h a l f b a k e r yNot just a think tank. An entire army of think.
add, search, annotate, link, view, overview, recent, by name, random
news, help, about, links, report a problem
browse anonymously,
or get an account
and write.
register,
|
|
|
Please log in.
Before you can vote, you need to register.
Please log in or create an account.
|
Cars equiped with analogue sensors require conversion routines to perform engine management and the engine management system MUST be calibrated for the sensors fitted to the engine.
For example narrow band lambda sensors produce a signal that varies between 0 and 1 volt depending on the amount of
CO2 in the exhaust gases. This signal is converted to an Air Fuel Ratio that can be used to adjust fuel delivery.
Another example is the Air Flow sensor that produces a signal between 0 and 5 volts depending on the amount of air flowing through the sensor. This signal is converted to a manifold pressure that is again used to adjust fuel delivery.
An engine management system that expects a digital signal requires no calibration .
For example the lambda sensor will produce an Air Fuel Ratio (a real value between 1 and 20 say) and the Air Flow Sensor will output cubic feet per minute (a real value between 0 and 500 say).
This scheme allows components to be produced for any vehicle, i.e. making it easy to swap in a high end after market Manifold Pressure Sensor without having to purchase an expensive remap (or after market ECU)...
Madness...
B-777 Aircraft Analog Sensors with Digital Outputs
http://books.google...oq8&hl=en#PPT410,M1
[Amos Kito, Mar 06 2008]
CAN bus
http://en.wikipedia...roller_Area_Network
[half, Mar 08 2008]
[link]
|
| |
But the sensors are sensing real world phenomena, which tend to be analog. There has to be A-D conversion somewhere in the chain. |
|
| |
(-) The calibration and mapping aren't a big deal - it's much more difficult to produce hardware that works without it. Such sensors would be far more expensive than analog ones. |
|
| |
Yes, but they'd be far more accurate. An
analogue sensor may give you a value of
about 5, but a digital sensor will give you
5.000000. |
|
| |
!)By displacing the D/A converter onto the sensor you would increse the cost of a sensor. |
|
| |
2)Many sensors live in very harsh conditions that would be hard on a microchip. |
|
| |
3)Interchangability of componants would not be increased because manufacturers refuse to use universal standards for threading, voltage, and polarity as it is. Having a D/A converter on the sensor wouldn't change this. |
|
| |
You would be paying for an expensive ECU on every sensor. Which would defeat the stated purpose found in the final paragraph. |
|
| |
//far more expensive//
The complexity and cost makes this impractical for the family car. But you'll find it in more sophisiticated vehicles. The Boeing 777 has "Air Data Modules" with built-in conversion of their analog signals (no air sense line plumbing necessary). The signals from many components share a data bus (only 2 wires in each bus), reducing weight. |
|
| |
Sure, and common bus technology is getting more common. Gauges and Ecu-Tcu connections already use this technque. However placing a microchip inside your O2 sensor or your coolant temp sensor is silly. Consolodating wiring by ganging multiple sensors through a bus is great (especially where long distances and weight are a factor) but there is little call for it in a car especially where it would needlessly increase replacement costs. |
|
| |
Some MAF sensors have frequency (pulse) output. That's reasonably digital, otherwise you need coding systems (CAN? Fieldbus?) to keep the wiring costs down. I assume you don't propose parallel digital signals? |
|
| |
But the MAF sensor with frequency output needs calibration at the PCM. I think you want to have the sensor say: this is normal 0% value, and this is normal 100% value, and right now I am at 45%, which is doable but more complicated than a frequency output. By the way, the MAF calibration depends on the installation, also. So even with a perfect digital system, it still needs calibration to account for bends in piping and resonance effects. |
|
| |
In a car we call these switches and they could hardly be called an innovation. Modern ECUs can do much better than "Coolant is cold/Coolant is hot". As for your asserition that this could be cheaper in some cases that may be true but analog sensors are generally cheaper than their switching counterparts. |
|
| |
When you need to measure somthing then use a consolodated ECU. If you need to process a logic measurement (waveform) then moving the D/A converter into the sensor can make sense. |
|
| |
Bianary? In a car we call these switches and they could hardly be called an innovation. Modern ECUs can do much better than "Coolant is cold/Coolant is hot". As for your asserition that this could be cheaper in some cases that may be true but analog sensors are generally cheaper than their switching counterparts. |
