Continue to Site

Welcome to our site!

Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

  • Welcome to our site! Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

Model Gas Turbine ECU

Status
Not open for further replies.

Barry

New Member
Hello all,
I want to build an ECU for a model turbine engine. Inputs would be RPM and Exhaust gas temperature. This data would be manipulated in a PIC, probably a 16F877, and would control the fuel pump which controls the turbine speed etc. to prevent over speeding the turbine etc. Model turbines idle at approximately 30,000 - 40,000 RPM and, depending on the model built, would run at 120,000 - 150,000 RPM.
Questions: 1. Has any forum member attempted a project like this?
2. To determine RPM one can use either a Hall affect sensor or a LDR system or possibly turbine pressure. At these speeds, which would be the better option? Worried about the affect of sunlight on a LDR system! How would this be connected to the PIC?
3. Temperature would be measured with a thermo couple, probably the K-Type. How would one connect the thermo couple to the PIC? Would one need some sort of amplification circuit? Temperatures to be measured up to 650 - 700 degrees Celsius.

I realise I am way out of my depth here as a complete novice but I think it would be a very rewarding project. Any comments or advice in this regard would be really appreciated.

PS. The PIC would have a 16X2 line LCD display connected to show temperature, RPM, battery voltage, fuel pump setting and a few other readings relative to the project.
Thank you all in advance.
Barry.
 
1. I've done engine gauges.
2. I don't know what an LDR is specifically, I assume some sort of optical device? I don't know the speed of Hall effect sensors, be sure to check. There's also the possibility of a simple pickup coil too, though the Hall still sounds right.
3. PIC's 10-bit ADC has a minimum size bit size of 5mV, and there's some significant offset error too. In short, you cannot use it for measuring the very low voltages of a K-type.
You will either need an amp, or an external ADC with either a programmable gain stage or a high bit count. Be sure to review the spec sheets carefully for offset.

If you have trouble finding an op amp with a low enough offset for the accuracy you need, look for a chopper amp, like the LTC1051. Very low offset, but be aware that if the output overloads, it takes a very long time- milliseconds- to recover.

Be aware that the thermocouple junction can't contact ground when you use an op amp or a nondifferential ADC. The best thing is a differential ADC, then you can tack the junction directly to a hot surface on the engine and float the two wires. As long as the input doesn't exceed 300mV, the differential ADCs by Linear (LTC) and many of the Burr-Brown Texas Instruments can accept it without having to generate a neg supply.

You will also need to have cold junction temp compensation unless you can deal with offset errors of 50 deg or more. K-type voltage is a difference between the hot & cold junctions, as the point where the thermocouple wires meet copper in the ECU heats up, the reading will decrease. Just use a thermocouple located near the cold junction to adjust the value. However, PIC's max 2.5K input impedance is sort of a problem for this. You can use an op amp to fix it, not worry about the self-heating and current loss, or not worry about the inaccuracy of failing to meet PIC's max input impedance spec. Or use a digital sensor like the DS18S20 (I think I got the number right).
 
Hello Oznog,
Thank you for your reply. I still have a lot of homework to do on this project and your advice is a start. Anybody else out there with an input or two?
Thanks.
Barry
 
150,000rpm sounds like a really high number, but it is only 2,500 revolutions per second, which is 4us per revolution. To put this into perspective, with an 8Mhz clock, a PIC or AVR can perform one instruction per us, so if you ran one of these deives at 16Mhz you could get 8 instructions per revolution.

Feeding this into something like a capture compare port and treating it like an incomming PWM you could just read it every few cycles ( say 10 cycles ) and prety much know what sort of speed it was spinning at. Your sample rate would depend on how quickly you needed to react to changes. Could you live with 10 or 20 revs ? ... In all honesty, I don't think you're going to be able to do much anyway, if it starts to run away. So 10 or 20 cycles ( or revoltions ) might give you a heads-up to start running :)

An IR opto disc should be able to read that sort of speed, I imagine it would depend on the speed of the receiver or the response time of the photo-transistor. I wouldn't bother pulsing the IR diode. Just run it on DC and count the revolutions. There are plenty of devices that could do this. I'm not sure about hall-effect devices, they may not be fast enough, but I'm sure a photo-transistor feeding a schmitt trigger input could cope.

