Page 1 of 1

Running two timers and not consistent

Posted: Sun Mar 06, 2011 7:23 am
by PWD
I have been having a weird problem where I was just not trusting one our timers. Decided to hook up a second timer. Was not really surprised the two timers don't show the same time but was surprised they are not that consistent in the difference between the two timers from one run to another. That I would have suspected. It varies by over 100 of a second. I did not think they would show the same time as the start switch is different technology.

Timer 1 is using a burglar alarm type magnetic, reed switch to trigger. It is connected to a spring loaded start gate
Timer 2 uses a slotted photo sensor to trigger. Also attached to the spring loaded start gate

The next part is very difficult to believe. But I have now verified. And apparently this has gone on for years. When you get below a specific time on the timer it flips to adding a tenth of a second to the time. So 3.063 shows up as 3.163. I thought it was weird when a time would be exactly as expected but 1/10 higher. Plus you can see 1/10 of a second difference in a run and more importantly the sound.

I just could not believe it but now have verified with the second timer. But how in the world could you have such a bug?

I am going to talk to the timer company on Monday. See what they know. So no names at this point. It is not one of the big names anyway. The backup timer, NewBold, is a bigger name. I guess the NewBold now become the primary timer.

I can understand a sticky start switch but a bug in the electronics adding a 10th of a second is very hard to understand.

I know, how could I have not noticed with this happening for years. Well it only happens when the car runs below a time on the track. Which use to not happen very often. It is happening more now. Basically the timer will do 3.163, 3.165, 3.168 then goes to 3.072. It will never show a time below 3.072. Between timers 3.163 was 3.0211 on the New Bold. Minus the 10th of a second there is a basic mapping between the timers but not very inconsistent but somewhat reasonable.

Could you imagine some on using this timer in an actual race? That is why it is really difficult to believe it is the timer and not something in my environment. But what could it be? And how could it perfectly be 10th of a second? But how do you mess up the electronics and add a 10th? Unless it was on purpose it just does not seem likely. I have the single lane version. Could this only be a problem in single lane configuration?

Anyone know how reliable a magnetic reed switch is in tirggering? Anyone else ever heard of anything like this?

Re: Running two timers and not consistent

Posted: Sun Mar 06, 2011 10:16 am
by Stan Pope
PWD wrote:I have been having a weird problem where I was just not trusting one our timers.
Anyone know how reliable a magnetic reed switch is in tirggering? Anyone else ever heard of anything like this?
Wow! What an investigation!

Reed switches, in general, are pretty reliable devices. I used one in a radio-telegraph keyer many years ago. The action of keying a transmitter to send just this one sentence (about 345 times, excluding this paranthetic) would actuate the reed switch more times than you would actuate one in a start gate (once per heat) in months of tests. And, I assure you that I transmitted much more than just a sentence or two over the years that I used that keyer. The currents and voltages (about 100 v) involved were much higher than would be present in your start gate (unless your timer is using mechanical relays to do the computation ... 1940's? technology!)

As I recall, the main failure mode for reed switches is akin to failure of other switches in that arcing during make/break causes pits to form in the contact surfaces. Eventually, either the contacts stick together or fail to make sufficient contact. Reed switches are supposedly more immune to pitting because the contacts are sealed in an environment which inhibits arcing.

Still, the reed switch is probably the only mechanical component and, so, the most likely to fail over extended operation. The next most likely failure point is a leaky capacitor somewhere on the logic board. Caps seem to degrade over time regardless of usage.

Re: Running two timers and not consistent

Posted: Sun Mar 06, 2011 10:40 am
by Stan Pope
If it is correct that the times errors are 10th or 100th digits, then you can probably exclude the start switch as the cause.

My first thought was start switch "contact bounce", but errors from the start switch would probably be random around one or more intervals that weren't associated with a digit in the time. That would be a low probability event ... one in one hundred (or, maybe, one in ten), and would not be consistent.

The manufacturer may have a failure history for the timer that would help you decide on "repair or replace."

Re: Running two timers and not consistent

Posted: Sun Mar 06, 2011 12:07 pm
by PWD
We have been doing a lot of testing on cars this morning. The timer definitely has a consistent 1/10th of a second problem. It is amazing. We now consistently see the problem. On four runs we will get 3.168, 3.170, 3.072, 3.074

It does hurt a little. We built a completely new type of car a couple of weeks ago. We covered the timer with a piece of paper exposing one number at a time. The time was exactly about what we would expect but 1/10th off. We ripped the car apart taking the tungsten for re-use.

