Integrated circuit for HID

Hi,


You mean you want to use a microcontroller to measure the current and voltage and calculate the power? You can measure samples of the instantaneous current and voltage and do the math, or you can measure just the time delay and assume a partial sine wave and do the math for that instead (a triac partial sine wave is very predictable). Either way you end up with the average power which you can compare to anything you want.

Hi,
At the moment I am just trying to find out, what would be the best option, so it is not yet decided whether it will be a microcontroller or something else.
Thanks for your suggestions.

I just want to confirm, whether I understand correctly what you suggested.
After my voltage measurement circuit for example if I measured 4Vpp (square wave). I can compare it by comparator with refrence that is for eaxmple 1V representing 73W. then the difference is sent to integrator circuit. the output of integrator circuit is inverted triangular signal because input is square wave. so output will ramp down when the difference is poitive that should increase the delay time to decrease power.

MrAl said:

Hi,

which causes the integrator to ramp up, which causes a longer time delay

Click to expand...

are you here suggesting that the triangular signal after integrator is sent to timer and the timer can adjust the delay of trigger pulse?
Please correct me if I am wrong,
thank you for you patience.

Hi,

I think you should draw up a circuit of what you have or what you intend to do and we can go from there. These things sometimes have to be very specific and can vary depending on the actual circuit used.

Normally a feedback system will contain an integrator/error amp and a reference. The reference sets the set point so that when the feedback signal is below the reference the integrator ramps up to increase the control signal (or ramps down to increase depending on the system). The control signal increasing means to increase the output, so if your delay has to be shorter to increase the output then your control signal has to make the delay shorter as it goes up (or down depending on the system).
So you are first making a measurement and subtracting the reference, then integrating that result. When the feedback measurement is less than the reference the integrator ramps up, when the feedback is more than the reference the integrator ramps down. In some systems the integrator ramps down instead of up and up instead of down, but the remainder of the circuit is made to accommodate that control signal regardless.
The reference can be as simple as a potentiometer voltage divider which you set to get the desired output at the time.

View attachment 62717
Hi MrAl,
This is the design I have made so far I heve left out the Triac and ignition circuit for simplicity. It looks so rough because I made all those blocks separately and now I put them together to show you. because of inducter in the lamp circuit there is some delay to lamp voltage and measured voltage is almost a square wave. but i think first i need to convert it to DC then to subtract it from refrence voltage. for now i used square wave as refrence voltage to see how it works. I got some ramp output after integrator. could you please have a look and tell me whether i am in right direction and how to use that ramp signal to get delay for trigger pulse and where to use zero cross detector.

Hi,

You're on the right track, but i couple points to review here.


What are those diodes D3 through D6 doing for you?

The schematic shows a sine wave reference, but you dont have to regulate on a cycle by cycle basis. What you would do is full wave rectify the measurement and filter it, then compare it with a DC reference and that becomes your error signal. The best way is probably to full wave rectify the voltage, then use a resistor divider after that to obtain a decent measurement level. This way the diodes variable drop dont play too much of a part in the signal measurement.

The 'Lamp' is shown as two voltage sources and two diodes, but there is always some resistance in there too, in series. It's not possible to regulate a source voltage, but it is possible to regulate a source voltage in series with a resistance and there's always some resistance.

Thank you MrAl,
I am considering already to full wave rectify the measure voltage. D3 through D6 are there to limit the current in the circuit to 1.4A.

I just dont undersatnd how to use that ramp signal for delay. Do you have any idea?

Hi again,


Well, once you get the measured signal rectified the ramp will move up higher to indicate that the delay should be longer and that makes the triac 'on' time shorter. So you have to use that with a variable one shot (triggered from the line) to vary the delay time.
My first thought was a 555 timer but im not sure yet if this is good enough. What you can do is do a simulation with a 555 and set the time of the one shot to 1/1000 seconds (a tenth of a half cycle) or shorter and see if you can vary the control voltage pin to get a delay of nearly 1/100 seconds (or maybe not that long as you probably never want to trigger that low).
If i get a chance today i'll take another look at the 555 too. I did a post about this one this forum somewhere but not sure where it is now. It was about using the control voltage to generate longer or shorter delay times in the one shot mode.

You also need a zero crossing detector to detect every zero crossing of the 50Hz line. This isnt hard to do though.

LATER:
Ok it looks like the 555 could work for this using the control voltage pin for the control. The ramp would then be connected to the CV pin and as the ramp gets higher the output conduction phase angle gets shorter. With R and C of 30k and 0.1uf, and limiting the input to the CV pin to about 0.7v to maybe 4.7v, we get a variable delay time from about 1ms to about 9ms. That's probably good enough. The input voltage to the CV pin does have to be limited though, so if the ramp naturally goes from 0v to 5v a diode from CV to output with a pullup resistor of maybe 470 ohms on the CV pin to +5v would get us there. If the ramp goes negative we have to prevent that with a limiting circuit or try to run the last stage from 0v and 5v instead of plus and minus 5v.
The 555 Trigger input would come from a zero crossing detector that generates a short 500us low going pulse for every zero cross of the line, and the 555 output going LOW would trigger the triac to turn it on.
The time constant of the output integrator should probably be much longer than 1 second, not as short as indicated on the schematic, and the gain limited to maybe 1000 instead of 100, as 100 might not be enough. Of course we also have to check the startup of the circuit to make sure we get a nice slow startup.
One question though, what is the nominal voltage of the Lamp and what is the normal line voltage going to be?

