Multiple bridge rectifiers

[SBrock]

I am trying to design a circuit with multiple full wave bridge rectifiers (with smoothing) to act as liquid level control.

I am using bridge rectifiers as I need AC to detect liquid level to avoid electrolysis in the liquid, but a 'DC' signal to a PLC.

A single bridge rectifier works very well, with a 13V AC input and 16V DC at the PLC input, using a 120 uF capacitor to smooth the signal.

The problems arise when I try to use more than one 'switch' (liquid level detection) on the same AC supply.

Whenever I connect the two together, I get what looks like a half-wave rectified from a switch that is not made. If I put the live AC connection on one rectifier and the neutral connection on the other then I get two steady signals from one input!

I have attached the current circuit diagram to this message. The 1000 uF capacitors are now 120 uF and I have tried both with and without the diodes on the 0V DC, with mixed but not successful results.

Any help would be appreciated - I am tearing my hair out!

 

[benta]

Should your PLC inputs be negative? Otherwise you have the bridges the wrong way round.

Benta.

 

[SBrock]

The PLC inputs should be (and are) positive.

It may be that the schematic is drawn the wrong way round - I've relied on the +,- and ~ printed on the rectifier hardware and they work OK...

 

[btrueblood]

"I am using bridge rectifiers as I need AC to detect liquid level to avoid electrolysis in the liquid, but a 'DC' signal to a PLC."

A bridge rectified signal is effectively DC, and the capacitor is filtering the signal further to give you more or less steady DC; you won't slow corrosion/depostion on the electrodes very much, if at all, using that scheme. You need fully reversing (AC or AC square wave) to minimize electrolysis effects. Bob Pease has/had a good AP note for a (now obsolete) liquid level detector IC that Nat. Semiconductor made for many years, discussing the whys and wherefores of conductivity sensors, you may also find a patent with his name on it for the device.

In any case, based on your diagram, once a single probe makes contact, the current flow must apparently spike (there are no resistors in your circuit diagram to limit current), and swamp the transformer, i.e. the AC voltage drops to nearly zero. Effectively, the voltage at all PLC inputs will drop to zero when contact on any one probe is made. You need to limit the current thru any one probe using a resistor, and sense the voltage "downstream" of the resistor with the PLC. See my quick and dirty sketch attached. The circuit shown (or quite a number of such probes) can be "hung" off of a single transformer/diode bridge/capacitor, without trouble.

 

[btrueblood]

Ugh, I need to reread my posts before hitting "submit". My point about Mr. Pease's IC design is that it provides a true AC square wave probe voltage and sensing method; the basic oscillator circuit is not that difficult to replicate in discrete components and/or using a microcontroller (or maybe even a PLC).

 

[roydm]

I agree with Btrueblood, your circuit is way too complicated.
To avoid electrolysis instead of DC as he shows you could use AC with a single diode and capacitor (½ wave) going to the PLC input.
Roy

 

[MacGyverS2000]

Look for app notes on electrolytic tilt sensors. They need A/C to avoid killing the electrolyte, and the driving circuits will be similar to what you're after.

Dan - Owner

http://www.Hi-TecDesigns.com

 

[VEBill]

The schematic is obviously wrong (the way the outputs are connected to the '0v' and the PLC backwards). So perhaps the actual hardware is wired-up all wrong too. Which makes troubleshooting over the Internet impossible.

But I'd redraw the schematic, and then triple-check the actual wiring against the corrected schematic.

 

[Skogsgurra]

Diagram wrong or not probably doesn't matter. Using one and the same AC source might be a problem since the liquid will conduct from electrode pair n to pair n+1 etc.

A separate AC voltage (one secondary winding for each electrode pair) is needed for reliable operation.

Gunnar Englund

http://www.gke.org

--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...

 

[VEBill]

That shouldn't be an issue. For clarity of reasoning, imagine that only the bottom-most sensor input contact is used. The input contacts are all wired in parallel anyway.

 

[btrueblood]

Well, yes, Skogs, for the diagram he showed...I think...

But multiple probes from a single power supply will work, with the current limiting resistors as shown in my first post, and I have a working model (two actually) in our shop, if anybody wants to come take a look at 'em. But it's a DC circuit, because our system stands empty for 99.99% of its life, so corrosion/plating is of little concern.

An AC circuit like Roydm describes is an even simpler method, and I'd forgotten about. Attached schematic is my interpretation of Roydm's description, and should work. The values of resistors and capacitors are complete WAG's by me, and would depend on the impedance of the PLC input circuitry, conductivity of the fluid, and the desired response time of the sensor. But value ranges shown should be good starting points for testing.

 

[btrueblood]

wups, the lower circuit, where it says "PLC 1 input", should be "PLC 2 input"...

Oh, should've pointed out...the response time of the AC sensor is also limited, and variable, due to the possibility that contact on the probe is made on the "wrong" part of the AC cycle, as well as because of the decay time of the anti-ripple capacitor (which can vary tremendously with temperature for some electrolytic capacitors, we found out the hard way), thus an AC sensor can/must have a lag in response that a DC sensor doesn't have. If you use a simple line transformer to generate 60Hz AC, you may be limited to a response time greater than several tenths of a second, or even more, depending again on all the variables. If you need faster response, or a very repeatable response, you either need to use higher AC frequency (e.g. the sensor circuit by Bob Pease or similar) or a DC circuit.

 

[BobM3]

If you have to have AC, then why not just an AC input photocoupler? Size a capacitor on the output to get you throught the AC zero cross.