Post by thread:[EE] Critique my project schematics

part 1 889 bytes content-type:text/plain; charset=US-ASCII (decoded 7bit) Hi guys, this is my first pic project and I'm hoping that some of you would be kind enough to review my schematics for mistakes and blunders and offer suggestions as you see fit. I don't have a background in EE so this is designed with knowledge obtained from the list, internet and some old electronics books. http://www.geocities.com/mr_goat2000/schem.gif The intent of this device is to control an electronic transmission. There are inputs from a pressure sensor, thermistor and 6 switches as well as the LM1815 chip which is meant for a variable reluctance sensor. Outputs are to solenoids and the LM7221 which controls a 4 digit led display. I have USART for debugging and possibly a bootloader in the distant future. Do I have adequate decoupling for the automotive environment? Any other comments, suggestions are welcome. Cheers, Milosz K. part 2 47590 bytes content-type:image/gif; name="Schem.gif" (decode) part 3 35 bytes content-type:text/plain; charset="us-ascii" (decoded 7bit)

Hi, The power supply is going to cause great problems in an automotive design. First off, your design should handle "load dump" (google it). Your regulators are only good for 29volt input and in an automotive system there are situations that can greatly exceed that limit. You will need some type of suppressor circuit. Second, it would be nice to have some sort of common mode choke in the supply to remove noise. A fuse would be nice as well. The 12V supply regulator won't have enough headroom when the ignition is off, or when a transient occurs (like lights being turned on). You may want to search around the net and look for good examples of automotive power supply designs. I didn't check the rest of the schematic. Ken spam_OUTklumiaTakeThisOuTadelphia.net {Original Message removed}

Automotive inputs need a hash choke and a fuse or polyfuse followed by a HEAVY 18V varistor or similar device before the regulator. The circuit will be unreliable without this input cleanup. In fact, I'd use a larger input electrolytic than 10uF, I'd use at least 100uF. Be sure to put it into a tight METAL box, to keep out RF noise (sparkplugs generate RF noise). You have an ambitious effort for just 4Mhz. Personally, I'd go to 8Mhz and use a ceramic resonator, NOT a crystal (easily damaged by shock). Be aware that in a hot environment, the lifetime of wet electrolytic capacitors is VERY reduced. Use tantalum slugs wherever you can. If you have external sensors, you must install an RF choke and capacitor to keep noise from entering the system that way for EVERY WIRE coming in. How are you planning to debug the unit? You have no provision for ICSP and/or ICD2. The MCLR line is especially unuseable by the ICD2. That's all I see for now.... --Bob Milosz Kardasinski wrote: {Quote hidden}>Hi guys, this is my first pic project and I'm hoping that some of you >would be kind enough to review my schematics for mistakes and blunders >and offer suggestions as you see fit. I don't have a background in EE >so this is designed with knowledge obtained from the list, internet >and some old electronics books. > >http://www.geocities.com/mr_goat2000/schem.gif > >The intent of this device is to control an electronic transmission. >There are inputs from a pressure sensor, thermistor and 6 switches as >well as the LM1815 chip which is meant for a variable reluctance >sensor. Outputs are to solenoids and the LM7221 which controls a 4 >digit led display. > >I have USART for debugging and possibly a bootloader in the distant future. > >Do I have adequate decoupling for the automotive environment? Any >other comments, suggestions are welcome. > >Cheers, >Milosz K. > > > > ------------------------------------------------------------------------ > -- Note: To protect our network, attachments must be sent to .....attachKILLspam@spam@engineer.cotse.net . 1-866-263-5745 USA/Canada http://beam.to/azengineer

