Post by thread:Sigma Delta converter

Russell McMahon wrote: > > Has anyone implemented a Sigma Delta converter in software, whether > with a PIC or otherwise. > I haven't but I am probably about to. > > For those who haven't met them, in their simplest form a sigma-delta > converter is an integrator followed by a comparator (effectively a 1 > bit A2D) which tracks the signal. The mean output (digital) signal is > the input signal. This can be measured by averaging o=r by using a > digital filter. Essentially any desired sensible level of accuracy > can be achieved given enough conversion time. Walter has a full explanation and pictures too : http://www.bytecraft.com/addaconv.html If you pick Ri, Rf, and the sample times wisely, you have zero maths to do, and the code becomes : PinA := !PinB -- Friendly Regards /"\ \ / Tjaart van der Walt X ASCII RIBBON CAMPAIGN spam_OUTtjaartTakeThisOuTwasp.co.za / \ AGAINST HTML MAIL |--------------------------------------------------| | Cellpoint Systems SA http://www.cellpt.com | |--------------------------------------------------| | http://www.wasp.co.za/~tjaart/index.html | | WGS84 : -26.0124 +28.1129 | | Voice : +27 (0)11 2545100 | |--------------------------------------------------|

I asked about this some weeks ago but a second query may be useful. I am trialling a Sigma Delta analog to digital converter. This SHOULD allow low cost high accuracy A2D with modest conversion speed. Results to data only get so good (about 1/2 percent accuracy - slightly worse than 8 bits)(plus a few worse results here and there) and then seem to offer rapidly diminishing returns. Using more conversion cycles doesn't seem top help much. I'm using 2048 counts at present and would EXPECT much better results. I have used both a standard microprocessor input pin as the sense element (PIC) and an internal comparator (Z8) and an external op-amp (LM324) as comparator. I am aware of the poor performance of the LM324 as a comparator but am compelled to use it if at all possible due to cost considerations. If it CANT be used then something else will be used but at around $US0.16 in volume for a quad op-amp it is very attractive. I also need to deal with a 0 to 0.2V input signal so an amplifier is desirable. Anybody had better than about 1/2% accuracy results from a Sigma Delta built discretely? Any thoughts on the process? This is a stunningly cheap and simple method for A2D if it would only work :-) - once sorted it would/will prove a boon to other PIC users. regards Russell McMahon

At 01:13 13/07/99 +1200, you wrote: {Quote hidden}>I asked about this some weeks ago but a second query may be useful. > >I am trialling a Sigma Delta analog to digital converter. This SHOULD >allow low cost high accuracy A2D with modest conversion speed. > >Results to data only get so good (about 1/2 percent accuracy - >slightly worse than 8 bits)(plus a few worse results here and there) >and then seem to offer rapidly diminishing returns. Using more >conversion cycles doesn't seem top help much. I'm using 2048 counts >at present and would EXPECT much better results. > >I have used both a standard microprocessor input pin as the sense >element (PIC) and an internal comparator (Z8) and an external op-amp >(LM324) as comparator. I am aware of the poor performance of the >LM324 as a comparator but am compelled to use it if at all possible >due to cost considerations. If it CANT be used then something else >will be used but at around $US0.16 in volume for a quad op-amp it is >very attractive. I also need to deal with a 0 to 0.2V input signal >so an amplifier is desirable. > >Anybody had better than about 1/2% accuracy results from a Sigma >Delta built discretely? >Any thoughts on the process? >This is a stunningly cheap and simple method for A2D if it would only >work :-) - once sorted it would/will prove a boon to other PIC users. > > >regards > > >Russell McMahon > > An LMC660 comes close to the bargin price, but the 324 should do. Some more information is required, are you using a split supply for the OPAMP, if not how close to 0V do you get. When you say 0.5% error at what voltage input level is this? Is the error at the same input level or over the entire range? The 324 has a 2mV offset that's up to 1% error in 200mV! Also are you using dual slope conversion or single slope? The number of counts not altering the value indicates that this is what the converter is reading, so I would dare say that the reference is also ok. But a Z8! Bad boy! Please tell me more Dennis

