New to forum / Custom WI system for AUDI

[simple]

Hello All,
I am new to this place. Have found it searching some info on W/I.
This place is great source of useful info!
My Location: Australia
Car: AUDI 1.8T Quattro B6
The reason for searching, I am trying to build a custom W/I system for my car. Still learning about it. My intension is to have a variable W/I with knock sensitive activation and self learning function.
I know that I am on a quest for the Holy Grail, but it is in my nature to explore (I lead R&D team at factory).

Val

:smile:

[Richard L]

It is a good idea. YOu need to find a good reliable knock signal or risk injecting too much water due to other engine noises at high RPM.

The holy grail is finding a true "knock signal", a "voice print" of your particular engine is necessary, may be require a high speed DSP (digital signal processing) chip to record and compare and filter out the background noise.

Good luck with your prioject.

[simple]

Thanks Richard,
I have parts already coming from different parts of the world. I will keep you posted on the progress.

[simple]

An update for those who interested :D

At this stage we are only testing the hydraulics, to get some data to refine electronic control module which is in drawings stage.
This is the pic of the one of the test set-ups

http://www.audizine.com/gallery/data/500/10126set_up-med.jpg

The pump is 5 L/m 160PSI. Happy workhorse

http://www.audizine.com/gallery/data/500/10126160PSI_pump-med.jpg

The jet we experimenting is agricultural 40 micron, low wear ceramic internals atomizing nozzle. Utilizes centrifugal force atomization principle. It has surprisingly stable atomization from 30-160PSI range. Picture taken at 160PSI

http://www.audizine.com/gallery/data/500/1012640micron_jet_400_cc-med.jpg

At this pressure it delivers 420 cc/m. And at 20PSI (min pressure) we get about 140cc/m.

We also need to try Aquamist jet, it has better adopted fitting. Just need to get my hands on one :wink:

[Richard L]

I am not sure how you are going to control the flow, If you vary the pump speed to vary the flow, please consider the following:

Your starting pressure = 30psi+ 20psi (assummed manifold pressure) = 50 psi to get a decent droplet side.
You final pressure will be 160psi-20psi = 140 psi

The actual pressure change is 50-140psi, approximately 3x.

Translate the presesure change to flwo change:

In order to double the flow, you need 4x the pressure. So you actual flow range is about +70%. Do you think it will be adequate for your application? Can you pump create more pressure?

If you put a 15psi checkvalve inline to stop the water from vacuum siphoning , your pressure span will be reduced further to 65psi to 125psi, giving you less than +50% increase between min and max!


http://www.aquamist.co.uk/forum/jets-cc2.gif



Richard

[simple]

Hello Richard and thank you for the input.
I did not allow for boost back pressure ? my bad. At this stage I have two paths to go, it is a twin stage pump system to generate 250-280PSI or to have twin stage nozzle set up when second nozzle valve controlled and opened when first nozzle is maxed out.
Both ideas need to be tested for suitability and no so great.

On the other hand I plan to have accurate knock detection and water will be delivered and volume of water accurately calibrated JUST to suppress the detonation. Based on that I guess that my requirements for water flow range will be significantly reduced due to the fact that amount of water needed to be injected not proportional to RPM at given load.

I have come across a person in Canada who tested SS fuel injectors for 6 months in a solution of 50/50 water/meth. Both, fuel pump and fuel injectors have survive the test and in a good shape. This is something to test and consider as well.

I also was considering HSV you sell and constant water pressure. But I would prefer analogue water delivery to pulsed spray as I plan to inject water 50sm away from engine head and you may not get homogenised mix of water droplets and air.

I guess that truly Ideal set up would be to have 4 nozzles (one for each port) and each nozzle controlled by HSV and constant water pressure. Basically we have to copy fuel injection system. Than we can vary HSV duty cycle to deliver necessary amount of water. This set up will also allow you to deliver water to cylinders as needed due to the fact that not all cylinder knock when engine under the load. Some are more stable and need a lot less water added. This set up will use a lot less water as it will not waste any.
The problem here is the cost of HSV as you would need 4 of them.

Looks like I get carried away a bit. Sorry for the long post
:smile:

[Richard L]

Depending on the final specificationa, the capability required.

