Hi Guise.

Lovely facility, this. Well done.

Okay - I've checked existing threads and posts and found nothing that specifically addresses my challenge. I have a 555 PWM circuit from Tony van Roon ( http://www.sentex.ca/~mec1995/circ/pwm555.html ) that I used some years back on a hydrogen project. Works perfectly.

Although Tony's design is primarily a motor control, I've also successfully used it as a dimmer for 12v lighting systems. But now, I have an added challenge. Here in South Africa we have intermittent and dirty utility power most of the time, so battery backup systems are the rage - if not absolutely essential.

I'm helping an artist friend who has a huge house (3 levels with guest cottage). He designs and builds amazing LED chandaliers and perspex lamps out of LED strips, which draw anything from 250mA for the smaller lamps to around 3A for the bigger arrays. His entire home lighting system was supposed to be designed for DC but our local electrical engineers got it all wrong and, as a result, the 30Amp 12v DC supply only delivers 9 - 10v by the time it gets to the lamps at the furtherest part of the house. There is voltage drop everywhere - even in the same room as the supply!

So we decided to use a 24V grid and build a small regulator (based on the same design) into each lamp to keep power rock steady at exactly 12.5v DC. I have changed the circuit slightly. An LM7812 is feeding the 555 circuitry with no problem or heat from the 15v difference. The FET is an IRFZ44 and the lighting load is connected directly to the 24v supply. Now here's the problem:

When utility power is available, the PSU puts out 27V because it has to maintian the 4x 60A/h batteries (configured for 24V). So any setting on the PWM regulator will hold steady as long as the supply remains at 27v. But when the utility power fails, the input to the PWM regulator drops to battery float (approx 25v) and then, depending on how long the utility power is out, the battery will drain steadily down to a minimum of 21v.

There is no compensation by the FET and so the voltage to the lamps also drops, causing the llamps to dim accordingly. You probably know that LED lighting is very sensitive to voltage.

So my question is: keeping component count down (and size), how do I get the 555 to compensate for the changing imput voltage at the FET. Remember that the 555 is steady at 12 v via the 7812 but the supply voltage to the lamps is varying.

To reiterate: circuit size and component count are considerations and parts are hard to come by in our small town - so this is pretty much what have to work with. I do have some 5w 13v zeners and a TL431 adjustable shunt on hand. I tried a few angles with them but without success - I could only get the 555 to change its frequency but not its duty cycle.

Help?

PWR2U

 

Welcome to AAC!

A 555 has a max limit of 15VDC. This basic concept can be done with other chips, but the transistor can be connected to 24V.

The LM393 is a dual comparator, the LM339 a quad. Comparators generally need a pull up resistor such as the 10KΩ, but have much cleaner edges than op amps.

LEDs are not sensitive to voltage, they are sensitive to current. If you use simple resistors then the current will change when the power supply voltage does, but there are ways to compensate nicely for this.

...

Have you read this?

LEDs, 555s, Flashers, and Light Chasers

 

Thanks very much for your response, Bill.

Being mostly self-taught, I'm not exactly sure of what you've given me. Meaning, I'm not sure if the 317 design is shunt or switching. The shunt route generates too much heat from the voltage diffs and Darlingtons are about the price of an average diamond mine here.

What I can tell you is that I've tried various versions of the shunt route (from circuits on the Net) and it generates too much heat to mount in the perspex lamps without melting the base material. The lamp designs are very dainty and see-thru, so there's no room for large components or heatsinks. They are also faceted, so any exposed circuitry is likely to be (undesireably) reflected in the panels. I'm trying to keep the circuit to a narrow strip of veroboard that can be easily concealed.

One other consideration is that we already have a few of the basic PWM dimmers installed in the house and would like to stick with that route if humanly (or electrically) possible. As I mentioned previously, components are not available in the small town where I live and everything - yes, everything - has to be shipped in from major towns. I have 555's, Z44's and 7812's up the wazoo, which is why I've gravitated towards this route. (Neccessity being the mother of invention and all that.)

I've attached a quick schematic of the circuit that I'm presently using (a modification of Tony's original design) in the hope that you can tell me how, if at all, I can get the 555 to compensate its pulse width to the FET for the varying input in its existing configuration.

(I uploaded the image - being my first time here, I hope it comes up as an attachment when I post this.)

Many thanks again for your (input).

PWR2U

 

Where is here? You'll note the old hands post their location as part of their profile, it helps with the parts issues.

Darlingtons are about $1 here. You can always make one using two transistors, there is nothing magical about them.

It is definitely an analog current regulator, with the transistors taking the bulk of the load. Since power transistors are designed for heat sinks, no worries.

PWM is a good way to cut voltage, but it is not regulated. Generally the opposite is true, you are likely to get more voltage swing, unless you use a feedback mechanism.

The original design I showed could be modified very easily into a SMPS. You'll note that the + input on the comparator is a voltage input.

555's are a specialty of mine,

Bill's Index

The 555 Projects

My Cookbook

BTW, here is how to put your images as part of the posts.

How to Display Attachments Full Size

 

Thanks again.

Q1: So can I take it that my design idea is obsolete - no way to regulate it at all?

Q2: Can a 741 (single supply) or a 358 be used instead of the 339? (simply because I have them and the 339 is 14-pin where the others are 8-pin)

Q3: I have some 2N3904 and MJE3055T (both NPN) - is this adequate Darlington material? (I understand that it takes a small TR to trigger the bigger one.)