>>> Be aware that in a hot environment, the lifetime of wet >>> electrolytic capacitors is VERY reduced. Use tantalum slugs >>> wherever you can. >> Better still, use SOLID Aluminium electrolytics (not the normal wet >> electrolytic type) . All of the advantages of tantalum and none of >> the disadvantages. > I'm not familiar with those, Russell. Got links? As ever, Google is our friend. here's a few. Vishay multi reference page is excellent. Philips do a good range of Solid Aluminium caps that are well priced and specd relative to Tantalum. Excellent Vishay page with many technical papers on Solid Aluminium/aluminum capacitors (also tantalum) http://www.vishay.com/capacitors/volt-10-below/related Vishay (from above page) EXCELLENT 2 page Solid Aluminium reliability paper Comparison with std aluminium electrolytics. eg lifetimes at 105C/85C/65C are 2000/2000, 20,000/8,000, 200,000/32,000 hours http://www.vishay.com/docs/90017/reliabil.pdf Vishay: Precautions using SA because they are so good ! :-) http://www.vishay.com/docs/90018/precaut.pdf Good page on basic capacitor construction and principles. Explains how a solid tantalum is made and why it is so good. (Doesn't explain why it is also so bad). http://www.ami.ac.uk/courses/topics/0136_ec/ Kemet on why they are doing solid Al (1999) Note how they introduce themselves :-) http://www.kemet.com/kemet/web/homepage/kechome.nsf/vabypagename/aluminum Kemet of course do both. They note that tantalum are superior volumetrically. I note that Solid Al is normally very acceptable unless you are utterly pushing the edge of packing. http://www.arrowasia.com/en/arrowtimes/focus_on_01.htm

Thanks for links. I musta been asleep while these were being created... On the other side of the world, maybe...? --Bob Russell McMahon wrote: {Quote hidden}>>>> Be aware that in a hot environment, the lifetime of wet >>>> electrolytic capacitors is VERY reduced. Use tantalum slugs >>>> wherever you can. >>> > >>> Better still, use SOLID Aluminium electrolytics (not the normal wet >>> electrolytic type) . All of the advantages of tantalum and none of >>> the disadvantages. >> > >> I'm not familiar with those, Russell. Got links? > > > As ever, Google is our friend. here's a few. > Vishay multi reference page is excellent. > > Philips do a good range of Solid Aluminium caps that are well priced > and specd relative to Tantalum. > > Excellent Vishay page with many technical papers on Solid > Aluminium/aluminum capacitors (also tantalum) > > http://www.vishay.com/capacitors/volt-10-below/related > > Vishay (from above page) EXCELLENT 2 page Solid Aluminium reliability > paper > Comparison with std aluminium electrolytics. > eg lifetimes at 105C/85C/65C are 2000/2000, 20,000/8,000, > 200,000/32,000 hours > > http://www.vishay.com/docs/90017/reliabil.pdf > > Vishay: Precautions using SA because they are so good ! :-) > > http://www.vishay.com/docs/90018/precaut.pdf > > > Good page on basic capacitor construction and principles. > Explains how a solid tantalum is made and why it is so good. > (Doesn't explain why it is also so bad). > > http://www.ami.ac.uk/courses/topics/0136_ec/ > > > Kemet on why they are doing solid Al (1999) > Note how they introduce themselves :-) > > > http://www.kemet.com/kemet/web/homepage/kechome.nsf/vabypagename/aluminum > > Kemet of course do both. They note that tantalum are superior > volumetrically. I note that Solid Al is normally very acceptable > unless you are utterly pushing the edge of packing. > > http://www.arrowasia.com/en/arrowtimes/focus_on_01.htm > > -- Note: To protect our network, attachments must be sent to .....attachKILLspam.....engineer.cotse.net . 1-866-263-5745 USA/Canada http://beam.to/azengineer

Hi, Is the DSub male or female? Is it going to be connected to a PC or look lika a PC (DTE or DCE)? If it is going to be directly connected to a PC it should be a female and pin 2 should be TX (output to RX in PC) and pin 3 should be RX (input from TX in PC). Pin 1 is an input in a DTE (PC) so it looks like you are going to use it as a (DCE). Also put a decoupling cap on the MAX232 (VCC to GND). They can be quite noisy. Besides what others have said, make a big fat groundplane on the component side and route as much as possible of signals and power on the solder side. When you have to make a jumper on the component (groundplane-) side, make it as short as possible (to interrupt the groundplane as little as possible). This applies for two side board with through hole components. If surface mounted components are used, they should be placed at the same side as the traces (opposite to the groundplane). If it is an automotive application, place the board in a metal case, connect the case to the groundplane (good connection) and make sure the whole box is grounded to the chassis. Preferably bolted directly to it, if not use a short braid (from a screened cable perhaps). Regards/Ruben {Quote hidden}> Hi guys, this is my first pic project and I'm hoping that some of you > would be kind enough to review my schematics for mistakes and blunders > and offer suggestions as you see fit. I don't have a background in EE > so this is designed with knowledge obtained from the list, internet > and some old electronics books. > > http://www.geocities.com/mr_goat2000/schem.gif > > The intent of this device is to control an electronic transmission. > There are inputs from a pressure sensor, thermistor and 6 switches as > well as the LM1815 chip which is meant for a variable reluctance > sensor. Outputs are to solenoids and the LM7221 which controls a 4 > digit led display. > > I have USART for debugging and possibly a bootloader in the distant future. > > Do I have adequate decoupling for the automotive environment? Any > other comments, suggestions are welcome. > > Cheers, > Milosz K. > ============================== Ruben Jönsson AB Liros Electronic Box 9124, 200 39 Malmö, Sweden TEL INT +46 40142078 FAX INT +46 40947388 EraseMErubenspam_OUTTakeThisOuTpp.sbbs.se ==============================