Progressive pump system analysised:
Apart from the limited dynamic range (unless you have a pump that can delivery between 50psi to 1000 psi), you will still fall short of the power range of an engine.

Progressive pump speed system is also quite slow to react to change fo transient load, for example: change up or down a gear could mean a change of delivery quantity of 1000rpm. A rotating mass (pump motor) will not be fast enough to react to this change. So you have one more things to consider if you want to go for a anaogue pump speed controller.

Pulsed delivery will not affect the engine dynamics that much, considering the engine only have one intake stroke per revolution (I think?). Since droplet size is a mixed bunch, smaller droplet to travel faster towards the throttle even it starts at the same point. So in real time, you have a continuous train of water droplets spread over the entire inlet tract.

I think you should really consided a system similar to the fuel injection system adopted by almost all car makers.


Richard

[simple]

OK, done some tests with fuel injector:

http://www.audizine.com/gallery/data/500/10126fuel_injector-med.jpg

Injector is off 3.0L Turbo supra / V-shape (440cc).
Result is disappointing. Atomization at 140PSI is poor as you can see on the picture. It is more like jet spray rather than mist (comparing to previous agricultural nozzle).
Now I need some professional input from tuners here, do all injectors have that bad atomization?

[Richard L]

Almost all injectors have narrow angle and the atomisation is about the same. Theyarenot design to cool inlet air but aim at teh inlet valves (also requires cooling).

Richard

[simple]

Richard, can you post some specifics for HSV valve.
I wonder how I can drive it. I believe it has 250Hz max operating frequency.
I need to know what signal needed to drive (like square pulse) and what is recommended signal duration to operate it at 1% duty and 99% duty.
What is it? internal resistance, so I can work out power consumption. Also what is it?s inrush current (if know)?

Car specifics as mentioned before:
1.8L Turbocharged motor / 4 cylinder
Four inlet ports/ 3 intake valves per port.

Considering using one HSV to operate two M3 nozzles (mounted in common section of plenum)

[Richard L]

Richard, can you post some specifics for HSV valve.
I wonder how I can drive it. I believe it has 250Hz max operating frequency.
I need to know what signal needed to drive (like square pulse) and what is recommended signal duration to operate it at 1% duty and 99% duty.
What is it? internal resistance, so I can work out power consumption. Also what is it?s inrush current (if know)?

Car specifics as mentioned before:
1.8L Turbocharged motor / 4 cylinder
Four inlet ports/ 3 intake valves per port.

Considering using one HSV to operate two M3 nozzles (mounted in common section of plenum)

The HSV's coil resistance is 13ohms. on time is 2ms and off time is 2ms, overall is 4ms = 250Hz max. cycle rate. I suggest drivng the valve at 30-80Hz

at 30Hz: the maximum duty cycle span is 88%
at 80Hz: the maximum duty cycle span is 68%

You can improve the span by starting your opening pulse at >2ms so the new span becomes:
at 30Hz: maximum span is 94%
at 80Hz: maximum span is 84%

Further improvement of the shutting time can be made by controlling the decay time of the HSV via the clamping of the back emf's peak voltage rise.

It order to achieve 99%, you need to consider using a "peak and hold" circuitry to minimize the residual current left in the coil after switching off.

In practice, I will just run it at 50Hz, starting pulse at 2ms and finish at 18ms - for 0-100% duty cycle control.

Richard

[simple]

Richard,
I would need HSV to try. I send you and email. Please check.

[Richard L]

Richard,
I would need HSV to try. I send you and email. Please check.

Have got your mail and replied.

Richard

[simple]

Currently looking in to how to drive HSV. Probably will test varying pump speed with conjunction with HSV. Calculation for water delivery will be a nightmare but this is a software guy problem
:D
By the way, Richard, ?peak and hold? drive principle is not suitable as HSV has high resistance. Do you supply them with 3-4Ohm coils suitable for ?peak and hold? ?

[Richard L]

Currently looking in to how to drive HSV. Probably will test varying pump speed with conjunction with HSV. Calculation for water delivery will be a nightmare but this is a software guy problem
:D
By the way, Richard, ?peak and hold? drive principle is not suitable as HSV has high resistance. Do you supply them with 3-4Ohm coils suitable for ?peak and hold? ?


