john banks

It is VERY quick - we are talking nano seconds. I don't think the ECU will even know it had its pants pulled down

dowser

I've just refreshed my electonics 'cache' - I think this'll work Just a matter of setting the 'window' to the right size and making sure the output swings quickly and completely to operate the switch.

You gonna' try it? I'm going to order the bits and have a fiddle, but knowing me it'll take a while, lol! Where can I get vero board from on the internet?

Richard

john banks

I use tripad board - saves a lot of cutting.

I think the analog switch I showed was surface mount too. Need a SPDT analog switch in a DIL package.

The LM339 looks to be ideal for the window comparator.

john banks

Looks like input range for the 339 is up to about 1V below Vcc - problem with 5V supply. Could use 12V - the output of the comparator drives a switch not the ECU directly.

If I could find a DIL packaged dual or quad comparator with single supply, rail to rail inputs and open collector outputs we would be in business very easily - but struggling to find such a beast.

Same with analog switch - need SPDT off single supply in DIL package.

Any ideas?

Paul_H

What we are trying to do is considerably more complex than the commercial FCD's! I found an excel spreadsheet analysis of the HKS version

http://www.mr2.com/ARTICLE/HKSFCDat.html

which shows that it is a lifter but looks like it is based on some sort of voltage divider, which gives an error right across the range. There is also a guy selling FCD's for £15 which consists of a zener diode.

john banks

What about this then? Should cost less than £2 and theoretically work perfectly?

dowser

I can't find anything either - I've asked Analog Devices tech support.

Richard

john banks

There is a MAXIM chip but not available easily in UK from Farnell or RS.

dowser

MAX394? Could've saved me some time and given the part#

Richard

john banks

Couldn't remember it sorry!

I also wondered if we could turn on the power to an op amp that would voltage clamp it with a zener in the defend zone. Would this work?

dowser

Yeah - a zener would do it, but it's a matter of finding the right value. I can't even remember what voltage window we want to defend at?

I prefer the chip based switch solution, if only 'cos you can then modify your defend voltage easily - I'm going to try and get some sample's through my company, wonder how many they'll let me have?!

Richard

john banks

Back home now - found the one I had the datasheet for - MAX4544. 8 pin DIL package - single SPDT single supply analog switch. Would be ideal - but how to get it!?

Is it not possible to make an adjustable clamp with a variable resistor and a zener diode with an op amp?

dowser

I've just ordered samples of the max4544 and max394 - I guess they'll send more than one of each?

A zener will always try to pull the voltage down to it's rating, I guess you could use a few resistors around the zener to adjust it, but you're then just wasting current. Think the chip solution is cleaner....if it works, lol!

Richard

john banks

MAX325 available from Farnell £2.04. 8 pin DIL. Single supply. Dual SPST analog switch one NO one NC easy to make a SPDT.

http://dbserv.maxim-ic.com/quick_view2.cfm?qv_pk=1313

john banks

4544 is undoubtedly the best for the task closely followed by the 325 as they are both 8 pins. With the 325 you just connect IN1 and IN2 together and COM1 and COM2 together to make a SPDT.

Maxim's site said they send 2 samples of up to 8 products.

[Edited by john banks - 2/21/2002 3:18:29 PM]

john banks

BTW I made up the resistor values on the sketch above. I think they are sensible. Also this capacitor - please see the bit in the comparator's datasheet about oscillations and see what you think?

The input range is fine with the comparator when used as a differential amp which is what we are doing.

LOL - even I should manage not to blow these up - two 8 pin single supply DIL chips, two pots, one resistor and a capacitor.

Any comments on the circuit please say. KISS is the principle I live by so don't make it complicated!

I can test out my lovely new regulated supply on this lot

I just hope if ******* works after all that brain draining.

If it does I think it will have considerable advantages over the original design. I would be much happier attaching a 5V single supply circuit to my car. It won't need the rapidly increasing number of protection circuits or expensive components that the other one had I hope.

Failure of the op amp would lead to the output being high because of the pull up resistor, so behaving like in the defend zone and attaching the ECU to the V1. Not a calamity - just to be aware of.

[Edited by john banks - 2/21/2002 3:31:29 PM]

Paul_H

John

Can you explain how the comparator works?

I can see that for normal operation below defend point:

OP1 = (-) supply = 0V, because (+) input is less than (-) input = 0V to switch, therefore ECU reads MAP sensor

In defend zone, wouldn't OP1 be (+) supply =5V, because (+) input greater than (-) input = 5V to switch? and trigger a fuel cut?

I'm sure I'm missing something!

john banks

The analog switch works basically like a solid state relay. The 5V is just a control. Note where the Normally Open (shown "NO" on diagram) connector on the switch is coming from - it is the same as the LOWER reference - giving the response curve I drew with hopefully a square notch in it. Note the COM on the switch is the output to the ECU, and the throws of the switch are the inputs. It took me a lot of brain draining to come up with this, but I just knew there had to be a simpler way than how it was done before.

5V to the analog switch in the defend zone connects the ECU to the lower reference and keeps it there until the upper reference point is reached. Then OP2 output is pulled low and the switch reverts to feeding the ECU with the actual MAP reading triggering a fuel cut.

The open collector outputs simplify the circuit MASSIVELY. It only takes EITHER OP1 OR OP2 to be pulled low (ie outside the defend zone) to connect the ECU directly to the MAP sensor.

