Notices
Drivetrain Gearbox, Diffs & Driveshafts etc

DIY AFR meter for £10 with excellent results!

Old 12 February 2001, 10:35 PM
  #1  
john banks
Scooby Regular
Thread Starter
 
Join Date: Nov 2000
Location: 32 cylinders and many cats
Posts: 18,658
Post

The penny-pinching fiddler strikes again:





... the bargraph - beauty is plug it onto a breadboard next to the IC and nine of the ten cathode connections are made. Plug the other side of the DIL into an anode rail and send one wire round from pin one on the IC to let the first LED also pick up its signal. Much easier than soldering/connecting 10 leds - disadvantage they are all the same colour.

Don't have digi camera to post a pic, but at the moment it is a toolbox tool and will not stay on the car. When I have fitted my boost controller and checked things it will go away in my toolbox.

The IC contains a precision voltage source and the reference voltages were holding at 700 and 900 +-1mV consistently - this is less than the error on my multimeter.

The resistors are two 4k7 pots. Total cost of the bits £10 from Maplin plus the £5 or so for a solderless 3x2" breadboard I already had.

This is a modified version of a circuit I found at http://www.students.tut.fi/~eppu/dev/EGO-bar.html which was 0 to 1 V in 0.1V steps (see below why not useful).

A few features cause me to post this info - firstly it was incredibly easy to build and could be fabricated on a PCB or veroboard within a 2" x 1.5" board. Secondly it is only two pots, a chip and a bargraph LED. Thirdly, I am running the meter from 0.70 to 0.90 V in 0.02V (20mV) steps, which gives readings that equate to 1% CO to 10% CO. At WOT I am seeing 7% CO which I think is ideal. Notably it reads higher than a multimeter did because even under boost, the lambda signal is slightly pulsatile and the multimeter was averaging it and underreading.

Most air fuel meters run a range from 0.0 to 1.0 V in ten steps. You really need it in the range above - 0.83 V at WOT full boost in a Scooby is on the lean side, 0.87 V is ideal and 0.89V would be way too rich. So a meter that gives you 0.8 or 0.9 V only is not very good. Notably the LambdaLink seems to have very sensible layout of LEDs.

Can easily read WOT AFR even in the dark - cover up all but the last four lights and out of the corner of your eye can see how many light up - on my car its 2-3 which suggests 0.86-0.88V - ideal AFR!

The order list:
WQ41 LM3914N
YG33 Green bargraph
UH02 Hor Encl Preset 4k7 x 2
BL85 Solid core bell wire 10m
PP3 battery and clip

One thing an electronics whizz could help with (I am only a dabbler) - I am using a 9V battery as it is very easy. If I wanted to use the car ciggie lighter socket that would be fine for the IC, but I would have to tie the lambda sensor ground to supply ground, and would be nervous about doing this. Can I decouple it in some way - eg a capactitor and where would I put it or does it not matter?

[Edited by john banks - 12/2/2001 10:44:52 PM]

[Edited by john banks - 2/1/2002 6:07:00 PM]
john banks is offline  
Old 12 February 2001, 11:11 PM
  #2  
john banks
Scooby Regular
Thread Starter
 
Join Date: Nov 2000
Location: 32 cylinders and many cats
Posts: 18,658
Thumbs up

http://www1.rswww.com/cgi-bin/bv/sea...quickSubmit=Go


Thanks Carl! You're a star! Looks made for the job. The top three LEDs being green would represent 6 8 and 10% CO, and all would be showing mixtures richer than 12:1. The bottom three would represent <=2% CO - major lean on WOT.

Now what about this common ground and decoupling then saving me using a PP3 battery?

[Edited by john banks - 12/2/2001 11:12:23 PM]
john banks is offline  
Old 12 March 2001, 10:29 AM
  #3  
john banks
Scooby Regular
Thread Starter
 
Join Date: Nov 2000
Location: 32 cylinders and many cats
Posts: 18,658
Post

True Carl. It worked on my computer at home (for a while) so I posted it. I hate these Javascript/framed sites where you can't post a proper link!

I was thinking of installing something where the ashtray is and picking up power from either the stereo or the back of the ciggie lighter.

Also considering making a boost gauge, oil temperature and knock display using a similar setup, and might put all four bargraphs in the one box for a neat compact permanent solution.

Knock display signal I will have to work on, but I think a sensitive voltmeter circuit would pick stuff up - after all it's only a piezo mike.

Oil temp - will have to install a sender to do this - any ideas of what and where for those that have done it?