I would not recommend vary the pressure of pump the AS WELL AS varying the duty cycle of the HSV. The only time when you want to vary the speed/pressure of the pump is to compensate the manifold/water pressure differential.

Life would be a great deal simplier if you deal with a valve-PWM system rather then pump-PWM system. As the early chart show, the relationship between the pump speed and flow in not linear at all. It will definately a nighmere to map/tune the water flow with some consistancey fuel flow. I based on the assumption that the purpose of water is the replace the necessarity of running an over-rich fuel mix and result in better power, cooler potential.


Richard

[simple]

Will consider your advice Richard.
Can I get you to reply to my Email with Proforma invoice?
Thank you!
:D
I really want to get my hands on HSV and see what it can do.

[Richard L]

Will consider your advice Richard.
Can I get to reply to my Email with Proforma invoice?
Thank you!
:D
I really want to get my hands on HSV and see what it can do.

I have just sent you any email, you omitted for the HSV but instead a standard low-speed on/off valve. Please correct or amend the list I posted.

Richard

[simple]

Thank you Richard, I appricate your time very much.
Order placed, payment arranged. Please let me know when order will get dispatch.
At this stage I cannot do much but wayt for parts to test so I can confirm our electronics configuration circuitry. By the way this is the specs for set up. Anyone's comment welcome :smile:
----------------
WI system specs

General:
System to have rectifier bridge and filter for electronics to prevent reversal of polarity feed and take out possible spikes (voltage regulator 12V for electronics?) and on board fuse. Also to have on board supercap for temporary chip?s memory back up.


User interface:
? UP & DOWN, RIGHT & LEFT, ENTER buttons for menu
? Backlight ON/OFF button
? LCD screen

System Memory:
? MS memory card (of similar)

System inputs:
? Injector duty cycle (two inputs)
? MAP sensor absolute
? MAP sensor differential
? EGT temperature sensor
? Ambient Temperature sensor
? Turbine discharge temperature
? Engine intake temperature
? Engine block temperature
? Knock sensor signal
? Water low sensor(selectable 12/ground/sensor pin)
? Ignition ON (activation)
? Power (12V/15A input)
? Pump power supply interruption ( build-in over pressure sensor)


System outputs:
? Pump voltage control
? HSV duty cycle control (two independent outputs for future use .CPU support? )
? Solenoid ON/OFF control (2 / one for IC spray, one for water flow cut off)
? System programmable output (two / +12/ground)

--------------

Specs to be confirmed as hydrolic hardware get tested.

[simple]

Done some graphs for ceramic atomazers water delivery VS preshure. This is done on SHUREFLOW type pump:

http://www.audizine.com/gallery/data/500/10126Water_delivery-med.JPG

[Richard L]

Very close to the predicted chart, almost identical.

Richard

[simple]

I was thinking I am the only ?smart? a** around and will invent something new, till I found this: http://www.aquamist.co.uk/dc/coollinks3/index/street/smdex/jankel/jankel2.html

Some one already has the system I am trying to build RUNING! Man, life is so not fear :cry:

Anyway, Richard how much is the set up like this? Do you make it?

By the way, funds where droven of my CC, this would mean I will see my order soon. Thanks Richard.

[Richard L]

I am sorry, but we are not supplying this system anymore. Production ceased on 1996. We supplied the system to RR Malsanne Turbo and Jankle only - it was a special system commissioned by RR for a batch of highly tuned car ordered by someone in Brunei.

Your parts were sent last week by registrered airpost.

Richard

[simple]

Wow, you guys are got to be top dogs in the field if had system like this running in 1996!!! The current systems you sell would look primitive when compared to this high tech staff. Today we have chips that can do and process 100?s times more than back in 1996. Not to mention that the sensors and the rest of the products moved forward a lot.
Richard, why did you stop offering the kits? With today mass production you can produce it relatively cheap.
:roll:

[Richard L]

I long for someone else wouldl take up this and continue to develop it further. We have to out our personal interest aside and make products that suit the pocket size of the majority. If some one can come up and give us a bunch of fund for keeping us going for five years without have to selling anything, I am "more" than happy to take this project on again.

It has been a number year since I looked at powerful processors other than little PICs. I hope you will continue this development and if you need any help on special parts, let me know.