Hi Ruben >If it is an automotive application, place the board in a metal case, connect >the case to the groundplane (good connection) and make sure the whole box is >grounded to the chassis. Preferably bolted directly to it, if not use a short >braid (from a screened cable perhaps). > >Regards/Ruben This is something I'm never sure about. I've pulled others' products apart and find that the metal case is insulated from the ground plane. If connecting to the chassis is an option to the end user, should it be relied upon? Or should it be the only path? Usually (from what I've found) ground is only through the designated ground pins on the connector, unless the pcb is part of something like an embedded injector control, where the body is absolutely grounded. Comments solicited.. Regards Roland Jollivet

On Mon, 31 Jan 2005 20:18:45 -0700, Bob Axtell <engineerspam_OUTcotse.net> wrote: > Automotive inputs need a hash choke and a fuse or polyfuse followed by a > HEAVY 18V varistor or similar device before the regulator. The circuit > will be > unreliable without this input cleanup. In fact, I'd use a larger input > electrolytic than > 10uF, I'd use at least 100uF. Be sure to put it into a tight METAL box, > to keep out > RF noise (sparkplugs generate RF noise). Would you put the 100uF cap across the input of each regulator or just across the input power. Ok, I think I selected a good MOV for the application, would part # ERZ-V20D220 from Digikey be a good choice? I could use some help with the inductor selection, I'm self taught and inductors are still over my head. > You have an ambitious effort for just 4Mhz. Personally, I'd go to 8Mhz > and use a > ceramic resonator, NOT a crystal (easily damaged by shock). I am planning on using a resonator, forgot to update the schematic. Apologize for that. Do you really think that 4mhz is not fast enough? > > If you have external sensors, you must install an RF choke and capacitor > to keep > noise from entering the system that way for EVERY WIRE coming in. I know very little about RF chokes or selecting the proper values. Do have a link to information on the subject or could you recommend some parts? > How are you planning to debug the unit? You have no provision for ICSP > and/or > ICD2. The MCLR line is especially unuseable by the ICD2. I was planning on debugging through a serial communitcation to a PC. For time being I'm just going to pull the chip and use a programmer to load firmware. At some point I'll try messing around with a bootloader. On Tue, 01 Feb 2005 08:30:38 +0100, Ruben Jönsson <@spam@rubenjKILLspampp.sbbs.se> wrote: > Hi, > > Is the DSub male or female? Is it going to be connected to a PC or look lika a > PC (DTE or DCE)? The DSub is female. Yes, I do plan on hooking it to a PC > > If it is going to be directly connected to a PC it should be a female and pin 2 > should be TX (output to RX in PC) and pin 3 should be RX (input from TX in PC). > > Pin 1 is an input in a DTE (PC) so it looks like you are going to use it as a > (DCE). > > Also put a decoupling cap on the MAX232 (VCC to GND). They can be quite noisy. Thanks for the suggestion. Does the standard rule for decoupling apply here, 1/3 of the harmonic? Ei. 0.1uF for 4Mhz?

