Many thanks for the wealth of answers on the drift of our MR-400.
Summarized below are the original email, and many of the responses I
received. The general consensus: we could have been luckier, but no
disaster here - we're going to continue to watch it along with Varian.
~David Sibley
david.sibley_at_furman.edu
On Jul 17, 2008, at 10:13 AM, David Sibley wrote:
> Hello All -
>
> We recently purchased a Varian MR-400, and the engineers brought it up
> to field back in November. Since Varian officially "certified" the
> installation complete in January, we've been noticing what, to our
> minds, is a fairly significant drift for the 400 MHz magnet - about 12
> DAC units/hr - which works out to about 4 Hz/hr. Varian says this is
> within spec - their limit is 8 Hz/hr.
>
> That said, we're having to reset the "lockfreq" parameter from an HDO
> standard about once every few months to keep all solvent systems
> accessible - a procedure we didn't have to repeat at all over several
> years on our other magnets. We've also had other reps express their
> surprise at the magnitude of the drift - even if it is within spec.
>
> What is your collective experience out there for drift on the MR-400?
> Have you experienced similar drifts in this magnet? Even within the
> Varian spec, what would you consider an "acceptable" amount of drift
> to be?
>
> We're a smaller, primarily undergraduate institution, so our
> experience is limited to only about 3 high field magnets - our Inova/
> Oxford 500 (which still runs like a champ), a now decommissioned
> Varian R2D2 300, and now this new Varian/Magnex 400 - so we're totally
> aware that this just may be a bit of paranoia on our part - that said,
> we've never had a magnet drift quite this significantly before.
>
> Your help is appreciated -- I will summarize responses and repost them
> to the list.
>
> David Sibley
> Furman University, Greenville, SC
> david.sibley_at_furman.edu
> David, Good magnets decay like 0.1-0.2 Hz an hour.
> 4 Hz is a lot by today's standards, but you have no legal recourse,
> as the drift
> is within the installation factory spec. This drift is caused most
> likely by an
> imperfect weld in connecting one end of the coil to the other. I
> used to take
> care of a 400 that would require a field bump from a power supply
> every year or
> so (it drifted about 40 Hz an hour or something outrageous by today's
> standards). It started with being unable to lock on certain
> solvents (upfield,
> as I recall, e.g., acetone-d6), and that was the cue that it was time.
>
> As long as you are not losing the lock during long nD runs (you
> probably use
> this instrument mostly for 1D's and the occasional 2D), there is
> really only
> the inconvenience of not being able to remember solvent-specific z0
> value.
>
> There is a command called setlockfreq (I think this is right) that
> you might
> want to run every so often to keep the z0 values in line, or you can
> manually
> edit the lock frequency using config (you are probably running
> VNMRJ, so I'm
> not sure how much of this information will translate from VNMR 6.1C
> -- what I
> am running here).
>
> It's just one more thing to do to keep the wheels on the bus.
> We've got two Oxford actively shielded magnets (I could look up
> the model # tomorrow, I'm replying to your e-mail from home) that were
> purchased back in 2001. The 500 drifts at about 2.5 Hz/hr. so every
> 3-4
> months I'm doing the lockfreq reset shuffle. The 600, well I'm not
> sure
> what the drift rate is on the 600, it's that low; I think I've only
> had to
> reset the lockfreq 2-3 times in its 7 year lifetime. We've also got
> an
> actively shielded 800 from Magnex (now a wholly owned subsidiatry of
> Varian NMR, Inc.); it drift rate is less than 1 Hz/hr. which is pretty
> astonding. The way one of the prof.'s here who's got a fair amount of
> magnet experience explained it to me, the magnet companies keep pretty
> good statistics on drift rates when they're ramping up production of a
> particular model. If 75% of magnets of a given model drift under,
> say, 2
> Hz/hr. and another 20% drift under 8 Hz/hr., and a few drift over 8
> Hz/hr., they set the spec. at 8 Hz/hr., minimizing the number of
> magnets
> that they have to take back to the factory. Maybe that's kind of
> cynical,
> but a large portion of the magnets are pretty far under spec. Is
> there
> anything out of the ordinary about where you have the magnet sited?
> We
> actually have a fairly annoying vibrational problem with our 800 that
> no-one can find the cause of. We've turned off all potential building
> sources of vibration, and still it persists. Varian says it's our
> site, but neither of the lower field magnets are picking it up.
