Feedback Control - How?

[Peaky 556]

As a total novice with modern R/C systems, I don't know what is possible or available, but I know what I need!  As a mechanical engineer I am aware that control systems can be complex and need to avoid oscillations/resonance etc when trying to use feedback from an output to control an input.
The scenario is speed control.  My application involves internal combustion engines with a servo on the throttle, but it could equally well be a live steamer with a servo on the regulator.  The difference may be the lack of system inertia with mine, as a small throttle opening can result in a large and rapid increase in speed, unlike in a steamer when it may take tens of seconds or minutes even for the loco and train to gather speed.
The question is, how does one use a speed related signal to feedback to the servo and cause an adjustment to be made, with the aim of maintaining a constant speed?
My output could most easily be a DC signal voltage rising in proportion to rotation speed.  There would be a need to adjust speed from the transmitter, so the action of opening the TX 'throttle' would need to allow a higher, but regulated, speed.  Likewise backing off the TX throttle would need to progressively close te engine's throttle servo until a new and lower speed equilibrium is reached.
I want lots of ideas please, don't hold back, the wild, the practical, the zany, innovative etc, along with any known easy solutions for live steam, but with the proviso that my application may need to be rapid acting and possess good damping.
Regards,
Tim

[cabbage]

You do seem to want things complex!!! Now coming from a very simplistic background this is how I would do it.

Your IC engine turns your dynamo. This generates DC. This then is smoothed via a "dustbin" Capacitor Bank and feeds a Voltage regulator (Vreg). The smoothed DC is them fed to the ESC and thus to the motors. This is all standard. If you look at the ELECTRONIZE series of ESCs for locos then there are a few control options!

What you want to do is examine the voltage produced by your dynamo into the capacitor bank. If this falls (i.e the Vreg is pulling more current then you need to increase your RPM from your IC engine. Similarly you need to lower the RPM of your IC engine if the Voltage held in the Capacitor Banks rises.

If you are with me so far -then this is the "complex" bit!!!

You have a "Reference Voltage". This is a SLA battery pack or similar. You then have a small electric motor operating a screw push pull linear actuator. It the voltage in the Capacitor Bank RAISES them current flows from the bank to the Reference Voltage and the motor turns. Similarly if the voltage in the Capacitor Bank FALLS then current flows from the Reference Voltage to the Capacitor Bank and the motor turns in reverse.

I would mechanically damp the system using thick greased felt pad on a fixed disc held against the disc fixed to the rotating shaft of the motor.

regards

ralph

[IanT]

I understand that R/C units are pretty clever these days Tim but I cannot advise you in this area. I see Ralph has suggested a more 'electronic' solution and I'm sure this would be a simpler approach in many ways.

However, (since you asked for "way-out" ideas) it seems to me that what you need could also be achieved via an element of computing. The traditional route for this would have been to programme a PIC or Stamp computer. To do this you also needed a "programmer" or ICE type device and some knowledge of PIC assembler or 'C'.

There is however another option these days and it is called "Arduino" - have a look at this site for an overview - http://www.arduino.cc/

I am currently investigating this technology via an Arduino "Uno" board. It cost me £18 and can be programmed directly from my laptop via a USB cable. The PC runs the Arduino programming environment and this can be downloaded FOC. Everything is in the public domain and there are many varieties of Arduino compatible boards available from multiple suppliers. Some boards are very small and some extremely powerful (32bit!).

The Arduino software comes with a library of routines that are (fairly) simply knitted together to give the desired end result. These include digital and analogue I/O and routines for DC motors, Servos and Steppers. Everything is programmed in a "simple" version of 'C' but it's not as intimidating as you might expect (apparently kids learn it pretty quickly - and "physical computing" as it is known is becoming very popular in schools!).

Once programmed the device can be used "stand-alone" using an external power source if required (e.g. batteries). The smaller versions (Nano etc) are about the size of the older "Stamp" devices but you can also get the Amtel chips pre-programmed to build your own Arduino PCBs if required. I am using a 'breadboard' to connect 'things' to the Uno at the moment and there is s/w out these on the InterWeb that will allow you to input your 'breadboard' design and that will convert this into a standalone PCB layout, which you can then get 'made' for you (all via online sources of course!). See :-  http://fritzing.org/

So there are lot's of possibilities with this technology but really too much to explain here. However, Arduino (and it's related tech) may offer you another approach to your problem.

