Intelligent glow driver?

[badcockr]

Hi chaps,

Before I get too far with this I would appreciate some feedback!

Attached is a rough schematic for an intelligent on-board glow using a PIC 12F675 and low R FET switch. And now for an explanation .......

JP1 is the RX channel output for drawing power fro a 4 cell NiCd and possible 'switch on' signal from the RX. The PIC checks for valid 'switch-on signal' and powers the FET (T1) on - enabling current to flow from B1 (single cell NiCd) into the glow plug on JP2.

R5 is a low ohm 'sense' resistor in series with the glow plug and enables the current through the plug to be measured via the single rail Instrumentation Amp (AD623 gain ~20). The output from the AD623 is fed into AN0 of the PIC.

Here comes the clever bit .....

Assuming a typical glow plug cold resistance of ~ 0.15 ohms, and a hot of ~0.5 ohms, a cell voltage of 1.2 Volts, FET on-resistance of 0.04 ohms and combined internal R of battery and lead resistance of 0.02 ohms then:

AD623 O/P voltage (Hot) = 0.85 Volts
AD623 O/P Voltage (Cold) = 2.45 Volts
Or a control Output of ~0.4 Volts per Amp in the glow plug

In order to optimise the glow power usage (a hot glow doesn't need power!) then the glow is energised for 10 milliseconds every second and the resistance and the current draw measured. If the current draw is high then the plug is cold and power is kept on until the measured voltage is below 1.2 Volts (current of 3 Amps or R of ~0.4 ohms)

By periodically checking the glow resistance then the plug is supplied power only when cold - optimising battery usage!

For safety I have fitted SW1 - a switch to manually turn the glow off and enable safe charging of the NiCd to take place.

Any comments on my design thoughts?

Rod

[Ironsides]

Rod:

That is quite clever.

My design is quite 'dumb' and is triggered strictly by throttle stick position.

http://www.rc-float-flying.rchomepage.com/RCWeb/PIC/PBP%20Glow%20Driver.htm

David

[badcockr]

Ironsides:

I've seen your design and think it suits it's purpose for single cylinders.

What is driving me towards a more intelligent design is a quad engined Lancaster - I don't want the overhead of too many battery packs but also don't want the risk of an engine dropping.

If we can make the intelligent scheme work then I can use the two spare I/O on the PIC to make a dual cylinder version using only one cell. The capacity requirements for the cell are greatly reduced and should ensure longevity in this scheme.

Any comments would be usefull before I start ordering components (especially from anyone else who has tried intelligent schemes)

Rod

[vintage1]

I am pretty sure this is how the best intelligent glo drivers work - they sense the temp of the plug and only heat when needed.

Good luck. Seems a great project.

[Mr.RC-CAM]

If you do not mind a little more programming work, I think you could Muntz the circuit by eliminating the current sense resistor. The MOSFET could be used in its place.

You'll need to accomodate "reversed" servos, so a user programming mode would be nice. That would only take a removable switch or jumper. This could also be used to calibrate the FET (if necessary).

Also, it would be really nice to have a status LED. It could wink out the "temperature," perhaps using one flash per 500mA of load current. If the glow plug was bad (open), it could report that too.

Looks like fun!

RC-CAM

[badcockr]

RC-CAM:

Good ideas!

Do you know how stable (and variable from device to device) the on-resistance of a FET is? I am assuming that the on-resistance is proportional to e^(-kT) so I plumped for the fixed R - however I am not sure! If it is then there could be problems due to self-heating.

Using the FET resistance would certainly improve efficiency - anyone got any info on resistance stability?

The servo signal I had assumed would be from an auxillary channel - just to switch on and off, not linked to throttle. Might revisit this now.


Status LED and 'learning' the plug are excellent suggestions. I'll think about the circuit mods for this and I'll post an updated suggestion this evening.

