-
Oscilloscope bandwidth - is jump from 100MHz to 200 MHz significant?
Posted by bmdaly on 02 Mar, 2018 11:28 -
Hi,
I'm a hobbyist proposing to upgrade from my 10MHz Picoscope 2204 to a Siglent SDS1104X-E, primarily for the 4 channels.
My main area of interest at the moment is analogue electronics, and I'm working my way through a series of lab tutorials in "Learning the Art of Electronics". I expect that being able to monitor multiple points on the circuits I'm building and view the signals' relative phases and amplitudes on a single screen will be a big help to learning.
I've also done a couple of microcontroller courses in the past (using TI MSP432 and TM4C123 boards), and also use Arduino boards.
I've read Performa01's in-depth review https://www.eevblog.com/forum/testgear/siglent-sds1104x-e-in-depth-review/ of the SDS1104X-E and the SDS1202X-E, including his chapter comparing the bandwidth of these devices, and have decided that the 100MHz model is sufficient for what I need - particularly given the 50% cost premium for double the bandwidth (but no corresponding increase in sampling rate), and limited (lack of???) anti-aliasing on the 200MHz model.
Before finalizing the decision, I wonder if there are any use cases where the extra bandwidth would really make a significant difference to understanding what's going on in a circuit?
I don't want to be in a position a few months down the road thinking "if only I had a little extra bandwidth...", but equally, I need to be pragmatic and draw the line somewhere.
Any comments from people who have been down this road would be appreciated.
Thanks in advance,
Brian
-
For your applications, you'll be fine with a 100MHz scope. My most used scope is an old and crusty 100MHz TDS220, next stop is the good old analog 7603.
4 channels is pretty useful, most of the time you'll find yourself using one or two channels.
Having more than 100MHz BW also requires more advanced probing techniques, and often 50 Ohm terminated inputs. Your vanilla 10x scope probe won't make any use of more than 100MHz BW - you'll just see more ringing. So 100MHz, or even as low as 50MHz is enough for the most tasks in debugging analog and uC circuitry. -
More important for analog work are things like high G differential preamp with
high CM rejection.
Or spectrum analyzer.
You can look at older Tek series 7000, its a mainframe that takes many different plugings.
Still very useful in today's work. Plugins like DMM, High G diff preamp, spectrum analyzers
(several), FET input plugin, curve tracer.........
http://w140.com/tekwiki/wiki/7000-series_plug-ins
Regards, Dana. -
It depends on what frequency analogue signals you will be using, what you will be looking for, and what logic family you will be using.
Sometimes analogue circuits can oscillate at a much higher frequency than you are expecting ("amplifiers oscillate, oscillators won't").
In some analogue applications, harmonics of the fundamental frequencies are important; factor that into your assessment.
A 100MHz scope will (just about) allow you to see a 3.5ns risetime; a 200MHz scope 1.8ns. Logic families from the early 80s (LSTTL, STTL) have that kind of risetime. Some modern jellybean logic families have sub-nanosecond risetimes (e.g. 74lvc). If using similar families, a principal use of a scope is to check "signal integrity", which needs all the bandwidth you can get.
Most "high" impedance *10 scope probes with a 6"/15cm ground lead self-oscillate at ~100MHz. A 200MHz scope will show you that, but a 100MHz scope will disguise it
At such frequencies construction techniques and probing techniques become important.
-
The question "How much bandwidth do I need?" comes up about once a week or so. There will be threads here in the Beginners forum as well as over in the Test Equipment forum. But mostly here...
For most projects using sine waves (audio) a 100 MHz scope is probably overkill. Where 100 MHz tends to come up short is looking at square waves. A 100 MHz scope can do a decent (but far from perfect) representation of a 20 MHz square wave and that is because it can only faithfully display the 5th harmonic (and even that will be attenuated). A 200 MHz scope can similarly display a 40 MHz square wave. For Arduino projects, the clock is just 16 MHz so pin signals won't be running anywhere near that fast. Serial decoding (all 4 channels of SPI) will be quite helpful.
Move up to FPGAs and the bandwidth requirement goes way up. Probably much farther than 200 MHz.
-
I agree with pretty much everything said. My minor quibble is with Dana. He's correct, but he's quite a bit ahead of where you are. And keeping a Tek 7000 series going is not for a novice. You'll also need test gear to do it.
I bought a used dual trace 60 MHz Dumont 1060 in 1990 for over $300 with a 30 day warranty. About day 35 the horizontal section failed. The only instrument I had was a 5 MHz recurrent sweep Heathkit scope, In retrospect it should have been "good enough", but I was too intimidated to try. A few years later I picked up a Tek 465 which sort of worked. It had lots of issues so I got it in trade for a couple of ESDI disk drives. It was enough to let me trace my way through the horizontal section until I found the bad solder joints. Once I had the Dumont working I went though and fixed the bad solder joints in the 465.
