1-Wire ADC

Not sure why there aren't any around ?

But we/I made our own for a battery shunt, in the negative lead A/D monitor. Single wire is the power and two way data but only about 10 conversions per second in this case. The shunt is the second wire.

Used a 35 cent micro and I2C A/D. I don't see why you couldn't address more than one of these on a single wire.

I suppose it all depends on how fast you need the data to be

Every few minutes will do.....

Common grounds are another obstacle. Hard to measure batteries in series...

--
A host is a host from coast to coast.................wb8foz@nrk.com 
& no one will talk to a host that's close.......................... 
Unless the host (that isn't close).........................pob 1433 
is busy, hung or dead....................................20915-1433

I know nothing of ADCs, but isn't this successive approximation? maybe you have the wrong search term? George H.

Any digital device can be a 1-wire device. I would just program up a cheap MCU with a built in ADC to be controlled over 1-wire. It just takes a resistor and a cap to power it. No?

--

  Rick C. 

  - Get 1,000 miles of free Supercharging 
  - Tesla referral code - https://ts.la/richard11209

At that sort of sampling speed it's almost always sigma-delta - pulse-width modulation of the probe voltage to match the voltage being measured. Typically good to about 20-bit accuracy if the software keeps on trying for long enough.

--
Bill Sloman, Sydney

Making a 1-wire slave can be a bit fiddly, but it's perfectly possible, yes. And if you are monitoring the power in the battery, you can surely use that to power the microcontroller (your microcontroller is mostly in sleep mode, and a tiny device will have lower power requirements than most batteries' leakage current).

If you don't need to conform to the Dallas 1-wire protocol, but merely need communication over one wire, then it gets a lot simpler. You can use a standard UART, tie TX and RX together with the TX pin in open drain mode, and you've probably got a good enough solution.

The MCU itself will cost practically nothing. The real cost is that you'll need to program them.

Power each measuring unit locally from local battery and connect the units in series into a single current loop using optoisolators. If more or less standard UARTs are used, use two current loops. Connect a scheduler transmitter to one loop and use it to address individual measuring units (with Rx pins connected to the loop with optoisolator). The addressed unit then sends measurement data by interrupting the current flow in the other loop. A central monitor then extract the data from that loop.

The drawback is that every unit must drive the Tx into Mark (20 mA) Idle state to allow current pass from one station to the next. These days with better optoisolators a much smaller loop current (say 2 mA) could be used.

To get rid of the polling circuit, some CAN (Controller Area Network) style current loop could be used (e.g. 2 mA recessive state and 0 V dominant state) and let the nodes themselves handle the arbitration on the same loop with both Tx and Rx isolators in the same loop for all stations.

I once invented a leapfrog matrix of current-limited SSRs to check a series string of batteries. Odd/even nodes are switched to opposite sides of a floating, bipolar ADC.

A couple more SSRs could make a flying-capacitor isolator.

--

John Larkin         Highland Technology, Inc 

Science teaches us to doubt. 

  Claude Bernard

You seem to be reinventing the telegraph. lol

No need for two loops unless you want redundancy.

--

  Rick C. 

  + Get 1,000 miles of free Supercharging 
  + Tesla referral code - https://ts.la/richard11209

Kind of like a Lithium battery BMS sounds like.

"Reinventing" ??

4-20 mA (and 0-20 mA) current loops are still in widespread use in industrial systems.

I just described it in more detail so that some younger readers of this newsgroup would also consider it.

Half-duples over a single pair has been used for nearly a century with teleprinters. There must be some conventions to allow only one end of the link to transmit at a given time.

In a multidrop current loop, only one station is allowed to transmit at once, thus arbitration between slaves (and optional master) is needed.

The arbitration can be done by a master station in a single loop and each slave must be able to distinguish between a master command and some other slave sending out data (since it is heard by all slaves).

With optoisolators about 2 V voltage drop will occur in each Tx and Rx. With a large number of nodes, say 30 (for a 30 x 12V =360 to 400 V system) the current loop constant current loop supply must be at least

120 V (= 30 x 2 x 2V) with a half duplex system, which is too much for some optoisolator transistor max Vce. Using two loops and the supply voltage needs to be only 60 V (30 x 2 V), which more optoisolators can handle.

Don't use optos, use capacitors. save lots of power.

You may need to use symbols with the same number of ones and zeros but that shouldn't be a big problem.

--
  Jasen.

If you are going to use Manchester coding, you could also use transformers for isolation, e.g. cheap Ethernet transferrers. Just make sure you have steep edges, if the transformer low frequency response is not very good.

When working close to big batteries, inductive or capacitance coupling can be an issue, so you can't reduce the bus signal power too much.

A current loop optoisolator is not that bad. About 3 mW (1 mA from 3.3 V) from each node isolated power supply should be enough to control a

10 mA current loop.