Rod

[Mr.RC-CAM]

Do you know how stable (and variable from device to device) the on-resistance of a FET is?I do not have any data on that. However, I expect that the variations are probably very stable within production lots, but wide enough outside that to require "calibration." It could be as clever as running the engine at full throttle, then toggling the R/C channel's switch in a set pattern (perhaps three rapid toggles), to tell the PIC to re-calibrate to the high temperature.

The sense resistor method makes it an easier device to develop, so you might wish to prototype it that way to prove out the basic concepts.

RC-CAM

[badcockr]

Okay,

Attached is the version 2 schematic. In terms of hardware the only changes are:
1. An LED to indicate the plug temperature (connected to GPIO4)
2. An LED to indicate that intelligent mode is activated (GPIO5)

This will require a different operation and software. The operation of the driver can be summarised as follows:

1. The switch at the output is turned on to enable power to the glow.
2. The signal from the auxilary channel is checked; if high then power is applied to the glow plug and the default intelligent routine is started.
3. LED1 indicates the plug temperature; permanently lit is a cold plug, flashing at twice a second is nominal and unlit is hot.
4. LED2 lights whenever power is applied to the glow to keep it warm.

I would like to put in an element of programming for plug termperature but some considerations make me hesitate:
I only want to use one receiver channel for the circuit (currently switch-on).
Most glow plugs have the same nominal resistance.
I want to keep operation simple.

I think the circuit will allow more advanced programming of the PIC to be carried out at a later stage.

I hope to be building and testing over the next two weeks.

Rod

[Mr.RC-CAM]

I suspect that you will find a wide variation in plug resistance vs. temperature (the elements vary in size and material content). If true, then the cal step will probably help out.

BTW, I think R6 is incorrect. It should go to Vcc if it is your A/D ref.

Please keep us posted on what you learn -- sounds like interesting project.

[badcockr]

RC:CAM:

Well spotted on R6!

(ahem) R6 should of course be connected to the supply rail as it is current limiting for the Zener Vref.

I'll be testing common plugs such as OS, Taylor and Fireball and looking at the resistance variation. In the first instance I'll probably use a trimpot set in a potential divider from the Zener and feed Vref with an adjustable voltage - this will give an adjustable Vref and hence allow plug resistance variations to be handled easily.

On the update loop I'll probably reduce the 10 ms 'hot' period and bring the cycle time down to 0.5 seconds (this is a stab in the dark as I have to find the thermal time constants experimentally).

I'll keep you posted over the next two weeks.

Rod

[badcockr]

OK Chaps,

Thanks for the feedback - I now have (what I think) is the final design and I'm now starting build / order / test.

The updated schematic is included below, and incorporates several new changes:
1. The PIC is powered from a LDO variable Volt Reg (LM2931) delivering 4V to the PIC. This allows 4 and 5 cell NiCd packs to be used with the circuit.
2. The fixed Vref (Zener) has been replaced by VR1 as a potential divider across the regulated 4V supply. This allows the unit to be set-up for any glow plug by trim-pot adjustment.

If you talk nicely to me I'll post the picbasic code on this forum once testing is complete :D

Rod

[odalaman]

Hello Rod,

Could you please post EAGLE files also.

Thanks

Osman

[Zeta Phoenix]

Well I am hooked!!! I definatly want to build me one of these! Please post the code!! (If begging is nessary, you shall get it!) Also, it would be neet if you could make it run off the 1.2v bat, so you do not cause radio glitches by a low battery during landings when the plug may get cold. Also I would like to adapt this to make a stand-alone starter using a 1.2v batt, and have it be a smart driver.

[clipclop]

P4 Hover,
The plug is driven from its own cell . also I think 1.2 volts will be too low for the Pic . the RX bat is only supplying the control cct not driveing the plug its drain on the RX bat would be low .
Stewart

[Zeta Phoenix]

Yea, you may be right on the low voltage for the PIC. I was hopeing to adapt this to make something to a Radio South Pro Driver.

[jhayden954]

Any updates Rod? I can't wait to hear on your progress.

