2

I'm looking into solid state relays for an application where I need to switch between two different transducers to allow two different modes of operation for an echo sounder. Basically a MUX that allows selecting either transducer #1 or transducer #2.

Short overview of how an echo sounder works: the echo sounder generates a short pulse (1-10 ms, 60 Vrms) which is emitted through the transducer. Then afterwards the echo sounder goes into listening mode receiving the returned echo which reflects of objects and the bottom as it travels through the water column.

Developing a MUX has some challenging requirements for this kind of signal:

  • transmit requires a relatively high voltage signal to pass through (max 60 Vrms sinusoid of 1-10 ms duration), whereas when the echo sounder is listening it will receive a signal which has a low amplitude (< 10 mV).

  • The signal should pass through without distortion (close to linear transfer curve).

  • The received signal should not be colored by noise (high S/N ratio, > ~90 dB).

  • Signal is bipolar.

Is this even possible with SSRs? A better device? With solid state relays I'm wondering if a device capable of relatively high current will not perform well for low signals during receiving (non-linearity or too noisy). Of course a mechanical relay would solve all these but might not last as long if switching frequently and reliability is important (i.e. years).

Drawing shows TX and RX circuitry switched during transmit and receive (listen) phase. Operator can switch the mux to use transducer 1 or 2 when echo sounder is off.

Assumptions: Both transmit and received signals have no DC component. Transmit is a pure tone of e.g. 35 kHz. The received signals will be more wideband, but will be bandpass filtered in the receiver. The switching between transducer 1 and 2 can be restricted to only occur when system is turned off. Transducers are ceramic. Impedance of transducer at transmit frequency is on the order of 1.5 kOhm. Assume max 40 mA inst. current during transmit pulse. Assume there is no ringing in existing circuit as transmit pulse has smooth envelope (ramp up/down). The transmit circuitry is only connected for as long as the pulse is transmitted, afterwards the receiver is connected and listens until the next transmit ping.

kashiraja
  • 123
  • 4
  • 1
    You need to fully articulate what the 60VRMS signal is in terms of AC and DC content, frequency and duration (you have mentioned that). A picture would be better of course. You will also need to state the currents that flow during transmit and how long the transducer will ring for before switching to the rx circuit. Much more detail is required. – Andy aka May 27 '16 at 18:50
  • We also need to know the relative locations of the power TX circuit, the transponders, and the sensitive RX circuit with the mux(es). – user2943160 May 28 '16 at 01:03

2 Answers2

1

I did something very similar (about 15 years ago) in order to signal over a 100 metre loop being used to inductively power multiple "stations".

Using just on-off electronic switches did not provide the isolation I required. I added shorting "switches" at the output of the power switches to effectively form a very high ratio voltage divider. ie simplistically Vout ~= Vin (Ron/(Ron + Roff)). Where Ron is the on resistance of the shunt switch and Roff is the off impedance of the main power switch. (I say impedance for the latter as while the on switch resistance dominates, the off switch capacitive coupling become significant).

For AC a normal MOSFET switch is not able to handle AC signals applied relative to the switch. This is because the body diode conducts when reverse polarity is applied even when the switch is off. The (or a) solution is to use two MOSFETs in series in opposing polarity. Source to source, gate to gate and the two Drains become the AC input terminals. For eg N-Channel MOSFETS driving the gate positive relative to the source turns both FETS on and connecting source and gate whether by a resistor or opto or whatever turns the MOSFETs off. The g-g an s-s connection works because MOSFETs are two quadrant devices and when turned on a resistive polarity independent connection is provided from drain to source.

A challenge is to provide power supply to the gate drive which floats at about AC zero crossing level when the FETS are off and which depends on AC signal level when the MOSFETs are on. This can be done with an isolated power supply - a cheap enough solution nowadays, or some care with resistor and diode feeds from supply rails or for experimentation a local battery.

Russell McMahon
  • 150,303
  • 18
  • 213
  • 391
1

Using some sort of FET-based solid state relay with isolation sounds like a great fit for this application. OMRON has a portfolio of FET SSRs (list) that probably will give you a reasonable output voltage and output on-state resistance. The G3VM series probably has the best options, with 100V output versions. Try a G3VM-202J1 at 8 ohms on resistance, 200V; single channel G3VM-201J1? Or G3VM-101ER 0.2 ohm on resistance, 100V?

From IXYS, the LCC120 could be another candidate at 20 ohms and 250V. From Vishay, suggested by another thread, the VO1400AEFTR, but it's only rated to 60V.

Additionally, if your signal duty cycle is low enough, you'll be less concerned with the continuous on-current limitation/rating of the FET SSR because of the lower energy dissipated by the device.

Community
  • 1
user2943160
  • 2,908
  • 1
  • 18
  • 32
  • Do you know if high power solid state devices introduce more noise in the signal being switched? (than what one would see for SS devices that switch much smaller currents)? – kashiraja Jun 03 '16 at 05:49
  • Not an estimation in the slightest. I would suspect that these devices wouldn't contribute much noise, but I can't tell you if they would be significant in your system's noise calculations. – user2943160 Jun 05 '16 at 02:49
  • 1
    These look like great devices for this application. Especially the ones with low resistance. With the low duty cycle I might be able to get by with lower power rating. The voltage rating probably should not be exceeded even for short signal bursts? I could test it, but still concerned about the device degrading over the long haul possibly. – kashiraja Jun 05 '16 at 20:38
  • 1
    For long term reliability: definitely do not exceed the voltage rating of the device. Because the device will be dealing with both pulsed currents (when conducting your signal) and pulsed voltage (when keeping a node disconnected from the signal), ensure that both device parameters are met by the part you choose. And, as always, do lots of testing if this is more than a quick, temporary, "good enough this time" project. – user2943160 Jun 05 '16 at 20:40