On Tue, 13 Jul 1999 01:13:23 +1200 Russell McMahon <.....apptechKILLspam@spam@CLEAR.NET.NZ> writes: >I have used both a standard microprocessor input pin as the sense >element (PIC) and an internal comparator (Z8) and an external op-amp >(LM324) as comparator. I am aware of the poor performance of the >LM324 as a comparator The 324's comparator brother the 339 (also offered in a dual unit, the 393) should be about the same cost. But then you need two chips unless you can make the 339 work well as an amplifier. There was a chip with 324 type amplifiers and 339 comparators on one chip. It's likely been discontinued. Probably the major problem with the 324 is its low bandwidth. If you slow the whole process down it may work better. Amplifying the signal before appling it to a MCU pin ought to improve that method. Metal-gate (4000 series) CMOS gates biased into linear operation make decent rock-cheap amplifiers. The 4069UB is the best choice, but other gates will work if you happen to also need a logic gate in the circuit. Also, amplifiers made from 2-input gates can be placed in low-power shutdown easily by a logic signal applied to the other input. These amplifiers are maybe not so good for DC precision but work well with small high-impedance AC signals. ___________________________________________________________________ Get the Internet just the way you want it. Free software, free e-mail, and free Internet access for a month! Try Juno Web: dl.http://www.juno.com/dynoget/tagj.

Dennis, I'm using a Z8 due to overwhelming cost advanateg at 10,000 up quantity with hardware 20 KHz PWM and 1 comparator. I'd have been happy to use a PIC but the price was nowhere near the Z8+ I'm using. I have control of the FET which is supplying the current being measured so can zero calibrate whenever I wish (subject to user not noticing ;-)) This allows me to compensate for offsets and bias currents subject to certain limitations. I can calibrate for fullscale (or any other 1 chosen operating point) during setup. LMC660 is about 4 or 5 times dearer than LM324 as far as I can see. So are most other "real" opamps. The 324 does some very nasty things which have to be worked around, BUT its sooo cheap. IF you can use it its the way top go. If you can't you shout for joy and go and use something real :-). regards Russell McMahon {Original Message removed}

Mike Thanks for the thoughts. I'm aware of using CMOS inverter s an amp - in this app its not too good - I'm right on one rail of a 5 volt supply (0 - 0.2 volts to sense). CMOS works best with voltage in refd to centre point. 339 is a MUCH better comparator (of course) but gain is not a nice trick - could make a class D amp (I considered it) but there would be more problems there. I have used a Schmitt gate as an amplifier before today - you get a rail to rail pulse length modulated signal out which can be low pass filtered - lots of fun. Russell -----Original Messag e----- From: Mike Keitz +ADw-mkeitz+AEA-JUNO.COM+AD4- To: PICLIST+AEA-MITVMA.MIT.EDU +ADw-PICLIST+AEA-MITVMA.MIT.EDU+AD4- Date: Tuesday, July 13, 1999 12:52 PM Subject: Re: Sigma Delta Converter +AD4-On Tue, 13 Jul 1999 01:13:23 +120- Russell McMahon +ADw-apptech+AEA-CLEAR.NET.NZ+AD4- +AD4-writes: +AD4- +AD4APg-I have used both a standard microprocessor input pin as the sense +AD4APg-element (PIC) and an internal comparator (Z8) and an external op-amp +AD4APg-(LM324) as comparator. I am aware of the poor performance of the +AD4APg-LM324 as a comparator +AD4- +AD4-The 324's comparator brother the 339 (also offered in a dual unit, the +AD4-393) should be about the same cost. But then you need two chips unless +AD4-you can make the 339 work well as an amplifier. There was a chip with +AD4-324 type amplifiers and 339 comparators on one chip. It's likely been +AD4-discontinued. Probably the major problem with the 324 is its low +AD4-bandwidth. If you slow the whole process down it may work better. +AD4- +AD4-Amplifying the signal before appling it to a MCU pin ought to improve +AD4-that method. Metal-gate (4000 series) CMOS gates biased into linear +AD4-operation make decent rock-cheap amplifiers. The 4069UB is the best +AD4-choice, but other gates will work if you happen to also need a logic gate +AD4-in the circuit. Also, amplifiers made from 2-input gates can be placed +AD4-in low-power shutdown easily by a logic signal applied to the other +AD4-input. These amplifiers are maybe not so good for DC precision but work +AD4-well with small high-impedance AC signals. +AD4- +AD4- +AD4- +AD4AXwBfAF8AXwBfAF8AXwBfAF8AXwBfAF8AXwBfAF8AXwBfAF8AXwBfAF8AXwBfAF8AXwBfAF8AXwB fAF8AXwBfAF8AXwBfAF8AXwBfAF8AXwBfAF8AXwBfAF8AXwBfAF8AXwBfAF8AXwBfAF8AXwBfAF8AXwB fAF8AXwBfAF8AXwBfAF8AXw- +AD4-Get the Internet just the way you want it. +AD4-Free software, free e-mail, and free Internet access for a month+ACE- +AD4-Try Juno Web: dl.http://www.juno.com/dynoget/tagj. +AD4-