[badcockr]

jhayden:

yep - progress has been slow due another project (charger interface (http://www.rcgroups.com/forums/showthread.php?t=214510)). Learning to program a Palm entailed more than I expected :eek:

With the Palm interface up and running I have more time available and I am back on the glow driver. The advantage now is that I can interface the palm to some of the PIC circuits and this opens up new possibilities!

Rod

[Michael Hall]

That is superb, nice work, Any possibility of a list of components and their values in a table? My electric knowledge is somewhat limited although i can follow your schematics. These are perfect for our helicopters as they tend to like a bit more heat than usual IC engines.

[n2wizard]

Guess no one on this forum is old enough to catch MR. RC-Cam's quip about
"Muntzing "the circuit. In the 50's during summer break I installed TV antennas for Mad Man Muntz. He made a very cheap TV set by eliminating about half of the circuits, and it was a bear to get the set working half way decently. Just a wild guess that that's what
RC was refering to.
Herman (DOB 1927)

[deh6]

Rod,

Your schematic looks fine. The idea is similar to what I posted in the following thread earlier this year:

Inflight glo-plug driver

I've got my design to the point where, for hardware, I have a bunch of PC boards, and some of the units completed. I included a number of features for experimentation, so I haven't gotten around to flying it, but that will be soon.

I too had considered putting in a 0.02 ohm sense resistor, but the lead from the FET drain to the glow plug will serve just fine (e.g. 6" of #26 (or was it #22?)) is equivalent). By doing this, a "sense" wire from the glow plug is needed, but unless one uses big (heavy) bus-bar wire, the resistance from the FET to glow plug is going to be large enough that one might as well include it in the circuit.

I did without a glow-plug battery switch. The off current of the FET is so low that it is not a factor in discharging the battery during storage (self-discharge is probaby greater). Also, the current is high enough that the switch needs to be "good." The first pulse through the glow plug when the drive starts is of the order of 7+ amps for a cold plug. (The time constant of the glow plug is around 1/3 sec, so in 1 sec it up to temp.) I felt it was better to leave the switch out.

So far, I find that at 1.2 - 1.4 volts the glow plug resistance is around 0.33 ohms. Your figure of 0.5 ohms looks rather high--unless it is getting nearly white-hot!

Rather than go with an op-amp feeding an A/D convertor I put in a comparator and bridge circuit. The op-amp raises issues of voltage-offset on the inputs versus temp, i.e. drift. Instead I went with a brdige circuit. A pot in the bridge circuit adjusts the point where the comparator switches.. Since the comparator came as a dual, I put an RC on the 2nd comparator and used the time for the cap to charge to the point matching the voltage across the sense resistor. I get a good resolution current measurement. Both schemes work, and yours with the op-amp A/D convertor should too. I didn't look up the specs on the AD623, but you might ponder what the effect of temp does to the setting.

Be careful with board layout. The voltage drops across the current carrying traces is substantial, so the traces all need to be viewed as small resistors.

Batteries: My first test was with a 1300 mah NiCd battery that came out of a glow-plug driver (Sanyo). One software program sends voltage, current, etc. to the PC over the serial port, so that I can measure battery resistance (including leads), glow plug resistance, etc. With the glow plug drawing roughly 3.3 amps I got the 1300 mah out of the battery, but that is only about 15 minutes (at full drive of course).

I then tried two 1800 mah NiMH cells in parallel (roughly the same weight). I got 2400 mah out of an "off brand" (Lenmar) and 2700 mah out of Rayovacs (from WalMart). For the same weight the NiMHs are in improvment (as one would expect).

One interesting measurement was to measure the battery resistance. This measurement included leads, and PowerPole connector, so the total was approx 0.035 ohms. Taking this measurement versus discharge time I could see how the battery resistance changed as the battery ran down. Surprisingly, the NiMH was so close to the NiCd that one could consider them equivalent. I hear the NiMH have higher internal resistance, which from my measurments appear to "slightly true," but the difference hardly measurable. The very tail of the discharage--when the battery goes from about 0.95 volts to 0.7 volts is where the difference shows up, but the time duration in this range is so short that it is only academic interest.

