Post by thread:Variable gain amp using PIC

Hello all. I need to read 50Hz signal with PIC16F877. Actually, 10-bit accuracy is enough, but the signal level is unknown, therefore PIC has to adjust to the signal. The level changes relatively slowly and there is some time allowed for adjusting. In this case four gain bands (1:2, 1:4, 1:8, 1:16) will do the job. Now, the problem is that I don't want to use variable gain amps, because they are too expensive and hard to get here, in Ukraine. Another way is to use analog switches which will change the gain of a usual op amp. Can the PIC's pins be used as analog switches, connected to ground? The datasheet says that ports use FETS and that the leakage current is as small sa 1uA. I would connect four resistors to PIC and made the pins go low or tri-state. Will the circuit work? __ -------| + \ | \ | >------------- To RA0/AN0 | / | ---|-- / | | ____ | |-----[_____]---- --------- --------- |---------------- | | | | [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] | | | | | | | ---------- RA1 | | --------------- RA2 | --------------------- RA3 -------------------------- RA5 Thanks in advance.

Hi, sorry, the circuit cannot be read. OTOH, you can try digital potentiometer from Xicor or from DALLAS... I hope this helps. Imre On Tue, 26 Oct 1999, Nikolai Golovchenko wrote: {Quote hidden}> Hello all. > I need to read 50Hz signal with PIC16F877. Actually, 10-bit accuracy is > enough, but the signal level is unknown, therefore PIC has to adjust to the > signal. The level changes relatively slowly and there is some time allowed > for adjusting. In this case four gain bands (1:2, 1:4, 1:8, 1:16) will do > the job. > > Now, the problem is that I don't want to use variable gain amps, because > they are too expensive and hard to get here, in Ukraine. Another way is to > use analog switches which will change the gain of a usual op amp. Can the > PIC's pins be used as analog switches, connected to ground? The datasheet > says that ports use FETS and that the leakage current is as small sa 1uA. > > I would connect four resistors to PIC and made the pins go low or tri-state. > Will the circuit work? > > __ > -------| + \ > | \ > | >------------- To RA0/AN0 > | / | > ---|-- / | > | ____ | > |-----[_____]---- > --------- > --------- > |---------------- > | | | | > [ ] [ ] [ ] [ ] > [ ] [ ] [ ] [ ] > | | | | > | | | ---------- RA1 > | | --------------- RA2 > | --------------------- RA3 > -------------------------- RA5 > > Thanks in advance. > >

It was a basic variable gain op-amp circuit (ascii schematics, ACK! I swear, I'm going to write a VGA Dos circuit renderer, if this keeps up...) Looked functional to me, though I'd want a little more flexibility than just 4 gain bands (as in, the jump from 1:8 to 1:16 is pretty large; how about using any 1,2,3 or 4 pins at once to give more gain choices?) Careful resistor choices should let this happen... When all 4 resistors are grounded, you have 1:2 (with my plan here.) Just another option <G> Mark Dr. Imre Bartfai wrote: {Quote hidden}> > Hi, > > sorry, the circuit cannot be read. OTOH, you can try digital potentiometer > from Xicor or from DALLAS... > > I hope this helps. > > Imre > > On Tue, 26 Oct 1999, Nikolai Golovchenko wrote: > > > Hello all. > > I need to read 50Hz signal with PIC16F877. Actually, 10-bit accuracy is > > enough, but the signal level is unknown, therefore PIC has to adjust to the > > signal. The level changes relatively slowly and there is some time allowed > > for adjusting. In this case four gain bands (1:2, 1:4, 1:8, 1:16) will do > > the job. > > > > Now, the problem is that I don't want to use variable gain amps, because > > they are too expensive and hard to get here, in Ukraine. Another way is to > > use analog switches which will change the gain of a usual op amp. Can the > > PIC's pins be used as analog switches, connected to ground? The datasheet > > says that ports use FETS and that the leakage current is as small sa 1uA. > > > > I would connect four resistors to PIC and made the pins go low or tri-state. > > Will the circuit work? > > > > __ > > -------| + \ > > | \ > > | >------------- To RA0/AN0 > > | / | > > ---|-- / | > > | ____ | > > |-----[_____]---- > > --------- > > --------- > > |---------------- > > | | | | > > [ ] [ ] [ ] [ ] > > [ ] [ ] [ ] [ ] > > | | | | > > | | | ---------- RA1 > > | | --------------- RA2 > > | --------------------- RA3 > > -------------------------- RA5 > > > > Thanks in advance. > > > >