MAP - thinking of putting 15-20PSI in 0.5 PSI steps - dead in the middle is where the action is on my car and I could get rid of the huge ugly boost gauge. The MAP voltage is proprtional to pressure within -1 to +1.8 bar with an offest - range being 0.88 to 4.84V. 4.1V is about where the action is on mine.

For different applications the circuit will need different pot values and some trial and error and/or calculation will be required to get the reference voltages to either end of the scale - hardly any of these values want a 0V bottom end. Note also that pot T2 controls the brightness of the LEDs! With more components according to the data sheet you can make the brightness control independent.

Whilst I am at it, I might put an outside temperature guage in as well. Need to have a think - see if I can be bothered as the Knock sensor display would probably add little value. There is a £10 unit from maplin with a captive temperature probe on a 3m cable which has a one year life even from a 1.5V watch battery. I'm sure it wouldn't take much trouble to do...

My car will soon look like Knight Rider if I am not careful.

[Edited by john banks - 12/3/2001 10:33:03 AM]
john banks is offline  
Old 12 May 2001, 11:17 AM
  #4  
john banks
Scooby Regular
Thread Starter
 
Join Date: Nov 2000
Location: 32 cylinders and many cats
Posts: 18,658
Post

Much more beautiful than mine!! But I'm getting some board this afternoon.

See guys how simple this thing is - the picture really shows that. Thanks Harry.

I had a thought - check the data sheet and see if you can reverse the lo and high ref points and then your display will be the "correct way up"?


your pots look in a similar position to mine so we must be doing something right. With no fiddling after a few days and a flattening PP3, my references are still within 1mv of 700 and 900mV.
I also found no resistance between lambda ground and vehicle ground, so I'm going to hook mine up the car supply.

I think I might convert it to three different coloured LEDs for 0.83,0.85 and 0.87V and mount it all discretely somewhere in the cabin. I think those three are all you really need, and having the led off the board could make it even smaller.

Even with the led bar, it is possible to build this within a 1.5x1" board.

[Edited by john banks - 12/5/2001 11:21:00 AM]
john banks is offline  
Old 12 May 2001, 02:49 PM
  #5  
john banks
Scooby Regular
Thread Starter
 
Join Date: Nov 2000
Location: 32 cylinders and many cats
Posts: 18,658
Post

I am monitoring at 0.83, 0.85 and 0.87.

See the voltages in my above post. At stoich (ie. light throttle/cruise conditions/off boost) the lambda sensor signal rocks back and forth from 0.2 to 0.6 or so volts to accurately indicate stoichiometric mixture (14.7:1 for petrol). It it this rocking behaviour which is caused by a virtually vertical response curve of the sensor at that point and the ECU controlling mixture with feedback.

All you will see below 800mV is flashing lights which are to my mind meaningless - let the ECU get on with it - it is very good at it.

Our concern is monitoring the mixture on boost. After a few PSI or 4700rpm which ever is first, the ECU goes into open loop mode. It then fuels dependent on MAF mainly, probably with some input from RPM and throttle position. As you increase the boost you require a progressively richer mixture for controlling temperatures and preventing detonation.

Recommendations I have are that for 3-9PSI boost you should see 830mV - just under 5% CO 13:1 AFR. Between 9-15PSI boost you should see 850mV - 5.5% CO 12.5:1 AFR. Over 15PSI boost you should see 870mV - 7% CO 11.5:1 AFR.

Therefore, after fiddling with 10 LED displays and blu-tak, I think you need three lights at these voltages, and if you want say another 2 - at 890mV and 810mV to widen it a bit. Below 5PSI I don't really care what it is doing, but I am wanting to see the 830mV LED lighting up. 900mV is over 10% CO - VERY rich. 1V would probably be impossible?

So not criticising your nice design, but consider what you want to read, why and at focus on that IMHO. Otherwise you will have a nice flashing display that WILL irritate you after 5mins in the dark, and will distract you from the important bit. Hope this helps.

[Edited by john banks - 12/5/2001 2:51:36 PM]
john banks is offline  
Old 12 May 2001, 04:07 PM
  #6  
john banks
Scooby Regular
Thread Starter
 
Join Date: Nov 2000
Location: 32 cylinders and many cats
Posts: 18,658
Post

A bypass cap can be used to good effect with noisy power supplies. http://www.seattlerobotics.org/encod...97/basics.html

[Edited by john banks - 12/5/2001 4:25:19 PM]
john banks is offline  
Old 12 June 2001, 05:32 PM
  #7  
Scott.T
Scooby Regular
iTrader: (2)
 
Scott.T's Avatar
 
Join Date: Feb 2000
Location: East Sussex
Posts: 6,174
Post

Sounds like that's spot on then John. Is that just with the MBC or are u bleeding/restricting anywhere.