This may not apply directly to your situation. I tried rolling my own ÊV to I converter to use as a single slope A/D converter with a PIC a year ago (Similiar to one of their app notes). It worked ok but could not get good stability and repeatability better than 0.5% to 1%. I needed 0.1% so was forced to go with a commercial dual slope A/D. I would like to get a sigma delta unit which will give me a good 14 bits. Understand Tel-Com is working on one but not available yet. If you come up with something I would be very interested. @ Sent by: pic microcontroller discussion list <PICLISTKILLspamMITVMA.MIT.EDU> 07/13/99 08:25 AM ZE10 Please respond to pic microcontroller discussion list To: .....PICLISTKILLspam.....MITVMA.MIT.EDU cc: bcc: Subject: Re: Sigma Delta Converter At 01:13 13/07/99 +1200, you wrote: {Quote hidden}>I asked about this some weeks ago but a second query may be useful. > >I am trialling a Sigma Delta analog to digital converter. This SHOULD >allow low cost high accuracy A2D with modest conversion speed. > >Results to data only get so good (about 1/2 percent accuracy - >slightly worse than 8 bits)(plus a few worse results here and there) >and then seem to offer rapidly diminishing returns. Using more >conversion cycles doesn't seem top help much. I'm using 2048 counts >at present and would EXPECT much better results. > >I have used both a standard microprocessor input pin as the sense >element (PIC) and an internal Êcomparator (Z8) and an external op-amp >(LM324) as comparator. I am aware of the poor performance of the >LM324 as a comparator but am compelled to use it if at all possible >due to cost considerations. If it CANT be used then something else >will be used but at around $US0.16 in volume for a quad op-amp it is >very attractive. I also need to deal with a Ê0 to 0.2V input signal >so an amplifier is desirable. > >Anybody had better than about 1/2% accuracy results from a Sigma >Delta built discretely? >Any thoughts on the process? >This is a stunningly cheap and simple method for A2D if it would only >work :-) - once sorted it would/will prove a boon to other PIC users. > > >regards > > >Russell McMahon > > An LMC660 comes close to the bargin price, but the 324 should do. Some more information is required, are you using a split supply for the OPAMP, if not how close to 0V do you get. When you say 0.5% error at what voltage input level is this? Is the error at the same input level or over the entire range? The 324 has a 2mV offset that's up to 1% error in 200mV! Also are you using dual slope conversion or single slope? The number of counts not altering the value indicates that this is what the converter is reading, so I would dare say that the reference is also ok. But a Z8! Bad boy! Please tell me more Dennis

Your goal is low cost--skip dual slope for now...1/2% is way above that... What strikes me from my lab bench experience with sigma delta front ends and from some DSP code for a sigma delta interpolator is the way this method always has a part in the signal loop that drives as well as measures. Look at the way the input terminals of the comparator and/or amplifier affect the quality of the voltage on the interpolating capacitor and the quality of the voltage of the measured signal. Either loading, or noise, or input current mismatch could cause a lack of accuracy. The 1/2% is likely an analog design problem. To keep your costs in the $US0.16 in volume for a quad op-amp range it is hard to add resistors and such...but that may account for the 1/2%. See if you can think of a way to use the extra amps in the package as mirror image or chained pieces in the signal path,(even if they cause an equal delay). Use them to create equal input offset currents--they are a matched set. Use them to reduce load on the measured signal. Create noise immunity. Where is their linear input operating voltage relative to the power supply voltage available?--can their function as comparators be improved economically by powering them from a separate voltage than the uP? (for 2 kinds of improvement potential--random noise, and DC operating point). This economic balancing againxt performance is the toughest and the most rewarding part of the art--do not expect much help from any cookbook anywhere... just trial, err, and observe observe observe... John G. Austin TX > {Original Message removed}

