Maker Pro
Maker Pro

Constant voltage load

(*steve*)

¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd
Moderator
Jan 21, 2010
25,505
Joined
Jan 21, 2010
Messages
25,505
I'm building a constant voltage load for a future project. These are standing in for a 12V lead acid battery so I can run tests without worrying about overcharging the battery.

12V Load.png

I've breadboarded this up and tested it briefly this afternoon. (Then we had guests, so my time evaporated).

Can anyone spot any gotchas before I design and etch a few boards tomorrow?
 

KrisBlueNZ

Sadly passed away in 2015
Nov 28, 2011
8,393
Joined
Nov 28, 2011
Messages
8,393
Hey Steve,

My only concern, and one I can't quantify, is that the circuit will have a very steep V-I curve above the threshold voltage of 11.8V. In other words, almost zero dynamic resistance.

The TL431 has a very low AC output impedance, then you're amplifying it with a common emitter amplifier with no degeneration. In my experience, adding gain and delay like this leads to instability and oscillation, but this is a subject I've never understood, so I'm just mentioning it in case you think it might be worth checking out.

Just wondering why you don't use the standard "amplified zener". I guess you need better accuracy.
 

Arouse1973

Adam
Dec 18, 2013
5,177
Joined
Dec 18, 2013
Messages
5,177
Not being stupid here. But what the benefit from this and just a resistor?
Adam
 

(*steve*)

¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd
Moderator
Jan 21, 2010
25,505
Joined
Jan 21, 2010
Messages
25,505
My only concern, and one I can't quantify, is that the circuit will have a very steep V-I curve above the threshold voltage of 11.8V. In other words, almost zero dynamic resistance.

Interestingly enough, in the breadboard tests (from 2 to about 200mA) the voltage varied by about 0.1V. More careful measurements showed that this was, to the limit of accuracy of my meter, entirely due to voltage drop across the thin wires that ran back to the power supply.

The TL431 has a very low AC output impedance, then you're amplifying it with a common emitter amplifier with no degeneration. In my experience, adding gain and delay like this leads to instability and oscillation, but this is a subject I've never understood, so I'm just mentioning it in case you think it might be worth checking out.

Yes, that was a concern of mine too. I need to invert the output of the TL431 to drive the N Channel mosfet -- I don't have a suitably rated P Channel mosfet on hand. Perhaps an emitter resistor for Q1 to reduce the gain might be a good idea? I will fire up the oscilloscope this morning and look for hints of instability. Perhaps I should add a little inductance into the power supply leads to make things worse?

Just wondering why you don't use the standard "amplified zener". I guess you need better accuracy.

Initially I wanted a variable voltage, and R1 in the final circuit will probably have a trimmer placed in series so I can set the voltage to exactly 12V, or maybe 13.8V. But yes, to eliminate certain inaccuracies, I want something a little more accurate and precise.

Not being stupid here. But what the benefit from this and just a resistor?

At somewhere between 1 and 2 mA of input current the circuit here maintains a fixed voltage across the input terminals. This does not vary significantly until the current gets to a value that destroys Q2. :) A resistor obviously has a voltage across it which varies in proportion to the current through it. I am playing with my own MPPT design that will (eventually) be capable of combining several strings of panels with varying voltages and orientations. I want a load which has a fixed voltage so I can accurately measure power by measuring current (and because it reflects one aspect of a lead acid battery).

Nitpick: The 2N3904 is NPN, I think you mean 2N3906.

I certainly did. Thanks.
 

cjdelphi

Oct 26, 2011
1,166
Joined
Oct 26, 2011
Messages
1,166
I've had 2 dollar ebay dc dc lm2596 supply 14.5v to a lead acid battery it will supply 3amps max before shutdown steps in..

had this setup going for about 2 years, you need to supply 13.8 - 14.5v or it wont fully charge..
 

KrisBlueNZ

Sadly passed away in 2015
Nov 28, 2011
8,393
Joined
Nov 28, 2011
Messages
8,393
Perhaps an emitter resistor for Q1 to reduce the gain might be a good idea?
I honestly don't know!
I will fire up the oscilloscope this morning and look for hints of instability. Perhaps I should add a little inductance into the power supply leads to make things worse?
Yeah, I guess that would make it more likely to become unstable. And mess with the circuit to see how quickly it stabilises when it's returned to normal, and whether there's any overshoot or ringing, I suppose.
I am playing with my own MPPT design that will (eventually) be capable of combining several strings of panels with varying voltages and orientations. I want a load which has a fixed voltage so I can accurately measure power by measuring current (and because it reflects one aspect of a lead acid battery).
MPPT interests me. Will you be posting your design here?
 

(*steve*)

¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd
Moderator
Jan 21, 2010
25,505
Joined
Jan 21, 2010
Messages
25,505
I've had 2 dollar ebay dc dc lm2596 supply 14.5v to a lead acid battery it will supply 3amps max before shutdown steps in..

had this setup going for about 2 years, you need to supply 13.8 - 14.5v or it wont fully charge..

Yeah, this design is to simulate the battery voltage. It makes testing easier. In real life, the MPPT charger simply attempts to maximise output current and essentially lets the voltage do what it will (until some point at which it determined it is at end of charge). In my testing I want to compare performance over time and so eliminating the variable of battery voltage means I don't have to worry about changes in it affecting (say) efficiency between two tests. I also can eliminate any need to cater for end of charge.

