Forums » General Electronics Discussion

Hi,

I find a for me interesting circuit here: http://www.nerdkits.com/videos/backlight/

But the SIM isn't really working:

https://www.circuitlab.com/circuit/vs6b7k/el-inverter-with-fet-not-working/

Voltage is very low - unless I change the L to a transformer where I leave the secondary open.

Is the standard indoctor SIM without iron core?

Good news, it works fine if you turn the time step down to like 0.1usec.

You need the time step to be at least ten times smaller than your clock period. I turned it down to 0.1u and then you get like 80 volts out after a msec or two.

It also helps to at first simplify things, remove the second FET and make the 100K load resistor in there all the time.

And yes, it can take a minute or two for the sim to run at those short time steps. Just think of our poor ancestors, who had to punch the node descriptions onto cards, then submit the cards to a big mainframe, and then maybe get results back the next day. Shed a small tear for the old EE.

It may be safer to think of the timestep in terms of edge speeds rather than clock frequencies for analogue circuits.

The MOSFETs have fast edges (the page and video @Circute refers to mentions 22ns) but the inductor rise time will be somewhat slower due to the capacitive loading of the MOSFET parasitics.

0.1u (100n) is a reasonable compromise for this example: it's much less than 10th of the clock period of approx 6.7us and comparable with the likely risetime of the inductor voltage.

@Circute, it helps clarify what you are looking at (and others if you need help) if you attach netnames. Then it's easy to see what you're probing.

Note that the NerdKits example is feeding the EL panel from the voltage across the 100k resistor so that is probably the voltage you want to look at. If you named the diode end of R1 "hot" and the other end "cold" then you could plot V(hot)-V(cold) and your eventual plot would be self explanatory.

:)

I used to have an axe to grind but there's not enough left to hold ...

I am still learning and need to digest what you two write.

But when I think about it more deeply I start to think that the whole idea is not very practical because of the 2 input signals (150 + 150KHz). That would make the circuit rather complex (and relatively expensive) I guess.

Yes, the EL load would be parallel to the 100K R. But I don't see any pulsating voltage there. EL is usually driven by 130VAC/400Hz (or something similar). I just see a DC of about 100V

By intermittently earthing one end of the EL panel, the 150Hz signal effectively converts the DC across C1 to a pulse waveform across the EL panels.

Bingo: AC.

This might help with a bit of background:

http://electronicdesign.com/article/components/transformerless-el-display-driver-uses-lower-volta

:)

Remember: the light at the end of the tunnel is probably just some other lost soul wandering towards you holding a torch ...

I don't quite see how that original circuit could ever work. The panel will charge up on the first cycle and then never discharge. You need constant charge.discharge cycles to get light out.

The circuit from EDN will "work", for small values of "work"-- the way it makes the final AC is by shorting out the power supply, which is frowned upon in some circles.

It would be a whole lot easier, but a bit heavier, to just have a small transformer to step up the AC from the oscillator.