If you want a good sigma delta converter take a look at http://www.linear-tech.com (LTC2400 24 bits sigma delta a/d). Marcos ----- Mensaje original ----- De: Jack Shidemantle <EraseMEJShidemantlespam_OUTTakeThisOuTYSI.COM> Para: <PICLISTspam_OUTMITVMA.MIT.EDU> Enviado: Martes, 13 de Julio de 1999 10:38 a.m. Asunto: Re: Sigma Delta Converter This may not apply directly to your situation. I tried rolling my own V to I converter to use as a single slope A/D converter with a PIC a year ago (Similiar to one of their app notes). It worked ok but could not get good stability and repeatability better than 0.5% to 1%. I needed 0.1% so was forced to go with a commercial dual slope A/D. I would like to get a sigma delta unit which will give me a good 14 bits. Understand Tel-Com is working on one but not available yet. If you come up with something I would be very interested. @ Sent by: pic microcontroller discussion list <@spam@PICLISTKILLspamMITVMA.MIT.EDU> 07/13/99 08:25 AM ZE10 Please respond to pic microcontroller discussion list To: KILLspamPICLISTKILLspamMITVMA.MIT.EDU cc: bcc: Subject: Re: Sigma Delta Converter At 01:13 13/07/99 +1200, you wrote: {Quote hidden}>I asked about this some weeks ago but a second query may be useful. > >I am trialling a Sigma Delta analog to digital converter. This SHOULD >allow low cost high accuracy A2D with modest conversion speed. > >Results to data only get so good (about 1/2 percent accuracy - >slightly worse than 8 bits)(plus a few worse results here and there) >and then seem to offer rapidly diminishing returns. Using more >conversion cycles doesn't seem top help much. I'm using 2048 counts >at present and would EXPECT much better results. > >I have used both a standard microprocessor input pin as the sense >element (PIC) and an internal comparator (Z8) and an external op-amp >(LM324) as comparator. I am aware of the poor performance of the >LM324 as a comparator but am compelled to use it if at all possible >due to cost considerations. If it CANT be used then something else >will be used but at around $US0.16 in volume for a quad op-amp it is >very attractive. I also need to deal with a 0 to 0.2V input signal >so an amplifier is desirable. > >Anybody had better than about 1/2% accuracy results from a Sigma >Delta built discretely? >Any thoughts on the process? >This is a stunningly cheap and simple method for A2D if it would only >work :-) - once sorted it would/will prove a boon to other PIC users. > > >regards > > >Russell McMahon > > An LMC660 comes close to the bargin price, but the 324 should do. Some more information is required, are you using a split supply for the OPAMP, if not how close to 0V do you get. When you say 0.5% error at what voltage input level is this? Is the error at the same input level or over the entire range? The 324 has a 2mV offset that's up to 1% error in 200mV! Also are you using dual slope conversion or single slope? The number of counts not altering the value indicates that this is what the converter is reading, so I would dare say that the reference is also ok. But a Z8! Bad boy! Please tell me more Dennis

