Return to the thumbnail page Display/hide file information See previous file See next file

18/20w T8/T12 12v DC MOSFET Real preheat/switchstart inverter

18/20w T8/T12 12v DC MOSFET Real preheat/switchstart inverter

Click to view full size image

My design for an inverter that drives 2, 18/20w T8/T12s in a preheat/switchstart configuration.
The transformer is wound in this order: secondary, insulation, primary, then feedback. An ideal size would be around 3cm x 2cm x 2cm
L1=50turns of 18swg wire on a ferrite rod

mosfet_invertse_fluorescent_srs.png small-inverter.gif mosfet-invertse-fluorescent-alt.gif mosfet-invertse-fluorescent-real-rapid-start-internal.gif

File information

File information

Download: Download this File
Filename:mosfet-invertse-fluorescent-alt.gif
Album name:eclipsislamps / Various DC Ballast Diagrams
Keywords:Gear
File Size:35 KB
Date added:Feb 18, 2013
Dimensions:2040 x 1540 pixels
Displayed:642 times
URL:https://www.lighting-gallery.net/gallery/displayimage.php?pos=-76348
Favorites:Add to Favorites
Comments
Globe Collector
Newbie
*
Offline

Gender: Male
Posts: 26
View Gallery

Preserving the Brightest Ideas of Our Age


View Profile Personal Message (Offline)
May 22, 2013 at 05:10 AM Author: Globe Collector
Hey, eclipse, you might need to try "blurring" the circuit diagram image slightly with a photo editor, this allows most traces to be seen at the different resolution levels of our computers' screens.
Two-tone images, particularly those where traces are one or two pixels wide, do not scale to different resolutions well and some traces can disappear entirely from the viewed image.
The file extension on this image says "....cent-alt.gif.jpeg" showing it was once a ".gif" then possibly later converted to a ".jpeg", but the old ".gif" was retained when it was changed.
Examples of image editors which would rectify this resolution image would be things like,
"Photoshop" or "Roxio Photosuite", but there is probably a whole multitude of others.
Here... http://www.lighting-gallery.net/gallery/displayimage.php?album=2496&pos=1&pid=75274 ..is an example of a couple of circuit images, generated initially with "express schematic" and exported as large .bmp images. They were then edited with "Microsoft Paint" and altered with "Photosuite", cropped, blurred, file size reduced and convered to .jpg for upload to this site. A "collage" function in "Photosuite" was used to combine the two circuits into one .jpg image. Not that the contrast of the two circuits are somewhat different.

Manufactured articles should be made to be used, not made to be sold!

Fee, Fye, Fow, Fum, A dead man's eye and a parrot's BUM!

Ash
Hero Member
*****
Offline

Posts: 3737
View Gallery


View Profile Personal Message (Offline)
May 22, 2013 at 07:28 AM Author: Ash
No need to blur, just open it in paint and drag the pieces closer together "linearly" to fit it in 640x483 px size, then the gallery won't attempt to resize it

Medved, the circuit is exactly symmetrical. When power is applied, how does it decide on which transistor to start ? can't it just hang in there stuck ? In a Royer based emergncy driver (8w T5 running at about 3w, 2xTIP31c) which i have, one of R1-R2 is missing so it is intentionally assymetrical
BG101
Full Member
***
Offline

Gender: Male
Posts: 165
View Gallery

EYE H80 Mercury Vapour


Brian TheTellyman
View Profile Personal Message (Offline)
May 22, 2013 at 02:35 PM Author: BG101
I resize all my images to 2050x1366 which the Forum seems to accept without resizing I use GIMP 2.6 which is free software.


BG

Say NO to DICTATORSHIP in the form of bulb/tube/ballast bans !!

Ash
Hero Member
*****
Offline

Posts: 3737
View Gallery


View Profile Personal Message (Offline)
May 22, 2013 at 03:01 PM Author: Ash
I resize to 1366x767 with Imagemagick (when resizing a batch of pics) or Kolorpaint (if i work on one pic more like making up one from few, cuting rotating setting gamma etc

But in this case just relocating the component list and titles, and maybe some relocating in the pic, will allow to keep hte sharp and exact lines so better
Medved
Hero Member
*****
Offline

Gender: Male
Posts: 4333
View Gallery

View Profile Personal Message (Offline)
May 22, 2013 at 04:07 PM Author: Medved
Circuit starting:
The circuit does not have to be asymmetrical to start.
When you turn ON the power on a fully symmetric circuit (assume that for the moment), both transistor activate. As a result, a current start to flow through them. As their drain voltage is forced high (or the input voltage drop), they become an amplifier with very strong positive feedback. Well, up to now nothing apparent to let the circuit oscillate. And when you would try it in a simulator, it may as well remain in such condition.
Well, what was missing in our thinking (and in the simulator) is one phenomenon, which is usually linked with problems: The noise.
Every resistive component and mainly every amplifier generate a noise - a random signal, what have nothing common with the signal fed to the circuit (here it is a steady DC voltage and currents).
As the amplifier is arranged with very strong feedback, it's noise is pot back to the input (gate, base,...) in phase with the original noise disturbance. This amplified noise is then amplified again and again fed into the input. To make the long story short, the noise, because it is random, is not symmetrical. And that is the main asymmetry in the circuit ultimately responsible for it's startup.

