T-uk

I'm not sure if a Dawes@18psi+top mount on a track day will be safe as the boost is held higher and longer,even @15psi with a Dawes I think the t/m will be struggling with temps after 5laps.this is why I am thinking of either a larger scoop on the bonnet or a fmic.

john banks

Another option is to check the intake temperatures and set the boost accordingly - only then you will know if you need to spend £££ to keep your desired boost level.

I wouldn't run 18 at K/hill either unless I knew they were OK.

CC showed me his intake temperature thermocouple - loosen the sleeve over the intercooler to throttle connection and slip it in. Looks easy enough.

T-uk

john,

I put a totally standard MY98 round K/hill and after 5laps the car was not pulling up the straights.we were told this was the ECU retarding due to the temps getting too high,possibly due to the standard exhaust.for the next track day,we fitted a mongoose fully decatted system and after 10laps the car was still pulling well up the straights.this was running standard boost and we know how the standard boost controlled cars tail off boost higher up.with a Dawes,even set low for the mid-range I think intake temps will go through the roof.

I hate to mention names when discussing problems,but you mention Dave,he has had problems and a tough learning curve.I would rather let Dave give you the details but lets just say he has taken things to the max to find out the "breaking point".I do not think he had access to a select monitor at the time and iirc it was put down to a faulty injector,but I would be very surprised if he runs high boost on track.

T-uk

I think the MY98 did have a slightly smaller i/c than later cars,but they also ran less boost did they not.I just used it as an example of my own findings.

"I wouldn't plan to do more than 5 laps anyway" ,
easy and sensible to say,but when you are out getting the best sport of the day with an equally matched car and driver you soon think just one more,5laps later you will still be saying it .

"How many laps do you usually do? Does your car drop off in performance?",
I try and do 1warm up,5hard and 1cool down.I felt my car drop off once.I had been out (as above)and was on my cool down lap just about to pit when I saw Dave's car coming round the hairpin and thought,I wonder how fast his car is,let him past and set off after him and about 5laps later I was still sitting on his bumper.I think both our cars got tired out that day .

"Why not buy a track car",
whats the point of having a car that does 150mph with a rally records like scoobies to go to the shops with.I could buy a 1.1Saxo to do that.a track is the only place to safely see what these cars can do.I bet you 100£ right now that after you do your first you get bitten by the bug.tell him Sam .

if I buy a fmic it is for safety and so that I can run the same boost settings on track as on the road,not to run more boost to get more power.I am not what you would class power hungry.I would rather have a reliable 260bhp than a gear box eating 320bhp.

john banks

260bhp is about what I will be running at the moment on the road at 1.2bar since I tightened the actuator a bit - now that is a thought - if I run with a Dawes on track I would be happy (in the absence of more measurements) to run 17 PSI midrange and loosen off the actuator again so my boost dropped like it did at the RR. I would basically be running like a PPP with only a bit more up top and at the bottom end, but a decat downpipe to hopefully offset the EGTs? I have started a Scotland thread specifically about this at K/hill to get more opinions and avoid cluttering this thread any more.

Sam Elassar

went out on saturday with derek K, done about 100miles of the twistist B roads in tayside, great fun . he had a thermocouple meassuring his intake temps and he did not get anymore than 35c peak on the road, although we were driving very very hard.
he has got link, 1.2bar and just fitted his decatted exhaust+down pipe last week.

on track the temps will be completely different. why don't you make your self a temp probe, it is much easier than all the stuff you are doing ! on the road the standard intercooler is totally adequate up to 1.3-1.4bar even. but on track it is a different matter all together. i saw 60-75 degrees with waterspray. which did not really help that much i thought.

plus more than 7-8 laps and you will start destroying tyres and pads etc... saying that i used to stay out for 20-30mins at the time, but i just wanted to race everyone!!!!!!!

i do think Dave car it fits the profile ! though.

dowser

Like John, I'm thinking about how the OE ecu with Dawes and FCD will compare to the PPP with just the Dawes. So will be testing with mine, thanks

Re. the track - last summer I did 700km's around Nurburgring over 2 days (with a/c on ) on my decat my00 PPP. No problems, and no drop off in performance - ate a set of tyres and brake pads though.....