Pretty one, already tested. DIP 8 small package... without nasty programming. Some small problems with grounding, but works nice. Wagner. Marcos Migliorini wrote: {Quote hidden}> > If you want a good sigma delta converter take a look at http://www.linear-tech.com > (LTC2400 24 bits sigma delta a/d). > > Marcos > > ----- Mensaje original ----- > De: Jack Shidemantle <RemoveMEJShidemantleTakeThisOuTYSI.COM> > Para: <spamBeGonePICLISTspamBeGoneMITVMA.MIT.EDU> > Enviado: Martes, 13 de Julio de 1999 10:38 a.m. > Asunto: Re: Sigma Delta Converter > > This may not apply directly to your situation. I tried rolling my own V to > I converter to use as a single slope A/D converter with a PIC a year ago > (Similiar to one of their app notes). It worked ok but could not get good > stability and repeatability better than 0.5% to 1%. I needed 0.1% so was > forced to go with a commercial dual slope A/D. I would like to get a sigma > delta unit which will give me a good 14 bits. Understand Tel-Com is working > on one but not available yet. If you come up with something I would be very > interested. > > @ > Sent by: pic microcontroller discussion list <TakeThisOuTPICLISTEraseMEspam_OUTMITVMA.MIT.EDU> > 07/13/99 08:25 AM ZE10 > Please respond to pic microcontroller discussion list > > To: RemoveMEPICLISTTakeThisOuTMITVMA.MIT.EDU > cc: > bcc: > Subject: Re: Sigma Delta Converter > > At 01:13 13/07/99 +1200, you wrote: > >I asked about this some weeks ago but a second query may be useful. > > > >I am trialling a Sigma Delta analog to digital converter. This SHOULD > >allow low cost high accuracy A2D with modest conversion speed. > > > >Results to data only get so good (about 1/2 percent accuracy - > >slightly worse than 8 bits)(plus a few worse results here and there) > >and then seem to offer rapidly diminishing returns. Using more > >conversion cycles doesn't seem top help much. I'm using 2048 counts > >at present and would EXPECT much better results. > > > >I have used both a standard microprocessor input pin as the sense > >element (PIC) and an internal comparator (Z8) and an external op-amp > >(LM324) as comparator. I am aware of the poor performance of the > >LM324 as a comparator but am compelled to use it if at all possible > >due to cost considerations. If it CANT be used then something else > >will be used but at around $US0.16 in volume for a quad op-amp it is > >very attractive. I also need to deal with a 0 to 0.2V input signal > >so an amplifier is desirable. > > > >Anybody had better than about 1/2% accuracy results from a Sigma > >Delta built discretely? > >Any thoughts on the process? > >This is a stunningly cheap and simple method for A2D if it would only > >work :-) - once sorted it would/will prove a boon to other PIC users. > > > > > >regards > > > > > >Russell McMahon > > > > > > An LMC660 comes close to the bargin price, but the 324 should do. Some more > information is required, are you using a split supply for the OPAMP, if not > how close to 0V do you get. When you say 0.5% error at what voltage input > level is this? Is the error at the same input level or over the entire > range? > The 324 has a 2mV offset that's up to 1% error in 200mV! Also are you using > dual slope conversion or single slope? > The number of counts not altering the value indicates that this is what the > converter is reading, so I would dare say that the reference is also ok. > But a Z8! Bad boy! > Please tell me more > > Dennis -- -------------------------------------------------------- Wagner Lipnharski - UST Research Inc. - Orlando, Florida Forum and microcontroller web site: http://www.ustr.net Microcontrollers Survey: http://www.ustr.net/tellme.htm

re: LTC2400 24-Bit ADC Ditto Marcos and Wagner's comments. I use two of them with a precision external reference hooked up to a PIC. Very easy to program. The results have been excellent. These are ideal for precision voltmeters which is what I'm doing (2-channel with differential inputs). Two status Bits provide Sign and Extended Range information which make it easy to implement an auto-ranging DVM. There are other features such as an internal oscillator, 110db 50/60Hz rejection, Auto Shutdown, 12.5% Overrange and Underrange, reference voltage range of 0.1V to 0.9 x VCC, and a 3-wire SPI interface. All this in a SO-8 package. - Tom At 09:53 PM 7/13/99 -0400, Wagner Lipnharski wrote: >Pretty one, already tested. >DIP 8 small package... without nasty programming. >Some small problems with grounding, but works nice. >Wagner. > >Marcos Migliorini wrote: >> >> If you want a good sigma delta converter take a look at http://www.linear-tech.com >> (LTC2400 24 bits sigma delta a/d). >> >> Marcos

