Hi Everyone!

I've created an account to hopefully get some help with this problem that I cant get my head around, and maybe help some others on the forum where I can.

I have a pressure transducer that is monitoring high pressure water and giving a 0 - 10 V output (see attached schematic). I added a 330 uF electrolytic cap to filter out some of the pulses caused by the pump which works well so i get more of a stable voltage when the pump is running (tried a LP filter as well but this came with its own problems). What i dont understand is that by connecting the capacitor seems to increase the voltage at the output.

I took the transducer out of the system and set up a test rig where exactly as the schematic to rule out any other influences, set the pressure to a static 40 bar which gave an output of 2.5 V without the cap (expected). When i placed the cap in the circuit the voltage went up to 3 V. I tried serval brand new transducers of the same type and they all do it.

I tried a couple of different transducers of similar spec but from different manufacturers and this problem did not exist. Can anyone help me understand what is different about this type and what is happening please?

Thank you
Dean

 

hi ph,
Welcome to AAC.
How stable and 'clean' is the 24Vdc input to the sensor.?
Have you tried placing the capacitor on the 24V supply.? [ the cap should be rated for 35Vwkg]
E

 

Since output of transducer a V you need to find out if that transducer is
signal conditioned with an OpAmp output. If so hanging a big cap on
its output can cause some serious stability issues, like RF oscillation
as one possible result.

The data sheet unfortunately does not show a block diagram of bridge/
signal conditioning.

You could use a RC LPF approach to see if that would filter adequately
your output. The R would decouple the cap (depending on its value)
from the OpAmp output. More R, less C better, to reduced effective
C load on OpAmp.


Regards, Dana.

 

... along the lines of ericgibbs's post, try putting a smaller capacitor across the + and - terminals of the sensor, maybe 0.33 uF, and no large capacitor on the output. The idea is to filter any voltage spikes at the sensor supply terminals.

 

hi ph,
Welcome to AAC.
How stable and 'clean' is the 24Vdc input to the sensor.?
Have you tried placing the capacitor on the 24V supply.? [ the cap should be rated for 35Vwkg]
E

Hi and thanks, the supply is from a SMPS so a bit of switching noise.. I tried your suggestion of putting the cap on the 24 V, but this did not change anything except the output did smooth out a fair bit. I did however return the circuit to the original configuration and used a different power supply, this time a linear type and the issue went away.

Does this mean anything? Unfortunately I cant use the linear power supply in the actual application.

Dean

 

hi,
SMPS supplies can be 'noisy', does the pressure transducer require a fixed 24Vdc or will it accept a lower voltage.?
E

 

... along the lines of ericgibbs's post, try putting a smaller capacitor across the + and - terminals of the sensor, maybe 0.33 uF, and no large capacitor on the output. The idea is to filter any voltage spikes at the sensor supply terminals.

Hi, and thank you for the suggestion I tried this and it did not have any effect. The trouble is, the transducer is responding to the mechanical system where there are a lot of pressure spikes so it is actually working correctly without the capacitor on the output. But the mini-PLC I am using doesn't have any filtering options on the input so it gets confused with the oscillation of the signal. So I am trying to average that out so the PLC can interpret the voltage signal.

Dean

 

I think the capacitor is averaging the output of the transducer. If you look at the noise and imagine an average line through it you might begin to notice that the average is a little higher than the RMS level. I said "I think."

 

hi,
SMPS supplies can be 'noisy', does the pressure transducer require a fixed 24Vdc or will it accept a lower voltage.?
E

Hi E

Yes it can accept from 12 - 33 V
Dean

Since output of transducer a V you need to find out if that transducer is
signal conditioned with an OpAmp output. If so hanging a big cap on
its output can cause some serious stability issues, like RF oscillation
as one possible result.

The data sheet unfortunately does not show a block diagram of bridge/
signal conditioning.

You could use a RC LPF approach to see if that would filter adequately
your output. The R would decouple the cap (depending on its value)
from the OpAmp output. More R, less C better, to reduced effective
C load on OpAmp.


Regards, Dana.

Hi Dana

I have been on to the manufacturer with all the info I have, so hopefully they can look at the internals and see whats going on. Thanks for the suggestion I will try using a resistor as well. I had used a 1 Hz LPF before but it dropped most of the voltage and the controller input is quite low-res.

Dean

 

I strongly suspect, as @danadak does, that putting a capacitor directly on the output of your sensor is causing its internal amplifier to become unstable and oscillate. Such oscillations are typically at a very high RF frequency; while they don't normally show up directly in a device's output, they do tend to cause large, unexplained DC shifts such as what you're seeing.

Try connecting a resistor (10 Ω to 100 Ω or thereabouts) in series with your output capacitor and see if the output returns to normal.