|
| |
When you need to make quantitative measurements then use a common ECU. If you need to process a logic measurement (waveform) then moving the D/A converter into the sensor can make sense. |
|
| |
Don't confuse bianary and duty cycle with digital. Maf sensors that use pwm are not digital as they do not produce a numerical output. Niether are optical mice. Infact I cannot think of even one single purpose sensor that uses numeric encoding to transmit data. |
|
| |
Whenever you are faced with an analog phenomenon and you want a digital signal you will need to have a D/A conversion. What you are proposing is not an actual digital sensor but rather to change the location of the D/A converter. You propose that each individual sensor have its own converter and send out its signal. |
|
| |
There is some benefit to such a system as well as a number of problems. |
|
| |
On the benefit side such a system could radically simplify the signal wiring in a typical vehicle , in fact such benefits are already realized in many modern cars. The advantage stems from the fact that a digital system can utilize a BUS to communicate with the ECM. Thus only one pair of wires is required to connect the engine and all its sensors back to the ECM. Each sensor would have its own "address" and all the signals would travel through the signal cable. Such systems are used today for items like steering wheel controls, so that the button press encodes the correct signal and sends it off down a signal line to the computer, error correction etc means that the message is received with no errors. |
|
| |
On the downside however you would need to power the electronics in the sensor with a separate filtered power supply, or incorporate the power conditioning into each sensor. For most situations this would mean at least 4 wires would need to connect to each sensor to do the job. So while reducing the number of signal cables connecting the ECM to the sensors you would have a corresponding increase in power distribution cabling, thus negating the reduction is signal cables. |
|
| |
Additionally the complexity of each sensor would be considerably increased, thus increasing cost to manufacture and the incidence of sensor failure.(engines are hot, shaky places, two things electronics do not like) |
|
| |
The practicality of designing one robust power supply, stored in a remote location that is dry, cool and free from vibration compared to 15 or 20 miniature supplies operating in extremely harsh environs are rather obvious. The reliability of modern sensors is really quite good in aggregate. |
|
| |
Existing systems are really quite robust and simple to work with. The signal cables are standard copper wire that can be readily repaired and are very rugged, the ECM has its own on-board power supply and each sensor is on its own powered loop, the only portion of the system subject to the harsh environment is the actual sensor head. Most large industrial sensor systems require on-site signal conditioning due to the need to boost the signal over large distances, this is not a problem in a car as the distances are short enough that no appreciable degradation can occur. |
|
| |
One last point is that the information that is provided to the computer is not sampled at a very high rate as these things go, for 99% of automotive applications one has more than enough information to work with having a signal of "about 5" There really is little to no benefit to knowing if it is 5.0001 as opposed to 5.0004. The difference between 4 and 5 is all that the machine requires to do its job. Sensors such as the O2 sensors actually change from 0 to 5 and back in fractions of a second. most are monitored over periods of seconds to observe any trend in their performance to permit adjustments. |
|
| |
In reference to interchangeability I think you would find it to be even less due to the need for each sensor to be coded to a particular type of BUS. Many modern sensors are actually nearly identical, such that minor variations from one maker to the next involve only the connection to the car, the sensor itself is largely identical. |
|
| |
The imagined benefit of having a fixed sensor output scheme is also not a real benefit as you would still need to tell the car what to do with the info, so telling to to set the fuel input when the O2 is at .75 volts is no different than telling it to set it when the ratio is 15. IT still needs to be mapped. After market components already provide compatible signals to the system, the remapping you discuss has nothing to do with signal type, it has only to do with the fact that car needs to know how to make use of the additional air it can now receive. Something that would be the same weather digital or analog. |
|
| |
In closing I don't really see this as a benefit if anything it makes things worse. |
|
| |
Shirley the solution is to use speaking
sensors. Each sensor has a small voice
synthesizer and speaker - not expensive.
