------Question for ADUM7442ARQZ-RL7 Please Put your question below------
These questions apply to iCoupler devices in general.
- ‘LINE’ refers to physical wires between PCBs
- 1st PCB supplies power & ground to 2nd PCB over 4-conductor cable: Pwr, Gnd, Tx & Rx.
- Line lengths <100m
- Speeds < 230Kb
Main purpose is data line protection
(1) Is it ok to use same supply & ground for both (isolated) sides of a single iCoupler device?
For PCBs, we normally use Optocouplers for data line isolation.
We found iCoupler tech as an alternative. Problem is all references only show two separate & isolated power supplies. We need to isolate/protect data channels only. Can Vdd1 be tied to Vdd2 and Gnd1 tied to Gnd2, where Vdd ranges 3-5VDC? Does that, in any way compromise the protection of the data lines? We are focused on this need to isolate two data lines:
(2) A second development effort requires data line level shifting - we HAVE found this sample circuitry, but again, none with the grounds tied together between the two PCBs. Is that okay?
Hope these are easy & quick for you!
Greetings,
The intended use of an isolation device of this sort is to allow communication between two different power supply domains that have no direct connection to each other. Yes, the device will still operate and transmit signals if the same power and ground are used for both sides, but doing so defeats about 80% of the purpose for which the device is made. What benefit remains could probably be achieved using $0.05 worth of multi-sourceable TVS diodes and such, compared to a $5 isolation IC that likely has no footprint-compatible alternatives on the market.
As for level shifting there are many ways to go about the matter depending on the circumstances, but if a person isn’t planning to isolate the power domains there’s an entire product family of parts designed for exactly that purpose which run about 1/10th the cost of a digital isolator IC.
1 Like
Excellent eye-opening point$$!
Thank you, Rick from 1976!
A follow-up, if you would, please:
a) What would be your go-to for a 5v line?
b) Would it be prudent to use two for each line - one to ground & one to 5V?
c) These are ground referenced signals - does it matter if bi- or uni-directional?
(this SMBJE5-0CAH is $0.15@1k)?
The rule of thumb for over-voltage protection devices is to select one with a reverse standoff rating equal or slightly above the maximum voltage normally seen in circuit, the idea being to provide a clamping effect that kicks in promptly when things go outside of normal, but not conducting to any significant degree under normal conditions.
5V being a common logic level, there are a lot of products available. There’s a bit of sloppiness as to how different suppliers relate the info, so I’d suggest making an initial cut by selecting for devices with figures in the 5-6v range.
The unidirectional types function essentially like a zener diode, clamping in the reverse direction to some defined level (5.5v perhaps) and having a low voltage drop of perhaps 0.6v in the forward direction. Careful examination will show that this behavior aligns quite nicely with the absolute maximums of many logic devices.
Note however, that devices such as the PRTR5V0U2F,115 exist which provide a more integrated and robust solution to a very common need.
Such devices work on the basis of “steering” an over-voltage event applied to a data line into the power network, where there’s usually capacitance aplenty to absorb it and a zener diode built into the protection device to clamp what’s left over. This “steering” principle is in fact incorporated into most ICs, partly for protection and partly to confuse undergraduate students who forget to turn on the power supply…
Yes, it matters. The bidirectional types exhibit a symmetric behavior, while the unidirectional types are asymmetric. The latter behavior is a much better match for the typical prescriptions of the Absolute Maximums box on the average datasheet for the average integrated device, which on average state that thou shalt not apply a potential to an I/O pin that exceeds the power supply rails by more than 0.5v or so. Why not? Because doing so forward biases those built-in steering diodes spoken of above, which are relatively puny and easily damaged by excess current flow. Adding a bit of series resistance to the I/O pin serves to limit that excess current under abnormal conditions, providing an additional layer of robustness that (for inputs at least) doesn’t make a great deal of impact in normal situations, since inputs are high impedance anyhow. (At least they are at DC… The input capacitance of the pin will form a low-pass filter with an external resistance, so one has to limit use of the trick as speeds increase.)
That just about does it! 
Very helpful, and I appreciate your suggestions -
as for me, to the next rabbit hole…