It is not the cost of the system but the complex aftersales service required to back your product up. Plus the cost of R&D require to acquired to product the algorithm for the DSP chip to identify true knock's voice print and deals with changes with aging engine components. Other signals from sensor possibility requires the same intense R&D to get it right.

[simple]

According to my calculation:
For HSV to deliver water properly you need to match its operating frequency to engine RPM. Since my set up is single HSV and its 4 cylinder engine with 4 nozzles (one per each intake) at 6000RPM HSV will operate at 50HZ (6000RPM -> 100RPS -> 50RPS to draw water from all 4 intake tracts -> 50HZ valve speed). This is not ideal (as ideal would be to have 4 x HSV operating at 50HZ each) but feasible compromise between HSV cost and performance.
The system must be matching frequency due to misting nozzles positioned next to intake valves and there is no chance for mist to be homogenized with air evenly. Offsetting HSV operating frequency in relation to intake strokes will cause over-delivery of water per some intake strokes and shortage of water in others.

Richard, am I correct in calculations?

In order to controll HSV using "peak and hold" style drive i need to know operating parameters. I can have only 12.5VDC for activation but what is recomended "hold open" voltage?

[Richard L]

I am afraid there are no easy way to deliver equal amount of water to each cylinder unless you are in sync with your individual fuel injector that happen to be a multi-point and sequantial type. Even the seqential system doesn't direct the fuel into an open valve. It only happens at low RPM - at high RPM and load, there will be fuel waiting at the closed inlet valve.

If you are going for port injection, there is no point using an atomising jet, I have worked with a few R&D labs - they all use a pencil shape jet, directly aimed at the inlet valve.

If your aim is inlet air cooling + incylinder cooling, it is better to use one single jet after the intercooler and four individual non-atomising jet on each port. You can set the flow ratio between the two.

Water will cool by evaporation until it reaches the saturation point. From that point onwards, no more cooling is expected.

Richard

[simple]

Good info Richard. Too much to comment at this stage as it will take me some time to think over it.

I this stage I just want to share surge arrestor results. I had one made from aluminum about 270cc capacity. It is membrane-less type.
Pressure variation at set points:

No surge arrestor
2.5VDC ? 4.2AMP-(30-50PSI)
3.8VDC-7.6AMP-(80-100PSI)
4.8VDC- 8.8AMP- (95-125PSI)
5.8VDC-9.7AMP-(120-150PSI)
7.4VDC-10AMP-(150-180PSI)

With surge arrestor (filed with water 100%):
2.5VDC ? 5.5AMP-(60-70PSI)
3.8VDC-8.1AMP-(105-120PSI)
4.8VDC- 8.8AMP- (128140PSI)
5.8VDC-9.3AMP-(140-160PSI)
7.4VDC-9.6AMP-(150-180PSI)

With surge arrestor (filed with water 50%):
2.5VDC ?AMP-(58PSI)
3.8VDC-AMP-(93PSI)
4.8VDC- AMP- (98-102PSI)
5.8VDC-AMP-(140-145PSI)
7.4VDC-7.6AMP-(160-170PSI)

According to test results, having surge arrestor is beneficial to power consumption and pressure variation reduction. Especially if you have proper (membrane type) or air filed arrestor it will make a huge difference. In my case it removed 90% of pressure pulsation and reduced power consumption by as much as 20%. Not to mention all the hydraulic system benefits such as increased flow and reduced stress of the system.

http://www.audizine.com/gallery/data/500/10126arrestor-med.jpg

[Richard L]

YOU DO GREAT WORK.

Could you explain th edifference betwen 50% and 100% filled with water?

I tend to use the smallest arrestor possible to do the job, in case a small leak on the valve, residual pressurised water left in the accumulator may leak in the engine overnight can cause hydo lock.

A sursge arrestor is essential for a good system.

Richard

[simple]

Richard,
Re 50% water filled, I meant ? surge container filled with water only 50% and the rest is trapped air which does not escape during operation and acts as pressure absorber.
100% water filed is when container filled with water only.

According to results you mast have air in container to efficiently absorb pressure spikes, otherwise container capacity has to be greatly increased. Just got to remember that air will dissolve in water. So container will have to be regularly checked. Or alternatively proper surge arrestor mast be used which incorporates membrane which separates water from air inside container.