That's on my todo list, thanks guys. Wagner Lipnharski To: PICLISTEraseME.....MITVMA.MIT.EDU <wagnerl@EARTH cc: LINK.NET> Subject: Re: Sigma Delta Converter Sent by: pic microcontrolle r discussion list <PICLIST@MITVM A.MIT.EDU> 07/13/99 09:53 PM Please respond to pic microcontrolle r discussion list Pretty one, already tested. DIP 8 small package... without nasty programming. Some small problems with grounding, but works nice. Wagner. Marcos Migliorini wrote: > > If you want a good sigma delta converter take a look at http://www.linear-tech.com {Quote hidden}> (LTC2400 24 bits sigma delta a/d). > > Marcos > > ----- Mensaje original ----- > De: Jack Shidemantle <EraseMEJShidemantleYSI.COM> > Para: <RemoveMEPICLISTEraseMEEraseMEMITVMA.MIT.EDU> > Enviado: Martes, 13 de Julio de 1999 10:38 a.m. > Asunto: Re: Sigma Delta Converter > > This may not apply directly to your situation. I tried rolling my own V to > I converter to use as a single slope A/D converter with a PIC a year ago > (Similiar to one of their app notes). It worked ok but could not get good > stability and repeatability better than 0.5% to 1%. I needed 0.1% so was > forced to go with a commercial dual slope A/D. I would like to get a sigma > delta unit which will give me a good 14 bits. Understand Tel-Com is working > on one but not available yet. If you come up with something I would be very {Quote hidden}> interested. > > @ > Sent by: pic microcontroller discussion list <RemoveMEPICLISTspam_OUTKILLspamMITVMA.MIT.EDU> > 07/13/99 08:25 AM ZE10 > Please respond to pic microcontroller discussion list > > To: RemoveMEPICLISTTakeThisOuTspamMITVMA.MIT.EDU > cc: > bcc: > Subject: Re: Sigma Delta Converter > > At 01:13 13/07/99 +1200, you wrote: > >I asked about this some weeks ago but a second query may be useful. > > > >I am trialling a Sigma Delta analog to digital converter. This SHOULD > >allow low cost high accuracy A2D with modest conversion speed. > > > >Results to data only get so good (about 1/2 percent accuracy - > >slightly worse than 8 bits)(plus a few worse results here and there) > >and then seem to offer rapidly diminishing returns. Using more > >conversion cycles doesn't seem top help much. I'm using 2048 counts > >at present and would EXPECT much better results. > > > >I have used both a standard microprocessor input pin as the sense > >element (PIC) and an internal comparator (Z8) and an external op-amp > >(LM324) as comparator. I am aware of the poor performance of the > >LM324 as a comparator but am compelled to use it if at all possible > >due to cost considerations. If it CANT be used then something else > >will be used but at around $US0.16 in volume for a quad op-amp it is > >very attractive. I also need to deal with a 0 to 0.2V input signal > >so an amplifier is desirable. > > > >Anybody had better than about 1/2% accuracy results from a Sigma > >Delta built discretely? > >Any thoughts on the process? > >This is a stunningly cheap and simple method for A2D if it would only > >work :-) - once sorted it would/will prove a boon to other PIC users. > > > > > >regards > > > > > >Russell McMahon > > > > > > An LMC660 comes close to the bargin price, but the 324 should do. Some more > information is required, are you using a split supply for the OPAMP, if not > how close to 0V do you get. When you say 0.5% error at what voltage input > level is this? Is the error at the same input level or over the entire > range? > The 324 has a 2mV offset that's up to 1% error in 200mV! Also are you using > dual slope conversion or single slope? > The number of counts not altering the value indicates that this is what the > converter is reading, so I would dare say that the reference is also ok. > But a Z8! Bad boy! > Please tell me more > > Dennis -- -------------------------------------------------------- Wagner Lipnharski - UST Research Inc. - Orlando, Florida Forum and microcontroller web site: http://www.ustr.net Microcontrollers Survey: http://www.ustr.net/tellme.htm

