TE-WB (v1.5) | Oz DIY-WB (v1.0) | TE-5301 | Wideband FMD | Silicon Chip FMD
Updated 5301 Unit now availableThis document describes the original Jaycar KC-5300 display that was modified and sold by us as the TE-5300 display. Tech Edge now sells the TE-5301 wideband display (shown at left) which is functionally identical (and uses the same firmware) but comes with double sided PCBs and some extra features. You can use the information in the following paragraphs to modify your existing KC-5300 for use with the DIY-WB unit or another wideband sensor interface unit. Order a 5301 display kit from us at Tech Edge. |
A suitable display for the DIY-WB unit is described here. It uses a modified Silicon Chip Fuel Mixture Display kit (the FMD unit). The standard unit features:
The modified unit features:
The Silicon Chip September and October 2000 issue features the original article about the FMD unit that is available as a complete kit (with case), or a set of two PCBs (prices correct @ 31 Oct 01) from:
|
The Jaycar KC-5300 |
As well as the basic modifications described here, we have added RS232 output using just three extra resistors (and a connector). As mentioned, the FMD kit is described here. The modified circuit below should be compared with the original circuit. The input circuitry has been changed to handle the voltage range of 1.4 to 3.1 Volts, rather than 0.0 to 1.1 for the Bosch LSM-11. This requires a new value for one of the presets (20k -> 5k) and a couple of 5% 1/4W resistor changes too. For your convenience we have included a component overlay for the two 5300 PCBs showing the location of the three control pots and the three option resistors R1, R2, & R3.
The display, when used with the DIY-WB unit, is most accurate in the rich end of the range (AFR 10:1) and easily gives 0.1 AFR accuracy (which is thus limited by the accuracy of the WB unit itself). At the AFR 25:1 end of the range, the accuracy is about 0.2 AFR, again limited by the accuracy of the WB output.
The original software for the FMD kit is available as zipped source form or zipped hex from Silicon Chip's web site.
The input from the NTK sensor is via a 1M and 470 k ohm resistor divider network. This converts the 1.4 -> 3.1 Volts input range into approximately 0.45 -> 0.99 Volts at pin 2 of the comparator IC2a. The capacitor across the 470 k resistor smooths rapid fluctuations in the WB output and stabilises the digital display.
The original charge pump circuit has been removed and replaced by a simpler +2.49 Volt reference that is divided by the 3.3 k resistor, the 5 k OFFSET preset, and the 1 k resistor to produce a voltage that will balance the divided WB voltage of 0.45 at the low duty cycle end of the PWM A/D converter's range. At the high PWM range, the resistor in series with the 250 k ohm SPAN preset has been changed to 560 k ohm.
Changes to the software involved modifying the calibration display to show a voltage from between 1400 and 3088 mV for the A/D count between 19 and 230. The actual circuit now measures 8 mVolts between counts rather than the 5 mV for the LSM-11 sensor. Tables for the bar and dot display have been compressed and use the AFR value in the lookup process rather than the raw A/D count. Lambda mode simply uses a different lookup table (this used to be the propane option selected with R3).
The Jaycar KC-5300 kit comes with very complete construction details including a copy of the two Silicon Chip articles describing it. Especially good is the black laser cut and professionally silk-screened panel with red perspex lens that is easily assembled with glue. See also component overlay for these two 5300 PCBs.
This is the processor PCB showing the 16F84A MicroChip processor and the LM358 acting as
a comparator and LDR buffer. The changes are to the bottom left area around the two blue preset pots.
|
This is the display PCB that sits atop the processor PCB.
The changes are simple and require a couple of links to be use in place of a bunch
of components, the re-orientation of the voltage reference, and some resistor changes.
|
The instructions optimistically suggest 3 hours to complete the kit, which is probably correct if you've already made one or two. In addition to the reprogrammed PIC firmware (or a copy of the HEX file to program yourself), you'll need:
|
Note: that these parts are supplied in the KC-5300 kit when ordered from Tech Edge. |
The three regions of operation are shown at right. A normal operating value (of 14.7 in this case) is shown as three digits. A rich condition is shown as 10. followed by an equals or less than symbol (=<). A lean condition is shown as 25. followed by a greater than or equals symbol (>=). Note: display is operating in dot mode. Other software changes are to:
The major changes have to do with how the dot and bar mode display intervals have been altered. |
|
Dot mode displays 48 unique LED configurations with a difference of 0.2 AFR between steps, from 10.0 to 19.6 AFR. Either one single LED is fully lit, or two LEDs are lit with a varying intensity, giving the impression the LEDs move slowly upwards as the display shows a richer mixture. The single centre LED is lit for AFRs between 14.6 and 14.7. |
|
|
Bar mode displays just 8 different LED configurations with a chunky difference of 1.6 AFR between modes. It does have the advantage of easily seeing how far away from stoic you are by comparing the total width of the lit bars. |
The startup display now shows four important items (it was originally blank)
These options are (refer to the kit's instruction on how to enable them):
Refer to the diagram at right showing which LEDs relate to which mode. The default configuration is normal mode, displaying AFR with a 13 step Dot display. |
|
Lambda mode uses a different lookup table to convert the A/D value from the sensor into a different display value. Lambda is simply the ratio of current AFR to AFR at stoic. For petrol this is commonly calculated as AFR/14.7 (although strictly it depends on the fuel, different fuels have different values for the AFR at stoic).