part 0 3161 bytes {Quote hidden}> -----Original Message----- > From: Mark Willis [SMTP:spam_OUTmwillisTakeThisOuTFOXINTERNET.NET] > Sent: 26 October 1999 10:13 > To: .....PICLISTKILLspam@spam@MITVMA.MIT.EDU > Subject: Re: Variable gain amp using PIC - possible? > > It was a basic variable gain op-amp circuit (ascii schematics, ACK! I > swear, I'm going to write a VGA Dos circuit renderer, if this keeps > up...) > > Looked functional to me, though I'd want a little more flexibility than > just 4 gain bands (as in, the jump from 1:8 to 1:16 is pretty large; how > about using any 1,2,3 or 4 pins at once to give more gain choices?) > Careful resistor choices should let this happen... When all 4 resistors > are grounded, you have 1:2 (with my plan here.) Just another option <G> > > Mark > > Dr. Imre Bartfai wrote: > > > > Hi, > > > > sorry, the circuit cannot be read. OTOH, you can try digital > potentiometer > > from Xicor or from DALLAS... > > > > I hope this helps. > > > > Imre > > > > On Tue, 26 Oct 1999, Nikolai Golovchenko wrote: > > > > > Hello all. > > > I need to read 50Hz signal with PIC16F877. Actually, 10-bit accuracy > is > > > enough, but the signal level is unknown, therefore PIC has to adjust > to the > > > signal. The level changes relatively slowly and there is some time > allowed > > > for adjusting. In this case four gain bands (1:2, 1:4, 1:8, 1:16) will > do > > > the job. > > > > > > Now, the problem is that I don't want to use variable gain amps, > because > > > they are too expensive and hard to get here, in Ukraine. Another way > is to > > > use analog switches which will change the gain of a usual op amp. Can > the > > > PIC's pins be used as analog switches, connected to ground? The > datasheet > > > says that ports use FETS and that the leakage current is as small sa > 1uA. > > > > > > I would connect four resistors to PIC and made the pins go low or > tri-state. > > > Will the circuit work? > > > > > > __ > > > -------| + \ > > > | \ > > > | >------------- To RA0/AN0 > > > | / | > > > ---|-- / | > > > | ____ | > > > |-----[_____]---- > > > --------- > > > --------- > > > |---------------- > > > | | | | > > > [ ] [ ] [ ] [ ] > > > [ ] [ ] [ ] [ ] > > > | | | | > > > | | | ---------- RA1 > > > | | --------------- RA2 > > > | --------------------- RA3 > > > -------------------------- RA5 > > > > > > Thanks in advance. > > > > > > Attachment converted: wonderland:sch-draw.exe (bina/mdos) (0000E77E)

Hi Nikolay. Your idea will work. If all pins are tri-stated total amplifying is equal to 1. If you ground some resistors it will increase amplification. You only thing you should take into account that true ground for op-amp (nor negative supply pin but really ground) and ground of PIC are connected together otherwise in case of connection negative supply pin of op-amp with PIC ground you will cause re-charging of cap every time you change code on pins. So in this case put some delay before beginning ADC read. The second point is PCB tracing. Do ground between PIC and op-amp and ADC as short and wide as possible. Else you get additional error because of ground level shifting. (Very easy to see it with 10 bit conversion ;-) WBR Dmitry. {Quote hidden}> On Tue, 26 Oct 1999, Nikolai Golovchenko wrote: > > > Hello all. > > I need to read 50Hz signal with PIC16F877. Actually, 10-bit accuracy is > > enough, but the signal level is unknown, therefore PIC has to adjust to the > > signal. The level changes relatively slowly and there is some time allowed > > for adjusting. In this case four gain bands (1:2, 1:4, 1:8, 1:16) will do > > the job. > > > > Now, the problem is that I don't want to use variable gain amps, because > > they are too expensive and hard to get here, in Ukraine. Another way is to > > use analog switches which will change the gain of a usual op amp. Can the > > PIC's pins be used as analog switches, connected to ground? The datasheet > > says that ports use FETS and that the leakage current is as small sa 1uA. > > > > I would connect four resistors to PIC and made the pins go low or tri-state. > > Will the circuit work? > > > > __ > > -------|+ \ > > | \ > > | >------------- To RA0/AN0 > > +-|- / | > > | |--/ | > > | _____ | > > |-----[_____]---+ > > --------- > > --------- > > |---------------- > > | | | | > > [ ] [ ] [ ] [ ] > > [ ] [ ] [ ] [ ] > > | | | | > > | | | ---------- RA1 > > | | --------------- RA2 > > | --------------------- RA3 > > -------------------------- RA5 > > > > Thanks in advance. > > > >