Well with the simulator: The simulator work with finite resolution, so it is not able to calculate the balanced state exactly accurate, so calculate it with some error. As the error is not really correlated with the signal, it should be considered as a noise and in fact it is really called "simulation noise". And even this noise is in most cases rather an enemy - as with some situations it could completely hide the useful signal. But in such circuits it could be the source of asymmetry to start the circuit in the simulator as well. But this is not reliable - many simulators have aids designed to find an equilibrium state (DC operating point,...), usually by introducing some "non-sensitive" window, so the algorithm stop iterating, so stop updating the circuit state. And these aids then usually tend to stop the desired circuit oscillations (that's why a regulation loop not oscillating in a simulator does not guarantee the circuit would be really stable in a real life). If the simulator finally does or does not oscillate then depend on the simulator settings (tolerance criteria, convergence method selection,...)
Usually the safest method is to deliberately add a "noise" source into the circuit: 10uV sinewave voltage source in series of the transistor base, some simulators even offer a random-number generator to simulate the transistor noise ("transient noise"; it usually follow the noise model of components, so the circuit behavior quite well copy the real life with things like oscillation startup times and so on). But these features slow down the simulation quite a lot, as a result I do not use these, but rather a single "kick-starter" source.

No more selfballasted c***

Ash
Hero Member
*****
Offline

Posts: 3737
View Gallery


View Profile Personal Message (Offline)
May 23, 2013 at 07:14 AM Author: Ash
Some contamination on components (moisture, dust) can drop down the feedback or maybe introduce negative feedback, i would not want to trust noise as a means of starting....
eclipsislamps
Newbie
*
Offline

Gender: Male
Posts: 17
View Gallery


View Profile WWW Personal Message (Offline)
May 23, 2013 at 09:19 AM Author: eclipsislamps
And also not to forget that the characteristics of the mosfets will vary slightly due to tolerances such as gate turn on voltages and drain-source resistance. That is all it takes to create an imbalance and start the oscillation and if you don't like it you could always use two different resistors to positive.

Keeping electrodes hot and gases ionzied.

Medved
Hero Member
*****
Offline

Gender: Male
Posts: 4333
View Gallery

View Profile Personal Message (Offline)
May 23, 2013 at 09:48 AM Author: Medved
@Ash: The noise is the most reliable starting aid, it is one from the very few phenomenons, what you can not block, it is of the ultimate reliability.
It is only of small and unknown power, but it is always there, you can not avoid it. So once your circuit is in the equilibrum state an unstable linear system, it would always start to oscillate. Of course, your design haveto be robust enough to really be unstable at component tolerances, supply spread, eventually the contamination (and so on), but when you have enough margin in your gain, it will start to socillate (or swing out of the equilibrum state in case of a flip-flop).

And further more, if any circuit designed to be an oscilator is designed so, it require an external event to start (so for small signal the gain is not sufficient to cause the positive feedback), it would be highly unreliable (an example is the two transistor astable multivibrator - with both transistors ON the circuit stay that way forever).

No more selfballasted c***

Kappa7
Jr. Member
**
Offline

Gender: Male
Posts: 58
View Gallery

View Profile Personal Message (Offline)
Dec 09, 2014 at 12:27 PM Author: Kappa7
What's the use of L1 and how affect the circuit parameters?
It works as ballasting inductor to reduce the current?
Thanks
Medved
Hero Member
*****
Offline

Gender: Male
Posts: 4333
View Gallery

View Profile Personal Message (Offline)
Dec 10, 2014 at 02:37 PM Author: Medved
It is an input filter. The voltage across the transformer is a sinewave, the transistors are just switches, which connect the lower potential to the negative input, while all the time one of them is ON. That means on the center tap is the pulsing voltage. The L1 then "connect" the DC component to the input, while keep the AC component isolated.

This arrangement means, the switching transitions happen only when there is nearly no voltage across the transistors, what means there are virtually no switching losses at all (with blocking oscillator inverters the switching transition losses are about 60%). That does mean higher efficiency not only because of eliminating one of the major loss contributor, but it allows the use of slower transistors (their slowness does not mean losses anymore), so one of the constraints less, so they could be better optimized for lower conductive losses (the only ones you can not avoid; practically it means mainly the base drive current could be way higher, what means slower switching, but as well lower Vce drop).
Other consequence is, the operation is steered mainly by the resonance frequency and battery voltage, so the parameters are insensitive to the parameters of the transistors (so no risk of thermal instabilities like with most blocking oscillator designs).

No more selfballasted c***

Kappa7
Jr. Member
**
Offline

Gender: Male
Posts: 58
View Gallery

View Profile Personal Message (Offline)
Dec 10, 2014 at 04:10 PM Author: Kappa7
Thanks for the explanation. I like this circuit, especially because is well balanced and the transistors run cool.
FrontSideBus
Jr. Member
**
Offline

Gender: Male
Posts: 59
View Gallery


View Profile Personal Message (Offline)
Dec 10, 2014 at 05:20 PM Author: FrontSideBus
Or you could just click on the image to view the original...

UK out of the EU!
http://www.lighting-gallery.net/gallery/index.php?cat=11271

© 2005-2019 Lighting-Gallery.net | Powered by: Coppermine Photo Gallery