How not to build an Ultima if you ever want to drive it
Discussion
As there seem to be some interest in my mods I will show what I do and did
in irregular order.
Got my kit in December 2000 and the car was legalized at the beginning of 2011.
I can drive it but it is far from finished if that ever will occur.
I begin with the new tanks I am working on. Hope to test them this summer.
The tanks were built because I want internal pumps and collectors,
want the two tanks to work as one including filling from both sides and the need for another shape
to create air channels under the doors from the front wheelarches to the engine bay.
Air channels only exist as cardboard mock ups until now.
All connectors except for the fill valves are in the top plates. They are that big to allow for easy servicing
and putting the collector tanks in.
The left tank will be the one with the high pressure pump, fuel will only be drawn from this tank.
The right tank will have a small lifter pump.
The signals of the two level senders will be compared and if the left tank´s level is lower the lifter pump fills it
up from the right one to roughly even out the levels.
In addition a level switch in each collector will switch off the pump if the fuel level is to low.
As internal collectors with trap doors from the usual manufacturers are very expensive and fixing the of the
pumps is difficult I try to make my own using chemically stable epoxide resin. They already exist including parts
of the lower pump fixings. The tall high pressure pump needs an additonal fixing at the top.
Both pumps are mounted in rubber to avoid as much noise as possible.
Collectors will be hold in place by the tank foam.
The pumps are made by Pierburg, the pressure pump can deliver up to 330 l/h at 4 bar.
The filler necks will be connected by a 1" aluminium pipe to allow filling from both sides. It´s the biggest pipe I can use.
The pipe runs from side to side under the upper chassis member. Baffels inside the filler necks will guide as much fuel
as possible to this tube while filling but the tank on the other side will fill slower.
Made some tests with water but could not reach the flow rate that the petrol nozzle delivers (approx. 36 l/min), so further test are needed.
The tanks will be mounted by welded in threaded inserts and a lug at the front.
What I now need to find is a pump controller as I don´t want the pressure pump run at full speed all the time.
That´s it for the moment.
Michael
in irregular order.
Got my kit in December 2000 and the car was legalized at the beginning of 2011.
I can drive it but it is far from finished if that ever will occur.
I begin with the new tanks I am working on. Hope to test them this summer.
The tanks were built because I want internal pumps and collectors,
want the two tanks to work as one including filling from both sides and the need for another shape
to create air channels under the doors from the front wheelarches to the engine bay.
Air channels only exist as cardboard mock ups until now.
All connectors except for the fill valves are in the top plates. They are that big to allow for easy servicing
and putting the collector tanks in.
The left tank will be the one with the high pressure pump, fuel will only be drawn from this tank.
The right tank will have a small lifter pump.
The signals of the two level senders will be compared and if the left tank´s level is lower the lifter pump fills it
up from the right one to roughly even out the levels.
In addition a level switch in each collector will switch off the pump if the fuel level is to low.
As internal collectors with trap doors from the usual manufacturers are very expensive and fixing the of the
pumps is difficult I try to make my own using chemically stable epoxide resin. They already exist including parts
of the lower pump fixings. The tall high pressure pump needs an additonal fixing at the top.
Both pumps are mounted in rubber to avoid as much noise as possible.
Collectors will be hold in place by the tank foam.
The pumps are made by Pierburg, the pressure pump can deliver up to 330 l/h at 4 bar.
The filler necks will be connected by a 1" aluminium pipe to allow filling from both sides. It´s the biggest pipe I can use.
The pipe runs from side to side under the upper chassis member. Baffels inside the filler necks will guide as much fuel
as possible to this tube while filling but the tank on the other side will fill slower.
Made some tests with water but could not reach the flow rate that the petrol nozzle delivers (approx. 36 l/min), so further test are needed.
The tanks will be mounted by welded in threaded inserts and a lug at the front.
What I now need to find is a pump controller as I don´t want the pressure pump run at full speed all the time.
That´s it for the moment.
Michael
Are you going to add the two level sensor signals together to drive the gauge? Otherwise you'll end up with a fuel gauge that rapidly falls to close to empty, then hovers there for a while, then falls to true empty.......... Which could be annoying!
Also, rapid filling is about managing the return paths of the displaced air, as much as providing a suitable fluid path. Generally poor filling and blow back are caused by insufficient return paths in my experience.
Also, rapid filling is about managing the return paths of the displaced air, as much as providing a suitable fluid path. Generally poor filling and blow back are caused by insufficient return paths in my experience.
Paul,
using one or two sensors for the gauge depends on how well the leveling system will work. The idea is to have even levels all the time.
If both tank levels are pretty much the same I may use the main tank´s sensor only for the gauge, otherweise need to create a sum signal.
Yes, the air paths need attention. Until now each tank has a rollover vent that can flow 48 l/min. These are connected to a carcoal canister.
Will see if that is sufficient to prevent blow back.
