Post by thread:Opto-isolator Question

I need to measure a voltage 3-30VDC using a PIC, but the PIC has a totally different ground reference than the voltage I'm supposed to be measuring (the two circuits _are_ connected, but I have no way of knowing how because each install will probably be different). I was thinking that this is an ideal situation for using an opto-isolator, and was wondering if anybody knew in a practical way if this would provide reasonable accuracy with varied temps, etc. Anybody have any other suggestions? Thanks! Andy ================================================================== Andy Kunz - Montana Design Go fast, turn right, and keep the wet side down! ==================================================================

How about a differential amplifier? Andy Kunz wrote: {Quote hidden}> I need to measure a voltage 3-30VDC using a PIC, but the PIC has a totally > different ground reference than the voltage I'm supposed to be measuring > (the two circuits _are_ connected, but I have no way of knowing how because > each install will probably be different). > > I was thinking that this is an ideal situation for using an opto-isolator, > and was wondering if anybody knew in a practical way if this would provide > reasonable accuracy with varied temps, etc. > > Anybody have any other suggestions? > > Thanks! > > Andy > > ================================================================== > Andy Kunz - Montana Design > Go fast, turn right, and keep the wet side down! > ==================================================================

Hewlett Packard make an opto isolated 4-20mA current loop in an 8 pin DIP - HCPL4100. I think there are also linear opto isolators for analog voltages, but I can't find a reference for them :-( Tony PS. I posted my 16F84 based Programmable Ignition system on the PicNPoke web site if anyones interested. PicNPoke Multimedia 16F84 Beginners PIC Tools. **PLUS** - PicNPlay - PicNPlan - PicNPrep - PicNPost PicNPort - DT Type Saver - *new* PicNQuiz. Recent addition - Hex To Source converter. http://www.picnpoke.com Email spam_OUTpicnpokeTakeThisOuTcdi.com.au

Thinking about it more, Maxim also makes a "flying capacitor" isolator that could be used. Then Analog Devices (I think) makes isolation amplifiers. Someone else also mentioned differential amplifiers. AD (again, I think) has one that will handle a couple hundred volts common mode. Harold _____________________________________________________________________ You don't need to buy Internet access to use free Internet e-mail. Get completely free e-mail from Juno at http://www.juno.com Or call Juno at (800) 654-JUNO [654-5866]

On Tue, 26 May 1998 21:03:50 -0400 Andy Kunz <.....montanaKILLspam@spam@FAST.NET> writes: >I was thinking that this is an ideal situation for using an >opto-isolator, >and was wondering if anybody knew in a practical way if this would >provide >reasonable accuracy with varied temps, etc. > I think CP Claire (or something like that) has an opto designed for linear operation. As I recall, there might be two LEDs and two detectors that are matched so you can get rid of nonlinearities by putting stuff in an op-amp feedback loop. They've also got some telephone DAAs that don't use a transformer, which is pretty neat! Harold _____________________________________________________________________ You don't need to buy Internet access to use free Internet e-mail. Get completely free e-mail from Juno at http://www.juno.com Or call Juno at (800) 654-JUNO [654-5866]

Hello Maybe set up a Voltage to Frequency converter powered by the 3-30 Volt i/p, use a standard opto isolator and count the pulses? You could look at one of these opto-couplers if you chose an analoug interface & A/D it on the uP's side. SIEMENS IL300 or IL300 Gain Stability +/- 0.005% Linearity 0.01% These analoug opto's have two separate isolated photo diodes, one you use as feed back on the input loop side. ( You could use a op amp that would accept a wide Supply voltage in (3-36V) to control a linear range of i/p voltage to feed back current.) HP HCPL-4562 Wide band 17 MHz Low Temp. Coeffiicient -0.3%/deg. C High Linearity at low drive currents High Gain Typicaly 2.0 >>> Andy Kunz <montanaKILLspamFAST.NET> 05/27/98 10:33AM >>> I need to measure a voltage 3-30VDC using a PIC, but the PIC has a totally different ground reference than the voltage I'm supposed to be measuring (the two circuits _are_ connected, but I have no way of knowing how because each install will probably be different). I was thinking that this is an ideal situation for using an opto-isolator, and was wondering if anybody knew in a practical way if this would provide reasonable accuracy with varied temps, etc. Anybody have any other suggestions?