Q4: Given the higher voltage and current rating of the Z44 (50A), can I use the Z44 instead of the 3055 as the main Darlington component even though it is a FET? (I really DO want to keep everything running cool.)

PWR2U

(Thanks for the display tip.)

 

Thanks again.

Q1: So can I take it that my design idea is obsolete - no way to regulate it at all?

Q2: Can a 741 (single supply) or a 358 be used instead of the 339? (simply because I have them and the 339 is 14-pin where the others are 8-pin)

Q3: I have some 2N3904 and MJE3055T (both NPN) - is this adequate Darlington material? (I understand that it takes a small TR to trigger the bigger one.)

Q4: Given the higher voltage and current rating of the Z44 (50A), can I use the Z44 instead of the 3055 as the main Darlington component even though it is a FET? (I really DO want to keep everything running cool.)

PWR2U

(Thanks for the display tip.)

If I understand you what you want to do be able to adjust the PWM to the virtual ground so that when the input voltage(approximately 24) changes you want to keep the output power the same. I'm not sure at the moment how to achieve this, but with PWM circuits there is feedback that is referenced and automatic adjustments made. Does this make sense?

 

If I understand you what you want to do be able to adjust the PWM to the virtual ground so that when the input voltage(approximately 24) changes you want to keep the output power the same. I'm not sure at the moment how to achieve this, but with PWM circuits there is feedback that is referenced and automatic adjustments made. Does this make sense?

Yes, that's exactly right.

The output is relative to the FET drain pin and the pos rail. I'm just stumped as to where the reference should go (and how much of it) on the 555 in this config. I'm wondering if pins 2 and/or 6 don't have anything to do with it. (But - what, dear Liza?)

I even tried hooking up a TL431 to pull down pin 5 (CV) at the set voltage but that just stopped the oscillator completely and it wouldn't continue till the jumper was removed. In all other attempts, the best I could achieve to was change the frequency, which made no adjustment to the output. This is a combination of inadequate tech and limited component availability.

PWR2U

 

A comparator is not an op amp, you really don't want to substitute one of the slowest examples of op amps for them either. PWM is digital, you need a faster component. Both the LM339, and LM393 are pretty common parts, have you looked to see if they are available?

The circuit in question shows a lamp, not LEDs. You must use a resistor to control current, LEDs are not voltage devices. You are trying to build a voltage regulator, which is OK, but you can also build a current regulator, which is what I showed in the second post.

I'm getting the feeling you are trying to treat these as lamps, which is why I mention the resistor. The schematic you show doesn't match the application you describe.

 

@PWR2U,

Maybe you have looked in the wrong direction.

What you need is a simple 24V to 12.5V switching power supply. This will maintain the voltage across the load(LED lamps) constant when the input voltage changes from 20V to 30V.

I would not try to build one if I can buy one as there is absolutely no way one can build one cheaper than buying one off the shelves.

 

@PWR2U,

Maybe you have looked in the wrong direction.

What you need is a simple 24V to 12.5V switching power supply. This will maintain the voltage across the load (LED lamps) constant when the input voltage changes from 20V to 30V.

I would not try to build one if I can buy one as there is absolutely no way one can build one cheaper than buying one off the shelves.

I appreciate the logic, but the cheapest 24 - 12v DC-DC regulator I've been able to find after a lot of searching and calling around the country is around 350 bucks. FIrstly, its too expensive to use for each lamp and, secondly, its way too big to be built in. What I'm trying to build is costing less than 20 bucks and fits on a strip of veroboard so that it can fit the lamp without being exposed.

Attached are some reference pics of the lamps my friend makes. Most of his creations are like img01 and 02 - dainty and see-thru, drawing about 250mA to 500mA each.

Img03 is of the 2nd biggest in his house - very flat (12mm height) with 380 LEDs (superbright 3.2v each) soldered and joined by hand as shown in img04. This sucker draws 2.7Amps with the LEDs joined as they are.

This is why the built-in regulator circuit has to be as small and unobrusive as possible. Img05 is my prototype pwm circuit.

So you see, my friend's an artist, I'm a muso - and between the two of us, this is a voyage of discovery.

So far, Bill has offered the most do-able solution.

Thanks again to you all for your asistance so far.

PWR2U

 

I have some other ideas, but check on the comparators. I think that will be the best idea. The 741 is a teaching tool, I would not use it. That leaves the 358, which is extremely usable.

We get people trying to use op amps as comparators all the time. I've done it myself. If you have a choice it really isn't a good idea.

The rest of it is reasonable.

I notice you didn't answer my question about the LEDs. I would really like the answer. A large part of the reason I wrote the article LEDs, 555s, Flashers, and Light Chasers was to cover the major mistakes people make using LEDs.

When I get a chance I'll show an idea I've had for a while for a current regulator.

Are you after 10ma through the LEDs?

 

Thanks, Bill. I did read your article on LEDs & 555s and found it very useful and applicable. Your tech and assistance is much appreciated.

PWR2U

 

OK, I doubt the OP will see this, but here is my first draft. It has never been built, but it would probably work.

Parts List
U1 - 555
U2 - 358
U3 - 7812

Interesting point, make Vcc 0V and Gnd -20V this would be a dandy negative voltage regulator.

This is one of those designs I'll eventually build for myself to see how well it works.