There was a long hiatus in my electronics tinkering. When I started back up I bought a Rigol DS1102E because I did not want to face fixing the Tek and the Dumont so I could do something else. I wanted something I knew would work when I turned it on.
As noted, the Rigol has a crap FFT. But my Instek MSO-2204EA doesn't have vernier adjustment on the vertical channels which is a huge nuisance.
FFT length can be important. The big failing in the Rigol scopes is short FFT windows make the FFT function pretty useless. The Instek scopes will do a 10 million point FFT and sampling at 1 GS/S gives you some spectrum analysis capability to 500 MHz. It's not great, but better than nothing. If you're playing around at HF you can look at the 3rd harmonic.
I think that if you *need* 200 MHz, you'll know it and why. -
I too would recommend you take the 100MHz option unless you have a specific need for more. I hacked my Rigol 2072A to the full 200MHz bandwidth but I usually leave the bandwidth limitation at 100MHz or even 20MHz for most stuff.
Save that money until you have a specific need and than buy what you need. Or get into a local makerspace which has equipment with higher bandwidth / specs than you might ever want to buy on your own. Might be a lot cheaper. You also get to know people which can give you inputs.
-
The OP will notice differences in signal fidelity using both scopes for the same measurement BUT with the level of experience already gained from owning a scope he will understand the limitations of having less BW.
The complete scope novice.....not so much.
Absolute trace fidelity is beyond the realm of 200 MHz scopes. -
For your applications, you'll be fine with a 100MHz scope. My most used scope is an old and crusty 100MHz TDS220, next stop is the good old analog 7603.
Yes, the ordinary x10 probe won't work well at those frequencies. Fortunately it's easy to make your own low-Z probe. I made mine from 75R co-axial cable and some resistors, giving a total input impedance of 750R. 50R cable will work too, but the total impedance will be 500R.
4 channels is pretty useful, most of the time you'll find yourself using one or two channels.
Having more than 100MHz BW also requires more advanced probing techniques, and often 50 Ohm terminated inputs. Your vanilla 10x scope probe won't make any use of more than 100MHz BW - you'll just see more ringing. So 100MHz, or even as low as 50MHz is enough for the most tasks in debugging analog and uC circuitry.
-
Absolute trace fidelity is beyond the realm of 200 MHz scopes.
That is beyond the realm of any scope!
There's no substitute for understanding your tool's limitations and how to live with them. -
For your applications, you'll be fine with a 100MHz scope. My most used scope is an old and crusty 100MHz TDS220, next stop is the good old analog 7603.
Yes, the ordinary x10 probe won't work well at those frequencies. Fortunately it's easy to make your own low-Z probe. I made mine from 75R co-axial cable and some resistors, giving a total input impedance of 750R. 50R cable will work too, but the total impedance will be 500R.
4 channels is pretty useful, most of the time you'll find yourself using one or two channels.
Having more than 100MHz BW also requires more advanced probing techniques, and often 50 Ohm terminated inputs. Your vanilla 10x scope probe won't make any use of more than 100MHz BW - you'll just see more ringing. So 100MHz, or even as low as 50MHz is enough for the most tasks in debugging analog and uC circuitry.
May I ask, why did you choose 75Ohm over 50Ohm? Do you have a 75Ohm input impedance on your scope? -
Thanks everyone for your comments.
It depends on what frequency analogue signals you will be using, what you will be looking for, and what logic family you will be using.
Sometimes analogue circuits can oscillate at a much higher frequency than you are expecting ("amplifiers oscillate, oscillators won't").
In some analogue applications, harmonics of the fundamental frequencies are important; factor that into your assessment.
A 100MHz scope will (just about) allow you to see a 3.5ns risetime; a 200MHz scope 1.8ns. Logic families from the early 80s (LSTTL, STTL) have that kind of risetime. Some modern jellybean logic families have sub-nanosecond risetimes (e.g. 74lvc). If using similar families, a principal use of a scope is to check "signal integrity", which needs all the bandwidth you can get.
Most "high" impedance *10 scope probes with a 6"/15cm ground lead self-oscillate at ~100MHz. A 200MHz scope will show you that, but a 100MHz scope will disguise it At such frequencies construction techniques and probing techniques become important.
Thanks for raising these limitations/pitfalls - some links that address some of them, that might be of interest to others:- some interesting info on probing here - https://youtu.be/zodpCuxwn_o - from You Tuber Alan Wolke (w2aew)
- ... and here, tggzzz's own blog - https://entertaininghacks.wordpress.com/library-2/scope-probe-reference-material/ (Surprisingly, Tektronix's http://w140.com/tek_ABCs_of_Probes_1990.pdf incorrectly uses S (Siemens) instead of s (seconds) at various points throughout the article!)