> I think CP Claire (or something like that) has an opto designed >for linear operation. As I recall, there might be two LEDs and two >detectors that are matched so you can get rid of nonlinearities by >putting stuff in an op-amp feedback loop. > They've also got some telephone DAAs that don't use a >transformer, which is pretty neat! I did this a while back, using a pair of optos, with the LEDs in series. The idea is that the photocurrent in the "B" transistor is the same as in the "A" transistor, since the LEDs have the same current. It is important to feed the LEDs with current, rather than voltage. I expected it to be very finicky, but in fact it was easy, stable, and very repeatable. No fancy pricey special optos. Four optos and a dual opamp, and I had a reasonable hybrid.

At 09:03 PM 26/05/98 -0400, you wrote: {Quote hidden}>I need to measure a voltage 3-30VDC using a PIC, but the PIC has a totally >different ground reference than the voltage I'm supposed to be measuring >(the two circuits _are_ connected, but I have no way of knowing how because >each install will probably be different). > >I was thinking that this is an ideal situation for using an opto-isolator, >and was wondering if anybody knew in a practical way if this would provide >reasonable accuracy with varied temps, etc. > >Anybody have any other suggestions? > >Thanks! > >Andy > >================================================================== > Andy Kunz - Montana Design > Go fast, turn right, and keep the wet side down! >================================================================== > > Ah Andy, You are about to undertake what people whom design Hi side switches do on a regular basis. Firstly you need to chose the OPTO carefully, as the operating curve is what you are after. If I remember this is for a battery powered system ,so look at the Sharp OPTOs as there are quite a few that operate very well in the low current range. Once you have characterised the OPTO, the rest is quite simple, but do look for OPTOs that have a largish linear type area. Be careful, as some OPTOS will just start to enter the 3V area, this would make switching and self calibration difficult) Unfortunately, I can't think of a way that will be cheeper than a POT in order to calibrate this thing, as each will be somewhat different too (You could use two OPTOS, two FETs and a reference to self calibrate). This of course may not be a problem depending on the accuracy required. By this, the gain should be quite OK, just the offset may need shifting. Note that you may find it better to alter the current flowing through the diode, rather than the current sinked by the transistor. The range that you need to measure is quite large, so some form of current scaling may be required (This may even be a simple resistor) The calibration point should remain fixed, so if the range is acceptable, then a simple 8 bit offset is all that's required As for stability over temp, what range? Most of these simple OPTO circuits will work quite well over a nominal temp of 0 to + 40C But as both circuits are connected together, then maybe a differential amp could be used, or even a current mirror. Dennis

At 21:03 26.05.98 -0400, you wrote: >I need to measure a voltage 3-30VDC using a PIC, but the PIC has a totally >different ground reference than the voltage I'm supposed to be measuring >(the two circuits _are_ connected, but I have no way of knowing how because >each install will probably be different). > Remember that you can use a external 8 pin AD converter and isolate the digital interface. You would need 3 of the cheapest optocouplers that you can find. In addition you will need a isolated PSU for the AD converter. In my application, this turned out to bee the cheapest solution. ( analog optocouplers are expencive ) As a bonus, there's no analog mess to calibrate. Haavard. ------------------------------------------------ HŒvard T¿rring .....htKILLspam.....neo.no EraseMExplisitspam_OUTTakeThisOuTneo.no simplisitspam_OUTneo.no Norsk Elektro Optikk A/S Tel: +47 67974700 PoBox 384 Fax: +47 67974900 1471 SkŒrer http://www.neo.no Norway

At 21:03 26.05.1998 -0400, you wrote: >I need to measure a voltage 3-30VDC using a PIC, but the PIC has a totally >different ground reference than the voltage I'm supposed to be measuring >(the two circuits _are_ connected, but I have no way of knowing how because >each install will probably be different). > >I was thinking that this is an ideal situation for using an opto-isolator, >and was wondering if anybody knew in a practical way if this would provide >reasonable accuracy with varied temps, etc. Hi, I'm measuring voltages up to 5000V with a simple diff.amp and a PIC. Cheap, reliable and LINEAR ! If you need some schematics, please let me know. But it is as simple as it sounds. So long Thomas =8-) -- ********************************************************** * Thomas Magin FON: ++49-761-4543-489 * * marquette-Hellige GmbH FAX: -507 * * M&E-FW / Emergency Systems email: @spam@maginKILLspamhellige.de * * Munzinger Str. 3 * * D-79111 Freiburg / Germany * **********************************************************