We have built cars for years and a very smooth sounding car, with rear wheels off the track riding the rail will not vary by 1/10th. We have had bad runs before and we always hear a loud scrape and further investigation it is almost always the rear wheels. Otherwise there is just minor variations in time. With out a rear wheel scrape maybe .03 seconds max.

Now the question is the time correct with the 1/10th added? Sure seems basically so. Because our backup timer is flacky. Not sure why. Works sometimes and others it does not. Must have a short or something.

After a while you can watch a car and guess about what time it will be. More from sound than anything else. Also wiggles but they are hard for me to see with my bad eyes.

Re: Running two timers and not consistent

Posted: Sun Mar 06, 2011 1:50 pm
by Stan Pope
Yes, it sure sounds like the timer logic or the read-out logic!

However, contact bounce could be involved if the errors are "in the vicinity of 0.100" rather than "exactly 0.100" seconds. And replacing the starter switch and checking the connections from start switch to timer are easier.

Re: Running two timers and not consistent

Posted: Sun Mar 06, 2011 4:31 pm
by PWD
Would a contact bounce make it so the 1/10th add on would only be for certain times? This is what I find very weird. The timer will never have a time of 3.071, or 3.070. If it is below 3.072 it adds a 1/10th. But 3.072 or above it never adds a 1/10th. So we see 3.171 but never 3.071. We probably ran cars about 30-40 times today and there has not been an exception.

The new cars are at this speed point this year so we are seeing it over and over again. A little bit more improvement and we might never be below 3.1 again :) But very happy.

Re: Running two timers and not consistent

Posted: Sun Mar 06, 2011 4:40 pm
by PWD
There must be something with the number 3.072 or 3.071. I don't think I have ever seen 3.071 or 3.171. I have seen 3.170 and 3.072. So there must be something significant with one of these numbers. I would have thought the problem was when the numbers were getting bigger not smaller.

Re: Running two timers and not consistent

Posted: Sun Mar 06, 2011 6:01 pm
by Stan Pope
PWD wrote:Would a contact bounce make it so the 1/10th add on would only be for certain times?
Depending on the timer, contact bounce could cause a timer reset/restart after the initial gate release. What this would do is to shorten the time between the last reset and the time the car crosses the finish line.

So, I don't see it adding a fixed amount of time, such as an even 0.1 seconds. It could subtract some times which would not be fixed but rather be dependent on how quickly and how many times the contacts bounced. These would be one times around an average plus or minus some variance.

Re: Running two timers and not consistent

Posted: Fri Mar 11, 2011 7:32 pm
by John Shreffler
I supply a reed switch with my timer, and you have me worried. Without naming names, does the flakey timer start with the letter "J"?

Reed switches are very reliable because they are hermetically sealed in a glass tube filled with argon or nitrogen, so the contacts do not oxidize. For this reason, they are great for switching very low currents like the 1 or 2 milliamperes typical of timer start circuits. Pitting would not occur at this current, so they are very reliable. I changed from microswitches to reed switches about 14 years ago after problems appeared. A microswitch is not hermetic, and the new bright orange BeCu contacts slowly turn brown, and then black, and finally stop conducting altogether. If it was switching a 25 watt lightbulb, there would be a sparking at the contact and fresh copper would boil up to the surface with each use, and there would be no problem. It is the low currents that are challenging, and reed switches are a great solution. At least in 14 years, I have never encountered a failure.

Switch bounce would only be a factor if the timer did not latch the signal. In other words if each subsequent bounce re-zeroed the timer. In your experimenting, it should be obvious is this is the case, by inducing some really long switch bounces. I would be surprised if any commercial timer would be susceptible to switch bounce.

Re: Running two timers and not consistent

Posted: Wed Mar 16, 2011 7:30 am
by PWD
No it does NOT start with a "j". I have talked to the company and the owner was extremely nice. He told me that he had never heard of this before.

Over the last days we have litterly made 100's of runs. Since our cars are now almost 100% of the time in the area where the problem exist we see it constantly now. It is very real and I am sure it has nothing to do with the start switch. It is a problem in the actual timer.

The owner of the company is a EE and he thought it was a chip that had gone bad. I told him I have a tough time buying into a chip going bad. For the chip to go bad in a way that the results make sense if you subtract a 10th. He suggested I send him the timer and he will look it over. Which I will ultimately do.

