DFH - Defacto FreeEMS Hardware in KICAD

[jharvey]

About LM1815 P45 P46 resistor, I can't seem to recall the value. I've put it in, but haven't specified a value.

[Fred]

Around 5k on the Toyota setup :-)

Also, what I previously said about decreasing the 18k one was probably inappropriate for the LM1815, so leave those other two input parts as is :-)

[Brian]

- In the VR cct, LED D103 and resistor R157 may not work as the LED is powered from 5V via R153 which is 5K6. Also SW8 needs 3 positions, centre off, as there are 3 modes, mode 1 is with pin 5 open. I have not seen R159 in the datasheet for the LM1815, what is it's purpose, to protect the 5V supply?

- What is the reason behind the seperation of digital and analogue grounds, are they somehow seperate or is it for PCB layout design considerations?

Some general thoughts on protection: When it comes to protection ccts, I have repaired a few ccts over the years in which the protection ccts caused the fault and made the device to stop functioning. So if we really want protection, at least make sure it is fail-safe, that is, if a component in the protection cct fails, the rest of the main cct can still manage to do its job. As far as possible that is. I don't like to use this comparison, but it's a good one, how many novice users managed to damage the MS1 processor on a V2.2 board? Has anyone seen a post that claimed an ignition output port was damaged by driving an ignitor directly? Lots of VB921's burned up, but not processors? Lots of people including me removed the zener diode protection from the O2 sensor input cct as the diode was leaking, and prevented the cpu from measuring the input correctly. Protection from EMI and RF is fine and good, but protection from poor installation/bad wiring practices is simply adding more stuff to the PCB that isn't necessary IMO. We have whole mega-corporations building igbt's for ignition and IRF540 or VNP5N07 etc etc etc for injection to make it easy for us, and publishing datasheets to show us how to use them as designed.

Fred > We need injector drive, but we want low Z drive. We need separate grounds and powers up front, we want an RTC on board. We need the 6 core sensors, we want the other 8 adc inputs. etc.

Did you mean to say "but we don't want low Z drive"?

jharvey > Changed AN's per Fred's request, but don't plan on them there,

OK I'm having a moment, what is RTC and AN? Real-time clock? Saves having power connected all the time, I guess I have missed some posts somewhere? AN I have NI.

[Fred]

Also SW8 needs 3 positions, centre off, as there are 3 modes, mode 1 is with pin 5 open.

I think he's not literally putting a switch there, just 3 pads so if you do nothing you get open, solder blob bridging to 5v gets you high and to ground gets you low.

I have not seen R159 in the datasheet for the LM1815, what is it's purpose, to protect the 5V supply?

Yeah, you don't need/want that there do you? I think it's designed to see rail/ground/nothing as it stands with direct connections.

- What is the reason behind the seperation of digital and analogue grounds, are they somehow seperate or is it for PCB layout design considerations?

To not have massive noise problems like MS has. The ground return to the block/battery for the switching devices (all fets and igbts etc) should DEFINITELY run separately from the signal/cpu ground. Not doing that causes noise to be present (as opposed to allowing noise).

You just made me notice something else too : The two switched 12v inputs are joined. There should be 5 connections in the top left corner, 3 power, 2 ground.

I don't like to use this comparison, but it's a good one, how many novice users managed to damage the MS1 processor on a V2.2 board? Has anyone seen a post that claimed an ignition output port was damaged by driving an ignitor directly? Lots of VB921's burned up, but not processors?

Probably none, reason being each pin could do about 20mA. On the contrary, with MS2 and FreeEMS each pin can only do 4mA with a bank of 8 doing 25 max. I've read about dead MS2 chips and ports from over voltage and over current conditions. MS1 was much more durable in that respect. Also, with MS2 the igbts werent being pushed into saturation fast enough because of the lack of current, so that helped cause the deaths. They need to be buffered with something that can push some current while protecting the CPU pin with a 1.6k resistor.

Lots of people including me removed the zener diode protection from the O2 sensor input cct as the diode was leaking, and prevented the cpu from measuring the input correctly.

Yes, that is another design flaw. Analog pins should not use zeners. I found this out the hard way myself, but it's actually well known so Bruce really had no excuse. The analog pins should be covered by normal diodes and the digital pins should be covered by either normal or zener setups. Heaps of guys have killed their table switching and two step ports on ms2 with 12v inputs to them...

Did you mean to say "but we don't want low Z drive"?

The thing here is the difference between what "I" want/need and what Jared wants/needs and what the "average user" wants/needs. I personally DO NOT WANT noisey low Z inside unless the design is VERY well tested (something we probably can't do) however some others will want it, so if possible it would be nice to have it. For support I think making sure it is outside and buffered from the CPU is a very good thing as it will reduce the number of noisy installations and problems.

What is RTC and AN? Real-time clock? Saves having power connected all the time, I guess I have missed some posts somewhere? AN I have NI.

RTC = Real Time Clock, but you still need power all the time. AN is the freescale CPU side name for the ADC ports. AN00 AN01 etc.

RTC = all datalogs are time stamped. This is nice if you are continuously datalogging to an hdd for example. It is also nice for shut down time stamping with the eeprom or sleep state ram for IAT temps etc on start up. Getting that stuff right could tell us a lot about heat soak and help get idles and startups a bit better than possible without a time stamp.

That's clearly a future thing though and even though it's one of my pet ideas, I'm not bloody writing any code for that at the moment either! :-)

Fred.

[Fred]

BTW Jared, if we aren't gong with low Z driving (which I'm happy to not include) we need a buffer IC of some sort on the injector channels too. Anything is fine, but 20mA drive would be nice so the current limiting resistors outside the buffer can be nice and low.

Fred.

[Delta]

I think you should include the low Z setup - quite frankly its not much extra space, and if you only use high Z injectors then it doesn't go into the switching zone anyway - so it won't make any noise at all for you, and others who need it can test to see if it works or causes crazy problems.

[davebmw]

The thing here is the difference between what "I" want/need and what Jared wants/needs and what the "average user" wants/needs. I personally DO NOT WANT noisey low Z inside unless the design is VERY well tested (something we probably can't do) however some others will want it, so if possible it would be nice to have it. For support I think making sure it is outside and buffered from the CPU is a very good thing as it will reduce the number of noisy installations and problems.
Fred.

Erm? yes we can, I have access to 3GHz Spectrum Analyzers and S parameter test sets in an RF Neutral chamber for a full RF/Harmonics analysis. I can test it to BABT approval too if required.

The low Z drive need not be a total pig as long as the drive is kept isolated from the CPU by optocouplers the grounding and supply is taken as far back to the point of origin as possible. the use of appropriate decoupling and snubber networks will quench all the nasty stuff.

[Fred]

Erm? yes we can, I have access to 3GHz Spectrum Analyzers and S parameter test sets in an RF Neutral chamber for a full RF/Harmonics analysis. I can test it to BABT approval too if required.

AWESOME!!!

The low Z drive need not be a total pig as long as the drive is kept isolated from the CPU by optocouplers the grounding and supply is taken as far back to the point of origin as possible. the use of appropriate decoupling and snubber networks will quench all the nasty stuff.

Damn good thinking batman! I was wondering about isolated grounds and power feeds and switching voltage differences it could cause. This should definitely occur, I hadn't thought about it, but I like it very much indeed.

[davebmw]

The best design for PWM injector control has to be that man Deltas comparator controlled work of genius.
With some real world testing it would be absolutely spot on i think

[Brian]

Thanks for the answers Fred, all much clearer now, especially the 4mA pin limit on the cpu. Thats not much room to play with.