Automotive 6 Volt Generator Transistor Voltage Regulator

[rjenkinsgb]

You could use a hall-effect current sensor, which does not need a shunt - but then it's another device that needs power..
It does avoid voltage drops and heat though.

eg. One of these needs around 20 - 25mA at 5 - 6V and gives 0-5V out for 0 to whatever rating current:

HLSR 40-P | LEM HLSR Series Current Transformer, 100A Input, 100:1, 40 A Output, 4.5 → 5.5 V | RS

uk.rs-online.com

There again, 5 milliohms at 25A is (I^2 * R) 3.125W - it should not be producing a lot of heat?
It should drop 125mV, I think you need a better resistor type.

Two of those open metal 10mR ones I mentioned early on, in parallel, would give you 5mR 10W, enough for over 40A?

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

uk.rs-online.com

edit - ps. While using a shunt resistor, the voltage sense should be from the battery end of the shunt, not the dynamo end.. Just spotted that on the drawings.

 

[Danwvw]

Yes I was just working on Ideas for the shunt. I like the resistors the best but probably try to get it working with copper wire and deal with that later. But yes the shunt plays a very important role doesn't it. It's the source of the problem when it comes to keeping the same voltage output under various charging conditions.
If I set the voltage at 7 volts without headlights when the battery is topped off it doesn't go over the 7 volts,
If set to 7 volts with battery low it will gain 1/4 volt as the battery charges.
If set to 7 volts with the headlights on then with them off it will gain 1/2 a volt as it charges.
I am not sure I had the current limit set perfectly though.

Schematic Final for Prototype 2:

Ordered one of these, devices, If it's anything like the voltmeter it will work, It will be nice to mount with the voltage regulator, It's shunt will double as our shunt. It's only 50 mv though!

 

[Danwvw]

 

[Nigel Goodwin]

Is it just me, or do these pictures look like he's making a still as well?

 

[Danwvw]

Had some troubles with getting it working just wiring missing still then a short then taking out a resistor. The way it test is to hook up all the (+ stuff) and the load (in my case a 6 volt LED Brake Light) to the + of two lithium ion's in series. The other end of the load goes to the DF wire everything else negative grounds.
Were now making all the Current Sense input voltage dividers 500 ohms instead of 1000. The 500 Ohm voltage adjustment pot is now a 250 ohm pot. And there is the 2 volt zener also the Motorola HDL Mosfet ! The E ic is switching with a pot setting of 100 ohms and the I ic is switching at 400 ohms on it's 10 turn 1K pot. It's ready to test on the Dynamo,
Has a 2 volt very high frequency oscillation on the gate at threshold.

 

[rjenkinsgb]

Looks promising!
I'd not worry about the oscillation, the voltages are probably varying a tiny amount when the load switches.

The only possible mod I can see before it's tested is move the voltage sense (2V zener) to the battery end of the shunt?

 

[Danwvw]

Well that would correct for any voltage drop at the shunt wouldn't it. But ideally there wouldn't be more than 50 mv if that is enough to detect? I should be able to measure that now with the test setup by looking at the voltage differential between pins 2 and 3 during switching by adjusting the current pot.
Looks like it can detect a change of 3 to 5 mv at the current inputs but through the voltage dividers it may a little less sensitive as they each cut the voltage in half.

 

[Danwvw]

Anyone know anything about cheap digital amp meters. This one has the Black Current lead Grounded. I was planning to use the shunt in the Positive D+ 7.55 Volt Circuit as my shunt. Is there a trick for floating the current inputs?

 

[rjenkinsgb]

That's annoying, though I don't think I have seen a combined unit with a "floating" current input, other than via a separate current transformer or transducer.

You could use two separate meters, or a programmable dual input meter, with one of the little current transducers I linked to earlier on.
eg. We use some little Lascar SGD-24M configurable TFT meters for various things, they are very convenient.

PanelPilot SGD 24-M | Programmable Touch Screen Panel Meter

Seeking a programmable panel meter? The SGD 24-M offers dual analog inputs, alarm outputs and a 2.4” touchscreen. Customize with the free PanelPilot M Software.

www.lascarelectronics.com

You can configure them for digital, bargraph, "analog" needle etc. and whatever colours and scale labels you like.

 

[Danwvw]

Wow! Impressive, I have looked at Current Sense Isolators with Analog output and some Hall Effect devices and This morning I noticed this little device that address this issue I think? They all seem to have some gain though. CS30 or MAX4173 or something.
Note! Ordered this one: AD22057

 

[rjenkinsgb]

That looks ideal - you can use a very low shunt value & one of those to convert to zero-based voltage, for both the current limit and ammeter display. It should make the current limit more accurate and possibly avoid other start-up glitches.

It may need a buffer amp to drive the meter if it's not directly at the regulator, though; or at least a resistor and capacitor filter to avoid noise pickup.

The gain 20 version would give 0-1V from the 50 mV max. shunt you mentioned, or you could go lower and use a higher gain device.

