Cable impedance problem with last MIMA systems.
|Where to make the connections|
As many of you know, we had a rash of bad behavior when the last 25 MIMA systems were installed. Turned out we got some bad DB-25 cables, that had extremely high (8-20 ohm) end to end resistance. The cables were replaced with some 10 foot DB-25 that had normal less than one ohm end to resistance, and we thought we were home free, but found that more strange behavior was still happening.
Some care full scope troubleshooting showed that the 20 KHZ MAMODE signals were not making it up to 5 V, because the rising edge was being rounded by the long cable impedance. The stock cables show some of this effect, but the much longer path and higher capacitance that the DB-25 presents was just enough to further slow the rise time so the car was only intermittently seeing the signals, as they did not rise into the high state. After two sets of cables, And the increasing difficulty in getting shorter and larger conductor cables, I found that simply adding a parallel pull up resistor to the 5V at the MCM brought the rise times back in line with the stock signals without the MIMA harness. All the problems went away, so this should fix the problem,and not let the high speed signals be so effected by small capacitance and resistance variables.
Beefing up the ground path
As described above, the db-25 ground wire on pin 3, is a 28 gauge wire, so we added some jumpers to the distribution board as well as the MCM board to decrease the resistance of this important circuit.
|beefing up the grounds|
Seems to make a very positive improvement in the signal integrity.
(Posted 4/25/2012 by mikey)
The real pulses
I went back out to the car armed with the storage scope, the MIMA kit from Trevor in AZ, and a 6 foot DB25 cable that I found in my cable stock.
|comparison of pulses|
Be aware that the short pulses are present only during the period where the car is booted up, but the car is not running, and I believe that this is where the car checks for reliable communication between the modules.
Once the engine is running, the pulse width gets to 20 KHZ 50% duty cycle.
Started by lifting one side of the resistor and used the 10 foot cable that was so bad the other day that the car would not even start.
The idea was to work up to the best pull up resistor value by putting a 20K pot in series with the added 1K and adjusting the pot while watching the signals and the cars behavior to determine where the system starts to work.
First step was no resistor at all, so I expected the same erratic try at starting the car and erratic MIMA behavior, once the problem was seen, I was going to reconnect the pull up resistor through the pot to see where things settled down.
I took a 20 mile run and the car started normally, and ran fine with autostop and MIMA all working normally????? whats going on here?
This is exactly the same setup and components that did not work at all the other day???
I snapped some photos of the signals. The stock signal just barely makes it to 4V, not to 5 as I had assumed should be the correct level.
The 10 foot cable showed the slowest rise time and lower max voltage as expected, and the 6 foot cable looked pretty similar in max V, but slightly slower rise time to the stock signal.
The car never acted up at all, but both Spence and Tim seemed to feel that problems got noticeable once things got warmed up.
I figure that the next test must be with the car interior as warm as I can get it. We are supposed to get some sun tomorrow,with temps near 60, so I may part it in the direct sun, and add some BTU's from my electric heater, and repeat todays test to see if anything changes????
Bottom line, if you have not installed the system best hold off until we can pin this down.If you have installed the resistor and see improvement, you may want to raise the value to more like 2-3K.
As soon as I get the new 6 foot cables I will compare the cables for effect on the pulses.
I did get to tweak the pot a bit and found that the pulse with 10 foot cable was able to be pulled to the same 4V as the stock signal with ~4K pull up.
Had to go down to ~ 1.5K to get the pulse to square off at the top like the stock signal, but an starting to think that the MAX volts reached is more important than the fast rise time?
Will report what I see tomorrow after more testing.
The first systems had no issues like this, and I know we used 6 and 10 foot cables, so we again have a mystery.
Hard to fix something that is working, so I have to get back to where it does not work to get any answer.
Note that the pulses on the top 3 traces were 5uS/DIV, and the bottom was 10uS/DIV, all are 1V /division on the voltage scale
(Posted 4/23/2012 by mikey)
Why the cable effects the pulse
Sending signals through unshielded wires in a noisy environment like the Insight requires that the signals have the full 0 to 5V swings that standard 5 V logic Ic's require. The cables have capacitance to ground and to adjacent wires, so keeping the overall impedance of the wires low helps to maintain the full swings.
|Cable capacatance slows rise time|
When the car designers look at the path of a signal, they consider all the effects that will effect this signal integrity.
One common approach, that is used heavily in the Insight is to use open collector driver transistors on the sending module, and a pull up resistor that can get the required rise time on the receiving side to pull the line high when the transistor turns off.
This resistor on the MCM MAMODE line was sized for the normal range of line capacitance expected in the car. The speed the resistor can pull the line high is directly related to how much capacitance there is between this wire and its neighbors over the entire length of the cable. Longer cable more capacitance, and as many will remember from basic electronics, capacitors store energy, so the low condition on this wire will act like a discharged capacitor. when it wants to pull high, the transistor turns off, and the resistors job is to as rapidly as possible charge the capacitor, as it pulls the line towards 5V. The MIMA cables add some capacitance to this signal path, and therefore it cannot charge to 5V as rapidly as the stock harness can. I know that earlier MIMA systems did not have problems with this, and I suspect that the cables we used then were of higher quality than the ones available now.
Almost all of the older style DB 25 and centronics cables are being made in China.
The only way tot be sure a cable will work is to buy and test many cables from many sources,I decided it would be better to simply add the 4 cent additional pull up resistor, and make the cable capacitance sensitivity issue go away.
(Posted 4/20/2012 by mikey)
Installing the resistor
The MCM plug in adapter board in the rear of the car is where we will do the modification.
|Adding the pull up resistor|
Start by uncovering the pc board just below the big grey connector.
Removing the label can be easy or hard, depending on how well the hot glue fused to the fishpaper and board. I have found that the fishPaper will unstick if you wiggle and work it loose, and the hotglue on the PC board can also be lifted off the board in most cases. A heat gun is a messier way to gain access, but can be used as a last resort.
The exact position of the two connection points can best be seen in the photo of the PC board artwork at the top of this blog.
The resistor can be any 4.7K 1/8 to 1/4 W resistor.
Bend the leads at a right angle, and tin the cut off leads.
The attachment points will also want to be freshly tinned.
Finally solder the resistor to the two points, being observant that you do not bridge the two close traces to the top resistor tie point.
After a close visual inspection, stabilize the resistor with a dab of hot glue, and also fold the fish paper cover back over the board and resistor, and plug it back into the car, and your IMA problems should go away.
(Posted 4/20/2012 by mikey)