[Edited by Scott.T - 12/6/2001 5:32:32 PM]
Scott.T is offline  
Old 12 July 2001, 03:45 PM
  #8  
john banks
Scooby Regular
Thread Starter
 
Join Date: Nov 2000
Location: 32 cylinders and many cats
Posts: 18,658
Post

Darren Dawes and I already discussed this in one of the other Dawes threads and he agreed that the device was not designed specifically around the Scooby sensor and had not been tested on Scoobies but he would recalibrate it for free.

My two main configurations have been 720 to 900mV in 20mV steps and 830 red 850 yellow and 870 green. The second one works the best and gives good results on my car showing fuelling is spot on.

These three values are based on information in the Link ECU manual for MY99/00 cars and the Link ECU default lambda targets for 3-9 9-15 and 15+ PSI respectively for the three voltages 830 850 870mV.

Unless you Dawes Devices AFR meter is recalibrated to these or near to these and you can tell that 20mV difference it is useless to you IMHO. A car showing 900mV would be running over 10% CO which is very rich, but occasionally seen on the dyno site. So crack it open and see if you can adjust it with your DVM I would say. Otherwise you will just be looking at a red light all the time unless your car does 10% CO on WOT which is not impossible, but hardly desirable for power and fuel consumption.

The values I have selected are not very different to the top three lights on a Lambda link, and most folk tune for the middle light to be on at full boost = 7%CO for safety, power and economy.

As per my original post, you have to be looking at the right range. A gauge that only shows 800 OR 900mV would also be useless IMHO and will just show you Knight Rider effects on stoich and that you are somewhere between say 3 and 10% CO - rubbish.

[Edited by john banks - 12/7/2001 3:50:00 PM]
john banks is offline  
Old 02 December 2001, 11:02 PM
  #9  
carl
Scooby Regular
 
Join Date: May 1999
Posts: 7,901
Post

RS components sell a 10-LED DIL array with 3 red, 4 yellow and 3 green LEDs. Stock number 246-5689 (can't link directly as all the pages are dynamically generated and have session IDs).
carl is offline  
Old 03 December 2001, 10:14 AM
  #10  
carl
Scooby Regular
 
Join Date: May 1999
Posts: 7,901
Post

John -- all very well but if you click on that link you get a session timeout error.

I may give this a go with the RS 3-colour led array. Instead of using the cigar lighter can't you just wire into the feed for the clock or something (I suppose it depends if you want a permanent or temporary installation -- I'd go for permanent. If you change car then build another one )
carl is offline  
Old 03 December 2001, 10:32 AM
  #11  
HarryBoy
Scooby Regular
 
Join Date: Jun 2001
Posts: 160
Post

John, what's the setup procedure for the two trimmers?

Harry
HarryBoy is offline  
Old 03 December 2001, 10:36 AM
  #12  
john banks
Scooby Regular
Thread Starter
 
Join Date: Nov 2000
Location: 32 cylinders and many cats
Posts: 18,658
Post

Set pin 4 to 700mV and pin 6 to 900mV. Then the first LED lights up at 720mV and each successive one in 20mV steps. Can confirm this by wiring a 220k pot across 9V battery and taking the centre tap and firing it into pin 5 with a voltmeter attached to simulate some trial voltages. Works very well, and within a few mV the correct LED illuminates at the predicted time all the way up the scale. And with the above pots it is quite sensitve, but the adjustments are about midway for one pot and 1/3 for the other, so pretty ideal for this application. However, it is not so fragile that breathing on it upsets the reference voltages.

Of course you need a DVM, but they are pretty cheap these days...
john banks is offline  
Old 03 December 2001, 10:38 AM
  #13  
Disco
Scooby Regular
 
Join Date: Jun 2001
Posts: 10,211
Post

John, I think you should go into business making and selling these things you keep creating, you obviously have a flair for it.
Disco is offline  
Old 03 December 2001, 10:42 AM
  #14  
john banks
Scooby Regular
Thread Starter
 
Join Date: Nov 2000
Location: 32 cylinders and many cats
Posts: 18,658
Post

Not worth it. I am OK at hacking things together, but the profit margin on each unit would be tiny, and my fabrication skills are not up to the job for production. I do it for the pure joy of tinkering believe it or not. (I am a frustrated Doc who wished he'd been an electronic engineer - funny how you can enjoy something that you are not paid to do so much more!)
john banks is offline  
Old 03 December 2001, 10:58 PM
  #15  
John Stevenson
Scooby Regular
 
Join Date: Jan 2001
Location: Powered by "SOLARIS"
Posts: 1,694
Post

If you use a capacitor then you will decouple the DC, exacly what you don't want, I'll have a wee think (when sober) and post tomorrow.