The next test "up" is do some running engine tests. Sometime back my measurements showed that the glow plug was about 0.23 ohms with the engnine at a fast idle and 0.35 with it running full tilt, leaned (Tower .46 engine). The key will be if the glow drive setting can be set such that it is virtually off while the engine is running normally, but high enough that it "loads up" it will keep it alive. If not, then the old scheme of turning it full on whenever the throttle is below some setting is probably good enough, and a heck-of-lot simpler.

Oh yes, turning the glow drive on. My first scheme, landing gear (ch5) switch: a) turn on rcvr, thus powering the micro. Glow drive is off, b) Turn on landing gear switch. d) glow drive is full on for "x" seconds (facilitate starting), then goes to a lower setting.
Variation: pushbutton triggers the driver on for "x" secs, followed by the "running setting." Another scheme yet-to-tried, a) as above, b) If engine is detected as running (hall effect sensor, magnet on spinner), then switch to "running setting," and if not running switch the glow drive off (so one can prime the engine without it kicking).

And another item. With regard to possible interference with the radio rcvr--I connect the engine frame to the glow battery positive and switch the glow plug to ground (via the FET). This way the "large" engine with associated capacitance is at a constant potential and the "small" wire to the glow plug (easily shield if necessary) switches. Final thought--since the rcvr input signal is available one could do the FET switching synchronously with the received pulse train, putting the switch point at a point late in the cycle so that interference would affect a non-existent servo (last resort stuff!).

Don

[badcockr]

deh6:

First - can we trade circuits? I am interested in your proto layout for comparison.

A few notes to ponder:

1. The glow plug battery switch is a safety and convenience feature - without it you cannot be sure the plug is off and it would make charging the battery problematic. The current is not an issue as the mean current (read heat) is still low.

2. Sense resistor - doh! Of course a defined length of wire *or track* works equally as well (and more efficient).

3. What make and type of plug are your resistance measurements for? How many types of plug did you measure and in what engines - can we trade data as the suppliers don't have a clue?

4. The AD623 is an instrumentation amp - very stable with temperature. I could not rely on an op-amp.

5. Ditto on the NiMh cells - ideal application for those dirt cheap 2000 mAh AA cells sold in Tandys e.t.c.

6. Noise - interesting suggestions! The idea of switching late in the Rx cycle is interesting but of course *much* more complex to implement.

[deh6]

Rod,

1. Trade circuits? Yes, and I'd be happy to send what I have to anyone else, though at the moment it is a "work-in-progress." (I got DSL recently and it comes with a web page. If there was interest I could put doing a web page on ny list of things to do.)

2. I just looked up the AD623. Indeed, the specs look quite good. Some of the older op amps have some ugly offset voltage v temps specs.

3. I have only checked a few glow plugs (3 ?) . Ones I happened to have in my field box. OS A8 was one of them. I would think that some of the more specialized ones could be substantially different.

I'm plan to do some "engine running" tests very soon. I'll post what I find.

4. Is the duration of the 10 ms FET ON test pulse to check if the plug heat set by the time required for the AD623 to settle and the A/D conversion be made? Currently I have my test pulse down at 0.1 ms, and that could easily be cut in half.


Don

[6pack]

Well fellas, what happened to the rest of this post?
Or can anyone recommend a good Intelligent glow driver?

6pack

[Acetronics]

Or can anyone recommend a good Intelligent glow driver?


Hi, 6Pack

What's your definition of an intelligent Glow-Driver ??? ( I'm not joking here ! )

Alain

[lazy-b]

Guys: Have saw a similar project on RCM magazine using a simple analog OP-amps circuit.

