Automotive 6 Volt Generator Transistor Voltage Regulator

[rjenkinsgb]

The inputs need to be at least 2V below the IC positive supply, from a quick look at the data. You also need an offset so it switches at the maximum current.

Just add pairs of resistors from the ends of the current shunt to ground and connect the 311 inputs to the junctions.
eg. 1K and 1k8 so you get 2/3 voltage or just below.

Make one of them at one side a preset, or a lower value resistor with a preset in series, so it can be calibrated for the inputs to be equal as it reaches the required current limit.

 

[Danwvw]

Ok, Well I don't think there is enough of a signal across my .01 ohm shunt to have anything left. The IC Supply Voltage I am planning is just 5.6 volts so for example 1/2 voltage would be 7.63/2=3.815 and 5.6-3.815=1.785 volts below IC supply.
Any suggestions for perhaps a more sensitive and isolated Comparator? I have some MCT5211 Phototransistor Optocouplers and some NPN and PNP switching transistors that could be used.

 

[Danwvw]

But yeah, maybe? I was mistakenly thinking the IR drop across the shunt was just 25mv, it's actually (25Amps * .01 ohms) = 250 mv. Also maybe I could raise the IC supply voltage just a little by changing the 2N2222A Constant Current Source 6.2 Volt Zener Diode to higher voltage than 6.2 however, IC supply voltage may already cause it problems at startup before D+ builds a voltage. I am hoping a 10 Ohm resistor from the Generator Field to Ground will override it not working at start up but I don't know. The Mechanical Voltage Regs have that Big Wire Wound resistor on the field. I bet it's pretty low in ohms. I think it may even be enough to drive the generator at full power at higher RPM which could mean it need to be removed for this application.

 

[rjenkinsgb]

You don't really need to regulate the supply voltage to the 311's, just the reference voltage circuit.

A simple R-C filter plus a higher voltage zener to catch any spikes (9 - 12V) would likely work fine, so you can keep the input range as near the positive voltage as possible.

You could also drop the voltage from the shunt ends via eg. a pair of 1.8V zeners, then use a much lower "upper" resistor & preset in the dividers below them, just enough to cover the voltage difference needed for the required current range.

 

[Danwvw]

Oh! Yeah, Great, I was also thinking about using some diodes in the current sense circuit.
Measured my 6 volt mechanical voltage regulator field resistor it was open looks like it should be about 2 ohms, It's messed up and I repaired it now it's 1.3 ohms, maybe the repair will fix some of it's issues?

 

[Danwvw]

ByTheWay! That repair of the wire wound resistor on the mechanical did make quite a difference. Now it's over charging at 8.25 Volts and with the 6 volt H4 60/55 Watt Headlights on High Beam it's putting out 7.4 volts at the B+. All the arcing in the field relay has quit, it just pulls in and stay in just above idle. B+ Voltage is no longer jumping around just builds smoothly to 8.25 volts or
(7.4 volts with the headlights on) as the RPM goes up.

P.S. I should measure the Voltage Drop now across the Field when it's putting out under a full load so we know what it's drawing. I = field voltage / 1.3 ohms.

 

[Danwvw]

Field voltage is 2.4 Volts with lights & reved. and 1.7 volts no lights reved. So minimum field current is 2.4/1.3=1.85 Amps or 1.7/1.3=1.31 Amps.
So what this means to me is that adding 120 watts load at say 7 volts about 17 amps to the Generator Output changes the field current about 0.54 amps.

 

[Danwvw]

Ok so here is Draft 2:

 

[rjenkinsgb]

Almost -
You have the resistors configured wrongly in the current sense; the dividers should be from each end of the shunt to ground, with the comparator inputs from the divider junctions.

The idea is that each input sees around 2/3 of the voltage to ground on the shunt terminals.

As there is now no real load on the 5.6V reference supply that could be simplified and have the transistor removed.

You only need one preset to calibrate each comparator; it's too easy to get in an impossible situation with two presets in the same section - eg. the lower the one on pin 2 of the voltage reg is set, the lower the reference one must be set and the lower the overall accuracy.

The pin 2 divider in that one should be fixed, with values that put it around eg. 3 - 4V with 7.5V output.

The same divider could be used for one side of the current sense part. Duplicate it at the diode end, but in that case add a preset in series with the lower resistor.

That's another general design principle - you normally use a preset and a fixed resistor, so when the preset is at zero the fixed resistor sets the actual lowest value in the circuit. It makes the preset easier to adjust and avoids things accidentally being shorted or set out of range.

 

[Danwvw]

Oh, ! Missed that! Thanks good catch! Yes, I will remove the 2n2222 just happen to have bought them as I thought they would be needed but testing showed the LM311 has the correct voltages to drive the MOSFET which it turns out does not draw any current at it's gate.

