Isolated current loop receiver

push pull or any way you like in series with the loop. Use the external power to drive the power stage gates.

Perhaps a simple-minded H-bridge driving a transformer, DC-restore on the other side.

maximum drop voltage

...Jim Thompson

sting

ed

to

hms

arge-

be

of

yeh something like an ad7401, will need a clock and an isolated supply

with calibration something like this:

-Lasse

are

of

box.

Totally cool part. Isolation, great linearity, with opamp maybe 3 V burden on the current loop, and good dc and ac accuracy and bandwidth.

?-)

existing

ohms

large-

be

Wow, so many ideas so little time. Since you have external power available you can make a version of Fred Bartoli's suggestion zero burden, dc and ac accurate to about 0.1%, and a couple kHz bandwidth real easy. A floating "virtual ground (0 v dc offset and controllable dynamics)" transporting power across the isolation boundary. Or perhaps similar Fred's circuit with a TL431 shunt regulator providing an fixed voltage drop of say 3.3 V (preferred for 4 to 20 mA circuits).

?-)

sting

ed

to

hms

arge-

be

of

Ooops, you're right and I used to design interfaces for TTY machines and their current loop interface.

K3, the opto CTR ratio, varies by almost 3:1.

re

of

|          |

ox.

but binned for +/-6%

hcnr201 is a bit better at +/-5%

but they both need calibration

-Lasse

It's not very good over temperature, either. Some PWM or digital scheme would be more accurate.

Le Sat, 15 Jun 2013 09:35:31 -0700, John Larkin a écrit:

VCC

???

K3 is given for 0.005%/K typ, 0.05%/K max. Pretty good actually.

Do these things age similarly? Can you add a reference opto pair to use in a feedback path to linearize the circuit? Drive both LEDs with the same current and you should get the same current on the other side.

Ah, I see that *is* what they are doing... lol

If you have got external power, and a +1V to +5V signal range (4mA to 20mA in volts) and don't need too much bandwidth, you could do what I did at Geo rge Kent back in 1975, which was to use a three-winding precisely 1:1:1 tra nsformer.

An op amp drove one winding, and sensed the voltage produced across a secon d winding, so that it exactly matched the input voltage (give or take the o p amps input offset voltage). The third winding was on the other side of th e isolation barrier.

The current through the driven winding was monitored and when it got too hi gh, the op amp output was clamped to the negative rail and held there until the current through the transformer had dropped back through zero to almos t too high in the opposite sense.

The output from the isolated winding was disconnected from its output filte r while this was going on. I think I used a FET for that job, with the gate tied to the output from the isolated winding.

It was all rather horrible, but exquisitely accurate.

...snip...

If you want to go digital, this design is screaming for a simple, low power ADC on the current loop. 100 Hz bandwidth means 500 Hz sample rate would be gravy. Using Manchester encoding to clock the data gives a 14 kHz bit rate for the 12 data bits plus a couple of framing bits. Something digital on the downstream side gets the analog back if that is what is needed or the digital can be used directly.

This can all be done in a small low cost MCU such as a PSOC with very minimal external circuitry.

Would that be simpler and easier than a typical opto-isolator? Once you go digital the opto doesn't need to deal with any of the analog stuff.

Le Sat, 15 Jun 2013 21:02:55 -0400, rickman a écrit:

Not exactly. There's no guaranty that leds would be aging similarly. But there's no aging for the photodiodes. So, only one led, shared between the forward and feedback PD paths.

sting

ed

to

hms

arge-

be

of

WOW! You're right that said 100Hz, not 100kHz !!!

Someday curiosity will outstrip vanity and I'll wear those glasses.

an

current)

ADC.

would

is

errors are

should

hundreds of

Vishay

junk box.

bandwidth.

would be

And the old hen is squawking and scratching at nothing much.

TAIFP

?-)

of

box.

be

So, you think it's accurate? And accurate over temperature? Do you admire trimpots?

Can't invent your own insults?

My first thoughts were something along the lines of a V-to-F converter, and I found such a scheme in US patent 7,477,080

Might also do a classic dual-slope A-to-D, use the signal current for one side of the integration, a reference current for the other side.

Trivial to do as a custom chip ;-) I'll have to ponder if it could be done with parts available off-the-shelf. ...Jim Thompson

Actually my problem is solved. My customer said "oh, I'll do that part".