Linear Technologies said about $14. LM324 + Zener + 9 resistors + 3 capacitors would take you close to $4 to $6, considering that the LTC2400 is 24 bits ADC, your cost is totally proporcional to dollars per bits of resolution. Remember that LTC2400 is quite new, less than 6 months old, some distributors don't have it available yet since it is not consumed in billions, but because its easy operation I believe it would has a nice place into the marketshare of ADC's. Wagner. Russell McMahon wrote: {Quote hidden}> > Yes, LTC2400 is a very nice product indeed but not suitable for my > application due to the need for ultimate low cost. The processor is > set due to other considerations and I must live withing its > constraints while costing approx nothing at all. What I have achieved > (see separate post) will do what I want (9 or 10 bits accuracy - 8 > probably OK) using an LM324, 9 resistors (including an amplifier and > "reference", zener and 3 capacitors. Assembly costs and board space > costs all add up but still far below LTC2400 cost I imagine - I > couldn't find a price at Digikey, Pioneer or Netbuy. How much are you > buying them for? > > RM

At 11:46 15/07/99 +1200, you wrote: >I shouldn't comment on prices but for SMD >I'd guess at $US1.18 for LM324, 9 resistors, 3 caps & a zener - (or $NZ2.40 >max) overall cost very dependent on pcb size, shape etc. > >Displays, PCB, connectors and metalwork are the major cost items. Standard >electronic component costs are pretty well negligable by comparison. > > >Of course if you buy everything from RS...... > >By the way - what temperature are you running at. We have had to change to >LM224s to meet high temperature requirements (and we pay more as a >consequance). > Rubbish! The cost of any component is always to be considered. It seems that you may be stuck in a less than 1000 production run world. It is the total cost of the electronic components that has to be cinsidered including the cost of one off components, size and other formfactot problems that such may induce incluing manufacture and test cost. If an IC costs $14 and the sperate components cost $2.40 (This indicates that you have not had real exposure to large scale production), then one can expect that the other costs including kitting storage and procurment will add arounf %10, or $.24 for one case and $1.4 for the otehr case. This is then altered by teh cost to manufacture $.02 per component, say $15.42 and $2.9, then add the test cost (Around 40% of the manufatured cost (Complete) and one can see that the componsnt cost is signifficant. To this assume that the otehr compents eg metal work etc increase by %1, so if the case cost $20 and the PCB $10, and packaging $3, the extra cost is now $6.2 vs $15.42 add test cost. I think that the difference is still large, around 1/2 the cost. Take this to the extrame with good component buying, and you can save the company over $1M per year Dennis

I seem to have been bitten! - Sorry - I don't think I made my point very clear. I was trying to point out the $4+ (US ?)estimate of the previous message was somewhat inflated from my viewpoint. Also I was trying to indicate that _standard_ components (commonly used resistors, capacitors LM324's etc. which are used in a number of different products) are so cheap that it is generally a cheaper option to use them rather than a more expensive single component bought in especially for a single product. This I think is your point also. I am fully aware of the costs of procurement, inventory, kitting & production - along with time to market and second sourcing questions. Possibly I am overstating the case on the basis of the simple items I spend most of my time working on but provided the PCB area requirement doesn't increase to the point where we get less PCBs / panel the cost of a handful of _standard_ parts is virtually nil. Richard > {Original Message removed}