The water/air system good side:
Very efficient pressure spikes suppression
System pressurized all the time (ready for instant use without delay for pump start up)

Bad side:
A lot of water will be discharged in case of valve failure
Slow pressure build up on system start-up after initial water priming

The problem is that if water/air container used (which is very efficient) it will discharge a lot of water in to the engine if valve fails (your earlier comment!).
If water only container used it will discharge only very small amount of water but it is not efficient for surge suppression.
So ideally you need water/air container with bleed valve which discharges system pressure back in to the main water storage tank when engine is turned off after operation to prevent hydro lock?

[Richard L]

Thank you for for clarifying the accumulator.

I belive you have include a 22cc accumulator on your order already. It should be with you anytime now. Those accumulators are air charged, see diagram:

http://www.aquamist.co.uk/forum/806-408.gif


We have always install an accumulator on Shurflo pump based systems: Here is our standard set up:

http://www.aquamist.co.uk/forum/shurflo45.jpg


Charging up the accumulator is only take a matter of seconds, especially the small 22cc one. The think the advantage far outweighes the disadvanages.


I wonder if you do a some test for me as you already have set up the shurflo and power supply, I have been asked to verify if a small nozzle (such as the one you are using) will response to a full scale voltage/current change, ie 0-12VDC or 0-100% chopped 12VDC (PWM). I wonder if you could do a plot of voltage vs flow and current vs flow on the same nozzle.

Please tell me what cam angle are you using on this shurflo pump?

Richard

[simple]

Hello Richard,

Re: test, I can test as per your instructions ? no problem.
However I did not order accumulator with you when ordered goods. As far as I know all what coming is:
1) 3 x 0.3 nozzles (threaded type) 4mm
2) 1 x 0.6 nozzle thereded type 806-309
3) 2 x Filter: 6mm hose. #806-257
4) 2 x Flow valve (not HSV): part number 806-234
5) 2 x Tank fitting: #806-270
6) 2 x HSV (806-244)
7) 4 x Nozzle adaptor with blanking plug 806-357
Please clarify!

I do not use Shurflo pump, instead I run China made (same as Shurflo) copy of Shurflo. Therefore operating parameters as well as results can be slightly different.
I can test 0.3 or 0.6 nozzle for you on my pump with or without accumulator for water delivery and voltage/current ratio when I will receive the goods. Please remember that I feed pump via Switch Mode Power Supply (SMPS) not PWM during my test. However I do not think PWM will perform any different to SMPS (according to my knowledge).

Please explain more detailed meaning of ?cam angle of the pump?.

Your set up as per posted picture of the pump and accumulator look very good as you have accumulator as close to the pump as possible using maximum diameter fitting.

[Richard L]

I will send you an accumulator if you like - in return or the test you are performing for me.

If you can stick to the same nozzle as your used before, it will be fine.
Shurflo as three cam profiles, cam angle dictates the flow and pressure. 2 deg, 2.5 and 3 deg.

If you detatch the motor from the pump head (three bolts), the cam angle is cast on the pump head part. This is very impotant as using the wrong motor may cause premature thermal cutout sensed by the internal thermal switch. All these information is based on the original shurflo specification, not sure if this applies to the copy.

Switch mode PSU is great, if you could test in 1 volt increments, it will be great, If you also record the current drain, you only need to do the test once. All I need is the pressure, I can calculate the flow from pressure.

Thanks.

Richard


Richard

[simple]

Richard,
Test is easy; I will get it done for you.
My current test was based on ceramic nozzle (0.3) flowing about 400cc @150PSI (as per graph in earlier post).
As for pump, this is the picture of internals after aggressive test we performed using acid:

http://www.audizine.com/gallery/data/500/10126pump-med.jpg

Vendor incorrectly stated that pump suitable for acidic environment, so ? we tested it in solution of HCL. Pump lasted about 10 minutes. I have not seen any indication of cam angle, but can tell you that pump run cool and barely warms up during the test. So far I did not manage to get it over 30-40 degree C. I will search for cam ange one more.

Pump specs 5.5L/160PSI (12VDC/10.6Amp)

Please let me know if hardware is sutable to perform the test you want....

:D

[Richard L]

WOW, specular !

This pump is not a copy of the Shurflo pump but a copy of a flojet pump. Shurflo has a three chambers and flojet has four. This is a flojet copy.