Ryan, thanks for the links.
I learned electronics a lot of years ago and thought about an op amp based solution too.
Michael
using one or two sensors for the gauge depends on how well the leveling system will work. The idea is to have even levels all the time.
If both tank levels are pretty much the same I may use the main tank´s sensor only for the gauge, otherweise need to create a sum signal.
Yes, the air paths need attention. Until now each tank has a rollover vent that can flow 48 l/min. These are connected to a carcoal canister.
Will see if that is sufficient to prevent blow back.
Ryan, thanks for the links.
I learned electronics a lot of years ago and thought about an op amp based solution too.
Michael
ezakimak said:
http://www.ecircuitcenter.com/Circuits/opdif/opdif...
http://en.wikipedia.org/wiki/Differential_amplifie...
Ryan
Whilst you can do that sort of thing in descrete analogue logic, when you can buy a small 8bit microcontroller with onboard ADC for around $1.50, i have no idea why you would want too! ;-)http://en.wikipedia.org/wiki/Differential_amplifie...
Ryan
Max_Torque said:
Whilst you can do that sort of thing in descrete analogue logic, when you can buy a small 8bit microcontroller with onboard ADC for around $1.50, i have no idea why you would want too! ;-)
Or maybe a robust circuit consisting of a wheatstone bridge formed with the two tank level sensors and a signal-soothing capacity triggering a FET or solid state relay with implemented optocoupler to switch the pump. So it would just need three or four elements to create. AndreasW said:
Or maybe a robust circuit consisting of a wheatstone bridge formed with the two tank level sensors and a signal-soothing capacity triggering a FET or solid state relay with implemented optocoupler to switch the pump. So it would just need three or four elements to create.
Wow, that sounds very much like rocket science to me. Are you sure you're still speaking English?Okay, I try to explain: Only two additional resistors are needed to realize a measuring-bridge with the level-sensors. The bridge-voltage (difference signal) switches an optocoupler at the input of the switching transistor/relay for the pump. Due to that the optocoupler works with a diode - it is only activated in one voltage direction.
So the circuit activates the pump only in the dedicated direction from tank a to tank b and not vice versa.
So the circuit activates the pump only in the dedicated direction from tank a to tank b and not vice versa.
AndreasW said:
Okay, I try to explain: Only two additional resistors are needed to realize a measuring-bridge with the level-sensors. The bridge-voltage (difference signal) switches an optocoupler at the input of the switching transistor/relay for the pump. Due to that the optocoupler works with a diode - it is only activated in one voltage direction.
So the circuit activates the pump only in the dedicated direction from tank a to tank b and not vice versa.
Whilst it all looks good on paper, once you start trying to add in Hysteresis, temperature linearity, voltage linearity, and allow for ageing and differences in the sensors themselves it gets really un-necessarily complex. Something like an ATtiny from Atmel (free tool chain except for programmer) allows you to do all that in digital logic to any level of complexity you like, and probably create a single voltage output to drive the gauge directly as well ;-)So the circuit activates the pump only in the dedicated direction from tank a to tank b and not vice versa.
Max_Torque said:
Whilst it all looks good on paper, once you start trying to add in Hysteresis, temperature linearity, voltage linearity, and allow for ageing and differences in the sensors themselves it gets really un-necessarily complex. Something like an ATtiny from Atmel (free tool chain except for programmer) allows you to do all that in digital logic to any level of complexity you like, and probably create a single voltage output to drive the gauge directly as well ;-)
Yes, I had even realized a lot with Atmels ATmegas and AVRs...But really, needs this task a signal conditioning or compensation?The accuracy does not have to be very precise also in my opinion...But when it comes to accuracy you will build a bridge...
I had been trying to think of a particularly simple solution. Building a circuit with a controller and an accordingly code requires some special knowledge and equipment.
But a clever combination of a few simple elements can already be sufficient for this (nevertheless I like the Atmel's probably as like you).
AndreasW said:
Max_Torque said:
Whilst it all looks good on paper, once you start trying to add in Hysteresis, temperature linearity, voltage linearity, and allow for ageing and differences in the sensors themselves it gets really un-necessarily complex. Something like an ATtiny from Atmel (free tool chain except for programmer) allows you to do all that in digital logic to any level of complexity you like, and probably create a single voltage output to drive the gauge directly as well ;-)
Yes, I had even realized a lot with Atmels ATmegas and AVRs...But really, needs this task a signal conditioning or compensation?The accuracy does not have to be very precise also in my opinion...But when it comes to accuracy you will build a bridge...
I had been trying to think of a particularly simple solution. Building a circuit with a controller and an accordingly code requires some special knowledge and equipment.
But a clever combination of a few simple elements can already be sufficient for this (nevertheless I like the Atmel's probably as like you).