Regards,

Ian T

[454]

Tim

The best feedback control system for models is the human hand eye coordination.

If the Tx stick is too coarse then get a stick expander & it will make it less sensitive.
If the servo to the throttle is too lively then slow it down with a servo expander.
It is just the viewpoint, the unit installed between the Rx & the throttle servo is fully adjustable.

You want wacky, you got wacky from me.

Just browsing through RC websites at the moment for suitable kit for the task.
Try "Turnigy" for their bits & pieces. IF you need one I got one you can play with.

I am sure it can be done without a lot of computer or bespoke electronic circuitry.
But if you want sophistication I am sure we can find it.

Dave
454

[Peaky 556]

I'll reply to Ralph first as I can relate to these kinds of electromechanical stuff easier! I need time to research the others!
So, presuming the linear actuator is then physically connected to override the servo, eg by sliding its mounting plate one way or another?  Then we have a damped way of keeping a constant speed. As I want to vary the speed though, wouldn't I need to vary the reference voltage in some way? Tim

[cabbage]

The speed of the loco is maintained by the ESC which is linked to the Rx of the system. The Generation System is quite separate. You would not actually control the IC engine at all. What would control it is the rise and fall of the voltage held in the Capacitor Bank. The fall in voltage moves the motor so it winds the screw and thus opens the butterfly more. And then visa versa...

Initially once you fire the IC engine it would rev at quite a high RPM until the Capacitor bank was charged and then it would idle until the ESC drew current from the Capacitor Bank to move the loco.

Thus the RPM of the IC engine of the loco does not depend on the position of the throttle lever on the Tx but on the Voltage difference between the Capacitor Bank and the Reference Voltage. As a guide I would suggest that your Vreg be 12Volts @ 15Amperes and your reference Voltage be 15Volts. You would ideally need 8 capacitors of very high "draw" but of actually quite low Farad rating (look for something with axial wires of 2,000uFarads @ 63Volts).

regards

ralph

[454]

Ralph

Basically that would be how I would do it but with the following modification:-

How about deleting the capacitor bank & substituting a bank of NiMH batteries?
The IC motor would just be a battery charger & the traction motors would behave like a regular battery powered loco without complication. The only caps you would then need are the tiny spike suppression ones to protect the ESC.

Feedback

How about using telemetry? So feedback is back to manual.

How did the early diesel electrics get their commands before the sophisticated age of electronics?

Have a driver's panel with a display of scale speed, volts, amps, engine revs & whatever else you can think of to take up the available channels of the telemetry unit.

Drive it manually using the display on the base station reception unit.
If you do want to be really flash, put a video camera module in the cab and have a driver's eye view also.

Probably this is a step too far.

Dave
454

[cabbage]

I opted for Capacitors as batteries would not charge/discharge fast enough and if you are using NiMH batteries -why not dispense with the IC generator? On tests with my Heilmann loco I found that the Dynamo could deliver 19Amperes at 5 Volts quite easily. I am not quite sure what the charging rate is for a NiMH but a typical charging rate for a NiCd is 10% of rating thus if it was 1Amp hour then the charging rate would be 100mAmps for 10 hours.

Another advantage of using Capacitors is the fact that you can "totem pole" them so that the load is shared by each of the Capacitors and there would be no "serial battery resistance" during the charging/discharging cycles. Remember that a NiMH can only be charged/discharged at most 500 times -thus the number of running "hours" that the loco would use on them would be quite high in a very short length of time.

regards

ralph

[454]

OK Ralph, I am convinced. Capacitors seem the way to go.

Cheers
Dave
454

[Peaky 556]

The speed of the loco is maintained by the ESC which is linked to the Rx of the system. The Generation System is quite separate. You would not actually control the IC engine at all. What would control it is the rise and fall of the voltage held in the Capacitor Bank. The fall in voltage moves the motor so it winds the screw and thus opens the butterfly more. And then visa versa...

Initially once you fire the IC engine it would rev at quite a high RPM until the Capacitor bank was charged and then it would idle until the ESC drew current from the Capacitor Bank to move the loco.

Thus the RPM of the IC engine of the loco does not depend on the position of the throttle lever on the Tx but on the Voltage difference between the Capacitor Bank and the Reference Voltage.