Its would be nice if you can eliminate a Single Cell Battery to Power a Glow Plug.....Futaba have this GH-1 On-Board Igniter that get its power from the Receiver Battery, this is just a very efficient DC-DC converter that convert a 4.8 Volts to 1.25 Volts.....Have been using this on my Scale Helicopter, have put a Micro-Switch near a HEX Starting Cone, this will automatically turn-on an Glow igniter when a Starting Shaft is inserted.

I Hope somebody can program a PIC controller to work as a glow plug switch and also work as a High Efficient DC-DC converter.

[Acetronics]

Hi, Ellion

I think such a device can be useful on a heli ... as idling the engine does not happens very frequently. And you fly close to the pilot, so there are no problems to know what's the batt level.

From my plane experience, I keep 2 things :

1) How does a switching supply interfere w/ a weak signal into the receiver ??? ( you fly at some distance ... and ground is common to Rx and Supply, moreover there's necessarily an electrical link between moving parts of the engine and the Rx ...)

2) Is it safe to add a , say 1 amp load ( 3A glow with 75% efficiency ) ..., to connectors, switches and wiring ( as best wiring is .3 sq.mm ) ???
For planes, the idling situation is quite frequent ...

I note there's a switch when you apply the starter ... so, I presume the device is used only once at startup, and under the tX antenna.

At the end, I've seen many opto isolators on such glow lighters ...may be they are here not only for the show ...

Alain

[lazy-b]

Alain, Yes, Your Right.....it is Intention of the FUTABA GH-1 Glow Plug Igniter just to Start-up the Engine only, they make a simple PUSH-ON button to Turn-on the Glow Igniter.

This Futaba GH-1 Glow Plug Igniter really make a lot of noises, I can not use my FMA Dual Conversion FM Receiver with this FUTABA Igniter, cause when I turn-on the igniter all the servo start Moving Back and Fort......the problem disappear when I use FUTABA Dual Conversion FM Receiver.

Have also tried using a simple 555 timer and Power Mosfet wired as Simple oscillator 3 KHZ with an adjustable duty cycle from 0 to 25 percent.....works perfectly, but have to connect it directly to the Reciever battery, due to very high peak current of 12 Amps, but has an Average current of 3 Amps

MAXIMS has this Single Chip DC-DC converter with a Built-in Power Mosfet inside the Chips, its a 3 AMPS 1 MHZ DC-DC converter with effienciency of 93% , just need an Inductor and a Couple of Capacitor and Resistor. have already asked for Samples a Year ago, but until now have not received any sample units.

Ellion

[Punkie]

http://www.southhertsmodels.co.uk/
These people do an intelligent glow driver, its avallable in the USA I think from Nelsons.
http://nelsonhobby.com/store/product_info.php?products_id=650
A twin version is avalable too.

[Acetronics]

Hi all,

In fact, I'm asking myself if a constant glow temp is a really a good thing ...

as it sets the ignition moment ...a constant temp glow will add ignition advance, compared to classical system, which is an open door to "knocking".

just a thought ... ;)

Alain

[nyxtech]

I would like to make Intelligent a Glow System for me, I alguem I could send the project for me with codigo HEX to record pic!

nyxtech@yahoo.com.br

[Acetronics]

I would like to make Intelligent a Glow System for me, I alguem I could send the project for me with codigo HEX to record pic!

kleber@krsmodelismo.com

Hi, nyxtech

I have to tell you these drawings and files are copyright protected ... I wanna be sure you confirm your NON-COMMERCIAL use ...

Thanks

Alain

[nyxtech]

The project is for same me, I am an enthusiastic one in aeromodelling, and would like something different. I do not obtain to find nothing similar to this system to buy then I decided to make one for me!

[Acetronics]

Ok, ... ;)

I just had found this very familiar project looking like a product on your site ...for sale.

Hope you'll understand my question .

Alain

[nyxtech]

Without problems, this project has a PIC 16F628, does not need to have program HEX to record it? This project is on in series with the servo of the accelerator?

[nyxtech]

Without problems, this project has a PIC 16F628, does not need to have program HEX to record it? This project is on in series with the servo of the accelerator?