 

[Danwvw]

Good to hear,
Prototype 2 installed and running on the car today. Using 2, 5.6 ohm resistors in series = 11.2 ohms for the field resistor now (the 7 ohm field resistor broke).
Prototype 2 voltages are more solid than 1st Prototype was. The dual coils helped a lot, will handle the headlights on now, only 0.15 volts difference at the battery when the headlights are turned on , the shunt coils still get hot though!
Voltage pot adjusting nicely as is the Current.
It doesn't have the 6 volt startup jumper wire to the car coil and isn't starting charging without revving the engine! Then it's ok, it holds in fine at idle, better than the 1st prototype.
The ammeter isn't hooked up yet. But it's nice to have the voltmeter mounted on top.
This one has the MOSFET and the Cutout diode mounted on the Heatsink. The 2 diodes on top are the Un-regulated supply feed diode (Same Kind as the CutOut diode STPS4015CW) and the Field Flywheel diode a MUR820.
It would be easy to use the un-used diode in the STPS4015CW for the voltage from the coil to see if that solves the start-up issue however, I still want to find a way that it wouldn't need it so it's a direct replacement for the Bosch Mechanical Voltage Regulator.

This Motorola MOSFET holds it's gate charge very will, perhaps there is a way to have it's gate set hi when shutdown or something so it's on at startup.
Prototype2:

Prototype 1:

 

[Danwvw]

Playing with the idea of trying to boost what little voltage is present on the field or the D+ before it starts to charge. The Field and the D+ voltage is 0.8 volts at idle and at starting charging it builds to 1.9 and it's doing this with the two 5.6 ohms which is with a 11.2 ohm load. So there is plenty of current there to use. There are these little boost power supply boards that do exactly this job. I think the LM311's need about 7ma each which restricts using an ultra low boost converter like the LTC3108-1 There are other boost converters though like the RP400K001A If I could find one that could handle say 50ma or even 100ma that could run on .8volts and make 5 volts out at 50ma that would power up the LM311 IC's at when the car is idling and solve the startup problem. Better yet find one that had an 8 volt or 9 volt out with a range of .8 volt to 9 volt input. and it could run things full time and regulate the un-regulated supply at 8 or 9 volts well above LM311 needs. Others: LT1307 LM2623

 

[rjenkinsgb]

Hi Dan,

with the modified circuit and pull-up on the gate of the FET, the power to the 311's should be irrelevant until limiting type regulation is needed; that's why I suggester those changes.

It just needs a feed to that gate pullup resistor, if anything, and the field will be running flat out.

Dynamos have always dropped out at low revs, that's one of the reasons they were replaced by alternators.

 

[Danwvw]

For sure! One thing I have not tried yet is to remove the Field resistor and see if the voltage will build higher easier at startup. Just did not want to take a chance on burning things out without it. I have one of these circuit boards on Order to try:
Article CE8301 Datasheet doubt these are correct for the E50D IC maybe like the BL8530 But the board below is rated for 1.5 Amps Actually it's 500 ma on 2 flashlight batteries.

 

[rjenkinsgb]

Interesting device!
I had to find the data on the IC... https://file.yizimg.com/390228/2012041215332322.pdf

Be wary of it's input voltage limits - but if it's only driving the FET gate resistor, the current requirement will be tiny and a zener + resistor should be OK.

I'd use a couple of small schottky diodes, one in series with the top of the gate resistor as it is and another from the output of the PSU board, so either can feed the gate.

Or just use the PSU output all the time? Though the current load will be a bit higher as it will have to stand the gate being pulled low when regulation kicks in.

 

[Danwvw]

I was hoping it would run the un-regulated supply however it's nice to see other options.

 

[Danwvw]

Next: The AD22057 a Sensor interface single supply chip for making the Amp readout work arrived however I must have it wired wrong! I installed a Gain resistor for a Gain of 1 (5.263K ohms) wired it up and wired as follows: and installed on the car. It's Not working.

Ammeter is drifting around 53 to 55 amps regardless!
0-60 mv across inputs (across shunt still using the wire coils)
The AD22057 Pin 5 Output is always .132 -.134 Volts regardless.
Reversing or Disconnecting the inputs changes nothing.
Adjusting 1K pot on pins 3&4 in the gain circuit does nothing.
Perhaps I didn't understand something about wiring the AD22057. Any Help and Chat would be much appreciated.

 

[rjenkinsgb]

You look to have the resistor from A1/A2 connected wrongly - it looks like it should go to the output pin, not ground.

See figure 1 in the datasheet; they have 200K for rather higher gain.

 

[Danwvw]

I did see that but configured it like figure 4 for gains less than 20.
Now have the chip in the house did some test, It appeared to only pass AC noise on the -input pin but then that stopped working after I put a 1.25 volt battery across the inputs. Guess it didn't like that. Apparently it needed input resistors?
Ordered the LTC6101HV That may work better anyhow?