Hi all,
The responses to this query on whether oscilloscopes are useful for
modern digital spectrometers were really great! Many thanks to the 19
of you who responded.
Briefly, the consensus opinion is that oscilloscopes remain very
useful for troubleshooting RF hardware. However, I was definitely
mistaken to think that one needs a scope with a frequency of at least
twice the spectrometer's 1H frequency for it to be useful. Many people
are using 500 MHz and 1 GHz scopes to troubleshoot even the highest
field NMR systems, and several respondents described doing basic
troubleshooting with lower-frequency scopes. This is very significant,
as there is a major cost difference for scopes above 1.0 GHz (2.5 GHz
was the next step up for the major manufacturers I searched).
The big factor one needs to consider when measuring peak-to-peak
voltages at high frequencies is calibration. The higher in frequency
one goes, the less able the scope is to accurately measure the Vpp.
But the drop will be consistent on the same scope, so one can
calibrate it with a source of known power and compensate when
calculating actual power.
Some respondents indicated that they have been using their scopes less
with more recent instruments. From noncomprehensive observation, this
seems partly due to there being fewer RF components for systems where
RF is digitally synthesized, partly due to the greater reluctance
(real or perceived) of manufacturers to withhold schematic-level
information that one needs to make scope measurements meaningful and
useful, and partly because on-board diagnostics are supplying more
information for troubleshooting, obviating the need to further probing.
It appears that consensus holds that these tools remain very useful,
but it is unlikely to be worth spending the money to get an optimally-
performing scope. For the relatively simple needs of an NMR facility,
a 500 or 1 GHZ scope seems to suit everyone well and only requires a
largeish, but not staggeringly enormous, investment. No one mentioned
a need for "deep memory" on their digital scopes, so it seems
worthwhile to live with the basic memory that comes with the device. I
posit that ease-of-use issues should take priority over RF performance
when selecting a scope for our needs; this increases effective
reliability and encourages rapid, simple application, while ultra-
precise power measurements are not that important.
The edited responses are listed below.
Thanks again for all your help!
- Josh
*****
While you may not need it as frequently as on older hardware, it is
still a very useful tool to have for all the modern digital systems
that I have worked on.
...
It is typically good enough for relative measurements (Vpp at input
vs. Vpp at output, or similar).
It has some measurement accuracy limits if you want to see whether
your amplifier output is 290W or 310W, I'd have to check the specs to
give you the correct tolerances here. But then again, this is of
typically quite limited practical value if your goal is to maintain a
working spectrometer.
The biggest rate limiter will be harmonics and sidebands riding on
your signal, as those are typically not relevant to the NMR, but will
show on top of the time-domain signal displayed by the scope. If you
want/need to get over that limitation, you would need a spectrum
analyzer and be frequency-selective.
Old, bandwidth-limited analog scopes responded in this case with a
'fuzzy' envelope, while modern scopes are a lot more revealing,
requiring a bit more interpretation from the user.
But for troubleshooting, the scope typically does just fine, and will
tell you whether you are losing 5% of voltage or 30% of voltage on a
given cable or connector or filter, etc.. In addition to tracking all
the control line signals etc., which is why I would make this my #1
tool (combined with a high-power 30dB attenuator and a few smaller
10dB units).
*****
In response to your question about scopes - I bought one (Agilent 1
GHz, cost about $35,000) about 2 years ago because we had money from
**** and we have used it about 5 times over two years on our Bruker
AVANCE II and III units, i.e. to date about $7000 per use, vs. $5000
per day for service engineer visit. I leave the scope with our
Electronics Technicians (in our Science Technical Center) who use it
daily and then I "borrow" it whenever I need it. Even in these days
with digital receivers and other digital components I still believe a
scope is essential to check frequency, look for pulse imperfections
etc (much easier with new digital scopes). It is so fast and easy to
trace signals at least for analog parts (amps, preamps, probes) and if
you are trying new configurations an easy (and safer) way to check you
got connections right without e.g. destroying cryoprobes etc. I also
need one available as our excellent Bruker Service engineers would
typically not bring one on the plane, and in fact I can only recall
one engineer doing so (Varian one who didn't trust my tests) bringing
his own scope in 20+ years of NMR repair supervision! That being said
most of the tests I've used a scope for could now be done using the
spectrometer itself with correct re-cabling, but this is tricky and
prone to damage and errors, and if the receiver is broken, it won't work
Bottom line = don't buy if you don't have surplus cash available and
find out if someone nearby has one and borrow that if you need to.