[nyxtech]

This glow plug intelligent that you it passed me possesss a PIC 16F628. It is necessary to record a code HEX in it? You it could supply this code to me?

[Acetronics]

:rolleyes: The scheme I passed to you is the copyright protected original scheme of the Glow driver YOU SELL on your site :

http://www.krsmodelismo.com/index.php?nOptnMenu=2&sViewNews=Visualizar&nCodeNews=27

... ( yours just has an opto-isolated output stage added... to cancel the original PWM glitches ...)

All features, DIL switches, pushbutton, led, program ...are EXACTLY the same as originally shown here :

http://lichtenberger.s.free.fr/GlowSL.htm ... and Sylvain Lichtenberger ( a friend of mine ... and University electronics teacher ) published this project some YEARS ago !!!


Understand here I CERTAINLY will NOT give you any scheme or soft I could see some weeks later on the market, shown as a " pure Brazilian product "!!! :mad:

Sorry for the " pure Brazilian product " ... :cool:

Alain

PS: Too much intelligent, that Glow driver ???? :D

rePS : No need to send me further mails ( including my mailbox !!! ) ... answer is definitly NO :censored:

[rodrigosim]

I saw this project GlowSL and had access the site where I made download of ASM and HEX, recorded the HEX in pic 16F628-04/P and followed all instructions, but the circuit does not enter in way programming of the firing points. Then I compiled the ASM and I generated another HEX, I recorded the same in the PIC and happens problem the same. I disconnect the receiver, pressure the button of the Glow Driver and bind the receiver, but the circuit does not enter in way programming. Somebody knows to say me what it can be happening?

[rodrigosim]

I saw this project GlowSL and had access the site where I made download of ASM and HEX, recorded the HEX in pic 16F628-04/P and followed all instructions, but the circuit does not enter in way programming of the firing points. Then I compiled the ASM and I generated another HEX, I recorded the same in the PIC and happens problem the same. I disconnect the receiver, pressure the button of the Glow Driver and bind the receiver, but the circuit does not enter in way programming. Somebody knows to say me what it can be happening?

[Acetronics]

Hi, Rodrigo

Fisrt, simply short R ...( 100 Ohms on scheme , 1k on list ) . Non working troubles come from there !!!

a simple 10 Ohms resistor might be enough, with a 100 to 220µF decoupling cap between PIC Vcc and Vdd;

corrected scheme here : http://www.rcgroups.com/forums/showatt.php?attachmentid=883244

I've tested this scheme for a friend having trouble, his radio was a Futaba one ...

I think he also had to put a level adapter on the input: PIC schmitt trigger inputs do not like Futaba "weak" signals.

YESSS ... I post it here.

Alain

PS : Pdf file much, much, much better !!!

[rodrigosim]

:) But I used this correct project to make my circuit. A friend mine made a time a GlowSL and he gave to me, I removed the PIC of it and placed in my circuit and with it I obtain to program pressing the programming button.

Already when I record my PIC 16F628, the writing of the PIC is effected successfully, but when I place in my circuit and I try programs it, it does not enter in way programming. I disconnect the receiver, squeeze the button of the circuit keeping pressured it and bind the receiver again, but it does not enter in programming way. :eek:

The Led is blinking and it does not function. I already tried some alternatives but without success! :confused:

[Acetronics]

The Last solution ...

[rodrigosim]

:( I am using MPLAB and the ICD2 to record the PIC. He will be that he cannot be some configuration in the hour to record? I used this project. :(

[Acetronics]

Exactly that I tested ... without any pain. I use MPLAB and PicStart+ ...

Is your PIC a 628 A ??? as stated in the .asm

I'd verify the hardware ... especially the pushbutton ( I once had one out of use ... despite it was brand new ... )

Alain

[rodrigosim]

I am using the PIC 16F628 only and not it 16F628A. In the .ASM this PIC16F628. But I go to verify everything again!