.  That is infinitely long, though the amplitude dies off pretty quickly.

In seismic work if one needs to properly handle the reflections that follow the first one, one uses the Z transform and evaluates the expression for the poles and zeros of the transfer function.  That is complex enough to merit publication as a professional paper by a prominent seismologist while I was in grad school.  As it happened, I had solved the same problem as a course exercise which I turned in shorty before the paper appeared.

The bottom line is, that as Dieter pointed out, until you get well above 1 GHz, a 50 ohm thru terminator on a scope which only has a 1 M ohm input will provide accurate results.  But you do need the 50 ohm thru to get accurate results if the source impedance is 50 ohms. Without it, the cable to scope interface is almost a perfect reflector.  So there will be a series of reflections without it.  What effect they have depends upon the timing of their arrival.

I did all the work in the previous post for my own benefit.  I've been using the 11801 to evaluate connectors, cables and terminators.  In the process  I discovered a bug in the 11801 FW that leads to spurious events until the scope is reset.  I came *very* close to posting an elaborate interpretation of what proved to be FW glitches in the response. Very plausible, but utterly wrong.

Have Fun!
Reg

[Apochrom]

Many thanks for all the information, even if I didn't understand everything completely.
Since a few weeks I have a new beginner DSO (1Gs/s and 200Mhz bandwidth) which doesn't have a 50ohm connection. I have played a lot with this DSO in the last weeks. For fun I connected a very simple radio FM antenna to the oscilloscope and did a simple FFT analysis. It is not so easy to try out all possible settings in the DSO and it took me quite a while to reach a certain optimum of the display. An experienced "Oscillonaut" would certainly achieve even better results, but I am quite satisfied with the result.
See appendix.
If I see the contributions correctly, are these results at my 1MOhm input very uncertain?

Jürgen

[Apochrom]

Supplement,

the antenna is something like this here:


https://www.ebay.de/itm/UKW-Antenne-mit-Koaxialkupplung-FM-Dipol-Radio-Wurfantenne-Hersteller-goobay/252085799381?ssPageName=STRK%3AMEBIDX%3AIT&_trksid=p2057872.m2749.l2649

[rhb]

That antenna is 300 ohms unless there is a balun in the connector, which is highly likely.  In either case there is still a large reflection coefficient, so you'll get better results with a thru terminator.

The screen dumps demonstrate the effect of the mismatch.  This is the same setup as the previous Instek dumps expect that there is 2 m of cable from the pulser to the DSO instead of a direct connection.

34 is the <40 ps pulser without the thru.  35 is with the thru.  There is a 2 m RG58, 50 ohm cable between the pulser which is a 50 ohm source and the scope which only provides a 1 M ohm input.

Because the pulser and cable are fairly well matched, the two way reflection from the DSO to the pulser and back is fairly small.  The reflection coefficient of the cable to pulser match is proportional to the ratio of the voltage of the first and last halves of the positive portion of the square wave.

As is clear from 34, the mismatch at the DSO input BNC to the cable seriously degrades the step response through reflections within the BNC connectors and the input of the DSO AFE.  The mismatch increases the apparent rise time by a factor of ~2x as shown in my previous Instek screen shots.

A comparison of 35 to the corresponding figure in my previous post using the Instek gives an indication of the effect of the cable.  Measured with 11801 it goes from 55 ps  to 341 ps as a result of the capacitive load presented by the RG58.

[Apochrom]

Many thanks for the explanation and the two diagrams!
I have measured the resistance of the antenna, it is only 3.8 Ohm.

[tautech]

Many thanks for the explanation and the two diagrams!
I have measured the resistance of the antenna, it is only 3.8 Ohm.

Tip for new SDS1202X-E user:
For screenshot Saves, use the blue Print button for direct to USB saves and in doing so the menu you have active still remains so we can see and advise about your settings.

[rhb]

That's a transformer winding.  What matters is the complex impedance.  To measure that you need either a fast edge pulse generator to do Time Domain Reflectometry or a Vector Network Analyzer.  The latter are rather pricey for a novice.  You can do TDR with a VNA or VNA with a TDR via the Fourier transform if you have the data in machine readable form.  Getting data off the 11801 is on my "To Do" list, but I've been having too much fun just messing around doing stuff like the sequence a few posts back.

I plan to write software and a detailed description of how to do VNA using a DSO and a fast edge pulse.  It's limited to the BW of the DSO, but for building HF radios it is quite adequate.  You can get 100 dB dynamic range from a single 20 Mpt, 12 bit record from an Owon XDS2102A.  You can also do it with an 8 bit DSO.  It just requires multiple traces to sum together.

[rhb]

Here are some spectral displays using the Instek MDO2000E series spectrum analyzer app.  You can't buy a license to use it on a different model in the GDS2000E line, but you *can* hack it.

