CW arc lamp PSU design, now the rat has moved out.

[Laser Ben]

Do you know what PWM signal it takes? The reason that I ask is that you may be able to integrate a PIC in there to provide the signals if you don't have any other way.

[mixedgas]

Do you know what PWM signal it takes? The reason that I ask is that you may be able to integrate a PIC in there to provide the signals if you don't have any other way.

Have to disagree here:


A PIC in that quasi RF environment would last about 5 clock cycles from when the main contactor closes. You can't take the risk of the micro not starting up, least you loose a lamp. Plus the arc lamp needs closed loop feedback and a lead/lag network of sorts to stabilize it. 8 bit PWM quanitization is also very bad in this case, from a feedback point of view. If he needs to change PWM frequency, and he WILL, he doesnt have to recode the 555.

He's better off with the 555s , mainly considering a 555 is inclined to survive the odd 500V spike, if not work better when spiked. 1970's semiconductors are tough, because of their large feature sizes.


Steve

[heroic]

Have to disagree here:


A PIC in that quasi RF environment would last about 5 clock cycles from when the main contactor closes.

I would suggest that whatever electronics he uses should be in a metal can, and communicate with the outside via an optocoupler arrangement. It may cost a little more, but I'm sure it'll be more reliable.

[dave]

He's better off with the 555s , mainly considering a 555 is inclined to survive the odd 500V spike, if not work better when spiked. 1970's semiconductors are tough, because of their large feature sizes.


Steve

What about the SG3525, IR2110 combo a lot of other big switching designs use??

[mixedgas]

What about the SG3525, IR2110 combo a lot of other big switching designs use??

Usually very tough, 3525 is done in a MSI process, and would be fine for this.

Steve

[weartronics]

Hi Guys,

Thanks for all the tips so far. Progress has been a little slow on this one, because we have so many other projects going right now.

In addition to the IGBT current control and the SCR triggering control, there is quite a lot of start-up sequencing and monitoring logic required. Originally, this was handled by a PLC behind the front panel, however the PLC is now defunct and I suspect the last time anybody knew where the program was in 1995. I will replace the PLC with a standalone micro, only because I am much more familiar with standalone micro programming than PLC programming. The PLC would have operated without any RF shielding, so I expect a micro will survive in the same environment. If inteference is an issue, it is possible to incorporate some RF shielding.

Although the IGBT switching was originally handled with a 556, it is very difficult to reverse engineer this circuit. The 556 seems like a very convoluted way to design SMPS, compared to e.g. TL1451 or SG3525. In this case, it is easiest to use the micro to produce the PWM signal for the IGBT and the trigger pulse for the SCR - of course these signals are opto-isolated.

So far, I did take a day to transplant the state switch, current pot, emission indicator and lamp hour meter from the original pendant to the front panel of the PSU. I also added an LCD display for the lamp current, interlock states etc... Since a micro is already used, it is actually cheaper to buy and simpler to drive this type of LCD than it is to multiplex lots of seven-segment and individual LEDs.

Here is a photo to keep you interested until next episode. Naturally, the new legends are laser marked, and the housing for the LCD is laser cut. I will post back in a few weeks once I have designed the controllers for the PSU and the cooling system. I am looking forwards to get the lamp running, because then it becomes the "simple" task of aligning the optics!



Best regards,

weartronics

[mixedgas]

After two weeks of working on a Positive Light Merlin, I would strongly suggest that if the shutter is closed, the lamp is idling.

Other things learned. Ramp down your lamp for a few minutes when done, DO NOT just shut it off, they get squirrelly at start and the next ignite is harder and the lamp voltage is up about 3 volts the next time you run it.
Lamps seem to last a lot longer if you ignite them at low current and ramp up, the Merlin takes a whole 5 seconds for a major current change, with a RC integrator hooked to the current pot wiper. The I pot is 10 turn, so its even slower if you try to go fast.

minimum current needs to be about 7 amps to sustain the lamp, then the supply can drop it as low as 3.6 once the cathode is hot and emiting electrons. I set the minimum on the merlin to 8.5 to account for aging.

Steve