Basic Troubleshooting

No lock signal in lock display mode: after a noisy or blocked fan, this is the most typical situation I hear when a phone call requesting service arrives; let’s try to outline a very basic approach to understand what happens without expensive test instruments.

A very basic block diagram BlockDiag2 of a typical NMR spectrometer shows a very helpful feature: there are 3 almost identical channels, independent each other except for power supplies and some clock signals: lock, observe and decoupler. (The decoupler channel often is only a transmitter and missing the receiving portion, but this can be easily changed via software; on more modern systems all channels are identical and each can be configured to operate as lock, observe or decoupler).

Therefore, no lock signal? No panic:

  1. Get a sample with a mix of D2O and H2O; the percentage is not critical, 50% each is fine
  2. Set a very simple 1H experiment: no interlock of any kind, nt=1, pw=5 uS (any value lower than PW90 will do), sw= maximum available, gain=midscale (typ 20)
  3. go (go, ga, zg or whatever is the “do one acquisition” command in your machine parlance)
  4. look at the FID or spectrum:
  • you see the H2O line: the magnet is at field, all the spectrometer works, the trouble is somewhere in the lock channel
  • no signal: the spectrometer is not working in some basic function: a) the Power pupplies or b) the Master Clock
  • Worst case: the magnet. Check with a small screwdriver (CAUTION!) for the presence of the field, but this is only a very rough test, the magnet could be sligtly out of field.

Up to this point, the method used is almost universal and can be used on any instrument type of any manufacturer. To proceed further on, the trail very much depends on the hardware and the software you are using. As stated somewhere else on this web site, I am a Varian man for good and bad, hence I will talk in Varianese; nevertheless, I hope the method is flexible enough to be easily adapted also to German-speaking systems (I don’t know anything about Japanese).

As stated by somebody (R. Pirsig, Zen and the Art of Motorcycle Maintenance) troubleshooting a complex system is one of the best expressions of human intelligence applied to problem solving; and for a service engineer, whose job is supposed to be to find and solve malfunctions, the knowledge and proper application of troubleshooting techniques is vital: if you are not able to find a malfunction, how cay you fix it?

To start with, a few essentials:

a) know your system; note as many details as possible when it’s working properly

b) very carefully look at the symptoms and elaborate them with a comparison with the proper operation of the system

c) change the operating conditions (i.e. observing nucleus, operating frequency etc) as much as you can and note instrument’s behavior

d) isolate the defective module/assembly using the above indications

Note that the above procedures don not use any test instrument; remember that the NMR spectrometer itself is a very accurate test instrument, and furthermore you have built-in the very best diagnostic tool: your brain.

Even if you are not able to isolate the defective module, the symptoms you detected will be of great help for the service engineer of the manufacturer; he will be very probably able to ship to you the proper replacement part.

To proceed further and locate the defective component you will need the electric schematics and some equipment, DMM (Digital Multi Meter) and Oscilloscope at least; but note modern instruments are very difficult to repair at component level, because SMD (Surface Mount Devices) are used everywhere, and SMDs are very difficult to replace without special (and very expensive) desoldering/soldering tools.

A very basic suggestion: when looking for a malfunction, start from the easy things: are all cooling fans working?

A) “know your system”:

this is, in my opinion, the very most important issue; and, for my experience, the most ignored from the standard user. Unless obliged by internal GLP (Good Laboratory Practice) rules, very few users run performance tests on a regular basis; almost no user knows the standard noise level of his spectrometer at various frequencies, and the noise floor is a very important indication of system performance. To roughly evaluate noise level, do one acquisition without transmitter (pw=0), save and measure the obtained noise. Varian software has the useful ‘noise’ command that displays rms noise value and an estimation of noise figure; use it and record the results for future reference.

The manufacturer usually displaced multicolor LEDs here and there, but very few customers knows what they mean, and many ignore at all that they exist.