*****
At our facility, we have 9 spectrometers 500MHz-900MHz. We also have a
network analyzer (1 emit/receive, 1 receive channel), a digital
oscilloscope
(1GHz) and a power meter. We use all three units on a regular basis to
test
and troubleshoot spectrometers. The oscilloscope is particularly
useful for
the following tests:
SGU output, router output, amplifier input (mostly peak-to-peak voltage
measurements, sometimes frequency/waveshape measurements)
Amplifier output (using 1kW load and directional couplers, fall/rise of
square pulses and shape of selective pulses)
Probe tuning/arcing (via directional couplers)
We find 1GHz digital oscilloscope to have just enough resolution for 1H
900MHz measurements. So far (8 yrs) we haven't find ourselves in need
to use
a 2.5GHz.
....
You are correct, absolute peak-to-peak measurements may not be perfectly
correct - you have to calibrate them. On the other hand, you have to
calibrate things anyway - at 900MHz, you lose 1dB (15-20% of peak-to-
peak
voltage) per meter of thin coaxial cable. I am not particularly
concern with
absolute voltage measurements though. We ran base measurements of all
channels at all points on all consoles to begin with. So, now I know
that
on, say, 800 spectrometer, channel 1, SGU output at 13C frq is 900mV,
router
1 output is 1000mV, amplifier input is 950mV, etc..., the outputs are
linear
between -6dB and +12dB within oscilloscope error. If smth goes wrong
with
the spectrometer, I start running test measurements. If I see that SGU
output is 500mV/-6dB or non-linear - here it goes. SGU is sent to
Bruker...
Network analyzer is a cool toy. It's basically a sweep generator that
measures signal loss. You can adjust frequency sweep range/center/span
quite
precisely and it has huge dynamic range (>70dB) of signal amplitude
measurement. Say, you connect it to a 400MHz bandpass filter, and
measure
signal transmitted through a filter. You can see at what frequency how
much
signal is attenuated. Also, you can use to measure signal loss in
cables and
attenuators.
It can be also used to tune probes. Bruker wobb routine is basically a
very
poor quality virtual network analyzer. The real NA is much more
precise and
more convenient to use. If you do some sort of probe
repairs/troubleshooting/testing, then NA is very useful.
Analog electronics folks use network analyzers all the time, so you can
probably find them at electronics or applied physics depts.
*****
First, our "digital" vnmrs spectometers have plenty of analog signals
flowing through them that a 'scope can usefully look at. It has made
troubleshooting much easier in my experience.
Second, don't buy the (marketing) idea that you need something like
2.5X or 5X bandwidth for the 'scope to be useful. Bandwidth indicates
the point where the oscilloscope's response is at -3dB; you can still
see the signal at or beyond that point, even using a 500MHz 'scope you
should be able to see a 600MHz signal. If you wanted to measure the
signal very accurately it would be a different story, but usually we
are just checking for signal bad or good. If I measure the signal
into and out of an amplifier (with appropriate power attenuators!) I
can still find the gain of the amplifier as the measurement errors
should cancel.
The sampling rate is important -- that needs to be sufficiently high
that you don't get artifacts; 2X is the minimum per Nyquist, but 4X or
5X would be better.
We have a Tektronix DPO4104, 1 GHz BW, 5 GS/s, 4 channels (1 would be
sufficient) with many features that I have never used although the ft
function was nice to show harmonics (on lower frequencies i.e. 13C on
a 600) that verified for me that the amp was having problems. It can
save data to flash cards, which I probably should use to save known
good examples to use for comparison when we have a problem. With
educational discount, I believe it was around $13k a few years ago.
********
I don't have a good scope here actually, but have not needed a fast
one yet (knock on wood). At XXXX, I found it sufficient to have a scope
with a trigger bandwidth that was at least as high as the spectrometer
frequency. My older 500 MHz digital scope would trigger above 800 MHz.
Clearly the voltages shown are not correct, but for a lot of purposes,
it
is enough to know if there is a signal there at the right frequency, or
not.
I have been intrigued by these pretty cheap scope interfaces where you
use
your computer as the display and the controls, and the hardware piece is
just a little box ... but I haven't bought one yet.
*****
I prefer having one at my disposal during troubleshooting. I use it to
verify proper routing of RF signals through the system.
PTS Continuous Wave output verification & switching of frequencies is
also important as well as RF pulse shape from the amplifiers.
You don't really need 2X the proton frequency. I have a Tektronix 3054
which is overkill.
I'd suggest a two channel model for _at_ $5 or $6K.
*****
I find my 500 MHz Tektronix TDS5054 4-channel digital phosphor scope
to be indispensable for troubleshooting my Avance III 400 console. I
also use it routinely for monitoring reflected power with an in-line
directional coupler as a way to detect tuning/matching drift during
long or variable-temperature experiments or sample/coil heating during
long pulse trains. I have not, however, had a need to accurately
measure signals greater than 500 MHz although such signals certainly
exist within the console hardware. If you already have a 1 GHz scope
and 2.5 GHz is the next step up in bandwidth, I would just stay with
what you have if funding is tight. After all, even if a 1 GHz scope
can't accurately measure the peak-to-peak voltage of a signal with
frequency greater than 1 GHz, you can still visualize the waveform,
accepting increasing signal loss with increasing frequency.
