Welcome, Guest. Please login or register.
Did you miss your activation email?
September 30, 2016, 03:16:46 PM

Login with username, password and session length
Search:     Advanced search
* Home Help Search Calendar Links Login Register
+  Machsupport Forum
|-+  General CNC Chat
| |-+  Show"N"Tell ( What you have made with your CNC machine.)
| | |-+  Linear Voltage Differential Transformer
Pages: 1   Go Down
Print
Author Topic: Linear Voltage Differential Transformer  (Read 138 times)
0 Members and 1 Guest are viewing this topic.
joeaverage
Active Member

Offline Offline

Posts: 202


View Profile
« on: September 27, 2016, 02:27:10 AM »

Hi All,
I've been fooling around making a lvdt for a tool height setter. Very amateurish
at present but coming along.

* lvdt2.pdf (28.53 KB - downloaded 16 times.)
* lvdt3.pdf (31.46 KB - downloaded 16 times.)
Logged

My wife left with my best friend...
     and I miss him!
joeaverage
Active Member

Offline Offline

Posts: 202


View Profile
« Reply #1 on: September 27, 2016, 03:01:49 AM »

Hi All,
will post some pics and diagrams of the transformer itself, as per usual I have to
make plans after I've made it!

The two circuit diagrams above are first the circuit I'm using currently and the second
is the next evolution, as yet unbuilt.

Refering to the first circuit the amp IC3A, biasing, gain setting and voltage limiting
components feeback into the LC tank C1,C2,L1 and R3.

R3 is a current sense and is amplified by IC3B by a gain of 50 to drive the lvdt primary
and the all-pass phaseshift amp IC3D.

The double balanced modulator is driven by the phaseshift amp via R16 and keeps the
ring diode currents to a few mA.

R15 and C13 lowpass the DC while damping out the LO+RF signal and IC3C is DC buffer.

The osc amp, gain amp and phaseshift amp all require a fast opamps, I chose OPA1654AIPW
with gainbandwidth of 20MHz. I ordered these only to realise that they are TSSOP with .6mm
pitch. Real sods to hand solder!

The net evolution will use similar amps but probably duals rather than a quad and SOIC being a lot
easier to handle. The only other difference is using two amps each with a gain of 7 rather than one with
a gain of 50. If I'm using two amps why not introduce some filtering.... hence the extra resistors/caps
as 4 pole Butterworth. Most if not all components on the 'evo' board will be SMD.

I'll draw some diagrams and be back when done.

Craig
Logged

My wife left with my best friend...
     and I miss him!
rcaffin
Active Member

Offline Offline

Posts: 650


View Profile
« Reply #2 on: September 27, 2016, 04:37:08 AM »

I have (successfully) avoided TSSOP so far. A 1 mm pitch is quite enough for me.

Have you considered using CMOS switches to do a synch demod? I like the technique.

Now ... linearity? Any measurements so far?

Cheers
Roger
Logged
joeaverage
Active Member

Offline Offline

Posts: 202


View Profile
« Reply #3 on: September 27, 2016, 05:29:10 AM »

Hi Roger,
hadn't thought of switches, they still require the phaseshift to make genuinely
synchronous. As it turns out the phaseshift tunes nicely to a repeatable peak
and once adjusted stays there unless the ldvt is altered. Do you think they offer
any advantage?. The noise floor of the existing design is pretty fair with potential
resolution in the region of 1um, about what I had hoped to achieve. Could I usefully
measure and use sub-micron resolution?

Don't have any reliable linearity measurements yet. The 'mental 'plot I had running
suggests good, but no hard evidence.

My development path is to get numerical data (zero crossing resolution, linearity vs stroke).
Next effort is about making it smaller and user friendly. After that thinking about suitable
outputs, probably zero crossing detector for straight probing and a 4 digit dual slope
integrating ADC similar to those used in multimeters and possibly 16bit digital output via
MODIO back to Mach.

Still a power of work/fun/learning to be had yet!

Craig
Logged

My wife left with my best friend...
     and I miss him!
rcaffin
Active Member

Offline Offline

Posts: 650


View Profile
« Reply #4 on: September 27, 2016, 07:05:26 AM »

Hi Craig

I have used the CD4066 for this sort of things many times, with good results. More often with a resistive load such as a strain gauge, but also with LVDTs. I found it crucial to use a frequency far away from resonance, to limit wandering phase shifts.

Often with an LVDT I drove the two outer windings out of phase, so the phase and amplitude on the inner winding was very useful when synchronously demodulated. That required that the two outer windings be exactly matched, but most commercial LVDTs had that sorted out. With the right winding geometry I am sure micron resolution would be normal. Linearity was a different matter: that required very carefully distributed winding on all coils. I suspect getting the winding distribution right was what made many of them rather expensive.

It sticks in my mind that Schaevitz once had some very good tech briefs on LVDTs, but that was 30+ years ago. (I THINK it was S, anyhow.)

Cheers
Roger
Logged
joeaverage
Active Member

Offline Offline

Posts: 202


View Profile
« Reply #5 on: September 28, 2016, 12:42:12 AM »

Hi Roger,
some good info there. The current model I'm working on shows no tendency to
waver in phase and guess therefore not close to any resonant condition. If that
is the case then it was by happenstance rather than design!

I run the centre coil energised with the outer coils wired in phase opposition, almost
the exact opposite to your description, of  course being in the southern hemisphere
will mean that it works fine, lol!

Your comment about linearity being related to winding distribution fills me with dread
however. I used Mach to wind the coils but is FAR from optimal. I may be very disappointed
when I get myself setup to measure it accurately.

I was largely the ease or otherwise of winding coils in 0.2mm wire that induced me to go
for such a high excitation frequency. The prospect of winding in 0.05mm sounds maddening
to me but your insight suggests that commercial makers have done just that and  I can well
imagine it to be expensive.

Likely to be a few days before I can get some numbers, I have to borrow the boss's 3-4" micrometer
so I really really want to do something useful for him before I touch him up for it!

Craig
Logged

My wife left with my best friend...
     and I miss him!
rcaffin
Active Member

Offline Offline

Posts: 650


View Profile
« Reply #6 on: September 28, 2016, 12:54:44 AM »

I suspect (suspect) that if you make the windings very uniform and neat, that will be enough. After that it might be a matter of adjusting the length of the ferrite core.

Cheers
Roger
Logged
joeaverage
Active Member

Offline Offline

Posts: 202


View Profile
« Reply #7 on: September 28, 2016, 01:08:14 AM »

Hi Roger,
the windings are reasonably neat but I suspect not particularly uniform.

The ferrite cores I bought of eBay, cheap and cheerful from a certain Asian
country! 10 cores for $15 so can't complain. Measured them out of the
packet and there is quite some variation, about 0.5mm length variation
on 25mm nominal and 0.25mm diametrical variation on 12mm nominal.
Given that they sintered after being pressed green about what I expected.
Sounds like I may have to take to them with a diamond wheel.

Craig
Logged

My wife left with my best friend...
     and I miss him!
rcaffin
Active Member

Offline Offline

Posts: 650


View Profile
« Reply #8 on: September 28, 2016, 02:27:25 AM »

Measure linearity first, untrimmed?
Then trim and repeat.
Cheers
Roger
Logged
Pages: 1   Go Up
Print
Jump to:  

Powered by MySQL Powered by PHP Powered by SMF 1.1.20 | SMF © 2013, Simple Machines Valid XHTML 1.0! Valid CSS!