>
> Hope someone else has a possible solution (or maybe at least deeper
> insite
> than I). Good luck.
> 4 Hz/hr is not bad for a 20 year old magnet. By today's expectations
> it is high.
>
> The drift rate can be influenced by magnet cycling on installation.
> Have you measured the drift rate as a function of time. Is the drift
> rate a constant, increasing with time or decreasing with time?
> We have a number of magnets, and none drift that much (although a
> couple did for a couple months following installation, but then
> slowed down in their drift). All our magnets drift <2Hz/hr.
>
> 6Hz/hr is a pretty standard upper value for magnet drift
> specifications; 8Hz/hr seems high. I don't think you should be
> worrying about 4 Hz/hr, if the following is true. You should not be
> having a concurrent lineshape (or shim) drift. If you shim to a good
> (it doesn't have to be exceptional, just good) lineshape, and then
> take a spectrum a few times a hour during an overnight run while
> unlocked (i.e., acquire data in a normal drift experiment, but
> insuring a start with a good shim), there should not be a
> "substantive" change in the line shape. The key is what
> "substantive" means. To some extent, it depends on the experiments
> being run, or possibly to be run in the future, on the
> spectrometer. Lineshape drift is not a standard specification given
> for magnets. I have included one on our last couple purchases, but
> had to discuss it in some detail with the vendors because it is
> unusual.
>
> For you at this point with your MR400, I suggest the following.
> Decide, in an objective manner, how much line shape drift is
> acceptable for your research; or alternatively, how much drift would
> negatively impact data you might need to collect on the spectrometer
> (now or in the future). If the drift is too large, you should then
> be able to make a case with Varian that they should assist in
> correcting the problem (and there are a number of ways to settle
> down" magnet drift). As an example, if some researcher needs to
> collect multiple hour data unlocked on this magnet, you then to have
> very stable drift. That is an unusual requirement, but certainly
> not unknown. But even for locked experiments, the lock can only
> compensate for the main field drift. If the lineshape drifts from a
> 0.2Hz half-height width to a 2Hz half-height width after 12 hrs,
> then any longterm data will be seriously effected. If the lineshape
> only drifts to 0.5 Hz, then case is harder to make (but may still be
> valid). We have one magnet that about doubles the half-height width
> after 12 hr (but there not due to main field drift, but rather a
> cryoshim drift). All our other magnets have very little lineshape
> drift.
>
> Even if the lineshape is reasonably stable over 12 hr, you should be
> watch you rt shim values. Plot the amount of DAC changes each
> month, and extrapolate how long it will take for the shims DACs to
> run out of range. Our one magnet that has a shim drift has a real
> problem with Z1 drifting such that the Z1 rt shim runs out DAC
> range. Once that happens, we'll have to go into the magnet to
> adjust the Z1 cryoshim, an expensive procedure.
>
> There is the possibility of having someone go into the magnet with a
> power supply, and cycling the current to try to reduce the drift),
> but likely Varian wouldn't pay and it would be expensive to do. If
> the data is not being effected by changes in lineshape, then the
> only problem is the occasional lockfreq correction. I would likely
> be ok with a magnet like that in our facility.
> I have two Oxford magnets, installed in 1987, and they have been in
> my custody since 1989. One is 7T and the other 11.75T (300 and 500
> MHz) and they have exhibited ZERO drift. Now on their second
> consoles, I have never had to reduce the lock frequency to follow
> their decay.
>
> However, my 400 MHz Oxford is actively shielded and has required
> repeated adjustment of lockfreq since it was energized. Another 400
> actively-shielded magnet upstairs has behaved similarly.
>
> FYI - I don't know what conclusions to draw, but I suspect there is
> a price to pay for active shielding.
> I think you could have been luckier with your magnet, but it's not
> disastrous. I currently have six magnets which range in drift rate
> from negligible up to about 3 Hz/hour, but that does not affect /
> reflect how well they work. A drift rate of 3 or 4 Hz per hour is
> a bit of a pain because you do have to reset lockfreq a couple of
> times a year, but aside from that inconvenience the lock will easily
> handle that drift rate and it will not affect the quality of spectra
> in deuterated solvents (if you want to run samples which are
> unlocked over long periods, that's another matter). What matters
> more is that the shims are stable (ie the drift really is just Z0)
> and that you can shim easily on going from sample to sample. The
> magnet I have that drifts at about 3 Hz per hour is used to run
> spectra under automation and performs very well (has done for
> years), so I would not worry too much.