Milosz Kardasinski wrote: >On Mon, 31 Jan 2005 20:18:45 -0700, Bob Axtell <KILLspamengineerKILLspamcotse.net> wrote: > > >>Automotive inputs need a hash choke and a fuse or polyfuse followed by a >>HEAVY 18V varistor or similar device before the regulator. The circuit >>will be >>unreliable without this input cleanup. In fact, I'd use a larger input >>electrolytic than >>10uF, I'd use at least 100uF. Be sure to put it into a tight METAL box, >>to keep out >>RF noise (sparkplugs generate RF noise). >> >> > >Would you put the 100uF cap across the input of each regulator or just >across the input power. > After the choke and MOV, right at the regulator input. JW Miller has sells hash chokes. Pick one that can handle the expected max current of the unit times 2. Hash chokes won't work if they are saturated. > Ok, I think I selected a good MOV for the >application, would part # ERZ-V20D220 from Digikey be a good choice? I >could use some help with the inductor selection, I'm self taught and >inductors are still over my head. > > Is that a good number? I thought all digikey parts ended in ND or TR... {Quote hidden}> > >>You have an ambitious effort for just 4Mhz. Personally, I'd go to 8Mhz >>and use a >>ceramic resonator, NOT a crystal (easily damaged by shock). >> >> > >I am planning on using a resonator, forgot to update the schematic. >Apologize for that. Do you really think that 4mhz is not fast enough? > > > Well, it might be. perhaps I was too abrupt there. {Quote hidden}>>If you have external sensors, you must install an RF choke and capacitor >>to keep >>noise from entering the system that way for EVERY WIRE coming in. >> >> > >I know very little about RF chokes or selecting the proper values. Do >have a link to information on the subject or could you recommend some >parts? > > What you are doing is to rid the DC or low AC signal of its RF. That RF was picked up when the wire acts like an antenna. Its a bit of black magic. I simply attach an LC until the noise goes away. My RF noise generator is a 110VAC relay that is wired so that it acts as a buzzer. When you wrap a wire to that relay around something, the RF gets imparted on it. {Quote hidden}> > > >>How are you planning to debug the unit? You have no provision for ICSP >>and/or >>ICD2. The MCLR line is especially unuseable by the ICD2. >> >> > >I was planning on debugging through a serial communitcation to a PC. >For time being I'm just going to pull the chip and use a programmer to >load firmware. At some point I'll try messing around with a >bootloader. > > Well, that's the hard way... but one has to start someplace. {Quote hidden}> >On Tue, 01 Feb 2005 08:30:38 +0100, Ruben Jönsson <RemoveMErubenjTakeThisOuTpp.sbbs.se> wrote: > > >>Hi, >> >>Is the DSub male or female? Is it going to be connected to a PC or look lika a >>PC (DTE or DCE)? >> >> > >The DSub is female. Yes, I do plan on hooking it to a PC > > > >>If it is going to be directly connected to a PC it should be a female and pin 2 >>should be TX (output to RX in PC) and pin 3 should be RX (input from TX in PC). >> >>Pin 1 is an input in a DTE (PC) so it looks like you are going to use it as a >>(DCE). >> >>Also put a decoupling cap on the MAX232 (VCC to GND). They can be quite noisy. >> >> > >Thanks for the suggestion. Does the standard rule for decoupling apply >here, 1/3 of the harmonic? Ei. 0.1uF for 4Mhz? > > > -- Note: To protect our network, attachments must be sent to spamBeGoneattachspamBeGoneengineer.cotse.net . 1-866-263-5745 USA/Canada http://beam.to/azengineer

Roland wrote: >> If it is an automotive application, place the board in a metal case, >> connect the case to the groundplane (good connection) and make sure the >> whole box is grounded to the chassis. Preferably bolted directly to it, >> if not use a short braid (from a screened cable perhaps). > This is something I'm never sure about. I've pulled others' products > apart and find that the metal case is insulated from the ground plane. > If connecting to the chassis is an option to the end user, should it be > relied upon? Or should it be the only path? > > Usually (from what I've found) ground is only through the designated > ground pins on the connector, unless the pcb is part of something like > an embedded injector control, where the body is absolutely grounded. > Comments solicited.. I agree with you. I don't think that the chassis in a car is a good ground connection, therefore the circuit should be connected through a good wire to the battery or a good ground point (electrically) close to it (not the chassis) whenever possible. Since a metal case usually will be mounted to the chassis, it follows (for me) that the circuit ground should not be connected to the case. Otherwise, the circuit ground would be connected to the car chassis, which seems to be rather noisy and not a good ground. Gerhard