I have never seen the other side of the window. So I have not seen it ever go back to the 10th being gone. Ultimately it does not matter. But I really wonder how many cars before this year, if any, that we put away because we thought they were running inconssitent times. I did destroy one car this year for the tungsten that was in this category.

Curious if I should share the company name. They have been very responsive. Maybe after they look over the timer I should. I am just sensitive in being fair to the company. If I was not 100% sure of the problem there would be no way I would even consider it. Also why I posted on the forum was for anyone to think of something I had not thought of.

But my conclusion after weeks of using the timer is that it has a very real problem. The problem has been 100% consistent. I am now just curious if it is a design flaw or somehow a one off problem just with my timer.

I have also now gone ahead and built a permanent mount of the start switch for the NewBold. I replaced the switch it came with with a reed switch. It is working really well. It use to be off and on for working but apparently it was the start switch because with the new one it has been 100%. Kind of nice having the two timers. I had a freak run last week where the time was 9 100ths of a second faster than the record. Having both timers with the same results was nice to have. Turned out the car was a little over weight. Still a very fast car even without the extra weight. It also had very little drift and must have made it down without a wiggle.

Re: Running two timers and not consistent

Posted: Wed Jan 25, 2012 10:12 am
by PWD
Wanted to post an update on this as we start back in PWD this year.

We now just use both timers all the time. I can easily recreate the problem as we have two cars that run between the times that cause the problem on a consistent basis. I have three ways to verify. The NewBold timer run in parallel accurately shows the problem. I can actually run the two cars next to each other and you can visually see they are not 1/10th of a second difference. Also if you have been doing this for any time at all you can see 1/10th of a car difference by sound and visually looking at the car run.

The company I purchased the timer from has been as helpful as they can. They explained never seen the problem before. Have no idea what is causing it.

I am not comfortable providing the company name on a public forum. But if anyone suspects they have an issue with their timer then feel free to PM me and I will share the timer brand. The timer was purchased about 6 years ago.

I do find it hard to believe if this was true with all timers provided from this company this would not be exposed. Even if only a problem in a few runs. The timer was purchased originally by us to be used on a 49' track. We left the NewBold on our 32 wood track. So cars were always above the problem time. But if anyone used it for a traditional 42 or shorter track it would always show the time 1/10th of a second high. That is just not something that would not show up. I only have the single track version but on a multiple track version you would see two cars nose to nose and register 1/10 of a second different.

Also I did not send in the timer to the company. I have so much on my plate and happy with our solution that I did not get around to it. Hopefully after this season I will do it just because of my engineering insane curiosity and not able to let things go :)

Re: Running two timers and not consistent

Posted: Wed Jan 25, 2012 11:05 am
by dna1990
I would first guess that exactly 0.100 has to be a software issue in the microcontroller. Some really flaky counter rollover that is too unusual to see in normal testing. But then every timer they stamped out would have the same bug. Now it may be possible there other customers just don't notice?

But I am curious around the testing side of this more. I have a homemade timer project for another venue and have always been perplexed as to how to calibrate or verify it is accruate. When working down to the 0.001 level, it becomes harder to have a standard event to run tests from. So were all your tests with normal cars runs from the top, or did you try some testing where you made a 'device' that you could consistently trigger thru the finish gate? Take the variability of a PWD car and 30 feet of track out of the mix and focus only on the timer. Could you make a 'two-lane' car where a flat nose plate could be affixed to the front and vary the step distance between lane one and lane that lane two always finished say 0.01 seconds after lane one. You then setup something to make lane one finish in the 3.07 range or whatever you said was the trigger point. At which time, lane two would show the 'bumped' 0.1 addition, instead of the usual small 0.01 difference.

Not really questioning your methods, really looking for more ideas for my own verification 'tools'.

Re: Running two timers and not consistent

Posted: Fri Jan 27, 2012 2:03 pm
by PWD
I could not figure out a way to easily simulate the timing intervals that I needed. So my testing was the actual running a car from the top using the 2nd timer. In isolation you could not be sure the 2nd timer was not actually what was in error. But the visual data of the behavior make me comfortable with the assumption that the 2nd timer is valid.

It would be interesting what the timer company would do to recreate. I doubt they have the infrastructure in place to simulate the event timing that is needed. They really need my track and cars to recreate. They would also need a timer without the problem to verify.

I wonder what other timer companies use in QA of their timers? I guess creating the electronics to simulate the closing of the circuit would not be difficult but the device would also need a timer. If the two timers do not line up then how do you figure out which one is causing the problem?