We can get rid of dc-dc converters, split rails, 12V, expensive sample and hold amp, all the zeners and all that logic as well as just having one comparator chip (both chips 8 pin DIL).

My hope is that the response should be very crisp - the MAP signal will be unaltered apart from in the zone where is should be flat as it is connected to a potential divider.

[Edited by john banks - 2/21/2002 9:12:36 PM]

Paul_H

Thanks - got it.

Should have no error whatsoever up to defend point. Very clever and yet simple, and won't break the bank either. Well done.

john banks

Are you and Dowser going to give it a go then? Or are you using the original design? What do you think? I'm going to think to see if we have any more thoughts on it for a day or two before I order parts and try it.

Paul_H

Have given up on original design - don't fancy +-12V into ECU! and not a good enough clamp.

Have built the other design you posted (fuel cut only, not a lifter) for pennies but not tested it.

Your design has got to be worth a go doesn't it. Input impedance should be high/output impedance low, and with 5V supply shouldn't fry anything if it goes faulty. Only other things that come to mind is input protection against a spike, and how much current is the switch rated for - assume this will be neglgible. I've already got some LM7805's which I think I would prefer to use rather than ECU supply. I'm happy to build it & post results, unless you want to. It's your baby

john banks

Feel free to be the pioneer. I know what you mean about the supply. Maybe do what I have done with the uC circuits - use your own power supply first and when you are happy with it consider the TPS supply.

I am in no hurry to try it as I already have a 19+ PSI fuel cut Be my guest, I will concentrate on the EBC if you are going to build it? It is the Drivetrain baby, not mine

Paul_H

OK I'll build it and see. Then maybe I'll have 19PSI too

john banks

19 felt good, but psychologically on a TD04L I feel happier at 18. I used to feel happier at 17 . 18 runs nicely with PPP (and hopefully with this). If the defend point is near to fuel cut then hopefully we will get the most sensible fuelling and advance possible. After a lot of hurdles I hope we come up with a pukka design.

john banks

Bill Putney kindly posted this reply on alt.autos.subaru:

You also may want to put a small amount of positive feedback in your
comparator circuit to create a few mV of hysteresis to prevent
oscillations at your V2 switchpoint in case of noise. Putting a
relatively high value resistor (maybe 500k or 1M ohms - you may have to
experiment or do some calculations for reasonable noise rejection
without affecting trip points too much) from comparator output to the
"+" input of the V2 comparator.

That 100pF cap is a good idea, but it will not totally eliminate
oscillations right at the switchpoint. It's effectiveness will be
dependent on how well you choose your ground point. You probably
already understand that you need to tie that ground point right at the
ECU - preferably a ground signal off the harness coming out of the ECU.
But, as I said, the capacitor won't get rid of all the oscillations, but
the feedback resistor for hysteresis will totally kill any remaining
oscillations.

Actually you need to do the same feedback to the V1 comparator that I
recommended to the V2 - but that's a little more tricky. You could put
a resistor in series with the "+" input to only that comparator, and
then tie a feedback resistor from the output to the "+" pin of that
comparator (in addition to the feedback to the V2 comparator). To
really do it right, buffer the V1 input with an op amp and a resistor
(5k ohms would be good - not critical) so you don't load the MAP sensor
output signal.

So what I've added if you do everything I suggested is add a dual op amp
(only one op amp if you don't buffer V1 from the ECU input) and three
resistors.

My reply:

Thanks very much for the suggestions Bill, most appreciated. I can see where
you are coming from. I think it might be worth a try with hysteresis on V2
only to start with and hope the capacitor copes with V1. I say this because
oscillations should not matter so much at V1 as the output voltage will not
jump like it will at V2. Need to try it and see. If I need it I can add it
later with the suggestions you give, but I hope I might just get away with
V2 since I only need one extra resistor. What do you think? I need to keep
it as simple as possible for my small brain to cope with! We had a previous
design using loads of gates and split rail supply and sample and hold amp
and it got overcomplex.

Also I wondered if the open collector outputs and pull up resistor would cause any trouble.

john banks

UPDATE:

I think the MAX983 would be a better comparator as it has hysteresis, and would solve all the oscillation issues.

Someone also helpfully pointed out the V1 is at a maximum of 2.5V with the resistors connected as shown. To solve this, the variable resistors should be in parallel to 5V and ground not series!

john banks

Still trying to work out the input voltage range for the comparator - it looks like it gets into trouble at supply minus 1.3V. Possible solution is to use a high impedance potential divider to halve the input voltage from the MAP sensor - adds two resistors.

john banks

Latest circuit. See MAX 973 datasheet for how to choose resistor values.

This circuit has no issues with hysteresis or input range. Only issue is having to set the defend zone voltage separately with a pot connected to 5V.

Paul_H

>>>Someone also helpfully pointed out the V1 is at a maximum of 2.5V with the resistors connected as shown. To solve this, the variable resistors should be in parallel to 5V and ground not series!

- had already worked this out

LM393 struggles with the jump at V2 on original circuit

john banks

How does it struggle? There are several more postings on alt.autos.subaru about this and possible solutions to look into and a bit of offline discussion too. I think V2 is too high for the comparator. Need a rail to rail input comparator with open collector outputs and single supply, preferably two in one package. Or use a higher supply voltage - ie 12V just for the op amp, 5V for the rest.

A bit of hysteresis is also required on V2 in particular to stop it oscillating at the jump point. Will look into it after Crail later.