Russell McMahon wrote: {Quote hidden}> > I asked about this some weeks ago but a second query may be useful. > > I am trialling a Sigma Delta analog to digital converter. This SHOULD > allow low cost high accuracy A2D with modest conversion speed. > > Results to data only get so good (about 1/2 percent accuracy - > slightly worse than 8 bits)(plus a few worse results here and there) > and then seem to offer rapidly diminishing returns. Using more > conversion cycles doesn't seem top help much. I'm using 2048 counts > at present and would EXPECT much better results. > > I have used both a standard microprocessor input pin as the sense > element (PIC) and an internal comparator (Z8) and an external op-amp > (LM324) as comparator. I am aware of the poor performance of the > LM324 as a comparator but am compelled to use it if at all possible > due to cost considerations. If it CANT be used then something else > will be used but at around $US0.16 in volume for a quad op-amp it is > very attractive. I also need to deal with a 0 to 0.2V input signal > so an amplifier is desirable. > > Anybody had better than about 1/2% accuracy results from a Sigma > Delta built discretely? > Any thoughts on the process? > This is a stunningly cheap and simple method for A2D if it would only > work :-) - once sorted it would/will prove a boon to other PIC users. A couple of years ago I tried this circuit: Vin o---/\/\/\---+---/\/\/\----> PIC | === | --- /// It uses the same charge balancing trick that has been discussed here before. The only twist is that only one I/O pin is used instead of two. The basic algorithm goes something like: 1) make the I/O an input 2) If the input high then make the I/O pin an output and drive it low else make the I/O pin an output and drive it high. 3) delay 4) goto 1 Theoretically the duty cycle - or the ratio the time the I/O spends high to the sum of the times it's high and low - is proportional to the analog input Vin. It works - sort of. I found that it was good for about 5 bits. But it certainly is low cost! There are few problems with this circuit, with which I'm sure Russel will agree. 1) It depends on the threshold of a digital input. The hysteresis could be fairly large and hence would necessitate a relatively large ripple on the capacitor. 2) The voltage on the cap will drift towards the input voltage while the I/O line is programmed as an input. This can be mitigated by using two I/O lines however (one programmed as an input and monitoring the capacitor voltage the other as an output). 3) The input impedance is fairly low, especially when compared to an opamp. There's another low-cost solution that substitutes these three problems for another: +V +V | | \ | / | \ R1 | / | | |V +----------| 2N3906 | |\ Ri |/ | Vin --/\/\--| 2N3904 +-----------> to Pic |V | R3 | +-/\/\--+ | /// | +---+----------/\/\------< from pic | | R4 \ | / === C1 R2\ | / | | | --- --- /// /// How it works: The 3904 buffers the input and drives the R2/C1 network. The input impedance is fairly high (~1meg if R2 and R4 are about 10k). If the voltage at the 3904's emitter is too large or (equivalently) Vin is too low, then the 3904 will be turned off. This turns off the 3906 as well. The main purpose of the 3906 is to remove most of the dependence on the pic's input threshold voltage. In other words, the 3906 is either going to be saturated or off and its collector will be either high or low. I realize that it's possible to balance this circuit so that the 3906 would be biased in the linear region, however the PIC is in the feedback loop and the control signal it provides has a much lower frequency response than the 3906. The R2/R4/C1 filter and divide the pwm signal from the pic. The problem is that now the circuit is somewhat temperature dependent. The Vbe drop is about -2mv/degree C. This means that for a given input voltage, the voltage at the 3904's emitter will have to be increased as the temperature rises. I haven't tried this circuit yet, btw. I'd be interested to hear comments and improvements (especially a way to remove the temperature dependence). Scott

FWIW, I re-tried the 2-I/O pin and also 1-I/O pin a2d converter again and got really good results. (The last time I tried a couple of years ago I failed miserably.) The circuit is this: +-----------> PIC (not used in 1-pin version) | Vin o---/\/\/\---+-+-/\/\/\----< PIC | === | --- /// The two resistors were 10k and the capacitor was a 0.56uF mylar film. I didn't experiment much with these values, I basically took some stuff out of one my junk piles. I was able to fairly easily get 12-bits of dynamic range with the 2-I/O pin version and about 8 with the 1-I/O pin version. I haven't attempted to verify linearity, temperature dependence, long term stability, etc. My conclusions on the dynamic range are subjectively based on how much flickering I can observe in the least significant bit. YMMV If you want to see the code go here: http://www.interstice.com/~sdattalo/technical/software/pic/a2d.asm but beware, the code looks almost as bad as my ratsnest prototype :). The trick I employed this time is a pre-charge algorithm. In other words, before beginning the actual a2d conversion I make sure that the voltage on the capacitor is right at the switching threshold of the pic I/O pin. Oh, one more thing; I used only a 4Mhz crystal. and one more thing: The 2 pin version could easily be scaled up for 4 analog inputs. The 1-pin version doesn't scale as nicely, but since it only has 8-bits of dynamic range it is certainly a candidate for vertical counters... Scott PS, the asin and arctan routines are now on my pic page: http://www.interstice.com/~sdattalo/technical/software/software.html