As John says, I think the important thing is to not be running too high boost at high revs - by 6k rpm the PPP is running under 1 bar as standard, this seems best given most of the time on track you're at WOT.

Regards
Richard

Paul_H

Boards posted today to Steve, Richard, Matt & Iain

mutant_matt

Cheers Paul.

john banks

Received thanks Paul. I will send you a cheque tonight (very nicely put together - someone - possibly me - will want it) and post results later using 10 bit ADC logging of input and output and graph it. The question I really want answered is how it compares to a PPP!

T-uk

john,

if you want to borrow my black MY99 ECU to try it with the FCD,I could give it to you on Sunday.my Knocklink and lambda are not here yet,but if you want the black ECU to test it with the PPP and green ECU's you can get it.

john banks

Thanks - I'll test it on the bench first and let you know. I'll still have more wheels than ECUs just

dowser

Got mine too, thanks Paul.

John, be sure to post your findings

Cheers
Richard

john banks

I would like to be posting my findings now, but they will be delayed due to technical reasons. I need to wait for some parts from RS to complete the job.

I intend to post 10 bit resolution 2 channel ADC logging to show the response between input and output. You'll be the first to know!

Paul_H

Please remember that this is a work in progress. Suggest you wait for the results before fitting it.

mutant_matt

Paul,

I know you've worked out the MAP output voltages from testing the FCD, but has anyone got or know where to look to get the actual specifications (read output volatages) of the MAP sensor?

I'd just like a backup of the figures as Pauls figs are dependant on the accuracy of his Boost gauge (no offence intended Paul )

Cheers,

Matt.

john banks

Bob Rawle quoted 0.88 to 4.84V for -1.0 to + 1.8bar. This tallies with what I have measured reasonably closely and is similar to Paul's figures.

I would personally want to put the defend point up to 1 PSI before fuel cut and then put the fuel cut up to 2 PSI above where you run the held boost in 5th gear. Subject to making sure fuelling and advance are OK of course.

In testing the FCD I used an (un)regulated "12V AC" power supply which usually puts out 12-14V after bridge rectifier and capacitor (so like a car battery) with previous circuits I have used. I had no idea it would give 18V DC and fry the most expensive chip on the board. So the £30 FCD has ended up £50 to order a new chip from RS and VAT and postage. Still it will be worth it as it is well under half the price of a commercial item and better.

A certain amplifier manufacturer quotes "Tested to destruction".

It is only the second chip I have blown up this year, and all on the bench whilst testing

I now know a bit more about power supplies

I won't tell you what the power supply was intended for, just that it came from Ikea! Shame on me, head hung low, should be banished to muppets forever.

But if you don't experiment you don't learn or innovate I suppose. Last year I could make a flashing LED circuit. Now I make microcontroller circuits and code that works, so a few blown chips are expected.

Perhaps we should start a thread - DIY botched nightmares...

I share this story to illustrate what can go wrong if you don't know what you are doing and even if you do or think you do. Test everything fully and be prepared to accept the consequences if you don't.

Paul has already tested this on his car and it works.

I was just trying to be clever and produce some graphs for you all

I am not trying to put anyone off if they are confident they know what they are doing with this.

dowser

John - you could nearly have bought a 'proper' PSU for the price of that chip

Richard

Paul_H

Now now, Richard - don't be naughty

I don't think the same situation is likely to occur on the car...i.e. 18V.....however I would like a bit more of a safety margin to protect against overvoltage. I asked Kostas for advice, this is his reply:

Paul the circuit is built to work at car level voltages. There is no need for any protection because in a car, whatever the overvoltage may be the batery itself takes care. If there is a destructive situation, then the ECU of the car will go too and this is a much worst situation than the FCD. The reference input voltage is regulated and this is important for proper operation. Any other regulation would just comsume power without any reason.
A small resistor of 10 ohms in series with the input voltage and a 12V zener after the resistor to ground can help but really its not required.

Regards, Kosta

So Kostas thinks that we don't need to worry, he designed the circuit so he should know!

Personally, I would like to hedge my bets and provide some protection against overvoltage, even if it's a little unnecessary. I'm going to add this to the circuit:



I haven't got the parts yet, and so haven't tested it, but the diagram is per manufacturers data sheet.

