Gravity measurement / Projects / Forums

Forums

4hv.org :: Forums :: Projects

« Previous topic | Next topic »

Gravity measurement

previous 1 2

Move Thread

LAN_403

Hydron

Wed Jun 18 2014, 10:58PM

Registered Member #30656 Joined: Tue Jul 30 2013, 02:40AM
Location: UK
Posts: 208

This article showed up on arstechnica today, made me think of this thread:

("Gravityâ€™s strength still an open question after latest measurement")

May be of interest, but as usual there is a damn paywall between you and the actual paper :(

Edit: in the comments on arstechnica this link came up:

, there is a free access figure showing the non-overlapping error bars of previous experiments.

Sulaiman

Thu Jun 19 2014, 05:26AM

Registered Member #162 Joined: Mon Feb 13 2006, 10:25AM
Location: United Kingdom
Posts: 3140

Dr. Slack, have you considered filling the apatatus with liquid to dampen oscillations?
I imagine that without damping you will have to perform your measurements frustratingly slowly.
I'd guess some sort of non-conductive, non-polar oil would be best, equivalent to transformer oil.
Even if you think it unneccesary it would be prudent to design for it, just in case.

Dr. Slack

Thu Jun 19 2014, 07:59AM

Registered Member #72 Joined: Thu Feb 09 2006, 08:29AM
Location: UK St. Albans
Posts: 1659

Yes, oil had crossed my mind, but I'll certainly keep the apparatus dry to start with. The bottom of the tube will be liquid sealed. It might be worth me doing some sums on the viscosity of air to see what damping I could expect with it the way it is. Smaller gap, more damping.

I was anticipating that one of the early bring-up experiments would be to verify the resonant frequency and measure the Q of the pendulum, driving it electrostatically.

Now if instead of measuring the position of the pendulum, I drive it closed loop electrostatically to keep its position constant, say drive for a null output, then that takes its dynamics out of the measurement. However, it moves them into the loop stability of the feedback loop.

thinking about damping a little more

a) For static measurements, I can tolerate a lengthy settling time, an hour need not be a problem
b) I'll be averaging, that can be made to filter out the undamped swing to give the underlying mean position
c) If the pendulum is moving, I will know that internal friction in the support wire is not giving it a static offset
d) I could not use a partial fluid fill, surface tension in the small gap would overwhelm any tidal forces
e) Even full of fluid, any bubble left in the gap would also swamp tidal forces, so I would need to be able to pull a good degassing vacuum on the tube.

So while I'll bear the idea of a fluid fill in mind, I think it will neither be necessary, nor straightforward to use.

Perhaps I should mention that I will use a PC sound card, audio captured using Python/PortAudio, to do a synchronous demodulation of the output with respect to the drive signal, so the data will be present as a complex IQ time stream, representing the xy position of the pendulum. That will need to be knocked down from 48ks/s to a more sensible data rate. I'm interested in two data rates, one suitable for instant display, perhaps 10s/s, and one suitable for storage and analysis over the long term, from lunar cycle to cycle, perhaps 10 seconds per sample. This lower rate will allow the expected pendulum natural frequency of around 1Hz to be totally filtered out. The under-damped pendulum resonance will make any sort of seismology interpretation essentially useless.

I think I've solved my electrode construction concentricity problem. I was attempting to get the inner and outer surfaces concentric by adding layers that I hoped were equal, by dead-reckoning. Doomed to failure. I've got a long mandrel that defines the internal bore, so I can build a simple jig to space a small grinder from its surface, that will be able to grind an over-size build up of car body filler down to a cylinder concentric with the mandrel. I can keep testing it in the bore of the outer, and stop when it's a push fit. Normally if you try that sort of process on a lathe, chuck repeatability can mean that the next cut is never quite concentric with the last. With a grinding jig, that's not a problem.

Ash Small

Thu Jun 19 2014, 04:01PM

Registered Member #3414 Joined: Sun Nov 14 2010, 05:05PM
Location: UK
Posts: 4245

Dr. Slack wrote ...

Normally if you try that sort of process on a lathe, chuck repeatability can mean that the next cut is never quite concentric with the last. With a grinding jig, that's not a problem.

You first machine a 'mandrel' which remains in place on the lathe when you remove the workpiece. This overcomes the problem of 'chuck repeatability'. (and can overcome other problems as well).

Don't, under any circumstances, remove the mandrel from the lathe until you are absolutely certain that you have finished machining.

Also, take your time machining the mandrel.

Jo Richards, former Olympic medalist (among other 'claims to fame') once asked me, one friday afternoon, to machine up a set of rigging bushes for his 1995 CHS 36 class (if I remember correctly) carbon yacht 'Full Pelt', he wanted 15 of them, and gave me drawings with no tolerances. When I asked him and his 'designer/engineer' what the tolerances were , he replied "the best you can do". I said "OK, then, but the setting up will take some time".

I took them back to him first thing Monday morning, He asked me the price, I said "Â£15 each", he said "That seems a lot", I replied "well, you wanted them as accurate as possible", He replied "I'll have to get my engineer to check them, come back on Thursday"

I went back on Thursday and he gave me the Â£225 and asked for another 15

I had to set everything up again

His laminator/boatbuilder later told me that the engineer hadn't been able to detect any measurable discrepancy from 'specification' whatsoever on any of them, but I knew the maximum deviation on any of them was less than a third of a thousanth of an inch.