This is because of the following:
1) simple digital circuits perform more consistently under changing conditions, and the automotive environment is a very variable one. Consider how you would add the necessary hysteresis in hardware? And how you would make that hysteresis respond to changing resistances of the sensors etc? Or what about if you build your device and find noise in the sensor signal (from say the ignition system) causes the pump to oscillate on/off all the time? In the digital domain, a low pass filter is 2 lines of code, added (with the help of a bit of googling maybe) in about 15min. In the analogue domain, well, all bets are off at that point
2) the allow you to easily calibrate the system to change how it works with just a software update. Juggling gains and offsets via resistor changes on a bunch of op-amps that get affected by heat and temperature is a lot more difficult and time consuming
3) Basic error reporting is easy for a digital circuit, much harder for an analogue one
4) Adding in extra functionality at a later date is fairly easy in software. For example, when the tanks are empty, you no longer want to run the transfer pump. That's 1 line of code in software, a right PITA in hardware
5) The basic digital circuit is simple. Just need a 12v(nominal, but needs to accomidate load dump etc) to 5v power supply for all the logic and sensors (unlike your op-amps, that might need split supplies or signal biasing to avoid limitations of those amps etc. The IC has it's own internal oscilator that will be plenty accurate enough. 1 pin to drive a logic level N channel Fet for the pump. 2 pins into two ADC channels on the IC (with a suitable pull up resistor to 5v, and connected to the two level sensors), and maybe 1 (pwm)pin as a voltage output to drive the fuel gauge with a composite signal
The entire device would be around half the size of a credit card, even using through hole components, and cost about $10 in components (minus enclosure / connectors etc)
No WAY would i be even starting that project in the analogue domain in 2014 ;-)
Edited by anonymous-user on Thursday 15th May 15:32
AndreasW said:
Okay, I try to explain: Only two additional resistors are needed to realize a measuring-bridge with the level-sensors. The bridge-voltage (difference signal) switches an optocoupler at the input of the switching transistor/relay for the pump. Due to that the optocoupler works with a diode - it is only activated in one voltage direction.
So the circuit activates the pump only in the dedicated direction from tank a to tank b and not vice versa.
But with the potential for slowly changing signals, that optocoupler will be operated in it's "Linear" region, resulting in the following transistor acting as a linear pass device, resulting in both varriable switching points and probably transistor death from the required power rejection.So the circuit activates the pump only in the dedicated direction from tank a to tank b and not vice versa.
If you do this you need a comparitor or schmitt trigger first to ensure you discretise that signal into only two states, and to prevent oscillation that comparitor/trigger needs a suitable degree of hysteresis.
Like i said, analogue looks simple on paper, but the practical embodyment of it rapidly becomes complex!
(which is why EVERYTHING is now done in the digital domain, especially as a single 8b processor will cost less than just the comparitor you need for the analogue version.......)
Max_Torque said:
But with the potential for slowly changing signals, that optocoupler will be operated in it's "Linear" region, resulting in the following transistor acting as a linear pass device, resulting in both varriable switching points and probably transistor death from the required power rejection.
If you do this you need a comparitor or schmitt trigger first to ensure you discretise that signal into only two states, and to prevent oscillation that comparitor/trigger needs a suitable degree of hysteresis.
Yes, of course, these transistor basics are also known to me. I also see that you know a lot about. I will not attack you and say that it makes no sense to bomb a microcontroller on it and solve the problems in the code. But maybe the problem is not big enough for someones to acquire the knowledge and accessories specifically for this one case. So maybe it makes sense, or even fun, to think about solutions that are, from this perspective, simply realized by solder together. Regardless of whether this would cost 1.50 Euro or at the end 50 Euro.If you do this you need a comparitor or schmitt trigger first to ensure you discretise that signal into only two states, and to prevent oscillation that comparitor/trigger needs a suitable degree of hysteresis.
There are now lots of integrated modules and smart switches in the analogue area that have everything on board. For example, Profet series of infineon, Logik Gate FET's from International Rectifier or many others that may could be on and off controlled directly with an optocoupler or any signal. Or, instead, you use a (reed) relay. This has a unique switching state and can also be turned on by an optocoupler (full activated).
This results in a lot of ideas can be developed that could be a smart solution without getting into learn microcontroller applications and assembler.
It looks like a good idea to me, to do a small "fueling ECU", in which all fueling related functions can be handled.
e.g.
-tank leveling function
-level sensor conditioning
-hp-pump fuel pressure controller to have a returnless system
....
in the end, it might be of interest for someone else, who has a "standard fuel system" and wants to use a single function only??? (e.g. pump-controller)
cheers
Cornel
e.g.
-tank leveling function
-level sensor conditioning
-hp-pump fuel pressure controller to have a returnless system
....
in the end, it might be of interest for someone else, who has a "standard fuel system" and wants to use a single function only??? (e.g. pump-controller)
cheers
Cornel
Gassing Station | Ultima | Top of Page | What's New | My Stuff