Parts
LM7812CT +12V fixed @ 1A RS stock no 648-438 £0.40
1N4002 rectifier diode RS stock no 261-154 £0.06
0.22uF electrolytic capacitor 25V RS stock no 267-5195 or 116-969 £0.07
0.1uF ceramic disc capacitor RS stock no 829-615 £0.08

RS have some minimum quantities, I think Maplin will sell single items.

My FCD draws 40 - 50mA only, so I doubt that a heatsink will be required.

The purpose is to protect the FCD, and thereby the ECU, against overvoltage, e.g. say from a faulty alternator. It probably isn't strictly necessary, but as it adds less than £1 to the cost I'm going to try it.

This does seem to taking some time doesn't it. When John's results are posted, hopefully this will confirm that we will have a truly universal fuel cut lifter.

john banks

Not good news I am afraid.

Received a new sample and hold amplifier and regulated 12V DC supply. Connected it in the car and the output followed the input by 90mV but there was no defending despite setting the pots appropriately. I connected the board inside the house and logged the data. Again there was no defending. I found by tracking the inputs and outputs of the gates that the second NAND gate chip was not functioning correctly - output was always high. The device had been plugged in for about an hour, but then again I noticed the DC-DC converter becoming very warm. A moment later there was a spark and a plume of smoke rose from the DC-DC converter. It no longer produces -12V so I cannot test anything else. Nothing had been shorted and the supply voltage is rock steady at 12.0V.

I really don't know why this has happened. Everything seemed to work perfectly and then it went pop for no apparent reason. I wonder if the faulty NAND gate had been drawing too much current?

Whatever, I can't keep affording £20 a pop from RS and hours wasted trying to get it to work.

I'm sure it was my fault for popping it in the first place and then there was an occult short in one of the chips?

Sorry to disappoint you guys, but I have to bow out here!

Paul I'll send you the bits back, please use anything you can salvage.

Sorry again, I have to admit defeat on this one.

However, I suspect that if I am having this sort of trouble others will too. If you are confident go for it, but I can't make this work.

The +-12V thing for processing a <5V signal worries me slightly, and it is so easy to get it wrong.

I have found microcontrollers on the other hand very easy to work with using a 5V supply.

If it works on your car Paul I would go for it and anyone else, but I have to be honest and admit my findings.

Paul_H

John

If you couldn't get it to defend then you've been testing a faulty circuit. It defended for me because I adjusted it prior to posting it to you. Sounds like you fried more than the sample & hold amplifier with your power supply.

Mine draws about 48mA only.

But I agree that you have to post your findings as you see them, and thanks for trying.

[Edited by Paul_H - 2/18/2002 10:46:35 PM]

john banks

This is what I plotted in blue. In red is what I would have expected to happen. The scale is 10bit 5V ADC - so 1023 is 5V.

At this stage I think the only fault on the circuit was that the sample and hold amp was not switching to hold mode at the defend point. I presume the response otherwise is what would have happened - note the increasing error after fuel cut as Paul pointed out but this is not an issue in practice.

Also I wonder if variable resistors for matching the input and outputs of the two buffer op amps would be a good idea - I did find offsets here - presumably to get a gain of one the feedback resistors have to be matched?

I think you are right. From testing it seemed like all the inputs and outputs of the op amps and the one of the NAND gate chips were behaving correctly. The sample and hold amp was new and seemed also to be functioning correctly. The only assumption I can come up with is the previously damaged NAND gate overloaded the DC-DC converter. Perhaps it is only the DC-DC converter and the NAND gate that is gone now.

If it was my own design I would have more of an idea how to proceed. Working with someone else's design or code is difficult. I can see that it is using two op amps/comparators to switch when each trigger point is reached and then the array of NAND gates produces logic to switch to hold mode when in between the trigger points. The other op amps look as though they are used like buffers.

I've already sent you a cheque Paul - thanks for all your help on this.

Another thought - the NMA1212 is only rated for +-40mA. I think this is OK given what you said about 48mA overall current drain - obviously something was drawing more on my circuit for the heat it was producing. Oddly just before the NMA1212 smoked the sample and hold amp started "drifting" like it did before What have I done to this beautiful circuit board you send me Paul!