Measuring the voltage and the current might get very tricky. I would therefore recommend measuring the intensity and controlling that instead.
In critical high intensity monitoring situations like 1000 W, what can be done is sampling a portion of the lamp with a fiber optic after all of the lenses. The sensor has to be thermoelectrically cooled as well.

It won't be trivial any way you go.

KeepItSimpleStupid said:

Measuring the voltage and the current might get very tricky. I would therefore recommend measuring the intensity and controlling that instead.
In critical high intensity monitoring situations like 1000 W, what can be done is sampling a portion of the lamp with a fiber optic after all of the lenses. The sensor has to be thermoelectrically cooled as well.

It won't be trivial any way you go.

Click to expand...

Hi,

I am assuming he wants to measure the voltage. What's so hard about measuring the voltage? Is there something about this lamp that makes this hard to achieve?

From Wikipedia: https://en.wikipedia.org/wiki/High-intensity_discharge_lamp

"At the end of life, many types of high-intensity discharge lamps exhibit a phenomenon known as cycling. These lamps can be started at a relatively low voltage. As they heat up during operation, however, the internal gas pressure within the arc tube rises and a higher voltage is required to maintain the arc discharge. As a lamp gets older, the voltage necessary to maintain the arc eventually rises to exceed the voltage provided by the electrical ballast. As the lamp heats to this point, the arc fails and the lamp goes out. Eventually, with the arc extinguished, the lamp cools down again, the gas pressure in the arc tube is reduced, and the ballast can once again cause the arc to strike. The effect of this is that the lamp glows for a while and then goes out, repeatedly."

What I experienced with the lamp I used was that over time (1000 hours) the Power would increase over time to maintain the same intensity.
This lamp operated at about 45A, 22V, I think. The ignitor, which was eventually moved into the housing put out a 40 KV pulse for starting.
V, I & P were metered. Intensity was calibrated (rotating calibration standards) and controlled each time. The spectrum of the lamp would also change over time. Over time there was probably a 10% increase in power. The envelope discolors as well.

Hi KeepIt,

Thanks for sharing that information. We knew that there would be issues so we'll have to wait to see what sacha thinks about all this. All i am doing right now is helping with the voltage regulation part of the circuit which seems to be what he wants to do right now. If he wants to change that we'll have to see.

Are you saying that the lamp required a higher voltage to get the same light out over time? That's to be expected right?

I only used this kind of lamp once and i didnt like how hot it got so i stopped using it.

I am totally in the dark when it comes to these lamps (pun intended, chuckle).

I vs T decreasing. Yes, expected.

This particular lamp didn't have arc stabilization. I did design a control circuit for a 100 W housing that had the input to do so.

The project involved two board based lock-ins, a rotary solenoid color filter wheel, a rotating variable neutral density filter, an arc lamp, a white light source and a differential amplifier.

The 24 position rotary solenoid wasn't my idea, but I was charged to drive it. My first generation rotary solenoid shutter driver was in this instrument. My second generation stand ard solenoid shutter driver was in the 1000 W source.

Thankyou guys for sharing useful information,

I was very busy in designing my circuit, so I did not have time to get back to forum. Mr. KeepItSimpleStupid is absoluteltly right about the strange voltage behaviour of HID lamp over time that's why I am designing this driver, so that when lamp voltage will increase with the time, the driver circuit should decrease the current, as a result lamp power will be constant.
I have already made power measurement circuit and i am already able to control the delay for triac and did not need to use 555 for that.

Now the problem is with zero crossing detecter circuit. there is an inducter used in series with lamp. so the zero crossing of lamp voltage is different than the input AC voltage and moreover lamp voltage is a square wave.
initialy I had idea to detect the zero crossing of curent, but it does not work fine when current is zero for longer time. the circuit is shown below attachement 1. it only works when input is full rectified unfiltered signal.
View attachment 63014

My second idea was to use an XOR gate to detect zero of square wave lamp voltage, but the problem over here is that the width output pulse is very small(4us), but i need atleast 100us. Does anyone knows how can I increase the pulse width. the design is shown in following attachement.
View attachment 63013

Hi again,

Remember you dont trigger the triac with the zero crossing detector directly, the zcd triggers the one shot, and the one shot triggers the triac. Most one shots will trigger with a much smaller pulse width.

If you have trouble syncing with the line then you might have to sense the line directly with another bridge rectifier, or with the present bridge rectifier with an added isolation diode.

Again, reminder:

P(t) = k does not mean I(t) = k1

P power
t time
I intensity
k,k1 constants

If i(t)=k then P(t) > P(0) generally.

I(t) also depends on temperature and where the arc strikes in the short term.

My data is based on traceable calibration data to NREL and
a 1000 W lamp being on about 8 hrs/day. Lamp lifetime was estimated to be 1000 hrs. Spectrum changes also occurred during it's lifetime.