Behind the dash is a very sensitive
microphone and some voice recognition
firmware (more expensive, but only one
per car). "Shit! Bloody oil pressure's low!!"
shouts the sensor, and the dashboard
firmware tells it to calm down, and puts on
the little red warning light. |
|
| |
See, you're talking Binary again? The robotic voice from the engine compartment should say "Oil pressure..3psi", and the dashboard should say "Isn't that a bit low?", and then the engine compartment should reply "Yes, you bloody idiot. I told you that last time", and then the dashboard should say "OK, I'll put that little red light on. I think I'll put the little engine light on as well for good measure". |
|
| |
I'm only interested in sensors for the purposes of signalling gauges, and having to go back to the shop a few times because they've given me a digital sensor when I'd expressly asked for an analogue one, as would invariably happen if there were the two sorts, would only be a nuisance. |
|
| |
Then there's the whole excuse - for - another - huge - factory angle. So [-]. |
|
| |
Lets take a look at your input... |
|
| |
--- Increased cost for extra ADC chips...
Nope each analogue signal requires separate real time conversion. |
|
| |
I am guessing the majority of you think there is just one ADC handling all the signal processing :) |
|
| |
--- Infeasible because can manufacturers will not adopt the technology...
Nope ODB-II. |
|
| |
Or better yet , in closing, think of it as the difference between a solid state LCD TV and one with valves... :) |
|
| |
No one said there were less A/D converters, every circuit would have one the difference is where they are located and the conditions they must endure. Rather than 2 wires per sensor you would need 4 and would need to have multiple power supplies rather than 1 central one. |
|
| |
The central point is that for an automotive application there is no real benefit as everything is in such close proximity(you will find a similar situation in large electronic devices with sensors such as appliances or scientific instruments. where distances allow for clean signal transmission) |
|
| |
There is no question of using digital technology in cars and for what its worth we are at OBD III now. |
|
| |
This would work just fine, it just would not be practical and your discussed benefits are largely non-existent. |
|
| |
//in closing, think of it as the difference between a solid state LCD TV and one with valves//
That's a pretty lousy comparison - a good CRT (valve) display still beats an LCD. |
|
| |
/I am guessing the majority of you think there is just one ADC handling all the signal processing :) / |
|
| |
Or, rather, one microchip with a number of A/D converters built in. Which happens, in my current ECU, to be the case. |
|
| |
I take issue with your last paragraph in the original idea, by the way. You want to be able to replace the sensor you've got with a better one. At a guess, it's going to have a wider range, to reflect the higher flow through a modified engine. This idea standardises the values that the ECU gets, so it understands that the engine is now receiving, say, 50% more air. However, you're still going to be off the upper limit of your fuelling map at the top end of the engine's speed range. You still require a remap. |
|
| |
The advantage is ensuring electrical compatibilty between the MAF in a skoda and the MAF in a mercedes... |
|
| |
So why go digital? It is required anyway and it is possible to use the appropriate SI unit... |
|
| |
Hmm - compatibility between models and marques is brilliant for the consumer (for instance, I did a cheap brake upgrade a few weeks back by fitting calipers from a larger model) but I can't see it being good for the people who make the decisions. Mercedes aren't going to standardise with Skoda just so that Skoda drivers can nick their components. |
|
| |
The problem of standardization is less one of compatibility and much more an issue of engineering demands of the application. While the MAF for a Skoda may do the same job as that for the Merc, the engineering and packaging requirements will be quite different, first the Merc engine is most likely larger and draws a greater volume of air thus requiring a sensor optimized for that condition, also the layout of the engine will determine the configuration of the sensor, what fits under the skoda hood may not fit in the space available on the Merc. As it stands now the parts compatibilty between makers is a lot greater than you seem to believe, almost all brands under a manufacturer use the same stock parts bin sets(all GM cars use the same setups for example) I know also that Audi, VW and BMW share a lot of cross compatibility for things like MAF sensors(i can even send links of some of these interchanges) |
|
| |