The cam angle is normally on the face of the head, between the motor and the light tan part.

Could you take a picture of the other side of the light tan part?

I would still like to test to go ahead if possible.

Richard

[simple]

Richard,
All parts just come in - thank you. Can you email me copy of origional invoice as i need to brake down the cost?
I will follow up with vendor about cam angle.
I will do the test for you as soon as i will get accomulator, unless you want me to test it with my home-made accomulator :D

[Richard L]

The test has to be done without an accumulator.

Your invoice should follow a few days on. We always send it separately.

Richard

[simple]

Sorry Richard, my English is pretty bad and I misunderstood your offer.
You offering to send me an accumulator as a gift for a test I will perform for you, am I correct?
I will do the test as per your request ether way. I found you been helpful and will return the favor.
:D

Please check test summary:
Current pump I have (160PSI max)
0.3 Aquamist nozzle or my original ceramic nozzle?
No surge arrestor.


Data needed:
Relationship- Voltage VS Amperage VS Flow/m

[Richard L]

Yes, an accumulation for your work. Advice is free for all, only on topics I know.

Please use the ceramic nozzle. This is because I have read all the previous test already so I have the basics.

your summary is absolutly correct:
Please check test summary:
- Current pump I have (160PSI max)
- my original ceramic nozzle
- No surge arrestor.
- pressure vs volt/current
- in one volt steps will be fine.

[simple]

Richard,

This is how I set all up:

http://www.audizine.com/gallery/data/500/10126test_set_up-med.jpg

Those are results:


Volt Amps.....PSI.........Water delivery cc

1.5.....3.6......40-50..........180
2........5.7.....50-65...........220
2.6.....6.8.....70-90...........280
3........7.5.....80-100.........300
4........8.3.....100-130.......335
4.5.....8.3......120-150.......350
5........9.2......125-160.......370
5.5.....9 ........135-170.......380
6........9 .......140-170........390
7........9 ........140-180.......390

System variables:
Voltage accuracy 0.1V
Amperage accuracy 0.1A
Pressure accuracy (non-calibrated gauge, suspect 10%)
SMPS used ? High power 70Amp/24VDC supply
Water capacity measurement accuracy 10cc
Water loss thry mist expected 1-2%
Nozle used 0.3 ceramic type

Comments:
Over all I am not happy with the test and will perform second test in 3-4 days to confirm all figures. My problem is that during low voltage range 1-2VDC pump was producing non-repetitive results and figures used to ?float? if I repeat the test.
Second problem is that at almost full pressure 6-7VDC there are no increase in pressure instead there is increase in pressure variation!

I will perform more accurate test with repetitive data collection and will do few changes in hydraulics system to double check results. As at this stage this is all I got.

[simple]

Richard,
We are now looking on driver circuitry for HSV. It is way more involved than I have expected. I was advised to have it done with current sensing feedback and controllable feed not to mention special discharge circuit which create ?dead short? across the HSV coil to reduce 2mS delay. In order for us to do it properly I need to know "hold" voltage for HSV! We plan to ?spike? 12-13VDC and then reduce the voltage for ?hold? time. What voltage is recommended for ?hold?? Or it is matter of testing?

[Richard L]

Thank you very much for the test performed. There is a discussion on the NASIOC forum about the effectiveness of a "pump speed controlled" WI system. Your data has helped a great deal on clarifying a few mysteries.

Driving the HSV:

We do have two "peak and hold" driver for the HSV, 12V and 24V. The 12V version will not speed up the cycle time but will reduce the current to 1/4 of the origuinal value -about 250mA from 1A @12V

The 24V driver will double the cycle time by about 100%, valve take 1ms to open. But there are no 24V supply on the car unless to add a voltage doubler to boost the 12V to 24V.

http://www.aquamist.co.uk/forum/PAH-HSV.jpg

[simple]

Richard,
Re test ? you welcome. I will do few more to double-check. If you will do the same test at some stage it will be interesting to compare.

Re HSV ? thanks for the info about the driver. I will keep it in mind.

