Tractorsport Flowbench Forum Archive

Posted: **Tue Jul 08, 2008 3:05 pm**

I need a computer card that can recieve information from my rolling road. It should have both analog and digital in- and outputs and it should have counters.

Is there anyone who has any suggestions on which card to go for? It should be a good card but I havn't the biggest budget on this so it should not be overpriced.

Posted: **Tue Jul 08, 2008 10:41 pm**

Hi Sandra, This is out of my field, but the PTS digital manometer "group" is using a Dataq 194 "device". The DI-148 device has 8 channels for analog plus digital conversion. Perhaps one of the members who has a lot more smarts than me can tell if that would work? A "starter kit" runs $50.

Good Luck! -- Doug

Posted: **Fri Jul 18, 2008 10:54 pm**

Sandra,

Doug is on the right track if you are in to DIY and can program in VB or VBA. It would help if you could give some more specific peramiters as to what your true needs are I/E.

Analog devices in
Digital devices in
Output need digital and analog

Rick

Posted: **Sat Jul 19, 2008 8:01 pm**

The problem is, that he requires high speed real time digital interval counters. Something like that usually requires some custom electronics built specifically to suit the application.

I really doubt if there is a simple off the self solution to this.

Posted: **Sat Jul 19, 2008 8:51 pm**

Tony,

I agree the Dataq products are great but require good knowledge of electronics (op amp circuits, votage dividers and clocking circuits) and some windows based programing skills.

Sandy a simple card and DIY budget lacking the above are not a easy find.

Rick

Posted: **Mon Jul 21, 2008 2:44 pm**

The card need to be able to read rpm on the enginge and the rollers and also the tourque that is a 0-5 V signal. The digital signals are up to 2kHz.

I realize that it is hard to get a finished card to buy but I am quite good at electronics and I have people who is really good at programming. I just want to know how to do or what to get. Any help you can give me is very appreciated!

Posted: **Mon Jul 21, 2008 6:55 pm**

The 0-5v analogue torque signal is easy, just use one of the Dataq products suggested by Doug.

The problem with measuring roller speed and ignition pulses are that you cannot just count pulses over a measured time interval. It will be far too slow to respond, and the resolution will be nowhere near good enough to be usable for monitoring a dyno.

You need to measure the exact time of arrival of each individual pulse, and then calculate rpm in software from that. That way, you get fresh new updated rpm information for every new incoming pulse. It responds extremely quickly, to rapid rpm changes, and you can measure the rpm to quite high accuracy each reading.

The only reasonably practical way to go about this is to use a microcontroller that has the input pulse capture feature. This records the exact time of arrival of each new incoming pulse. Some software to make this happen will need to be written for the particular microcontroller. Ask your software friend what he thinks about that.

The only alternative may be to buy one of the commercial dyno monitoring software packages. Not exactly a low budget solution, but everything will be provided to do the job.

Posted: **Mon Jul 21, 2008 8:44 pm**

Posted: **Tue Jul 22, 2008 1:02 pm**

Tony

I have talked to my software geek and he think your idea sounds good but we are not sure on how to make tha calculation for rpm in the software, do you have any suggestions? We have 20 pulses per revolution as it is right now, but that can easily be changed if some other number of pulses is more optimal.

Posted: **Tue Jul 22, 2008 6:35 pm**

First thing you need to know are the maximum and minimum roller rpm, let's assume a 10:1 operating speed range.
Maybe 500 rpm to 5,000 roller rpm ?
You should almost certainly be able to select a suitable vehicle gear to work somewhere within that roller speed range.

If there are 20 pulses per revolution, that would give 10,000 pulses per minute, on up to 100,000 pulses per minute.
Dividing by sixty we get 166.6 Hz to 1,666 Hz.

We need to convert that to time between pulses.
166.6 Hz converted to milliseconds is 1,000/166.6Hz = 6.0 milliseconds.
And 1,000/1,666Hz will be 0.6 milliseconds.

The next thing to look at is the particular microcontroller you plan to use, and how it can best measure that period range. There are two aspects to this, resolution and time overflow.

Let's assume we have a sixteen bit counter that counts up to 65,535 and it is clocked from a 1Mhz source. It will increment every microsecond, and overflow back to zero, and continue counting up to maximum every 65.535 milliseconds. It does this continuously, it just keeps counting up to the maximum time, than overflowing through zero and counting up again, over and over again all by itself.

What happens is that an incoming pulse from the dyno takes a snapshot of this counter and records the exact count at the time of arrival. You then must store that value away in ram. The next incoming pulse takes another snapshot of the same counter. If you subtract the old value from the most recent value, you get the time interval in microseconds between successive pulses in microseconds.

At exactly 500 roller rpm, we end up with a measured time interval of 6 milliseconds, and a value after subtraction of 6,000 (microseconds)

At exactly 5,000 roller rpm we end up with a time interval of 0.6 milliseconds and a value after subtraction of 600 (microseconds).

To convert microseconds back to rpm, just divide three million by the period in microseconds.

3,000,000 divided by 6,000 microseconds gives 500 rpm.
3,000,000 divided by 600 microseconds gives 5,000 rpm.

This method gives excellent speed resolution, and a very fast speed update rate.

The only problem is that the speed readings will very likely jump around due to mechanical jitter and vibration, but as we are receiving from 166 to 1,666 true speed readings every second, we can afford to average over many readings, and still get very fast response to real speed changes.

My own dyno works exactly this way, and this method of rpm measurement works very well.

Posted: **Tue Jul 22, 2008 7:32 pm**

Tony;

I understand your logic clearly i am still unclear why you feel that a Microcontroller must be used to do the work over the PC and ADC that can communicate sufficiently to perform these tasks and the software and PC should have no problem keeping up.

Am i missing something?

Rick

Posted: **Tue Jul 22, 2008 8:00 pm**

Rick, the problem is latency. How long before your PC can finish whatever it is doing, and recognise and measure the arrival time of a random asynchronous incoming pulse ?

This simply cannot work as well from an interrupt driven PC architecture, even a very fast one. It absolutely must have very fast dedicated time capture in hardware.

Once the exact arrival time of these pulses have been logged, the subsequent number crunching can be relatively slow and intermittent, and a PC would work fine for that part of it.

I am using a Motorola HC11 for doing this myself. It has three suitable input time capture channels, and some excellent 16x16 bit division instructions including an extremely useful fractional division instruction.

My chassis dyno system uses two chopper discs (separated by a torsion bar). With two optical pickups, and my HC11 microcontroller. It measures both rpm and torque digitally. Rpm is measured pulse by pulse as mentioned above, and torque is measured by how much one disc time lags the other, as torque twists the torsion bar.

There is no analg circuitry involved at all in torque measurement, it relies entirely on time measurement, and is completely digital. It also has extremely fine resolution and rapid update, as does the rpm measurement.

Posted: **Wed Jul 23, 2008 3:09 pm**

Thanks, now we know how to calculate it, great. Although we do realize that the card we are using is not nearly good enough, we probably need something more like your card. Do you know were to find anything like it?

Posted: **Wed Jul 23, 2008 7:25 pm**

I did all this quite a few years ago, for a guy that was building an engine dyno. I designed a circuit board and wrote the software for it, and he paid for the cost of the bare boards. I kept one for myself, and he had the other.

Unfortunately Sandra, these circuit boards are not something that you can just go out and buy from somewhere. There are no more except what we made at the time for ourselves.

Posted: **Sun Aug 17, 2008 8:37 am**

Sandra;

In doing some research i came across this company reasonably priced cards, you program.

Tony how do these look?

Rick

Tractorsport Flowbench Forum Archive

Computer card