Looking for Ultra-Low Leakage Current Relay


I am looking for a relay for a new design with some specific requirements, and while I suspect something of the nature I’m looking for exists, I’m having a hard time nailing down a specific product capable of matching our needs.

Hard Requirements:

  • 200V AC/DC Typical (so I assume max should be somewhat higher)
  • Max load current can be as low as 5mA
  • Ultra-low leakage current/noise (Looking for something in the sub-nA, ideally sub-pA range)
  • Low coil current (in our tests a higher coil current exhibits greater leakage into the operation side, which increases the noise floor of the relay. There is no minimum or maximum requirement on this though)

Softer Requirements:

  • 5V Coil Voltage (Based on our current design, but can redesign for the right relay)
  • 1 or 2 Form A Style (Again based on our current design, if Form B or even Form C can satisfy the above requirements we will design to suit)

I’m not stuck to a specific kind of relay (SSR, OptoCoupler, Reed relay). we have a fairly low volume and cycle requirement (probably only a few hundred cycles per year is my estimate at absolute maximum), and since it’s a new design we have a fair amount of flexibility in what we can use. I’m also open to higher voltage relays if they provide the required leakage current requirements.

Some Relays We’ve Looked at Already:

  • Panasonic AQW224N (and others in that series)
  • Gigavac 15kV Relays (Need to test this one again, difficult to estimate the leakage current when only 200V are applied)
  • Sensata 3kV Reed relay (5V and 12V coil) (120mA coil current)
  • Coto 1kV Reed Relay 5V Coil (Significant coil current leakage)
  • Comus 5kV Reed Relay (Coil current 300mA, too high)

Thanks in advance for any help!

Hello jackson.b,

Please tell us more about your application.

As I understand the requirements, you are looking for low pin-to-pin leakage for a set of normally open contacts. As a first approximation, a high vacuum glass ampule reed relay would offer good performance. A representative part may be a high voltage the Sensata DAT200-05.

Also, have you considered using a two-step drive. One step with full voltage to activate the relay and then a lower voltage for holding the contacts in position. For example, the 5 VDC coil in the Sensata relay has a resistance of 28 Ohm requiring approximately 180 mA drive. However, once the device is activated, we can drop the voltage to a considerably lower holding voltage. As a conservative design, we may test 2 VDC. Assuming this works, the holding current is reduced to a manageable 75 mA.

Again, please tell us more about your design. It would be especially helpful if you describe the nature and source of the noise you are attempting to eliminate.

Best wishes,



Apologies for my delay in response, I’ve been out sick.

The 2-step drive is a very interesting suggestion. We have some Sensata DAR705-05 reed relays on hand I can use to test this and see how it works.

Some info on the circuit I’m trying to build:

  • I have two input channels, called HV_IN and INPUT: HV_IN applies 200V AC or DC (upper frequency of about 10kHz) and the INPUT channel measures the return current from the applied 200V to a test object further downstream. The INPUT channel is capable of measuring DC currents up to 10mA and down to about 10pA, and AC capacitance down to 10pF.
  • I have two output channels, called HV_OUT and OUTPUT.
  • In one configuration, the connections are straight through, where HV_IN → HV_OUT and INPUT → OUTPUT.
  • I also need to have a second configuration where I swap the output channels, so HV_IN → OUTPUT and INPUT → HV_OUT.
  • Currently I had been testing this using a number of 1 Form A relays. However, I recognize that a Form C relay (SPDT) would be more effective here, however I haven’t been able to find one that has been capable of meeting the performance characteristics I’m looking for.

From our testing to date, noise has been identified from two sources:

  • From the isolation between the open contacts (more pronounced in the semiconductor SSR/Optocoupler relays)
  • Leakage from the coil current onto the measurement path (primarily observed in the reed relays we’ve tested)

Also, it is worth noting that in this design there will be no switching under voltage, if that has any impact on available options.

I hope this answers some of your questions above. I am open to any kind of suggestion as well. Thanks again for your assistance, it is much appreciated.


Hello Jackson,

Hope you are feeling better.

Here is a new article concerning the two-step PWM drive. It would be interesting to see if this PWM drive adds additional noise to your measurement circuit. The Arduino based PWM is only 1 kHz. I have slowed the system down using a 1 k gate resistor. However, the switching wavefronts are still steep. Perhaps it would be better to use 10 k or even 100 k to “round” out the waveform there by reducing the change for interference.

The leakage in the SSRs seems reasonable especially with the 10 kHz signal and anticipated capacitance in the SSR. I could be wrong, but I wonder if you are seeing an internal SSR snubber circuit.

The coil leakage is surprising. With a reed relay and a large glass enclosure, I would not have expected this problem. My first suspicion is capacitive leakage through the PCB trace.

I look forward to your analysis of the high voltage Sensata reed relays.



Hi Aaron,

Thank you again for this information, it will be very helpful. In your linked guide you use a relay with a 720Ohm coil resistance and 12V of DC Coil voltage, for a maximum of 16mA coil currents. This is achievable with a microcontroller like the the referenced Arduino.

For this case, the Sensata relay requires ~180mA (slightly different for the one we have but same range) of current at full voltage, so using a PWM microcontroller like the Arduino would be insufficient for this case. Do you have any recommendations to compensate?

As an option to eliminate the coil current noise altogether, we can also consider latching relays, as well. I hadn’t thought of this prior. This might also achieve what we are looking for in eliminating coil current noise as a source. In any case I still plan to test the two stage voltage influence on our sensate relays using a power supply, at least to further my understanding of how the coil current/noise influences the measurement.

Thank you,