These were machined out of 316 'marine grade' stainless steel on an old victorian treddle lathe that I'd bought for next to nothing and converted to electric.

(I imagine that the extrusions you have will require 'truing up' before they are usable)

Dr. Slack

Thu Jun 19 2014, 04:16PM

Registered Member #72 Joined: Thu Feb 09 2006, 08:29AM
Location: UK St. Albans
Posts: 1659

(I imagine that the extrusions you have will require 'truing up' before they are usable)

nah, I might consider an annealing step so they don't wander around, but I'm only really after stability, not precise dimensions. I'm only making one off, so am doing what the 'fitters' used to do a century ago when 'manufacturing', trimming each rough piece to fit the others, so that nothing was interchangable.

Sulaiman

Thu Jun 19 2014, 05:43PM

Registered Member #162 Joined: Mon Feb 13 2006, 10:25AM
Location: United Kingdom
Posts: 3140

If I was doing this my main obstacle would be
finding a stable secure location for the device where it can remain without risk of knocking it
yet having a reliable power supply for the thermal management and instrumentation.
You may need an enclosure for your enclosure !

Dr. Slack

Fri Jun 20 2014, 10:31AM

Registered Member #72 Joined: Thu Feb 09 2006, 08:29AM
Location: UK St. Albans
Posts: 1659

Attempting to do some very approximate sums on damping due to air, it turns out that there is a very strong proportionality to the size of the gap. Say the gap halves. The air that the moving bob pumps back through the gap now has to go twice as fast, and is squeezed through half the gap, so the energy that it dissipates is going up at least as third power of the gap width, maybe fourth. That's a third reason to minimise the gap size, along with distance to volts scaling, and output impedance. It's actually a well studied problem, albeit in a slightly different geometry. Air in a MEMs system damps microphones, accelerometers etc with what's called 'squeeze film damping', though the gaps are in the low microns. I'll make several electrode constructions with different internal bore, to investigate the difference in damping. That will give me a justification to make the first proper one large and easy to line up

Somewhere stable and out of the way to put it? I was speculating that an old mine or nuclear bunker would be ideal once it was working. However, in my concrete-floored garage, I have a number of steel shelving racks. The plan is to remove a couple of shelves at the bottom of one to give me a space. It's on the floor, against the wall for power, with a shelf above for oscilloscope and laptop, in a space that is already out of the way and not walked through. The top, bottom and back are already supplied, with the shelving frame making a nice attach for the front and sides.

Initially, I'm not going to attempt temperature control, just absence of fast temperature changes, and monitoring what it is for the moment. It's built in thick alli, so I'm hoping equilibrium will be fairly quick to establish. Anything that will influence the level of the working bits is being made from balanced lengths of the same materials, which should mean that the absolute temperature is irrelevant, only the non-uniformity across the apparatus should matter. It will incorporate small heaters on various bits of the structure which I will use to investigate its susceptibility to having a pulse of heat dumped into it here or there. Then from knowing the thickness of the insulation round it I ought to be able to put an upper limit on the non-uniformity of temperature it is experiencing.

It has crossed my mind about how I would control the temperature, and it's not obvious. Where would I heat it? At one point, multiple points, everywhere, or stir the air with a fan? I need to know how susceptible it is to temperature gradients before I begin to consider heating. Running a fan or two in the enclosure would be a good test of susceptibility to disturbance, on the way to considering temperature control.

Leaving aside heating, I'm not sure that there's anything about the sensor that needs to run continuously, at least not initially.

Andy

Sat Jul 05 2014, 03:58AM

Registered Member #4266 Joined: Fri Dec 16 2011, 03:15AM
Location:
Posts: 874

@Dr. Slack Hi
Have you thought about the protons or electrons effecting the measurements,like from the devices?.
I think using chips for laser range rangefinder might have a offset designed like GPS satellites that have a slower or fast clock to account for drift.

Dr. Slack

Mon Jul 21 2014, 10:11PM

Registered Member #72 Joined: Thu Feb 09 2006, 08:29AM
Location: UK St. Albans
Posts: 1659

Now making the next, close tolerance, electrode construction.

First, wind the 200um thread onto the cyclinder as a removable spacer

Next, the BeCu electrode patches, each 19mm square, pre-bent to roughly the curvature of the former

I've now wrapped several turns of very thin polythene over the thread as a release layer. It took me some while to figure out how to get the electrodes lined up properly. Well, first I held the first rank down with small elastic bands, and coarsely adjusted them for position. Then wrapped a few turns of thread round them tightly, while trimming the position. Then wrapped a few turns of epoxy-soaked thread, and finally adjusted position before it set. This holds them firmly in position while I assemble the other rank, and wind on the first few turns of glass cloth, before applying resin.

My mistake with the first attempt had been to try to use an additive process and expect to come up with the outer concentric with the inner. No chance. So, I applied excess glass and resin, and then made a jig for my router, with meccano amongst other things, to cut the GRP back to a repeatable distance above the surface of the former. The four screws that support the two meccano brackets are M4, with a 700um pitch. I figure I can control about 30 degrees of nut turn, so can achieve around 60um resolution in the height adjustment of each screw, or 15um if I tweak just one of them. Then followed a long slow skimming down of the surface until it was a nice snug fit inside the outer tube.

previous 1 2

Moderator(s): Chris Russell, Noelle, Alex, Tesladownunder, Dave Marshall, Dave Billington, Bjørn, Steve Conner, Wolfram, Kizmo, Mads Barnkob