---john---
John Stevenson is offline  
Old 04 December 2001, 08:21 AM
  #16  
john banks
Scooby Regular
Thread Starter
 
Join Date: Nov 2000
Location: 32 cylinders and many cats
Posts: 18,658
Thumbs up

Thanks John. Was thinking to use 12V power a cap across the supply would be best according to the datasheet, but info on what to do re the signal ground would be great.
john banks is offline  
Old 04 December 2001, 08:33 AM
  #17  
Gridlock Mikey
Scooby Regular
 
Gridlock Mikey's Avatar
 
Join Date: Jan 2001
Location: http://www.facebook.com
Posts: 15,233
Question

COR BLIMEY GUVNOR,

A bit teckie for me

Mikey
Gridlock Mikey is offline  
Old 04 December 2001, 02:30 PM
  #18  
John Stevenson
Scooby Regular
 
Join Date: Jan 2001
Location: Powered by "SOLARIS"
Posts: 1,694
Post

Having a look at the circuit diagram John, why do you think you need to worry about the Earth. Do you know whether the -ve lead from the probe is true vehicle earth (measure resistance between here and car body ?)
John Stevenson is offline  
Old 04 December 2001, 05:47 PM
  #19  
HarryBoy
Scooby Regular
 
Join Date: Jun 2001
Posts: 160
Post

John, just completed the AFR meter. It works!

Unfortunatley the coloured LED bar graph from RS has the colours the wrong way round for easy back to back wiring/mounting with the LM3914 so red means good on mine and green is bad!

I will try it on the car tomorrow.....

Harry
HarryBoy is offline  
Old 04 December 2001, 06:00 PM
  #20  
Scott.T
Scooby Regular
iTrader: (2)
 
Scott.T's Avatar
 
Join Date: Feb 2000
Location: East Sussex
Posts: 6,174
Post

Let us know how you get on HarryBoy.
Where do u wire it to, is there an easy connection or do u splice into the Lambda wires.

As my last post seems to of vanished....here's one I made earlier

Only Kidding....about $90, I think.
Scott.T is offline  
Old 04 December 2001, 06:30 PM
  #21  
john banks
Scooby Regular
Thread Starter
 
Join Date: Nov 2000
Location: 32 cylinders and many cats
Posts: 18,658
Post

I didn't want to ground the signal ground if I wasn't supposed to. I'll measure resistance as you suggest. Obviously for the chip input one input is pin 5 the other goes to power supply ground (ie effectively vehicle battery - still using PP3).

LEDs Harry,mmm Anyway, as long as you know what they mean. Course it works

Scott - I spliced in at the ECU - see the Link manual. http://www.mrtrally.com.au/performance/docs/WRX-V5.pdf
john banks is offline  
Old 04 December 2001, 06:41 PM
  #22  
pat
Scooby Regular
 
Join Date: Aug 1999
Posts: 679
Post

The chassis and Lambda grounds should be about the same.... after all the lambda ground is via the downpipe / headers depending on MY of car.... there are sometimes slight offsets in the loom but this can be compensated for by adjusting Vref- and VRef+ accordingly....

The real problem comes from trying to get the temperature compensation right. But none of the commercial units seem to do this either so it won't be any worse than (say) a LambdaLink....

The "right" way of doing it is to load up the output of the sensor (a little) to determine its impedance. This will give an indication of the sensor temperature and hence the level of compensation required. I'm looking into ways of doing this without spending megabucks, think I have a reasonable idea, just need to test it and get the temp correction curves for the Subaru OEM lambda sensor...

Cheers,

Pat.
pat is offline  
Old 04 December 2001, 06:57 PM
  #23  
john banks
Scooby Regular
Thread Starter
 
Join Date: Nov 2000
Location: 32 cylinders and many cats
Posts: 18,658
Post

Thanks Pat.
Correction curve sounds interesting.
john banks is offline  
Old 05 December 2001, 11:00 AM
  #24  
HarryBoy
Scooby Regular
 
Join Date: Jun 2001
Posts: 160
Post

Quick pic of my (John's) AFR.