I may also think about another approach to designing this with a 5V supply - probably using a uC. I tend not to blow them up for some reason well not very often anyway - version 15 of the code and the 5th version of the circuit and I've blown up one

[Edited by john banks - 2/18/2002 11:18:43 PM]

Paul_H

John

This has always been a work in progress, so some ups and downs were to be expected - no problem. Although when I fry chips, I prefer them to be made out of potatoes

I decided to get to the bottom of this, so I set up a PP3 battery and pot, with the intention of recording input & output and producing a chart of the results, as John had intended to do. It confirms what I had learnt through watching the voltmeters and using it on the road - but adds a twist. Here it is:



It shows the output tracking input, with a small glitch where output jumps to the defend point slightly ahead of input - followed by the defend zone (horizontal line) and a new fuel cut where the output rises. At the top end of the scale there is increasing error, but this is after fuel cut.

This is what I had experienced on my car - it works on my car.

However there is a BUT.

During this testing I have discovered a behaviour of this circuit that I had not previously noticed:

When I held the FCD in the defend zone for a sustained period of time, using the PP3 battery, I noticed the ouput voltage rising very slowly. There seems to be a leak in the circuit - the sample & hold amplifier cannot accurately hold for a sustained period of time.

So for example, when input is say 4.1V, output is 3.7V (defending) but very slowly rises so that eventually input & output are the same i.e. no defending. It takes over 5 minutes for this to happen.

I did not notice this when I road tested it, it provided an effective fuel cut lift. Arguably this bug would not show up on the road. However it definitely would show up if you were for example using a rolling road to map an ECU or similar.

For this reason regrettably I think that the circuit is not really good enough - I want something that clamps the output more effectively.

So I think that this particular project is drawing to a conclusion, and not the one I wanted. It does work on the road as I have demonstrated, but has enough potential problems for me regrettably to decide not to use it.

Thanks go to John for his help and willingness to experiment. I now know a lot more about fuel cut defenders, anti boost chips, whatever you want to call them, and intend to carry on the research and try for another solution. John do you fancy working together on this?

dowser

Paul

Is it a linear rise over the 5 minutes? I don't think I'd ever want my car held at the boost necessary to have the FCD active for that long anyway.....I'd consider this feature an active benefit...so send me your assembled one too!

Re. the rolling road - are you saying that they keep your car at up over 1 bar of boost for 4 minutes or more? I find that scary

Richard

Paul_H

Richard

Yes it is linear. I watched the meter rise in 0.01V increments, and timed it on my watch - it took 5 mins 11 seconds before output & input were equal. But it might take less time to reach the fuel cut level for your ECU - depends where it's set.

I hadn't thought of it as a safety feature - you would make a good salesman!

Probably it would be OK on the road because it is surprising how just a little feathering of the throttle drops the output of the MAP sensor. Also surprising is how with a flowing driving style you can make rapid progress with surprisingly little boost!

There seem just too many potential problems when you stand back and look at it coldly, which is a little difficult to do when you've got your head down working on it.... It's been modified by (1) bleeder resistor on input/resistor & zener on output to prevent overvoltage to ECU (2) It may well need a regulator on the supply, because the chips are rated at 15V max, which is a little too close for comfort when the alternator puts out 13.5 - 14V. John has already demonstrated that too high a voltage WILL fry the chips, and if this happened on the car it would damage the ECU. It may be safer to go for a 5V circuit. (3) There is a bug described in my last post which prevents the output being totally clamped at defend point. Add it all up and what have you got?

john banks

I am wondering about a microcontroller solution. I am just no good with these analogue circuits - they get too complicated too quickly - I think I was just dogged with bad luck with this one - innocent mistake then it caused it to go a second time.

I don't think there would be an issue on the road, but there are offsets and peaks in the response which I don't like.

I think these things are best keeping the component count down and everything physically VERY simple then it minimises the chance of me blowing things up. Doing it with code is enticing. I'll try and come up with something later.

john banks

IDEA:

Use an analog switch such as ADG736 £1.67 from Farnell.
http://www.analog.com/productSelection/pdf/ADG736_0.pdf

This would be instead of the dual rail sample and hold amplifier. It is basically like a solid state relay which would connect either the real MAP sensor reading to the ECU, or in a hysteresis zone send a lower signal. This could be done with an 8 pin package the MAX931 which also runs off 5V and needs only a couple of external resistors to setup. http://pdfserv.maxim-ic.com/arpdf/MAX931-MAX934.pdf £1.81+VAT from RS

The 5V supply could be taken from the TPS feed which is how the commercial FCDs seem to do it.