AFR |
10 | 11 | 12 | 13 | 14 |
14.7 |
15 | 16 | 17 | 18 | 19 |
20 | 21 | 22 | 23 | 24 | 25 |
LAMBDA |
0.68 | 0.75 | 0.82 | 0.88 | 0.95 |
1.00 |
1.02 | 1.09 | 1.16 | 1.22 | 1.29 |
1.36 | 1.43 | 1.50 | 1.57 | 1.63 | 1.70 |
mixture |
rich |
stoic |
lean |
In Lambda mode the rich range is always shown with a 0 in the left display position. Lean is shown with a 1. This makes it easy to see immediately if a rich or lean condition exists.
The actual changes to the PCB are simply to move the decimal point from its fixed position between the middle and the right digits, to between the left and middle digits. This is done by unsoldering the right end of the 150 ohm resistor on the display PCB, and resoldering it to the now vacant hole close to the left seven segment digit. The trace on the solder side should be cut with a suitable blade, and a wire run from the resistor's pad to the left display's unconnected pad (this is the pad closest to the dimmer pot's adjustment hole - look at the pad on the first display that the resistor was connected to if you're unsure). Also make sure you route the wire away from the large hole so that the dimmer pot on the processor PCB can still be adjusted when it is all assembled. |
The calibration procedure is similar to that described in the construction manual. The processor PCB is first wired temporarily with R2 -1.8 k ohm, on the solder side of the CPU PCB. The display will now show the SENSOR INPUT voltage in volts x 10, or alternatively, an out of range condition. You'll see either Lo, a value between 14.0 and 30.8 (representing 1.400 through 3.088 Volts), or Hi.
The calibration mode display actually shows 3 1/2 digits, the 1/2 digit is the dot/bar display, and is interpreted using the image at right (note that 0 is blank). However, as the firmware A/D converter is accurate to only 8 mVolts, the dot/bar display will give the appearance of "jumping all over the place" as it displays (say) 14.00 -> 14.08 -> 14.16, etc. - the dot/bar will jump from [blank] -> 8 -> 6, etc.
A suitable voltage source for calibration can be obtained from an external 20 or 50 k ohm potentiometer wired across a couple of fresh 1.5 volt batteries (which will show at least 1.6 Volts). An accurate DVM measures the voltage input to the FMD unit (note: leave the meter connected during calibration to ensure the test voltage is not influenced by the meter itself). If you have a multi turn pot, then setting the test voltage becomes easier, although this is not essential.
The FMD unit is calibrated first at a minimum input voltage of 1.400 Volts (1400 mV) by adjusting R3, which is the new 5 k ohm potentiometer (VR3 = OFFSET) on the display PCB. The display should read 14.0 and the dot/bar should be blank. Ideally, the OFFSET pot should be set to almost display Lo as this will be closest to 1400 mV.
For the other end of the range, it is calibrated at a maximum input voltage of 3.000 Volts by adjusting the 250 k ohm potentiometer (VR2 = SPAN) on the processor PCB. This is done through a hole in the display PCB which is closest to the bottom left corner. The display should read 30.0 and the dot/bar should be blank. Ideally, the SPAN pot should be set to almost display 29.9 as this will be closest to 3000 mV.
This procedure should be repeated as each adjustment affects the other slightly. A quick check at other points in the displayed range of 1.40 to 3.08 can be made, but remember the DVM you use may have some linearity errors as well as the FMD circuit. The two displayed voltages should track such that the rightmost FMD digit is never off by more than one from the DVM's corresponding displayed value.
As with all kits, there are variations in the components supplied. We note that some people have reported being unable to set the lower calibration limit of 14.00 with 1.4 Volts across the sensor input, this can be "cured" (assuming there are no construction errors) by either replacing the 3.3 k resistor connected to VR3 with a larger value (say 6.8 k) or by replacing VR3 with the 20 k pot in the kit (though this will make adjustments to VR3 a little more "touchy").
The modified software may be made available in source form if copyright details relating to the original DIY-WB licencing agreement can be resolved, but at present the software will be made available only:
This software is copyright © 2001 Tech Edge Pty. Ltd. and I reserve the right to refuse any individual or organisation the supply of this software for whatever reason. In general the software will be made freely available if you want to use the FMD with the DIY-WB kit and you are doing this for your own personal non profit use. If you are a commercial organisation, then it may be possible for you to licence the software from Tech Edge Pty. Ltd for your own purposes. We can modify the software or the hardware design, and will do so at out normal consultancy rates.
As this software is supplied for free, there is absolutely no warranty either expressed or implied. You must decide for yourself as to the suitability of this kit for use with any other device.
Use of an air-fuel meter for tuning purposes represents a potential hazard to the safe operation of your vehicle. Use of any display device within a moving vehicle present a safety hazard. Be warned you should consider having an assistant when using this device.
Assembler/Disassembler | 1227808 ECM | 8192 Baud ALDL | 160 Baud ALDL
Last updated 11th June 2003 (links)
Statistics by
www.digits.com
Shows approximate unique hits since 03 November 2001.
This document is Copyright © Tech Edge 2001 and
contains reference to material that is Copyright © Silicon Chip.
Previous |
Home |
Feedback |
Copyright