*******
We've never been able to afford a high bandwidth scope; there have
been times when it would have been helpful to have a scope that could
display the Larmor frequency, but in over 30 years of troubleshooting
spectrometers we've always been able to diagnose RF problems just
using a 100 MHz scope, a crystal detector and the resources of the
consoles themselves (an NMR spectrometer is a very good spectrum
analyser!).
Our experience with digital consoles obviously only goes back a few
years so we've had relatively few problems. The lack of circuit
diagrams for modern spectrometers, the lack of access to internal
signals, and the proprietary nature of a lot of the chips, effectively
mean that the main components that users can troubleshoot are just
from the low power transmitter through the RF switching, probe, and
preamp to receiver.
******
we have found our digital )-scope quite useful working with our Bruker
AVIII's and AVIIs you really should not need need a scope above
1GHZ. , what might be very useful is a power meter
ther are several very good usb based power meters out there for a very
reasonably price. We had a Bruker service course about a year ago and
Bruker used the scope extensively on our AVIII to show us various
diagnostic signals.
*******
An O'scope's "frequency" can be one of 2 numbers: Bandwidth(BW) or
Sampling Rate(SR). It is important to distinguish between the 2. I
assume you are speaking of the SR, which is only a specification for
digital 'scopes. BW is the term which applies to all 'scopes
regardless of operating mode.
For a 600MHz spectrometer, a cheap 300MHz BW analog 'scope is
sufficient to measure every aspect of the spectrometer except 1H
frequency and amplitudes. 1H frequency signal presence can be
verified at the lower BW, but not accurately measured. This is not
true for an equivalent 660MHz SR digital 'scope, because the digital
sampling rate is too low to get even the correct 600MHz waveform. To
use a digital 'scope on a 600 takes a 600MHz BW, and at least 1.32GHz
SR digital 'scope. Thus you may see the distinct disadvantage to
using digital 'scopes from a cost perspective.
To troubleshoot a 900MHz or 1GHz spectrometer will only require a
500MHz BW analog 'scope. To measure 1H amplitudes and frequencies,
you can buy a cheap power meter and counter for much less than the
upcharge to a 1GHz BW (at least 2.2GHZ SR) digital 'scope. Even a
500MHz BW digital 'scope will work if it has at least 2.2GHz SR (many
500MHz BW 'scopes are 4GHz SR).
Now I'll put out another distinction - between RF and high clock rate
digital signals. In the Inova spectrometers, the fastest digital
signals are clocked at around 30 MHz, IIRC. In a digital
spectrometer, the CPU chips could run at several GHz even on a 300MHz
spectrometer! There is no way the average lab can afford a 'scope to
see those digital signals. That is why high clock rate digital
troubleshooting is best done through self-testing software and board
swapping, rather than 'scope monitoring. You won't see an
oscilloscope in a PC repair shop for that same reason.
So, first one must decide which signals are of interest, RF or
digital? If RF, then I would get a 'scope with 500MHz BW and at least
2.2GHz SR to troubleshoot a 1GHz spectrometer, plus a power meter and
counter if accurate measurement is desired at the 1H frequency. If
you only really need to verify signal presence without accurate
measurement, then the power meter and counter are unneeded. Some RF
troubleshooting can be done at lower frequencies by swapping
channels. To troubleshoot digital signals, I recommend board swapping
because the cost of an appropriate 'scope is prohibitive.
*******
I don't troubleshoot my spectrometers with an oscilloscope. On
Brukers, the idiot lights are enough to follow the problems down to
the board or fuse.
I'm an idiot and these are easy to follow. I have more problems with
my Varian (Inova). If your good with scopes, it may be worth it.
IMHO and background, I don't need a scope. Varian TAC lines may
disagree.
*****
We have a Bruker AvanceII console. The only time I've found a scope
helpful is when trying to diagnose a problem from the RF generation
board through the amplifier to the preamp and the probe. As we have
multiple probes and signal generation boards, we can swap probes or
boards to check these components. The scope is useful for determining if
problems are in amp or preamp. However, we have a parts replacement
service contract with Bruker and they will send us both amp and preamp
to swap out. Without the service contract, a scope would pay for itself
when an amp was lost.
The internal diagnostics on the digital consoles, in my experience, are
sufficient to diagnose most other problems.
*****
Well, my simple impression is, as long as the RF radiations used in an
NMR spectrometer comes out via connectors and cables that are humanly
accessible, an O-scope is always a valuable and preferred tool. But,
how long will it take before that situation changes, I cannot fathom...