This way there is no way it can send outside the range 0 to 5 V to the ECU and there should be no distortion of the MAP signal as it passes unaltered outside the defend zone with no distortion from buffers and need for matched resistors. Need to flesh out the details obviously.

I think we'll need to look at chip selection a bit for the right packages (ie not SMT). The MAX933 makes a good "window detector" which is what we want to activate the switch. The problem is input range though - within 1.3V of +V it doesn't go! There are loads of window detector circuits about - even a variant on the original design with the two op amp comparators, but attach it to an analog switch rather than the sample and hold amp with all its problems. If the window is detected the analog switch could connect the lower threshold (defend point) to the MAP input on the ECU.

Comments?


[Edited by john banks - 2/19/2002 10:43:08 PM]

Paul_H

Trying to get my brain around it - have printed the datasheets off.

Assume the switches would not trigger a check engine light?
5V has got to be the way to go for safety, but may need a separate reference supply -easy- as we don't know what the ECU can safely supply. What about a new fuel cut?

Keep going!

john banks

The 5V supply for TPS is for the engine sensors, and is used by Superchips and the HKS FCD AFAIK. We'll need to check. But if not a 7805 could be used with one capacitor to make 5V reg.

I think I've got the basis for a new design. Using the 49p LM339 quad comparator in a 14 pin DIL package we can make a window comparator by tying the open collector outputs together and using a pull up resistor. So one chip, two variables, one pull up resistor and an analog switch (need to find the right package with SPDT single rail supply) replaces the whole lot and gives raised cut out.

"Consider the conventional 'window comparator' bridge configuration below. As long as resistor R5 is within +- 50 ohms
of resistor R4, the circuit is 'balanced'. In other words, we are within the window's do-nothing detection range, and the Led
indicator is OFF. Now, let's change the resistance of R5 to 1051 ohms (actually any value above 1050 ohms for this test).
If we measure the voltage difference between pins 6 and 7 of U1B, we would see that pin 6 is now slightly more positive
than pin 7. Ideally, pin 1 will 'snap' in the direction of the common ground. One of the window thresholds has been
exceeded, and the Led will be ON.
Now change the R5 value to 949 ohms (or any value below 950 ohms). Notice that pin 6 is now
less positive than pin 7. This will cause comparator U1B to 'release' it's hold on the Led, but now the voltage difference
between pins 4 and 5 on U1A has changed, and pin 4 is now more positive than pin 5. Because of U1A's comparator
action, pin 2 of U1A 'snaps' to ground, and once again the window threshold has been exceeded (in a different direction)
and the Led is ON. To tie this to uP 'logical' terms, we say that:
As long as: (the voltage at pin 7 is more positive than pin 6 AND the voltage at pin 5 is more positive
than pin 4)
THEN we are in a 'dead band' (within the assigned limits of the window), and the Led if OFF.
This illustration uses two of the LM339 quad comparators."



"A window comparator is useful in particle spectroscopy where we want to count only voltage pulses whose peak height lies within a narrow range. Drive it with a sine wave and demonstrate how the output triggers only when the sine wave is within a window of voltage. Since the 339 outputs are open collector, it is possible to wire them together. You can't do this with conventional op amps ! You may need to bypass the voltage divider resistor with, say, 100 pF to avoid oscillations."



The analog SPDT switch will be fed from the output- note the second diagram has a NOT gate - we can get around that. When in the window the SPDT will connect the lower reference to the ECU. When outside the window the SPDT will pass the MAP signal unaltered.

Should be very crisp response, very few cheap components and 5V supply.

[Edited by john banks - 2/20/2002 8:29:09 AM]

dowser

Looks good John - I can't comment on the comparator circuit, too many years since I was in school

But I wonder whether the switch's 'break-before-make' operation will confuse the ecu? Easy enough to test, I guess.

Richard