 

Yes it can accept from 12 - 33 V

hi,
Is there 'space' to place say a 7815 Vreg in series, between the 24Vdc in and the sensor, with its associated caps.?
E

 

Well, after reading other posts my approach may not be the best. However, I was thinking of mimicking a regulator circuit using a few capacitors such as in this illustration:

 

I strongly suspect, as @danadak does, that putting a capacitor directly on the output of your sensor is causing its internal amplifier to become unstable and oscillate. Such oscillations are typically at a very high RF frequency; while they don't normally show up directly in a device's output, they do tend to cause large, unexplained DC shifts such as what you're seeing.

Try connecting a resistor (10 Ω to 100 Ω or thereabouts) in series with your output capacitor and see if the output returns to normal.

Thanks for the suggestion unfortunately this did not help. But I did notice the voltage fluctuating with a resistance in there. Maybe this supports the oscillation theory.

hi,
Is there 'space' to place say a 7815 Vreg in series, between the 24Vdc in and the sensor, with its associated caps.?
E

Its not ideal to have any more components in there from an assembly point of view as its all DIN rail mounted terminal blocks but I will test the theory. Do you think a linear regulator would be better so it has less noise or is the 7815 good enough?

 

tried a LP filter as well but this came with its own problems

What problems?
What type of LP filter?
That would normally be the way to go to remove noise and spikes..

 

Since the sensor is designed to operate from 12 to 33 volts, dropping from 24 to 15 means 9 volts that have to be dropped. That means extra heat somewhere having to be dealt with. If it were at all possible to eliminate the SMPS for a straight transformer bridge rectifier circuit with a few caps as shown in my diagram then I would suggest considering the possibility. You DID say a 12 volt wall wart is not an option - so I really don't know what avenues are available for you. Using a 24 volt transformer puts you at 33.9 volts after rectification - minus the diode forward voltage drops. You're right at the ragged edge of allowable input voltage. Using a 12 volt transformer means 17 volts potential. That's well within your allowable voltages. No regulation needed. Just filter caps to clean up the power.

 

Do you think a linear regulator would be better so it has less noise or is the 7815 good enough?

hi,
A 7815 is a linear voltage regulator.
E

 

Its not ideal to have any more components in there from an assembly point of view as its all DIN rail mounted terminal blocks

They make 24 volt linear regulators for DIN rail mounting.
SG

 

... If you are correct in diagnosing the problem as coming from a mechanical origin, then one option is to try mechanical damping. At any hardware type store, you can buy a 3 inch long copper tube with a pneumatic bladder inside, the purpose is to dampen any water hammer or pressure variations. Typical household applications are at a dishwasher water inlet that shuts off rapidly when a solenoid valve cuts off. All you need is a tee adapter and some Teflon sealing tape. If a single device shows some signal smoothing improvement, then add one or two more until you get the signal you want. ... Think they are known as 'accumulators' but not absolutely sure.
like this device:
https://en.m.wikipedia.org/wiki/Hydraulic_accumulator

 

Thanks for all of your input its very much appreciated.

I feel it is to do with the capacitor destabilising the amplifier inside the transducer as OBW0549 has said. So I have two things to address. 1. Going forward I will make a change to the system. I have got a 7815 Vreg on order and the capacitors that Tonyr reccomended to see if these help, also I am looking at a better controller that can do software filtering to get rid of the troublesome capacitor all together which would be my preferred solution. And 2. what to do with similar systems out in the field.

I tested a sample of 10 similar transducers today. Each one has this d.c. offset voltage... But not all the same value - most aren't out by enough to affect the performance of the system, but some clearly are. Does anyone know if the offset is likely to change over time? I can accept some margin of error as long as it will perform consistently. i.e. if there is some offset will it get worse?

Crutschow, the main problem with the low pass filter was (if i remember) to get the cut off frequency low enough we had to use a large capacitor or a high value resistor, the capacitor took up more room than was available and a high value resistor dropped too many volts. I will have another look into it though.

Dean

 

... If you are correct in diagnosing the problem as coming from a mechanical origin, then one option is to try mechanical damping. At any hardware type store, you can buy a 3 inch long copper tube with a pneumatic bladder inside, the purpose is to dampen any water hammer or pressure variations. Typical household applications are at a dishwasher water inlet that shuts off rapidly when a solenoid valve cuts off. All you need is a tee adapter and some Teflon sealing tape. If a single device shows some signal smoothing improvement, then add one or two more until you get the signal you want. ... Think they are known as 'accumulators' but not absolutely sure.
like this device:
https://en.m.wikipedia.org/wiki/Hydraulic_accumulator

I agree, we could dampen it mechanically but the operating pressure is up to 150 bar and I believe the components aren't 'budget friendly'!

Dean