As an example, on the right is a photo of the board rack of the Varian Gemini spectrometer; if anybody is interested, I can list the meaning of LEDs and their behaviour. Conventionally, no red LED should be on: green is normal condition, yellow stands for attention, red is bad. One exception are the LEDs monitoring power supply voltages that sometimes are red, like shown on the left image.  In this case, all LEDs must be on.      But remember,  these LEDs  only shows if the voltage is present! If you have a spectrometer not working in several basic parts, you should better check all supply voltages with a DMM.










Photo courtesy Prof. S. Chimichi Univ. Florence



A very good example of a well designed power supply panel is this one on the right, from the Varian Inova spectrometer: green LED = OK,  red = voltage is out of proper range or missing; as final bonus, test point at hand for every voltage.



Also, listen to the console! If the room is more silent than usual, one or more of the cooling fan does not turn anymore; the reverse, if you hear a strange, noisy sound, it comes out the bearings of a fan (fans are capable to produce an unbelievable range of sounds) that will for sure turn slowly. In both cases, replace the defective fans immediately. I strongly suggest as routine to have a look, every month or so, to all the cooling fans (some of them are carefully hidden: use a small light pen) for proper operation. And remember: bad power supplies and cooling fans are responsible for the vast majority of malfunctions.

So that, a suggestion: spend a couple of hours checking your instrument carefully throughout; open the doors of the rack, look wore the fans are and note the status of all LEDs. The very best time to do that is at installation; ask questions to the installation engineer, he will usually be happy to answer. But do it when the installation is completed, since during installation the man (any service-girl around? if she exists, please introduce herself!) will probably have some other problem to think about.

B) Symptoms:

a typical malfunction is the lack of NMR signal, and here is where the noise level comes of importance; if you know the noise level of your spectrometer, compare it with what you have now:

1) The same level but no signal? > Most likely the observe transmitter has no power out at all or very low power.

2) Very low noise level? > Receiver channel with low gain or completely dead

Some obvious suggestion:

1) try to increase pw (pulse width) parameter and look if the signal comes out; use a concentrated sample. If it appears, transmitter has very low power. Check probe tuning, if not already done; if probe is consistently out of tuning (and you are 100% sure nobody tuned it to a different nucleus!) suspect the observe frequency synthesizer is out of frequency.

2) use another observe channel, i.e. observe 13C instead of 1H; almost all spectrometers have 2 fully independent observe channels   BlockDiag2    namely HighBand (1H + 19F) and LowBand (all other nuclei); if one is working the problem is somewhere in the other one.

Note that with the #2 test we are using the c) method: “change the operating conditions”

Some experience will suggest a number of methods; please note that is very difficult to damage a spectrometer (i would say it’s impossible) only changing the operational parameters, unless you disable the hardware and/or software safety protections that are always preset and enabled at the installation. Feel free to change parameters, and in case of doubt call service. As rule of thumb, the average service engineers much more prefers to spend some time on telephone than several hours of travel to fix a simple trouble.

As we proceed further on more detailed troubleshooting procedures, they become really instrument-dependent; deciding what board or module is more suspect needs to know the structure of your particular model, what is the function of each board and were the boards are physically positioned. This is the job for manufacturer’s service engineers, but the above suggestions and checks will help service people quite a lot.

I have in my basement an operational console of a Gemini spectrometer, and have easy access to Mercury and Inova instruments (all Varian gear, as you know) and i can easily produce a detailed troubleshooting guide for those spectrometer; but this a several hours work, and I’ll do it only if it’s a real interest in the NMR community. Mercury, Inova and even more Gemini are obsolete machines, and I really don’t know how many of them are still on the field; but I’m not a youngster anymore, and these are the “top models” of my times. Hence, write a post in my vpnmr website asking for help, and I’ll do my best.

Well, that’s enaugh by now. Feel free to reply if you would like to have some more information on troubleshooting techniques, useful equipment etc.





Leave a Reply

Your email address will not be published. Required fields are marked *