# 8-channel LVDT measurement using AD598

#### abicash

##### Member
I will try to check more about the LVDT.
But why doesn't no one recommend such a scheme?
I saw few application notes but these will ask to parallel the primaries, not connect in series.

EDIT: few more details for the probe attached

#### Attachments

• 28.2 KB Views: 2
Last edited:

#### ronsimpson

##### Well-Known Member
EDIT: few more details for the probe attached
I did not learn much.
Voltage: 3V Frequency: 13kHz Ohm: 2KW
This comment might say; put 3 volts on the transformer. That is about the smallest amount the IC can do. (3x4=12V) and driving 4 in series is less than the max amount. BUT; 3V/2k = close to 1.5mA and the IC can do 6mA so 4 transformers in parallel.

If I had 4 or even 2 transformers:
Connect the transformers 2 in series, use a audio signal generator to drive the transformers. Measure the outputs with a oscilloscope. Set one of the transformers to center and one for max or min and see if the voltage divides nicely. We know the current will be the same. I do not know if the voltage will divide to 1/2. Another way to measure; with an inductance meter measure the transformer, at center and at max. I know the resistance will be the same but I do not know if the inductance will be the same. If the inductance changed 10 or 20% there is probably not a problem but if it changes more than maybe they should not be connected in series.

#### abicash

##### Member
Thanks Ron

Coming back to the original schematic, why don't you think it is not better than connecting series/parallel primaries?

#### ronsimpson

##### Well-Known Member
why don't you think it is not better than connecting series/parallel primaries?
price and complexity only. That is why I would not "mux" the center point on the transformers. It gets connected to ground either through the mux or not through the mux.
I talked to another engineer. We noticed that some transformers are rated for 60hz to 20khz. At high frequencies there is some unknows like inductance that worry me. At low frequencies the current in the transformer is a function of wire resistance. We think that at low frequency there is no problem with series. At high frequency we would need to test.

#### ronsimpson

##### Well-Known Member
The safest thing is to use 10 ICs. Any change you make increases the risk of failure. Any part you add increases risk.
You probably can not find where anyone uses mux(s). This adds risk unless you make a prototype. (example: your signal goes above and below ground, so check if the way you are using the mux will work)
In my work we could make millions. I can afford to make a prototype. Before that I would make a "proof of concept".
In high volume production we try to save every "$0.01". If I can leave a resistor off the board I will. If I had a month of time, I would consider using a faster micro or low end DSP and doing all the math in software. (do not use the AD598) But that has high risks! Add$1 to the computer and save \$50. That only works if you are going to make 10,000 boards.

You need to decide risk/reward. Can you afford not to make a prototype? I think your MUX idea is cleaver. It will work. For a test you could bread board (AD598, + MUXs with 1 transformer) Set the MUX to "0". Do not use the computer hardwire the A,B,C to 0. Just to see if inserting the MUXs will work.

#### abicash

##### Member
Fantastic reply ... Clears a lot of things
Thanks a million!

#### abicash

##### Member
Hello Ron

Coming back to this thread since now I have a working prototype on a printed board.

I have ditched the idea of using multiplexers, and have used a more safer approach of Relays.
I have a Half-Bridge LVDT and now I am using the AD698 in lieu of the AD598.

Please check the scematic for each stage attached to this post.

Now my queries will be AD698 specific, so I will request you to check the datasheet for the AD698

Please note that SIG REF pin on the AD698 is connected to Circuit Ground...or did I commit a mistake
and i was supposed to keep it floating and acquire a differential signal ?

I need..

Vo = 0V to +5V (which i am unable to get. I get a value from (+) ~3v to ~6v or so, which I am unable to understand)

Freq = 13 KHz (which i do get , so please ignore the capacitor value, since it is corrected on board)

Also please throw light on how do I achieve 1 micron resolution with the setup?
Do i need to set a span of only a few tens microns for the entire output voltage swing on the LVDT to acheive this?
For example : If i set a span of 100 microns for a total of 0 to +3.3v then with a 12-bit ADC and 11 useful bits
I get 50mV per micron = 31 counts

Requesting and expecting a walk through!

#### Attachments

• 45.8 KB Views: 4
• 4.6 KB Views: 3
• 44.6 KB Views: 2
Last edited:

Hi