I hear what your saying, I think? Wouldn't both (E compairitor potentiometers) be needed to fine adjust the voltage as well as adjust how the output of the (I compairitor) effects things. Just need to build a prototype with LED indicators at outputs etc... figure out what really works with the dividers etc...
Not sure what I am going to use for a .01 ohm resistor? In the car at 6 volts on the headlight circuit every connection drops 0.1 volts and connections get hot on the headlight circuit if they drop 0.5 volts. So probably just a stainless steel screw or something would work for the shunt?
2nd Draft Revised:

 

[rjenkinsgb]

We use this style for some things:

Arcol 10mΩ Metal Film Fixed Resistor 5W ±1% MSR-5 R01 F | RS

uk.rs-online.com

One of those would be OK for 20A continuous at 10 milliohms. Four in series - parallel would be OK for 40A.

The drawing is starting to get a bit confusing.. I'll try and have a proper look through that in the morning.

 

[Danwvw]

I think it should be just like what your said. I will re draw it but am building the prototype now don't bother to try to figure it out till I get something better drawn. Thanks for all your help! I will look into the resistors.

 

[Danwvw]

 

[rjenkinsgb]

That looks pretty good!

I'd remove the capacitor on pin 3 of the current sense 311, or move it to between the inputs. As it is, it makes the input dividers asymmetric and could cause voltage changes to be briefly seen as current changes.

One may be worthwhile on pin 3 of the voltage comparator to reduce noise effects there.

I'd also increase the gate-source resistor on the FET (right-hand 330R); the gate voltage should ideally be at least 5V and anything less than 10V is OK.
Possibly a 1K there, so a spike that is only limited by the 12V zener cannot exceed 10V.

And if possible use a fast-recovery or schottky diode across the field; normal 1N series are fine for lower frequency but are relatively slow switching, so not that good for flyback suppression, where the voltage changes can be very fast..


Edit - ps. And add a resistor between the reference voltage divider & cap on pin 3 of the voltage regulator and the connection from the current sense 311.
470 ohm or 1k should be OK.

That limits the current through the current comparator output & prevents trying to short out the reference supply.

 

[Danwvw]

Thanks for the tips! The 5 mfd cap was in Max's original schematic but we have changed it by adding the current sense etc... I will play with the values once the prototype is operational. Yes a resistor between the current LM311 out and the voltage LM311 pin3 is probably a good idea. I think the way I have it the voltage ref becomes the positive for the current LM311's output. Not sure how well they will play together, but output from the current comparator only needs to drop the voltage at pin 3 of the voltage comparator slightly to shut it down due to the current limit being met so I will start with the 1K and see how it behaves .
First I will do some bench testing before installing it on the 6 volt gen in my car. Actually it will be amazing if it works at all. I do like those 10 little m ohm resistors you linked above. My shunt will likely just go on the D+ Post of the generator and not be inside the voltage regulator housing. I plan to use an old original mechanical voltage regulator housing for this solid state one.
I would imagine the IN4148 would be too little to use on the field flyback circuit?
It's raining on the Oregon Coast. Usually we don't get rain this time of year, I have quite a load of hops out in the rain, time to make beer.
Thanks again for your help "rjenkinsgb" and Max and everyone. I welcome any feedback.

 

[rjenkinsgb]

The 5 mfd cap was in Max's original schematic but we have changed it by adding the current sense etc..

It's not the 5uF on the reference, it's the 0.1uF on the current sense comparator that concerns me.

And keep the direct connection from pin 7 of the current comparator to pin 3 of the voltage one, the resistor should be between that and the ref. divider.


For the flywheel diode, look at something like a 3A "fast recovery" diode such as the UF5400 series, eg. UF5402 or UF5407 etc.


I think it's god a good chance of working, I'll be interested to see how it goes!

 

[Danwvw]

Oh, Ok yes, I'll probably just leave out the LM311 input Capacitors to start with, I did play with putting the .1's around in various places on my LM311 Test Circuit. They seemed to shift the threshold slightly but everything still worked. Electrolytics just made it slow.

 

[Danwvw]

 

[Danwvw]

The prototype is wired still have not tested it!

This is just a Mock-Up:

Prototype Schematic:

 

[rjenkinsgb]

Looks good!

The only thing I can see is that you may need a rather higher value field bias resistor; 10 Ohms may cause overcharging, when there is minimal load.

It only needs to provide enough current to maintain minimal output and power the electronics; I'd aim for just around 1A output at most, with the resistor alone.

There again, I'm guessing and it may well be OK.

Just keep an eye on the gate voltage of the IRL540 when the battery reaches full charge & nothing in the car turned on, other than the engine & with that it typical driving revs.
If the FET shuts off totally, you need a rather higher value bias resistor.