The photo shows the antenna I'm using which is a 5 cm diameter H field probe I made.  Except for a very thin slot at the top, the inner conductor is completely shielded from the electric field, so it just senses the magnetic field.

The noisy display in the photo is because I had not yet set the averaging.  I first did this a week or so ago when I made the H field probe out of a RG402 semirigid solid Cu cable jumper I cut in two.  I'll make a probe with a smaller loop later.

The following screen dumps are with the probe on a longer cable about 1 m above the DSO.  I moved it there and when the noise on the display didn't clear up I realized setting the DSO at 1 mV/div was not a good idea.  So I changed to 5 mV/div.

37 is the spectrum using a 50 ohm thru to match the loop.

38 is the spectrum without the thru.

39 is the spectrum with the DSO set to 1 mV/div and the same averaging as 37.   I checked all the peaks and they correspond to the US FM allocation spacing.  I'm in a hilly, rural area, so there are a lot of stations 75 miles away that I cannot get because of distance and topography.

My MSO2204EA is a 200 MHz BW DSO, so very comparable to your Siglent.  Notice that the spectra are not the same.  The mismatch loses most of the signal from the antenna.  It also creates resonances which modify the amplitudes.  The peak on the left in 38 missing from 37 may or may not be real.

If you were trying to listen to a distant radio station it would matter a lot, though not as much as a comparable mismatch on a transmitter.  Many transmitters, if keyed without an antenna will immediately destroy the output power amplifier stage because of the reflected energy |(1e6-0)/(1e6+0)| = 1.0.  That is almost  the same as the 50 ohm to 1 M ohm mismatch, 0.9999.

The better ham transceivers have circuitry to detect an open antenna line and turn of the transmitter.

I've added a couple more displays.

41 shows the spectrum without the thru, using peak detect acquisition and 256 sample averaging in the spectrum.    40 shows te same thing, but with the BNC thru in place.  Otherwise everything is the same so far as I know.  It's raining here and close to sunset, so propagation may be changing.  The length of the cable from the join of the loop to the BNC is ~1.7 m.  It would require a good bit of experimenting to determine what's going on.

Have Fun!
Reg

[Apochrom]

Thank you very much for your answers!

@tautech: I had already asked myself how to save the settings bar. With the print key, that goes super thank you very much for the tip!

@Reg: Many thanks for the explanations and the interesting diagrams.
Is your h-Probe similar to the one Dave Jones recently presented?
https://www.youtube.com/watch?time_continue=3&v=nImoQcoqkuQ
I think I'll dare to take a look at it.

So, by connecting my simple antenna, the results should be enjoyed with caution. However, with this simple wire I could make the local radio-transmitters visible without further amplification. In fact, the frequencies of the big peaks are the lokal radio frequencies.
That was my main goal to make the radio frequencies visible, even if some of the client  peaks can only be reflections.

Here a few diagrams in appendix. The setting was average calculations with 20 samples.

Jürgen

[dietert1]

When we try to understand your "measurement", your screen dump seems to indicate that you are using a 10x probe. Is this correct?

Then if i read the 100 MHz peak at about -73 dBm, this would be 50 uV RF. If the antenna is connected directly to the scope input socket, the result is 5 uV instead. As far as i remember a usual FM receiver works above 1 uV.

Regards, Dieter

[Apochrom]

When we try to understand your "measurement", your screen dump seems to indicate that you are using a 10x probe. Is this correct?

Then if i read the 100 MHz peak at about -73 dBm, this would be 50 uV RF. If the antenna is connected directly to the scope input socket, the result is 5 uV instead. As far as i remember a usual FM receiver works above 1 uV.

Regards, Dieter

you're right, it's not a real mesurment, I'm just playing around with my new oscilloscope, trying to find the right settings and interpret the results.

And also correct, the antenna is directly connected to the Scope input.

[Apochrom]

If the antenna is connected directly to the scope input socket, the result is 5 uV instead

The measurement with the scope by means of curser shows me with the setting 10x probe approx. 100mV, thus with 1x 10mV for the largest peaks.

[iMo]

When we try to understand your "measurement", your screen dump seems to indicate that you are using a 10x probe. Is this correct?

Then if i read the 100 MHz peak at about -73 dBm, this would be 50 uV RF. If the antenna is connected directly to the scope input socket, the result is 5 uV instead. As far as i remember a usual FM receiver works above 1 uV.

Regards, Dieter

That would be true with 50ohm matching everywhere, imho. The OP does some basic experiments without any care on an "impedance matching", what is the keyword in this topic..
So the values of the spectral peaks could be anything.
The antenna is a cheap joke, I doubt there is a balun inside the connector. He would get the same result with 1m long piece of wire put into the oscope's 50ohm input.