EGT would be pretty straight forward, just get a small K-type thermo couple.
 
Sorry Multiple Post ... I was playing with my Hosts file after seeing calls to banner-x.com which it turned out were coming from this forum ...
 
BartSimpson said:
150,000rpm sounds like a really high number, but it is only 2,500 revolutions per second, which is 4us per revolution. To put this into perspective, with an 8Mhz clock, a PIC or AVR can perform one instruction per us, so if you ran one of these deives at 16Mhz you could get 8 instructions per revolution.

2500 revolutions per second -> 400µS per revolution
A pic running at 4Mhz could do 400 instructions in between...
 
Cool. Even easier then.

BTW, it's about 1:15am where I am, and I've had three or four glasses of red (okay, so it's been eight, but who's counting) , so don't pick on a bloke for pressing the wrong button on the calculator :) Besides, I've been studying datasheets for the last three days and using this forum for comic relief, so my mind is pretty much jelly at the moment :(
 
BartSimpson said:
Cool. Even easier then.

BTW, it's about 1:15am where I am, and I've had three or four glasses of red (okay, so it's been eight, but who's counting) , so don't pick on a bloke for pressing the wrong button on the calculator :) Besides, I've been studying datasheets for the last three days and using this forum for comic relief, so my mind is pretty much jelly at the moment :(

Im' not picking on you, just correcting :p
 
Hi guys,
Thank you very much for your input thus far. Much appreciated.
The loads of questions and cries for help will come later on :lol:
Barry.
 
Exo said:
2500 revolutions per second -> 400µS per revolution
A pic running at 4Mhz could do 400 instructions in between...

There are also much faster PICs. PIC18 series runs @ 40 MHz clock, 10 MHz instruction cycle. If you have to measure 150,000 rpms on an interrupt basis, you probably want that. Even still, there's only 66 cycles between interrupts. With that high of a speed, you might want to put some sort of hardware counter in front of it.

I never understood how to use that Capture function of the CCP for the tachmeter I did. It doesn't reset the TMRx counter, so the captured value only has meaning if you manually compare it against the last captured value.
 
Hi,

I am new to this forum. My name is Joseph Saliba. I found this forum by a search on Google for Model gas turbine ECU.

I so have an ECU, a comercial one and it uses an 16F876. It is like a 16F877 but have less input and output. I would like to build one myself. The ECU I have uses Infra red transmitter and receiver and works good. To avoid the sun, I did a tube long about 15mm and insearted the infrared in the tube. There will be some `shade' to the infrared receiver hence, it won't be affected too much. I know several ECU's that uses the hall effect type, but I have also read that these, sometimes, are affected by digital servos. One point to mention is that to use an infrared, you must do a hole in the spinner of the turbine. So the PIC will be counting two pulses per revolution. I spinned up my turbine to 122,000 RPM, therefore 244,000 pulses without a problem. One must mention that the crystal used in my ECU is 4Mhz only. I use a basic compiler. I can mention the name here of this copiler? I already begin something on the ECU software, but, I have a larger project before this ECU.

Regards

Joe
 
Oznog said:
I never understood how to use that Capture function of the CCP for the tachmeter I did. It doesn't reset the TMRx counter, so the captured value only has meaning if you manually compare it against the last captured value.

Exactly! The timer is free running and every time you have an "event" on your input pin, it stores the current timer value. By comparing it to the time of your last "event" you can extract a lot of useful information. All you have to do is ensure that the time it takes for the timer to overflow is longer then the longest period of the "event" you are capturing otherwise your data will be useless.
 
salibaj said:
I am new to this forum. My name is Joseph Saliba. I found this forum by a search on Google for Model gas turbine ECU.

OK, you're new so you probably don't know here. It's considered a bad idea to resurrect a thread that has been dead for 9 months. That conversation is done. It's ok (and quite preferred) to simply start a new thread even if it's related to an old thread. That way we're not reading archived comments like they're fresh conversation.
 
Status
Not open for further replies.

Latest threads

New Articles From Microcontroller Tips

Back
Top