Thanks Ralph, this all sounds very practical and sensible. 
I must admit my first assumption was to use batteries to start the engine, and controlling loco speed was by controlling the engine throttle and having the generator hard-wired to the traction motors, which I believe was more or less how the prototypes did it (I will ask some more questions of my ex Derby loco shed electrical supervisor, who worked on the BR diesel-electrics all through the 60's & 70's).  Obviously circuit breakers and polarity changers would be needed.  The batteries would have been 'floated' via a voltage regulator from the generator output.  Generator overspeed would be sensed with the output from a tiny generator mounted on its shaft (ie a small DC motor) which would provide the feedback I've been asking about.
Capacitors sound an interesting idea, but I need to calculate if they could practicably store the charge to start the engine(s) with plenty of reserve.  Leave that with me.
So, if I were to use a ESC for the traction motors, driven by the TX 'throttle', that would give better driveability, I agree.  The output of the ESC could be the variable 'reference voltage', and the actuator on the engine throttle would be driven by a difference in signal between the reference voltage and the output of the generator.  That way there is no engine racing to charge up the capacitor bank initially, and no risk of throwing windings.  However we'd never end up with a fully charged capacitor bank for subsequent starts, and the voltage drop through the ESC would not be accounted for.  Hmmm....
Sorry there's a lot of thinking aloud going on here, so I'll go away and rationalise some thoughts, do more research and come back with a wacky, but not too complicated proposal.
Thanks to Ian and Dave for their contributions, all grist for the mill!
Regards,
Tim 

[cabbage]

Where abouts in Derby are you? Because I live on the Chaddesden and Oakwood border! Perhaps it might be better for the pair of us to sit down with a cuppa and slice of cake and finally hammer out what it is that you want to do. Rather than use another dynamo -I would use a photo diode and an LED shining through a slotted wheel and use that. It wouldn't consume any power from your IC engine and could be simply made (ie bore a pair of holes in the flywheel!)

regards

ralph

[Peaky 556]

Another advantage of using Capacitors is the fact that you can "totem pole" them so that the load is shared by each of the Capacitors and there would be no "serial battery resistance" during the charging/discharging cycles.
regards

ralph

Ralph - I'm not familiar with the phrase "totem-pole" as applied to capacitors (or anything but red indians), but it sounds like a series connection.  Was this meant, as parallel seems more appropriate?
Cheers, Tim

[cabbage]

Tim, sorry -my age is showing!!!

"Totem Pole" is a method of RADIAL distribution often used in Valve Amplification. This ensured that all the lengths were the same and that any induction hum that occurred was cancelled out by the lead in the other direction(!) The collection of solder tags and nuts was said to resemble a "Totem Pole". The modern phrase would be "Buss Bar"

Yes it is a parallel connection with a FEED side normally a huge bolt from which the smaller capacitors radiate off the centre. The reverse where the negative leads all converge on one point is called "Star Earthing".

The advantages of "Totem Poles" are they are fast and can deliver a very high current at short intervals -or absorb them. "Ripple Bank" capacitors are capacitors with a small (typically 1 Ohm) between the +ve terminals -these damp down stray (bad) voltages to produce a smoothed output.

My Guitar amplifier (designed and made by me) has a Totem Pole 630uF supply rated for 650 Volts  @ 64 Joules...

I normally bake cakes on a Friday for the weekend!!!

regards

ralph

[Peaky 556]

Feedback

How about using telemetry? So feedback is back to manual.

How did the early diesel electrics get their commands before the sophisticated age of electronics?

Have a driver's panel with a display of scale speed, volts, amps, engine revs & whatever else you can think of to take up the available channels of the telemetry unit.

Drive it manually using the display on the base station reception unit.
If you do want to be really flash, put a video camera module in the cab and have a driver's eye view also.

Probably this is a step too far.

Dave
454

Dave I like these suggestions!  Where can I get the bits to make one?
Cheers, Tim

[454]

Try Technobots.

and

Some RC Flyer trader websites have telemetry gizmos that could be used or adapted.

A bit of electronics thinking will be needed though if nothing from the trade is forthcoming.

Ralph & I have been deliberating how we are going to MU our 2 Peaks as a coupled unit on a single Tx.
Added complication. I'm definitely not going down the telemetry route on my Peak.

Cheers
Dave