Hello PIC.ers, >Date: Tue, 26 May 1998 21:58:57 -0500 >From: David VanHorn <KILLspamdvanhornKILLspamCEDAR.NET> >Subject: Re: Opto-isolator Question >> I think CP Claire (or something like that) has an opto designed >>for linear operation. As I recall, there might be two LEDs and two >>detectors that are matched so you can get rid of nonlinearities by >>putting stuff in an op-amp feedback loop. <snip> {Quote hidden}>I did this a while back, using a pair of optos, with the LEDs in series. >The idea is that the photocurrent in the "B" transistor is the same as in >the "A" >transistor, since the LEDs have the same current. It is important to feed >the >LEDs with current, rather than voltage. I expected it to be very finicky, >but in >>fact it was easy, stable, and very repeatable. No fancy pricey special >optos. >Four optos and a dual opamp, and I had a reasonable hybrid. .. ditto the above. I've been using this same dual-opto arrangement for many years in analogue circuits. Originally I used the TIL113 optocoupler, but later found that 4N25s were much better matched for characteristics. The TIL113s would only work if you started out with a box of them and swopped them about till the isolated feedback behaved acceptably. 4N25s don't need this, every installation works a coconut first time. I suppose others will be equally good or better, but I've had no need to try further. Best regards, John .. email from John Sanderson at JS Controls, PO Box 1887, Boksburg 1460, Rep. South Africa Manufacturer & purveyor of laboratory force testing apparatus and related products and services. Tel/fax: Johannesburg 893 4154 Cellphone 082 453 4815

Andy, I've never used these parts, but it is possible to buy isolation amps off the shelf. Prices are relativly high but they might save you a lot of time messing about with opto isolaters. Analog devices make some of them (AD102JY, AD104JY, AD202JY and AD204JY), as do Burr Brown (ISO103, ISO122JP, ISO165P and ISO212JP). Here in Ireland Farnell components stock them. You can contact them at http://www.farnell.com to get a rough idea of prices from their online catalog. If you check out their site maybe you could let me know your impressions of their prices? Joe P.S. NOT affiliated with any of the above companies! At 21:03 26/05/98 -0400, you wrote: {Quote hidden}>I need to measure a voltage 3-30VDC using a PIC, but the PIC has a totally >different ground reference than the voltage I'm supposed to be measuring >(the two circuits _are_ connected, but I have no way of knowing how because >each install will probably be different). > >I was thinking that this is an ideal situation for using an opto-isolator, >and was wondering if anybody knew in a practical way if this would provide >reasonable accuracy with varied temps, etc. > >Anybody have any other suggestions? > >Thanks! > >Andy > >================================================================== > Andy Kunz - Montana Design > Go fast, turn right, and keep the wet side down! >================================================================== > >

Andy Kunz asked about measuring isolated voltages. You could buffer the 3-30VDC to drive a Hewlett Packard 4-20mA loop optoisolator. I once had to measure an isolated voltage from a pressure sensor of a pneumatic valve block. Initially I tried using the IL300 linear isolator, because I thought they were clever devices and I wished to learn how to use them for future ref. When I couldn't get it going in a hurry I fell back to the plan of opto-isolating the digital signals to/from the serial ADC (clock, select, data) and this worked fine. That was a cheap 3-wire serial DAC from TI. You could use two HP2630 dual optoisolators to isolate an I2C bus (as I have done), and have multiple isolated devices. Squeezing I/O data through opto-isolated serial link uses far fewer isolators than you would if you had to isolate every digital I/O signal. Its thus a lot cheaper. I believe Arcom Control Systems do this on one of their PC104 I/O cards.