MPPT interests me. Will you be posting your design here?

Assuming it doesn't explode, yes. :)
 

(*steve*)

¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd
Moderator
Jan 21, 2010
25,505
Joined
Jan 21, 2010
Messages
25,505
Here's the updated circuit. I have added R4 which is currently a link, and R6 to allow setting the voltage between about 11.6 and 15.4 volts.

Screenshot from 2015-02-22 17:49:37.png

And here's my first go at a layout.

Screenshot from 2015-02-22 17:49:13.png

Various events have conspired against me getting these boards made today so they'll have to wait a while... It will also give me a chance to test the stability of the breadboarded version.

In the meantime, I'd appreciate any feedback on the board design.

edit: oh Q1 is now a BC857, because I have some :)
 

Arouse1973

Adam
Dec 18, 2013
5,177
Joined
Dec 18, 2013
Messages
5,177
Here's the updated circuit. I have added R4 which is currently a link, and R6 to allow setting the voltage between about 11.6 and 15.4 volts.

View attachment 18798

And here's my first go at a layout.

View attachment 18799

Various events have conspired against me getting these boards made today so they'll have to wait a while... It will also give me a chance to test the stability of the breadboarded version.

In the meantime, I'd appreciate any feedback on the board design.

edit: oh Q1 is now a BC857, because I have some :)

Looks good to me Steve :) Oh I guess the holes in the pads are the right size, they look quite small.
Adam
 

(*steve*)

¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd
Moderator
Jan 21, 2010
25,505
Joined
Jan 21, 2010
Messages
25,505
And I've made a couple of boards using a laser cutter. See here.

I made a few more adjustments to optimise the clearance of traces on the board, but otherwise it's identical.
 

(*steve*)

¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd
Moderator
Jan 21, 2010
25,505
Joined
Jan 21, 2010
Messages
25,505
OK, so the first prototype is complete. And it worked... second time around.

IMG_6619b (Small).JPG

Nasty and uncleaned. Can you read any part number on the left hand SOT-23 device?

Note to young players: DO NOT leave your last container of solder paste open and in a hot workshop for a few weeks and expect it still to be useable.

I tried mixing it with some flux and pounding it into a paste, but (after a lot of trying) it was a lost cause. I ended up soldering these my hand with a soldering iron.

First test showed that the voltage across the input was 3.3V and did not vary with the pot.

Second note to young players: Always check the exact datasheet for the chip. The TI datasheet (which comes up first on Google) shows the pinouts for the TL431 and the TL432. These differ only in that 2 pins are transposed in the 432 (and I have a 431, right?). OK, so I have an AS2431 which just happens to follow the TL432 pinout :-(

Fortunately the fix is as simple as taking the device off, bending the leads around and placing it upsidedown on the board. I say "simply", but the leads get even smaller as you approach them with even the finest tip pliers you have.

And then it works! The adjustment range is 11.32 to 15.02V. That range is a little different to what I calculated, but is fine for a nominally 12V device.

Third note to young players: Make sure you get the connections to the trimpot around the right way. I wanted the voltage to increase as the pot is turned clockwise, but it's the other way aound. Sigh. I read the datasheet and I thought I had it right. And it's really tricky to probe the tiny pads on these things. Oh well, I have to redesign the board to handle the AS2431 pinout, so I can change this too.

Fourth Note to young players: If you're going to breadboard a circuit and you need to use an SMD device, use one in the same package that you plan to use in the final design. I used an SO-08 TL431 in the breadboarded test (because it was easier to pull out of my collection of parts) and the SOT-23 one for the final design (because it is smaller). If I had used the SOT-23 part originally I would not have made the major mistake on this board.
 

(*steve*)

¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd
Moderator
Jan 21, 2010
25,505
Joined
Jan 21, 2010
Messages
25,505
So here's the improved layout.

Screenshot from 2015-03-02 13:23:03.png

The mosfet will sit the right way, and I have included a mounting hole to attach the whole thing to the heatsink.
 

AprilSteel

Feb 2, 2017
111
Joined
Feb 2, 2017
Messages
111
I'm guessing if you wanted an LED on this you would put it in series with R5 on updated circuit
 

AnalogKid

Jun 10, 2015
2,665
Joined
Jun 10, 2015
Messages
2,665
As the source voltage increases, the effective circuit resistance decreases, making this a negative resistance device. Some power supplies really don't like driving these.

ak
 

(*steve*)

¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd
Moderator
Jan 21, 2010
25,505
Joined
Jan 21, 2010
Messages
25,505
It's not a negative resistance, it behaves almost exactly like a zener diode (in both the forward and reverse direction). The main differences are that the knee is very sharp, and the is a "leakage" of about 1mA from memory.

But it's true you shouldn't connect it to a power supply. The current through it will rise from about 1mA to as much as the power supply can supply as you try to increase the voltage past the knee. This will probably cause smoke to come out of the MOSFET if you have a beefy power supply or a wimpy MOSFET (or poor heatsinking).

The failure mode is likely to be a short.
 
Top