- some useful articles on unexpected/unwanted oscillations and how to deal with them in this series of articles - https://e2e.ti.com/blogs_/archives/b/thesignal/archive/2012/05/23/why-op-amps-oscillate-an-intuitive-look-at-two-frequent-causes
In the absence of any way of seeing high frequency oscillations (frequency beyond limits of scope/probes), are there any standard tell tale signs to look out for? Is it possible the circuit would still function, but waste power at unobserved high frequencies, or are the signs more obvious?
-
Thanks everyone for your comments.
It depends on what frequency analogue signals you will be using, what you will be looking for, and what logic family you will be using.
Sometimes analogue circuits can oscillate at a much higher frequency than you are expecting ("amplifiers oscillate, oscillators won't").
In some analogue applications, harmonics of the fundamental frequencies are important; factor that into your assessment.
A 100MHz scope will (just about) allow you to see a 3.5ns risetime; a 200MHz scope 1.8ns. Logic families from the early 80s (LSTTL, STTL) have that kind of risetime. Some modern jellybean logic families have sub-nanosecond risetimes (e.g. 74lvc). If using similar families, a principal use of a scope is to check "signal integrity", which needs all the bandwidth you can get.
Most "high" impedance *10 scope probes with a 6"/15cm ground lead self-oscillate at ~100MHz. A 200MHz scope will show you that, but a 100MHz scope will disguise it At such frequencies construction techniques and probing techniques become important.
Thanks for raising these limitations/pitfalls - some links that address some of them, that might be of interest to others:- some interesting info on probing here - https://youtu.be/zodpCuxwn_o - from You Tuber Alan Wolke (w2aew)
- ... and here, tggzzz's own blog - https://entertaininghacks.wordpress.com/library-2/scope-probe-reference-material/ (Surprisingly, Tektronix's http://w140.com/tek_ABCs_of_Probes_1990.pdf incorrectly uses S (Siemens) instead of s (seconds) at various points throughout the article!)
- some useful articles on unexpected/unwanted oscillations and how to deal with them in this series of articles - https://e2e.ti.com/blogs_/archives/b/thesignal/archive/2012/05/23/why-op-amps-oscillate-an-intuitive-look-at-two-frequent-causes
In the absence of any way of seeing high frequency oscillations (frequency beyond limits of scope/probes), are there any standard tell tale signs to look out for? Is it possible the circuit would still function, but waste power at unobserved high frequencies, or are the signs more obvious?
Thanks for the video URL; it is a useful practical demo of the effect of the ground connections. (I also like his taste in using a Tek 485 as the demo vehicle ) If you are interested in the theory behind that, plus practical examples, have a look at:
https://entertaininghacks.wordpress.com/2015/04/23/scope-probe-accessory-improves-signal-fidelity/
and the later higher bandwidth variant:
https://entertaininghacks.wordpress.com/2016/09/17/scope-probe-accessory-higher-frequency-results/
I'm no longer surprised at Tek's historic misuse of SI units; I was appalled about them 40 years ago
In extreme cases even a probe's tip capacitance can cause an amplifier to oscillate, but in such cases that is probably a benefit since it highlights something that is too close to the edge of the envelope.
Don't forget that a 15pF capacitor (think probe tip) has an impedance of ~100
at 100MHz (50 at 200MHz). The 0.7pF "low" impedance Z0 resistive divider probe is still ~450 at 500MHz)
-
The "digital black magic" books by Graham and Johnson go into considerable detail about the probing problem. Their recommendation is to design in test points that can be enabled with 0 ohm resistors. They discuss using pins, but U.FLs seem a better choice to me. With an ARM clocked at 200 MHz, the harmonics extend to well over 1 GHz.
-
Ignore the frequency since it is only indirectly relevant. Concentrate on the rise/fall time since that is directly relevant.
Why? Because if you have an Xns risetime, the bad effects are the same with a 1Hz or 100MHz frequency. -
Before finalizing the decision, I wonder if there are any use cases where the extra bandwidth would really make a significant difference to understanding what's going on in a circuit?
Not hugely, no.
Signal integrity on a 10MHz square wave for example you get some more harmonics adding to the detail, but nothing to write home about. Probing is vastly more important at this point.
Maybe if you were actually measuring amplitudes of >100MHz signals it helps a lot of course, but few people actually have that requirement.
Most people simply won't notice any real usable difference between between 100MHz and 200MHz bandwidth. -
It might be worth pointing out that in the 60's the 50 MHz Tek 547 was the main work horse for the Apollo program. So you can do a lot with a 50 MHz scope. And the 100 MHz Tek 465 was the standard for a long time after that. When I worked for Amoco in the early 80's there was typically a row of 2-3 465s by the window where I took my punch card decks.
-
Before finalizing the decision, I wonder if there are any use cases where the extra bandwidth would really make a significant difference to understanding what's going on in a circuit?
