Voltage Regulator for bulb filament. Need ideas.

[rob8748]

I need to design a voltage regulator for this purpose.
I want to control the voltage across a bulb filament from 0 to 2 volts with a current capability of 3 amps. I want to use a DAC for the control.

My questions is, what is best, transistor, mosfet, and I need some ideas for the circuit design.
I'm new here so don't bash me if this is the wrong posting area.

 

[jraef]

Maybe repost it in the Circuit engineering forum. I would just tell you to go buy a voltage regulator.

 

[rob8748]

I wish it was that easy, but if you understand filaments and how to extend the life of them, you would quickly understand how complex it can get. Thanks for the reply, I'll check the other area.

 

[itsmoked]

Why use a DAC? Why not just PWM it like usual? If you regulate as you are suggesting you are talking some nasty waste heat!

 

[bacon4life]

Since you are new, perhaps you are referring to DAC as simply a “Digital to Analog Conversion” instead of a resistor ladder type circuit? A 3A DAC might be hard to find as they are most often used as voltage reference than a voltage source i.e. connected to high impedance inputs. A Pulse Width Modulated signal with a large capacitor would allow you to digitally control the average voltage without the nasty heat waste.

More information, i.e. information on understanding filaments and how to extend the life of them would also allow us to make the best suggestions.

 

[rob8748]

Hi everyone,
I'm not new to design, just looking for some opinions.
The use of anything switching is out of the question due to noise produced. Noise produced results in a reduced filament life. Noise fluctuations cause heat fluctuations and this causes short life.
Filament science is complicated due to many factors, heat, heat time, filament post size, post length, filament thickness, type, vacuum, many, many other factors.
99% of engineers use a constant current source for this and that greatly reduces filament life. Even if it's ramped at every use.

Here is what I mean.
I have a 12bit DAC that has to be used for the control. It must control a linear circuit with the following capabilities. It can be series, shunt, or other.
0 to 2 volts max.
3 amp max.
I need this to test filament life. I can build it now, but I'm looking for better ideas from you guys. You know, one head, one way of doing things. Many heads, fresh and better ideas.
Many thanks.
BTW. please excuse me, I'm new to forums and not sure what all I need to post.

 

[itsmoked]

I presume you are going to run many of these and not stand around watching one lamp a year burn out. This means you are going to generate a lot of heat in your test system if you don't get around the pure analog waste.

If you don't want the noise I would use a filter fed by a high frequency PWM. You can filter virtually all the noise out.

Use an small PIC. Set it up to generate the pwm. Read back the voltage which is well within the PIC's A/D range to close the loop. You can talk to the PIC to tell it what voltage to run at. You can monitor the current and read it back too. Ask for the info as often as you want it. Step and repeat for as many lamps as you want. Just assign each PIC with its own address.

 

[fsmyth]

Try this:

http://people.txucom.net/fsmyth/pages/sw-regs4.html#pwm24

I have more that have not been posted to the web site if
that does not suit.
Also: google for "home power" +PWM
<als>

 

[rob8748]

Thanks for the info everyone.
But it looks like the best way to go is, just buy a Power-One supply and modify it to accept my analog input. By doing this I will reduce my build time and save money also.
So, the old trusty LM723 regulator series is the way to go.

 

[itsmoked]

Not in my opinion.

 

[Warpspeed]

rob8748, I can see where you are coming from, sort of ?

Back in my days at Westinghouse Signals we had a railway level crossing flasher that used a very simple cross coupled multivibrator using a pair of ancient old TO3 cased germanium power transistors. The thing about this was the very soft turn on characteristic that gave wonderful lamp life. The railway companies loved it.

Later this was redesigned using mosfets and hard switching, designed by a team of very young engineers that SHOULD have known better. It was not greeted with any enthusiasm by our customers, and we sold almost none.

Itsmoked speaks the truth, PWM is definitely the way to go, a high switching frequency WILL NOT thermally cycle the filaments, and anyhow, some very simple low pass filtering after your buck regulator can reduce the ripple to sub millivolt levels if you wished. A microcontroller can generate the PWM and do any slow ramping, monitoring, or feedback easily enough, and that is the obvious way to do it these days.

 

[rob8748]

Warpspeed, Itsmoked,
I'll get one up and running with a linear. Then do it with PWM. I just finished a floating high voltage pulser with mosfets. So, doing it is not a problem and certainly more efficient. The noise produced has to be in the 10's of milivolt level. As you guys know, switching mosfets produce alot of noise.
I have to use the analog control. I wish I could use a microcontroller, but I have to use what is actually going to be used for the testing.
Thanks