****
I have a 350MHz analog scope that I have used many times to successfully
troubleshoot problems with my instruments ranging from 300-600MHz. We
did
buy a higher frequency scope for the 800/900MHz labs, but haven't had
much
occasion to use them. A good analog scope can be used to troubleshoot
frequencies well above the rated frequency; however, the levels will no
longer be quantitative. One can "calibrate" the response (using a known
amplitude signal) at various frequencies, and still use the slower
scope to
measure power fairly accurately above its rated frequency limit.
With digital scopes, it gets more complicated because you have to worry
about aliasing and the Nyquist limit of the sampling rate of the scope;
however, with experience, you can learn to work with the aliased
signal and
still figure out what is going on. The amplitude calibration issue is
still
present with digital scopes being used above their frequency limit.
Bottom line, every lab that hopes to troubleshoot instrument problems on
their own should have a scope available (and somebody who knows how to
use
it properly), and it is always better to have a scope that is rated at
(or
above) the highest Rf frequency to be measured. But, as is often the
case,
we are sometimes forced to "make due" with less.
*****
I no longer troubleshoot, can only talk about the days past. A scope
is hardly ever used to look at the radio frequency. It is used to look
at pulse sequences, switching signals etc. A 100 Mhz scope has a
resolution of 0.01 us, good enough to look at pulses, if RF via a
detector diode. To look at RF, the actual waveform, you would need
a sampling rate of at least 5 times the frequency just to see the sine
wave. without any harmonics.
So any decent scope, together with an RF diode, a calibration chart
and a handful of attenuators worked fine for me. Look for multi channel
inputs and dual, delayed time base options to compare pulses and
enlarge portions of a sequence when needed.
A decent calibrated RF generator can come handy for doing
substitution measurement.
*****
I don't have a lot to contribute to this topic, but I would
make a couple of points.
- I has been several years since I have had a need to pull out
an oscilloscope to troubleshoot any of my systems;
- I have probably found myself wishing for a spectrum analyzer
more than a high frequency scope;
- With the lack of schematics, and documentation in general,
for the newer consoles, there is even less that one can
do to troubleshoot the systems. It has been years since
I have done any component-level troubleshooting for anything
past a power supply. I am happy when I can localize the
problem to a board to swap out with the manufacturer.
Obviously, there is some troubleshooting that can still be done,
and in some cases a scope might be useful, but then are getting less
frequent. It may be hard to justify the cost of a high end scope.
I know if I had a total lack of signal, I would want to look at
pulse and local oscillator signals first off, but.....
****
In our experience I would say that we use the scope less and less as
we move toward newer consoles. In fact in recent years I think we have
only used this for older systems for amps with parts that can be
desoldered and replaced and not at all on our more modern Bruker
consoles (AVIIs & AVIII between 3-6 years old). We considered whether
to invest in a new scope, but decided we could not justify this for
the amount of use it would likely get.
****
I don't think I could run an NMR lab without a decent scope. There are
two which are permanently installed on our 400 solids and 500 liquids
machine and are on all the time.
I'm not sure that a 2x spectrometer frequency is needed. I would think
that a 1 GHz scope would serve a 900 fairly well. What is important is
a LCD scope. Electron bean scopes are nice but nowhere near a magnet.
ORIGINAL POST:
Hi all,
I'd like to know whether you think an oscilloscope is a useful tool
for troubleshooting modern high-field digital spectrometers.
I'm performing my due diligence for my new lab, which currently does
not have a scope with sufficiently high frequency for our systems. We
are considering purchasing a scope that not only meets our current
needs (systems up to 600 MHz), but will also meet the needs we aspire
to have in the future (900 MHz, for the sake of argument).
I found my old 1 GHz scope to be very valuable for troubleshooting
older 600 MHz Inovas, but we haven't yet had to do much RF
troubleshooting for our relatively new Avance-III and VNMRS consoles.
I don't want to be in the position of buying a $30K 2.5 GHz scope
today, only to find years from now when our Inovas are all retired
that the digital consoles aren't amenable to being probed by
oscilloscopes.
If you have digital consoles that you need to troubleshoot, please let
me know whether you find oscilloscopes of appropriately high frequency
(~2X spectrometer frequency and above) are useful for you, or whether
you've found you do not need one.
Thanks.
Josh Kurutz, Ph.D.
Instructor and Senior Scientist for NMR
IMSERC, Chemistry Department
Northwestern University
2145 Sheridan Rd.
Evanston, IL 60208-3113
847-467-1949
fax: 208-978-2599
Facility: www.chem.northwestern.edu/imserc
NMR Blog: www.imserc-nmr.org
Twitter feed: twitter.com/imserc_nmr
Other: www.joshkurutz.com
Received on Mon Sep 27 2010 - 20:05:33 MST