David, Any chance we could see a schematic? Joe At 21:58 26/05/98 -0500, you wrote: {Quote hidden}> >I did this a while back, using a pair of optos, with the LEDs in series. >The idea is that the photocurrent in the "B" transistor is the same as in >the "A" >transistor, since the LEDs have the same current. It is important to feed >the >LEDs with current, rather than voltage. I expected it to be very finicky, >but in >fact it was easy, stable, and very repeatable. No fancy pricey special >optos. > >Four optos and a dual opamp, and I had a reasonable hybrid. > >

On Wed, 27 May 1998 10:30:54 +0100 Keith Howell <RemoveMEkeithhTakeThisOuTARCAM.CO.UK> writes: > When I couldn't get it going in a hurry I fell back to >the plan of opto-isolating the digital signals to/from the serial ADC >(clock, select, data) and this worked fine. To stay on topic, consider a PIC with ADC on the isolated side. It could measure several voltages (not isolated from each other) and pass the results back through one cheap optocoupler. It would be quite simple to use an asynchronous serial format. Passing analog signals, especially where DC counts, through optocouplers is fraught with problems. A circuit using two optos where one provides a reference feedback is essential. Even then, the two must be well-matched. I had considered isolating through a pulse transformer. The input side chops the DC voltage to pulses of proportionate amplitude. The output would rectify the voltage. An identical rectifier on the input side with a third winding could be used for linearizing feedback. If some connection between the two sides is permissible, a differential amplifier would likely be overall best. _____________________________________________________________________ You don't need to buy Internet access to use free Internet e-mail. Get completely free e-mail from Juno at http://www.juno.com Or call Juno at (800) 654-JUNO [654-5866]

-----Original Message----- From: Joe McCauley <spamBeGonepmcculeyspamBeGoneTCD.IE> To: TakeThisOuTPICLISTEraseMEspam_OUTMITVMA.MIT.EDU <RemoveMEPICLISTTakeThisOuTMITVMA.MIT.EDU> Date: Wednesday, May 27, 1998 9:05 AM Subject: Re: Opto-isolator Question >David, > >Any chance we could see a schematic? > >Joe I hate ascii schematics... Here's what you do: Configure an opamp as a voltage to current converter, at 1V = 10mA In the feedback loop, put the two opto diodes, in series. This means that you have 10mA through each diode. Now connect the collector of one opto's transistor to your + rail (+5V?). Connect the bottom of the diodes to ground, and the emitter replaces them to provide feedback current. (At this point, the current through the LEDs drops off to 10mA / the opto transfer ratio) If everything is working, you have 10mA through the transistor of the first opto for 1V input. The second transistor will also provide very close to 10mA regardless of supply and load (within broad limits) This gives you the basics, if I explained it clearly :) and you can adapt as you like. I did find that dual optos were better matched than singles, but I didn't have any trouble with it even using different optos. The output current may be different if the optos are different types, but it is still proportional and linear. Think of it as sort of an optically isolated current mirror :)

Andy, my first thought is to also to use a V/F or VCO. While there are several V/F and VCO chips you might want to take a look at classic old 555 timer circuits. One circuit that comes to mind is to use a voltage controlled current source to control the charge of the 555's timing capacitor. TI has an application note using a 555 and their TL011 current mirror which is incredibly simple. It only requires 2 caps, 2 resistors, and a diode. The reference design uses a 10-50V input and a 15V supply but you can easily change it. The advantages are low cost and a single opto-isolator. My next choice would be a serial A/D. Another option here is the PIC12C67x though I'm not sure of current pricing vs low-cost A/Ds. It would allow pre-processing if needed. In both cases, you can use a dual opto-isolator. Someone else mentioned isolation amps from AD and Burr-Brown but they are fairly expensive. However, they are very accurate and rugged. - Tom At 09:09 AM 5/27/98 -0400, Andy Kunz wrote: {Quote hidden}>At 11:59 AM 5/27/98 +0930, you wrote: >>Hello >> >>Maybe set up a Voltage to Frequency converter powered by the 3-30 Volt >i/p, use a standard opto isolator and count the pulses? > >That's a pretty neat idea. I'll have to investigate that. > >Thanks! > >Andy > > >================================================================== >Andy Kunz - Statistical Research, Inc. - Westfield, New Jersey USA >================================================================== > >