0

My design needs a very low contact resistance electro-mechanical device, such as a reed relay. The solution needs to be small (PCB mount). It needs to have low leakage, with a contact resistance smaller than 10 mΩ, as that is the largest resistance value in my circuit (shunt resistor).

My "goal" would be:

  • 1 mΩ contact resistance.
  • Voltage less than 40 Vdc,
  • Current from 1A to ~1 μA.

Some have suggested using mercury wetted reed relays, which may be a possible solution, although they are not easy to find (currently manufactured ones).

Could I use multiple low contact resistance relays in parallel, to both increase current capability and decrease total contact resistance? I have seen that at the small currents that my device sometimes experiences there can be an issue with relay contacts.

The design cannot use MOSFETs, the entire point of the device is that it has very low resistance.

Basically the same question was asked here, but I have other requirements than the ones in this other topic.

In basic terms, it is a current measurement device. It goes shunt -> buffer -> x10 -> PGA -> ADC. The circuit is (almost) outputting the expected values (3-digit precision down to μA level; I want 5-digit precision).

There are two relays in my design. I'm using two 0.01 Ω shunts for mA/A/inaccurate μA, and 10 Ω for accurate μA/low-end mA. Relay #1 shorts out 10 Ω shunt, not used in most measurements. Relay #2 shorts out fuse. Relay #2 could be eliminated along with the fuse.

EDIT: I edited the question to best of my ability so that it does not ask for specific manufacturer parts.

Voltage Spike
  • 75,799
  • 36
  • 80
  • 208
super7800
  • 21
  • 4
  • 1
    I think you might have an issue with the contact resistance increasing over time as the relay contacts degrade. You might need a mercury wetted relay. I just looked and you can still get them. – Drew Feb 17 '22 at 02:01
  • 1
    I would suggest you rearrange your circuit so the contact resistance of the relay is not critical. It is usually possible to do that in switching shunt resistors. – Kevin White Feb 17 '22 at 02:24
  • Relays have not only a maximum current, but also a minimum current ("wetting current") to prevent corrosion at the contact points. So mercury wetted reed relay may be your best bet. If I understand correctly, you're using a 10mOhm shunt to measure supply current of a device-under-test (DUT), and want to use the relay to bypass the burden voltage of the shunt when in normal operation, correct? Does your DUT power supply have remote sense (Kelvin sense)? Maybe you could leave the 10mOhm shunt always in circuit, and just regulate the DUT voltage after the shunt? – MarkU Feb 17 '22 at 02:56
  • Super, I agree generally with @MarkU. But. You are looking to operate over six orders of current, too. That's quite a span. I'd like to see more of the surrounding concept circuitry (behavioral is sufficient.) But without that, I'd generally go with what Mark said to you already. – jonk Feb 17 '22 at 03:22
  • jonk, edited my post with basic operation. mercury wetted relay seems to be a good option, but looking online only found one current manufacturer (or reseller), misensor. If i go this route, it seems i may need to contact some chinese manufacturers (i need a part currently in production, or would need to use a readily available surplus part). A possible issue would be the required orientation, but if it's the only option, i would live with it. Am i overlooking a manufacturer of these? Seems people don't like mercury... – super7800 Feb 17 '22 at 03:33
  • 2
    @super7800 [Here](https://i.stack.imgur.com/HeVna.jpg)'s what I use, for example. COTO relays. These do work well over a wide dynamic range. They are NOT cheap. But I could not afford leakage and they provide sub-picoamp leakage for me. These are used over about eight orders of magnitude current range. But not up to an amp, granted. Perhaps up to some milliamps. (Sorry about the dusty board... but I pulled it from storage to get the photo. It hasn't been used in about a decade, now.) – jonk Feb 17 '22 at 05:48
  • 2
    You will very greatly improve the presentability of your text by following basic English rules: Capitalise the first word in sentences, capitalise "i" , Use proper English construction so sentences are 'complete thoughts'. – Russell McMahon Feb 19 '22 at 11:40

1 Answers1

1

There are MOSFETs that have on-resistances of less than 10m\$\Omega\$. They will have high capacitance in the off state. Relays are finicky since the contact resistance will vary and low currents won't break oxide layers on the contacts. Mercury wetted, as you found out, are hard to come by due to RoHS restrictions.

Perhaps use two amplifiers as shown below. You can sum the outputs of U1 & U2 (high level symbols for instrumentation amplifiers) or multiplex using an inexpensive analog switch. This gives you 4-wire (Kelvin) connections on your sense resistors.

schematic

simulate this circuit – Schematic created using CircuitLab

qrk
  • 7,474
  • 1
  • 5
  • 20