New install - RPM weirdness

paulbutcher · Post by **paulbutcher** » Fri Mar 20, 2015 2:25 pm

I'm installing an RCP into a Jedi (motorcycle-engined single seater). All the analog sensors (water temp, oil pressure, etc.) are working fine, but we can't get anything sensible for RPM.

The engine is a K8 Suzuki GSXR 1000.

We've got the ECU's tachometer output wired directly to the first Pulse/RPM input of the RCP. That input is configured in "RPM" mode, timer speed medium, 1 pulse per revolution, and a 10Hz sample rate.

The behaviour we're seeing is very odd indeed - when the RCP is initially powered on, the RPM reading is zero. When I flick the ignition on, the RPM reading changes to some random value (anywhere between zero and several hundred thousand - yes, really) which it then holds at (i.e. it never changes). Running the engine has no effect on the RPM reading, but when I flick the ignition off, the RPM reading changes to some other random value. Each time I flick the ignition on or off, I get some other value (a typical sequence will be something like 800, 240650, 14000, 400, 136900, ...)

Any idea what the xxxx might be going wrong? Any suggestions for how to diagnose it?

gizmodo · Post by **gizmodo** » Mon Mar 23, 2015 1:52 pm

What version of the firmware are you running? I was having a lot of trouble with RPM until version 2.7.8 (on an MK1).

paulbutcher · Post by **paulbutcher** » Mon Mar 23, 2015 4:54 pm

We just managed to fix this—turned out that we were using one of the coil outputs as the input, not the tacho output as I thought. We've just switched to the tacho output, and everything's working fine

Post by **brentp** » Tue Mar 24, 2015 9:36 pm

Oh hey - can we get a picture of how / where you have it connected to the engine and/or ECU?

paulbutcher · Post by **paulbutcher** » Fri Mar 27, 2015 11:32 pm

So I got to test this "in anger" for the first time yesterday, and the good news it that it works. The bad news is that I don't know *why* it works, which worries me.

It turns out that to get an accurate reading, we had to go with the following settings:

Channel Mode: RPM
Timer Speed: Slow
Pulse per Revolution: 10

10 pulses per revolution makes no sense at all given that the engine is a 4 cylinder—the number of pulses has to be a multiple of 4, surely?! Nevertheless, this gave good, accurate RPM readings.

What does "Timer Speed" mean, exactly? Changing it to anything other than "Slow" resulted in wildly inaccurate readings (it would dramatically under-read, by an order of magnitude or more).

I'm happy that it's working. But worried that I don't understand it.

Post by **brentp** » Fri Mar 27, 2015 11:57 pm

The timer speed sets the dynamic range of the counter that is being used to clock the RPM signal. Fast signals would use the fast timer, slower frequencies would use the slow timer. Most cases uses the medium speed timer.

Without an 'scope trace it is pure guesswork. What is it connected to on your engine?

paulbutcher · Post by **paulbutcher** » Sat Mar 28, 2015 12:05 am

brentp wrote:Without an 'scope trace it is pure guesswork. What is it connected to on your engine?

Yeah, unfortunately I don't have a scope to hand. I'll see if I can source one.

It's connected to the ECU's tacho output (i.e. the output that goes to the tacho on the bike).

Post by **brentp** » Sat Mar 28, 2015 1:13 am

A scope would be super handy to see exactly the pulse rate at a given RPM.

What does 4 pulse per revolution give you with medium timer?

paulbutcher · Post by **paulbutcher** » Fri Apr 03, 2015 6:22 pm

brentp wrote:A scope would be super handy to see exactly the pulse rate at a given RPM.

Mystery solved—I got a scope on it, and the output was a perfect square wave, one pulse per rev. I've attached the output from the engine while it's idling at around 1200rpm (which gives 20 pulses per second, exactly as you would hope).

I've now got the RCP2 configured with one pulse per rev, medium timer, and everything's great.

So that's all good. I'm still rather mystified why the configuration I worked out before I put the scope on it (10 pulses per rev, slow timer) worked (it gives *exactly* the same results as 1 pulse per rev, medium timer).

It would be really helpful if the documentation gave more help about the difference between "slow", "medium", and "fast". How slow is "slow"? How fast is "fast"?

Post by **brentp** » Fri Apr 03, 2015 7:51 pm

Thanks, Paul for the update! What is connected to each of those channel inputs, just to be clear?

The slow, medium and fast setting selects how fast the clock runs for the internal timer. Think of it as a dynamic range for measuring RPM/Frequency. I'll get it on my list to document exactly what range each mode is optimized for.

Thanks again!

paulbutcher · Post by **paulbutcher** » Fri Apr 03, 2015 8:42 pm

brentp wrote:Thanks, Paul for the update! What is connected to each of those channel inputs, just to be clear

There's only one channel input—it's the tacho output from the ECU. The top trace is in the digital domain, the bottom trace in the analog domain.

There are some strange artefacts in the digital domain, which don't seem to have corresponding features in the analog domain—I'm not sure that I'd worry too much about them at the moment, the analog domain is the more interesting one for our purposes (and the configuration suggested by that trace, of one pulse per revolution, is working perfectly).

toga94m · Post by **toga94m** » Sat Apr 04, 2015 2:11 am

I'm betting the spikes that show up on the digital samples are really there in analog land, but your analog samplerate of 625khz is too slow to show the noise clearly. You can start to see a disturbance at the low side of the analog trace, that usually syncs up with a spike in digital-land. Also (without knowing how your scope works) if the analog trace is averaging multiple sweeps to get one displayed trace, that'll hide random noise to some degree.

It also depends on what the digital sampler's set to for a high/low threshold. Could be a fixed 2.5V level or some other threshold voltage, could have noise immunity (rising above 3V sets the bit high, falling below 2V sets the bit low, noise between 2-3V is ignored)... you might be able to turn on noise filtering for the digital channel too. But if the noise really exists, then ignoring it on the scope isn't going to help RCP hardware sync on the signal.. better to see as much noise as possible so you know the enemy.. then figure out how to quiet it down so your real measurement device (the RCP) can trigger on it cleanly.

paulbutcher · Post by **paulbutcher** » Sun Apr 05, 2015 8:50 pm

Here's a capture at 5MHz instead of 625KHz. There's clearly quite a bit of noise around zero, but all seems to be below 1V. Given that the setup is working, I guess the RCP2 is filtering it out?