You wrote {Quote hidden}> I need to read 50Hz signal with PIC16F877. Actually, 10-bit accuracy is > enough, but the signal level is unknown, therefore PIC has to adjust to the > signal. The level changes relatively slowly and there is some time allowed > for adjusting. In this case four gain bands (1:2, 1:4, 1:8, 1:16) will do > the job. > > Now, the problem is that I don't want to use variable gain amps, because > they are too expensive and hard to get here, in Ukraine. Another way is to > use analog switches which will change the gain of a usual op amp. Can the > PIC's pins be used as analog switches, connected to ground? The datasheet > says that ports use FETS and that the leakage current is as small sa 1uA. > > I would connect four resistors to PIC and made the pins go low or tri-state. > Will the circuit work? > > __ > -------| + \ > | \ > | >------------- To RA0/AN0 > | / | > ---|-- / | > | ____ | > |-----[_____]---- > --------- > --------- > |---------------- > | | | | > [ ] [ ] [ ] [ ] > [ ] [ ] [ ] [ ] > | | | | > | | | ---------- RA1 > | | --------------- RA2 > | --------------------- RA3 > -------------------------- RA5 > The problems are thus; 1. The signal cannot swing below ground in the present configuration. 2. The noise introduced by connecting the analog signal path to the digital signal path would be seriously detremental to youre 10 bit resolution. SO..... The nicest application I have used was a Light Dependant Resistor that replaced the resistors controlled by the micro thus __ -------| + \ | \ | >------------- To RA0/AN0 | / | ---|-- / | | ____ | |-----[_____]---- --------- --------- | [ ] LDR [ ] | | ___ Agnd A LED shines onto the LDR whose intensity is controlled by your 4 digital control lines, giving 16 levels if you like. The led drive can be decoupled with a suitable capacitor to remove noise from the digital circuit and now you have an unpolluted signal train with common components and digital control. It does work. _____________________________ Lance Allen Technical Officer Uni of Auckland Psych Dept New Zealand http://www.psych.auckland.ac.nz _____________________________

part 0 5405 bytes <META content=text/html;charset=koi8-r http-equiv=Content-Type><!DOCTYPE HTML PUBLIC "-//W3C//DTD W3 HTML//EN"> <META content='"MSHTML 4.72.3110.7"' name=GENERATOR> </HEAD> <BODY bgColor=#ffffff> <DIV>Hello!,<BR></DIV> <DIV> </DIV> <DIV>Thanks Rob! This editor is what I needed.</DIV> <DIV>I hope, this drawing is better readable:</DIV> <DIV><BR><FONT face="Courier New" size=2>       +1.5                  <BR>  2.5V -1.5 |\               <BR>   ---------|+ \          RA0<BR>            |    >----+----  <BR>          +-|- /      |      <BR>          | |/        |      <BR>          |     R     |      <BR>          +---\/\/\---+      <BR>          |                  <BR>         --- C               <BR>         ---                 <BR>          |                  <BR>          |  R1=R    1:2  RA1 <BR>          +---\/\/\---+---   <BR>          |  R2=R/3  1:4  RA2 <BR>          +---\/\/\---+---   <BR>          |  R3=R/7  1:8  RA3 <BR>          +---\/\/\---+---   <BR>          |  R4=R/15 1:16 RA5 <BR>          +---\/\/\---+---   </FONT></DIV> <DIV><FONT face="Courier New" size=2></FONT><FONT color=#000000></FONT> </DIV> <DIV><FONT color=#000000>Right, I didn't take into account the noise from current going through PIC</FONT></DIV> <DIV><FONT color=#000000></FONT>Besides there is a protection diod on any pin with anode tied to ground! The diodes will switch on all resistors when output swings below virtual ground (2.5V)  8-(  Because the capacitor starts discharging in that condition  )  Will this diod disable any possibility of the plot???</DIV> <DIV> </DIV> <DIV>I'm thinking about LDR or digital potentiometers now... Nice thing about LDR is that it can be controlled with only one digital line if you use PWM - needs only RC network for filtering.</DIV> <DIV> </DIV> <DIV>Thanks to all who replied.</DIV> <DIV>Nikolai<BR></DIV></BODY></HTML> </x-html>

You could use small MOSFETs to switch the resistors to a nice quiet analog ground. However, parts count is starting to go up quite a bit. If you are tempted by a digital pot, I have used the Xicor Quade2pot range quite a bit, they seem pretty good, although the datasheet is pretty obscure on some aspects. As an aside, I have the preliminary datasheet for a new Xicor part that contains: 3 digital pots (one 64 tap 50k, one 100 tap 10k and one 200 tap 10k), reset supervisor with power on reset, low voltage reset with adjustable trip levels, a manual reset input, three voltage monitor pins with user adjustable threshold (one with a hardware output, the other two readable through the serial bus), 2K eeprom memory and 2.7 to 5.5 volt operation. Phew. Sounds like a very useful part though, it's number is X6402 The only thing I would check about the LDR arrangement is linearity. I don't know how straight the transfer curve of a typical LDR is and combined with the non-linearity of a typical LED, you would have to use some kind of lookup table to get specific gains from the circuit. Certainly do-able though. Regards Mike Rigby-Jones > {Original Message removed}