> One thing to consider is different shim systems. Your 500 Inova has
> the Varian made "Tim" shims that run +/- 32K. I believe the MR400
> has a 14 channel shims with a range of +/- 2048. 4hz/hr is not bad
> and you may find it's still decreasing over time.
> I've never had a magnet drift more than 1 Hz / hr after a few
> months. A spec of 8 Hz/hr at 400 MHz seems very high.
> Actually, I do not think that 4Hz/hr. is unusual. Are you sure it is
> really 4Hz/hr. The real way to measure it is to take your 2Hz D2O
> (Doped D2O) sample, or the lineshape sample, lock and shim it and
> then turn the lock
> off. Then set up a long (say 12hour) experiment where all you do is
> to record a new spectrum say every 5 minutes for the whole 12 hours.
> Look at the peak position as a function of time, and that will allow
> you to directly calculate the drift in Hz/hr. If it is really 4Hz/
> hr. this is not a problem. Each magnet is different. We have some
> which drift much less than that and some which are higher than that.
> The reason you may have to reset your lockso often is that it is not
> very broadbanded.
> Our MR was brought to field in October and had a magnet drift of 6
> Hz/hr then. At this time it is still drifting at >2 Hz/hr. Varian
> tells us the same thing about the specification of 8 Hz/hr, but I
> would argue that 8 is a value for a magnet that has just been
> brought to field, not one that is several months old.
>
> We, too, have an Inova (400) with an Oxford magnet that has never
> needed the lockfreq adjusted in 10 years. We also had a stable R2D2
> 200MHz magnet that was shut down prior to installation of the MR.
> We also have had some Varian representatives think the drift is
> excessive. Initially rather than tell us this was normal, they came
> to find out what was wrong with it, when they couldn't find
> anything, then they said it was within spec.
> It's okay. Is it a shielded magnet or Helium long hold? You could
> be drifting because the magnet hasn't fully reached 4 K. Though
> it's annoying, it may still calm down to a much lower drift rate
> over the next couple of months. I started at UConn in 2000. I had
> a magnet drift up (as it cooled to 4K) for 3 years. It hasn't
> drifted since 2004. There is still a chance you'll see yours calm
> down.
> Your numbers don't seem excessive - I too was confused about the
> differences in DAC drift versus magnet drift. Our wide-bore Oxford
> 500 has only a 3 Hz/hr drift but Z0 must be updated weekly and the
> lockfrequency changed regularly. The narrow bore Oxford magnets
> hardly change. Here is the response from Mark Stevenson of Varian
> some years back to a question about our wide bore 500 :
>
> "Regarding your concern about Z0 moving 270 DACs per day. The
> approximate strength of the Z0 coil is 7 KHz for a DAC change of
> 32,768. This means that for every 1 Hz change in the main field the
> DAC setting will change by approximately 4.7 DACs. For a drift rate
> of 3 Hz/hr one would expect a change of approximately 4.7 X 3 X 24 =
> 338 DAC/day so your observed changes of 270 DAC/day are normal."
>
> It may be that the Z0 coil strength is different in the Magnex
> magnets compared to the Oxford design.
> I am sure you will get many opinions on both sides of this issue.
> I have seen many magnets during my days at Purdue and here at USC
> but have no direct experience with the magnet type Varian uses with
> the MR 400. However, all supercons behave similarly in that each
> is unique and the performance can vary widely. There are not many
> magnet out there have drift rates less than 1Hz/hour. While your
> drift is not ideal it is far from making the instrument useless/
> problematic. At USC we operated a Mercury 400 with an older magnet
> that drifted over 25 Hz/hr. The magnet was one of the first 400’s
> built and behaved this way for it whole life. The overall
> performance with the Mercury console was OK. The lock system could
> keep up with the drift so long term experiments were fine.
>
> We were adjusting the lockfreq every couple of months on the Varian
> console. Before this console the magnet was part of a Bruker
> system and the console had no ability to adjust the lock frequency.
> Therefore, the magnet had to be brought back into the lock range
> every two years. As you know this is a more difficult adjustment
> than changing the lockfreq.