HarryBoy is offline  
Old 05 December 2001, 11:23 AM
  #25  
Scott.T
Scooby Regular
iTrader: (2)
 
Scott.T's Avatar
 
Join Date: Feb 2000
Location: East Sussex
Posts: 6,174
Post

Have just costed up to build one in it's own Black ABS Box, with 10 3mm LED, 3 Green, 4 Yellow and 3 Red (but are clear when not lit).

Total Cost so far is about £12.50 (and that includes the PP£ and ordering more than required due to min order qty's)

Might even get some white lettering to mark it up, so will end up looking like the Lambda Link.

Not sure what box to put it in yet, will decide lunchtime as it depends on where I want to locate it.

I shall post a component list as soon as decided.
Scott.T is offline  
Old 05 December 2001, 02:14 PM
  #26  
Scott.T
Scooby Regular
iTrader: (2)
 
Scott.T's Avatar
 
Join Date: Feb 2000
Location: East Sussex
Posts: 6,174
Post

Right, order the parts required from Farnell www.farnell.co.uk
So it should look something like this :



10 clear 3mm LED's (3 Green, 4 Yellow, 3 Red), mounted in a 60mm x 35mm x17mm Black ABS Box.

I intend to mount it either on top of the steering column, in the cubbyhole by ya right knee or 'IN' the ashtray.

Parts List :

PART ORDER CODE
Red Led 322-489
Yellow Led 322-507
Green Led 322-519
LM3914N 405-863
4K7 POT 614-671
PP3 Clip 536-982
Stripboard 451-058
PP3 Battery 152-059
ABS Box 353-6427

Total Cost approx £12 incl VAT & Postage.

Some of these items have minimum order qty's but £12 takes this into account.
Scott.T is offline  
Old 05 December 2001, 02:17 PM
  #27  
John Stevenson
Scooby Regular
 
Join Date: Jan 2001
Location: Powered by "SOLARIS"
Posts: 1,694
Post

Earth sorted then !! Could i suggest you regulate the supply voltage also. From RS something like an MCT7809CT (part no: 177-5288) £0.26

---john---
John Stevenson is offline  
Old 05 December 2001, 02:34 PM
  #28  
john banks
Scooby Regular
Thread Starter
 
Join Date: Nov 2000
Location: 32 cylinders and many cats
Posts: 18,658
Post

John - the LM3914 takes 3-20VDC. Is regulation necessary? The self contained source seems very accurate.

Scott - why not run it off the car battery if you are building a permanent "attractive" solution. Also consider the colours you have selected and consider what AFR you want them to represent. Consider that the LM3914 will produce a linear scale and the lambda output is not linear.

ie.
mV %CO AFR

<600 <1% >15:1
720 1% 14:1
760 2%
800 3%
840 5% 13:1
860 6% 12:1
880 >8% 11:1

The only bit I am interested in if I don't want to watch it pulsing up and down at stoich is the 4-8% CO range - this is the area I am monitoring for safety. I went for 10LEDs originally, but I am going to put 3 in at 830 850 and 870mV red yellow and green.

In fact my original setup has the first 6 LEDs covered up with blutak so I can see at a glance how many of the 840,860,880mV lights are on. 900mV is also irrelevant.

This will tell me what I need to know and keep the thing very compact - may even be able to mount in an area where a blanking plate it for a switch.
john banks is offline  
Old 05 December 2001, 02:39 PM
  #29  
Scott.T
Scooby Regular
iTrader: (2)
 
Scott.T's Avatar
 
Join Date: Feb 2000
Location: East Sussex
Posts: 6,174
Post

John,

I thought the Lambda would sit at 1V.
Are you just monitoring 850mV to 870mV as this is the area of concern i.e it's getting a bit lean
Scott.T is offline  
Old 05 December 2001, 02:53 PM
  #30  
HarryBoy
Scooby Regular
 
Join Date: Jun 2001
Posts: 160
Post

Guys,

John S. has a good idea with the 7809 but it won't work with a 9v PP3 battery as it needs to see at least 2v greater on it's input than it's output for it to regulate properly but it's a great device to use if running of the 12v car supply....

Harry
HarryBoy is offline  

Thread Tools
Search this Thread
Quick Reply: DIY AFR meter for £10 with excellent results!


Contact Us - Archive - Advertising - Cookie Policy - Privacy Statement - Terms of Service

© 2019 MH Sub I, LLC dba Internet Brands

We are a participant in the Amazon Services LLC Associates Program, an affiliate advertising program designed to provide a means for us to earn fees by linking to Amazon.com and affiliated sites.