I shall comeback to the drawing board and test bench. Progress is slow as I doing these off-work hours. And there is a lot to be tested and calculated. I seem to like the idea of varying pump pressure, I may use it in conjunction with HSV. I am now waiting for samples of pressure transducers and flow meters.
More I look in to it more I realize I am trying to copy fuel injection system?
:roll:

Just found this NASIOC forum. This is a HOT topick you run Richard :cool:
The general public need to sit down with pencil and do some calculations - they will eventially see what you are trying to tell. I guess i went thru the same process :oops:

[simple]

Well I got the chance to do some more tests.

http://www.audizine.com/gallery/data/500/10126New_test_flow_vs_volts-med.JPG

I am happy with the results as they much more stable and can be reproduced time and time again. I have altered hydraulics set up by relocating pressure gauge from pump to nozzle.
Top graph is a relation of voltage VS flow on setup as per Richard request.
Bottom graph is the same setup but with added accumulator in the system.
The numbers speak for themselves.

Richard I have tested 0.3 Aquamist nozzle, it flow 112cc @ 162PSI (pump with accumulator) It looks like thad by the end of the tests I will be ordering another 1-2 (0.3) nozzles from you with ?mounting sits?. As my hydraulics and motor dynamics get put together few adjustments are obvious.
By the way this is how intake track looks on my car:

http://www.audizine.com/gallery/data/500/10126intake-med.jpg

The runners are over 300 mm long and I plan to fit one nozzle per runner just in the beginning of it. This way water will get to travel a bit before it hits valves.
Unfortunately AUDI engineers have designed the intake in such a way that the flow is violent with 120 degree turns ( towards cylinder1-2) which makes one nozzle injection useless. So the installation of ?cooling? nozzle before the throttle body is questionable as all not evaporated (and large) droplets will end up in third cylinder.

Comments anyone?
:roll:

[Richard L]

Thank you very much for performing test. May I have a copy of the excel file as I like to converted it to duty cycle % by dividing all the V range by 12 and mulitpluy it by 10%. After I have done that, I will post it to the NASIOC board. I have just finished building a simple PWM controller today, so I can repeat the same test here using the original shurflo pump. The results are good, Pity that you can use an accumulator on a PWM pump speed WI system. It appeasr that you are now able to control the water pressure to compensate the manifold pressure change instaed of compensates it by corrected water map on the EM.

It sounds pretty good on the 0.3mm, we rate it at 745cc/min at 7 bar without surge arrestor - the aquamist pump is only a two-stroke pump with a great deal of peaks.

On order to get a steady gauge reading, you can put a small ss set screws inside the hose to stop the flicker on the gauge.

You're right about the intake tract. You have to use port injection. It won't hurt by putting a small jet just after the IC, it will allow the a bit more cooling along the inlet track. Droplets will be small by the time it reaches 120deg. bend.

Richard

[simple]

Richard, E-mail sent. attached is a copy of Exel file. Good luck with NASIOC board.
Please post results of your test, I wonder how much different will they be.

[Richard L]

Richard, E-mail sent. attached is a copy of Exel file. Good luck with NASIOC board.
Please post results of your test, I wonder how much different will they be.

Thank you for the excel file, I managed to extract some good information and some inetersting findings.

It appeared that the non-linear-ness of the pressure/flow ramp again power is limited to the power of the water pump. after 3-4V ish, the pressure/flow begins to flatten out, it appeared that if a more powerful pump is used, you will get a better performace, more predictable and more linear with each addition volt fed into the pump.

I could be wrong, but not able to pin point why the drop in flowrate.

http://www.aquamist.co.uk/forum/pump-power.gif

If the power of the pump can be used to product pressure rather than flow, you might have a cmpletely different flow curve. The Shurflo pump is designed for high flow with moderate pressure, you really need to look for a low flow gear pump so that you get sustainable pressure ramp. Altenatively, get a more powerful motor?

Richard

[simple]

Just a speculation here, from what I know the only pump that will have linear output is turbine/centrifugal type. From what I see plunger/piston/positive displacement pumps need torque to operate and it is not proportional to output. This is part of the pump design. Stronger motor will have the same curve but moved to the left of the graph and less powerful motor will shift it to the right ?.

Just my limited knowledge speculation?

I have recived the accomulator you sent me - thank you. I may do few test later and see how it performs in comparison to my air filled accomulator.
By the way, your products seem to have nice finish and good quality. I like it! Bit expencive for mass production thou.