>
> When we moved our lab into a new building the old 400 magnet failed
> and we had to replace it. The new magnet (an unshielded Oxford)
> drifts at about the rate as your magnet. We are changing lockfreq
> about twice a year on the Mercury console. We are using CDCL3 as
> our standard lock solvent. When I make the adjustment I put the new
> “correct” Z0 close to +1000 dac units. This is 3/4 of the max value
> (Z0 can span +/- 2048 on the Mercury). This gives the drift a wide
> range. When it is time to change the lockfreq, the Z0 for CDCL3 is
> about –1000 dac units but higher field solvents like acetone can
> just barely be locked. So in about 150 days we have moved about
> 2000 Z0 units (~13 units /day).
>
> The biggest issue of having to do this is that different operation
> frequencies will yield different sets of spurious artifacts in
> spectra. Each time we adjust lockfreq brings different artifacts.
> Some frequencies are cleaner than others. At this drift rate I
> think I will have to adjust the magnet once every 8-10 years. Even
> with these issues the instrument is fully functional and the
> performance is good. Your system may or may not have issues like
> mine but I suspect the system will be fine.
>
> I am surprised that at your drift rate you need to make lockfreq
> adjustments so frequently. I would have thought that the MR-400
> console would have similar span to the Z0 coil output. There is no
> reason you need to “zero” Z0 when you make the adjustment. Can you
> make larger changes with lockfreq and do it less often?
>
> As far as dealing with Varian on this issue. The bottom line is
> that the agreed-to specs were met so you really do not have much of
> a beef with them. Even with this drift the instrument is fully
> operational. The bigger issue is the magnet homogeneity and its
> stability. Does the lineshape fall apart as the magnet drifts? If
> this is a problem then you can make a case for Varian to address the
> issue.
> The Varian spec is reasonable, and your drift is indeed within it.
> The
> spectrometer is intended to run with that drift, which is why the
> features
> to lock and to reset the lock frequency are there in the first
> place. You
> were simply fortunate to have sweeter magnets in the past. Is your
> drift
> above the center of the distribution for 400MR drifts? Probably,
> but Varian
> won't help you as long as it performs as advertised.
>
> The significant issue I would be concerned with is: how is the drift
> changing? If it were mine, I would plot the drift over time to see
> the
> drift profile before I made any conclusions. I would also monitor
> the LHe
> boiloff carefully to ensure it is within spec. If not, then
> negotiate with
> Varian for compensation due to excess cryogen costs.
> I had a similar problem with a Bruker AMX-400 WB. It was within
> specs but
> we had to boost the magnet every 10 months. Bruker sent us the
> power supply
> and we did it ourselves after a training secession. Good Luck
> We have two MR400, installed in February. One of them has been
> stable from the beginning at about 1.6 dac/hr. The drift on the
> other magnet was initially about 16 dac/hr but has slowly come down
> to about 3 dac/hr. Hopefully yours may still also come down further.
>
> We also have an old 500, 400 and 300 Oxford magnets, and the drift
> on them is almost negligible.
> They must be crazy. You are not paranoid. This is an absolutely
> unacceptable drift rate.
> We had a MR-400 installed last August; it's drift was about 1 Hz/hr
> after a month, and now is less.
> I don't have a MR-400, but I do have a Varian VNMRS-600 that has a
> drift rate of 600 DAC units/day or 8 Hz/hr. Varian also says that my
> magnet drift rate is within specs (10 Hz/hr). I change the lock
> frequency once a month.
>
> I thought this drift rate was excessive because I have a 13 year old
> INOVA-500 that had the lock frequency changed once at year eight, a
> two year old VNMRS-400 that has not had the lock frequency changed,
> and a GEMINI-200BB that had the lock frequency changed once at year
> 11.
>
> After many discussions with Varian, I have resigned myself to the
> fact that I will have to live with the drift. Welcome to the club!
> I don't have any experience with Magnex 400's but Oxford shielded
> 400's drift quite badly for the first year or so. We have an Oxford
> 400
> circa. 2000 which barely made specs after 1 month but had no drift at
> all after about 18 months and only about 1-2 Hz/day after 10 months.
>
> My impression is that new magnets take a very long time to truly
> settle down to zero drift even though the initial drift is sometimes
> quite bad. This puts the manufacturers in a tough position because
> they
> can't really wait a year to see what happens, they need their money
> now.
> They've set the specs quite high so that their engineer can leave town
> and we are left hoping that we don't have a bad magnet.
>
Received on Thu Jul 24 2008 - 07:59:13 MST