Engineering is the key to this question. In order to meet the engineering needs of all the myriad types, sizes and demands of engines you would require a whole array of different sensors in different sizes and configurations to meet all these demands. The fact the signal is compatible is really secondary and as stated before in most cases the signals are compatible(for example the Sensors on my 97 528 operate using the same output signal, and operating voltage as those on my 89 Pontiac Sunbird. I could splice the connectors on almost every sensor on the car and the ECU would get the same information as it does now, the problem is that the sensor does not physically fit where it needs to be). As far as I know there are currently only 2 main types of O2 sensors, heated and unheated, other than that they all put out the exact same signal regardless of application, Further there is pretty much only 1 form factor, the only difference lies in the Plug that is affixed(and you can buy universals as well with no plug). There are only 2 or 3 styles of Gas tank level outputs as well, however there are hundreds of different physical configurations that are required to meet the design demands. |
|
| |
For certain types of sensors(pressure switches) you will find that many cross over to multiple companies as well. |
|
| |
The greater issue of compatability is one of technical advancement. This makes interchange between different years difficult as newer products come out. |
|
| |
I would beg that you go back and read my annotations carefully as I have addressed these items you are citing in some cases 2 or 3 times. |
|
| |
Your last and greatest failure will occur when you discover that people who own Mercs or BMWs or Accuras, are really not interested in having a MAF sensor that is okay for a Skoda, or hyandai or whatever. These people will want the part designed for their car. |
|
| |
I agreed with you up until you wrote "The problem with standardisation..." |
|
| |
I am all for competition but there comes a time when standardisation is the only way forward... For example power transmission --- imagine the problems if a few engineers suddenly decided it would be a good idea to plug a 60Hz generator into a 50Hz power grid... |
|
| |
Standardisation is an economic driver not an engineering one --- so when I say standardise the engineers say how much... |
|
| |
//Yes, but they'd be far more accurate. An analogue sensor may give you a value of about 5, but a digital sensor will give you 5.000000.// |
|
| |
The question is not one of whether or not standardization is good, it is a question of whether it is feasible. The industry has largely standardized on a particular set of sensors and mode of operation. The problem is that what works in one design may not work in another. To use your example, there is a standardized set of specs for electrical systems, however there are literally hundreds of different configurations of generators using different motors, generators, windings, rectifiers, chassis, covering, coloration, fuels etc etc etc. Different setups are used to meet different needs while all operate and output power to the standard format. |
|
| |
Back to the idea, as I stated before and you apparently refuse to acknowledge, the industry has already largely standardized on the modes of communication and signal transfer, Almost every car on the road today is using a 5volt signaling system for their sensors and as mentioned many of the form factors are also consistent from one make to the next. |
|
| |
Your example is flawed and your argument is weak you will need to try harder. |
|
| |
Well obviously you missed the point... |
|
| |
--- I agree most sensors produce 0 to 1 volts (or 0 to 5 volts where more accuracy is required (ie a wideband))... |
|
| |
--- THE POINT IS THIS... even if two MAFs produced the standard 0 to 5 volts the air flow associated to each voltage will be different... |
|
| |
Hence standardisation REQUIRES a scientific unit and this is NOT currently available... |
|
| |
// THE POINT IS THIS... even if two MAFs produced the standard 0 to 5 volts the air flow associated to each voltage will be different... // |
|
| |
I think you are making an unsupported assumption here. If the 2 sensors share the same throat length and diameter then the signal correspondence will be quite high. If they are different shapes or sizes then the correlation will not hold. |
|
| |
Yip so the output needs to be calibrated at the sensor (and the output will be digital or quantised if you prefer...) |
|
| |
Clearly the actual digital encoding of some analogue voltage is left to some other interested party... (probably named alex). |
|
| |
You keep glossing over the other points that were made. |
|
| |
Well they are moot, in my opinion. |
|
| |
The continued advance of mankind is required and what is required is digital sensors ... |
|
| |
I admire your fortitude, if not your idea. |
|
| |
Yes --- my mind is like a fortress... its hard to get stuff in and equally hard to get stuff out :) |
|
| |
I am not sure fortitude is a good thing... |
|
| |
I find it hard to believe that sensors with built-in ADCs didn't
exist prior to 2008. |
|
| |