Not hugely, no.
Signal integrity on a 10MHz square wave for example you get some more harmonics adding to the detail, but nothing to write home about. Probing is vastly more important at this point.
Maybe if you were actually measuring amplitudes of >100MHz signals it helps a lot of course, but few people actually have that requirement.
Most people simply won't notice any real usable difference between between 100MHz and 200MHz bandwidth.
The frequency (or more accurately baud rate) of digital signals is completely unimportant w.r.t. signal integrity. I've even seen a problem on line with <<1Hz "frequency".
How so? The line signalled an infrequent error condition. When an error was signalled the risetime was sufficiently fast to cause overshoot which caused internal diodes to conduct - and circuit malfunction.
If an anthropomorphic explanation helps... When a circuit "receives" a transition it responds to that transition. Since it doesn't "know" when the next transition might be arriving, it doesn't "know" the frequency and hence doesn't "care" about frequency.
Here's a quick, dirty and somewhat unrealistic example designed to emphasise the problem. It is modern jellybean logic (3* 74lvc1g gates in parallel) closely decoupled, 5V supply, driving a <1inch unterminated line. The probe is a 1.5GHz low impedance Z0 probe (0.7pF) and a 350MHz scope. The scope is set to 10ns/div and 1V/div.
You can see that there is a 1.6V overshoot, which is more than enough to turn on diodes in a receiver (causing malfunction and/or long-term overstress) The overshoot occurs on a transition, and the frequency is so low that it is not visible in that trace; in fact it ~100kHz.
-
My oscilloscope has an input impedance of 1M, in parallel with 13pF.For your applications, you'll be fine with a 100MHz scope. My most used scope is an old and crusty 100MHz TDS220, next stop is the good old analog 7603.
Yes, the ordinary x10 probe won't work well at those frequencies. Fortunately it's easy to make your own low-Z probe. I made mine from 75R co-axial cable and some resistors, giving a total input impedance of 750R. 50R cable will work too, but the total impedance will be 500R.
4 channels is pretty useful, most of the time you'll find yourself using one or two channels.
Having more than 100MHz BW also requires more advanced probing techniques, and often 50 Ohm terminated inputs. Your vanilla 10x scope probe won't make any use of more than 100MHz BW - you'll just see more ringing. So 100MHz, or even as low as 50MHz is enough for the most tasks in debugging analog and uC circuitry.
May I ask, why did you choose 75Ohm over 50Ohm? Do you have a 75Ohm input impedance on your scope?
I chose 75Ohm, because it would result in less loading, on the device under test, than 50R. If your 'scope has a 50R input, then use the built-in resistor and 450R on the end of the cable.
One thing to note is that the input impedance of the 'scope becomes a factor, at high frequencies. In my case 675R in parallel with 75R is 67.5R, forms a low pass filter with the 13pF capacitance on my oscilloscope's input FC = 1/(2pi*13*10-12*67.5) = 181*106 = 181MHz, which is a non-issue for a 100MHz 'scope. -
From my experience, few people need more than 100Mhz.........
Usually the extra money is forked out because the bigger the bandwidth number the "Cooler" it looks to have one.......same goes with most test equipment actually...... -
From my experience, few people need more than 100Mhz.........
Usually the extra money is forked out because the bigger the bandwidth number the "Cooler" it looks to have one.......same goes with most test equipment actually......
Just so, but it is worth knowing the conditions under which you do need more bandwidth, and the consequences of having less bandwidth. There is no substitute for bandwidth.
(Actually, in some circumstances there is, but it requires skill imagination and knowledge. But you can buy bandwidth) -
Found this page recently, and was surprised how good a 500 MHz signal looks on a 100 MHz scope...
https://hackaday.io/project/4327-stretching-the-limits-of-a-rigol-ds-1102e-scope -
Found this page recently, and was surprised how good a 500 MHz signal looks on a 100 MHz scope...
https://hackaday.io/project/4327-stretching-the-limits-of-a-rigol-ds-1102e-scope
I wonder what the phase response is; I would have preferred he did the experiment with a square wave.
My Tek 485 (350MHz) will happily display and trigger on a signal that is >1GHz. The amplitude is somewhat reduced, of course
-
To avoid aliasing, an 8 bit, 1 GS/S DSO needs to be 48 dB down at 500 MHz. So a 1 Vpp sine wave input displayed on the 100 mV/div scale needs to be reduced to 4 mV or less at 500 MHz. But you'll be able to view it if you switch to a lower input range.
I would not describe those traces as very good as there is a lot of obvious distortion. -
Wow. Not the answer I was expecting. I thought people would say if you can afford it go for it. But their rationale was very logical. I just purchased a 100Mhz 1104 and was considering returning it for the 1204. No I think you have me convinced 100 Mhz is adequate for me the hobbyist. Of course this is an old post. porches are now $499 vs $759