[simple]

Where data-logging my ECU a lot in the last few days. Now have a 3-4 weeks worth of data to analyse. The whole think is really advanced staff ? most of the data is beyond my knowledge. There are complicated relationships between air temperature, AFR, Ignition advance, Ignition correction factor and so on?..
Anyway most important for me at this stage this data:

http://www.aquamist.co.uk/forum/cfvssvba6.jpg

CF = Ignition Correction Factor. Measured in degrees of timing pulled back (activated by knock sensor activity)
V = knock sensor voltage (looks like amplified by factor of 10)

Graph is 4-th gear pull from roll. Ambient temperature 20C, low humidity, sea-level, 98RON. ECU been fully adopted.

Based on this data we will try to extract formula of CF activation. If successful we will be able to activate WI just before ECU pulls timing under the load.

I also have mounted knock sensor on the engine head the other day and watched it?s performance on oscilloscope while running the motor. It?s mirrored the graph in output voltage activity with it rising as RPM rising.

[Richard L]

I wonder you can resize the image to about 1000 pixels across, it may be too big for most of pur screen size.

Interesting plt. I wonder how much of the knoock signal contains backgroubd noise, it is normal for the engine to make more noise at higher RPM.

I wonder if you can log a trace wthout load and compare it against the fiull load run, The difference will be a bit more representitive of the knock. I am expecting to see more know at maximum torque region rather increase with RPM.

The human ear is excellent for detecting knock, may be using a stethescope to confirm the actual knock activitity charted.

Richard

[simple]

Sorry for the large plot. I will resize next time!
Richard, I found, after looking at a lot of different data logs of different people (same model car) that our engines do not seem to have the most knock during the max torque power band. This can be due to ECU been properly programmed and ignition timing been equally aggressive allowing only ?minor knock? thru out whole RPM range.
Setting timing to more aggressive setting will indicate knock thru out whole RPM range unless better fuel used. It looks like ECU drives engines accurately right on the maximum timing all the time and rellies on knock sensor to stay tuned. There are only minor indications of increased CF (correction Factor) during max torque. Max torque on the stock engine is in the range 2500-3500RPM.

As you suggested my next step would be to data log ?sensor noise? I will drive the motor thru whole RPM range with only 5-10% load. Then we will compare ?background noise? graph with ?full load? graph.

It looks like on 2001-2006 model AUDI engines to have WI properly operating you need to inject water thru whole RPM range if your timing set to more aggressive tune.

[Richard L]

Do you mind if I resize it for you? theris no problem.

The plot is interesting. Do you have the tool to re-program the ECU at will?

It will be an intersting plot at 5-10% load, although the noise will be smaller then full load, with or without knock.

Richard

[simple]

OK, done more test.
Graph is 4 test. Two are ?idle? done spinning the motor thru rev range without any load at all. Other two ?low load? are second gear very gentle pull thru the rev range.

http://img129.imageshack.us/img129/6344/multytestmp0.jpg

Richard please resize any of my posts is necessary.
According to the data collected ?V? which is knock sensor voltage rises predictable. Rise factor can be calculated and depends on load (0-100% load). As the load increased, voltage output decreased proportionally. For the actual knock we need to look for ?spikes? in voltage output rather than voltage itself.
BTDC is ignition readout.
Test 1 ? low load
Test 2 ? idle
Test 3 ? low load
Test 4 ? idle
All test done 5-10 minutes apart.
Richard I have fully stock ECU and have no means to program it, only read.
If I will upgrade ECU software to say REVO (Europe tuner of AUDI) software I will have limited programming access, such as timing, boost, fuel. At this stage no tuner is interested in what I do therefore I cannot get program uploaded to play with. What make it worse is there are no local tuners in Australia.

[simple]

Done few more tests.
Here is the picture of HSV running 9Hz. He is one happy valve clicking his way thru?

http://img111.imageshack.us/img111/3711/img2560hp3.jpg

This picture is a PWM pump controller test. To run WI continuously (over 3 minutes) you need a fair size heat sink. There is a lot of heat generating by Mosfet trying to handle 12VDC @ 10 Amp. We are looking how to unload Mosfet to run it cooler now.

http://img76.imageshack.us/img76/4331/img2559dh3.jpg

Next step is in car text of knock sensor to see if we can pinpoint knocking?.