Using UC3843 as oscillator to control MOSFET

[sonnytiger]

Another user on the forums recommended me this to use instead of a 555 timer as the 555 timer didn't have the required bandwidth. I have spent hours studying the datasheet and trying different circuits, but can't get eh the output to toggle, I have a capacitor and resister selected to give me the frequency range I am looking for and have pin 1 directly connected to 2, to make the error amplifier constantly output 2.5V as based on my current understanding this is what I should do. Can anyone help me get this working correctly?

http://www.ti.com/lit/ds/slus223c/slus223c.pdf

[c4757p]

Oops - my apologies if you asked on the old thread and I missed it.

Remember that you need to have something happening on pin 3 (current sense). Find a way to shift the level of the Rt/Ct ramp so that it's consistently under 1V, then apply this to pin 3.

Sorry if I'm not being very helpful - I'm not exactly sure what you've done. Have you tried this already?

[sonnytiger]

No problem, I tried using two base emitter juntions and a 10k resistor, but the ramp flattened out and the lowest volatge went up to 1.5V and not above 2V. Basically the ramp became much more shallow. I am not sure what is going one. I guess I could bout this combination to ground and then connect it to the other pin?

[c4757p]

You're putting too big a load on the ramp and upsetting the oscillator. The resistor should be huge. Alternately, since you're using B-E junctions for your voltage drop, just power the collector so you get an emitter follower.

AcHmed99, I didn't even think of that circuit.  Thanks.


By the way, the reason I suggested 3843 is that you want to run pulses of current through a coil, and that is its purpose. Normally, the current sense pin's input is exactly that. You put a small resistor between the FET source and ground, and connect that to the pin, so that it sees a ramp proportional to the coil current. When you've got your coil stuff up and running, try that. You can, of course, generate pulses with a MCU or a 555 (for high frequencies, use 7555), but depending on what you're doing with the coil, a PWM controller with an analog feedback loop could be very useful, and it still does decent PWM without that.

[sonnytiger]

I don't need feedback, I just want to drive current through an air core coil at 100+ kHz, and if a 555 would have been sufficient I would have just used that. I am really not understanding what you guys are talking about I feel like an idiot. grrrrr

I currently have the resistor and capacitor set up and a jumper going from pin 1 to pin 2. that is it. just so you guys are caught up.

[c4757p]

I am really not understanding what you guys are talking about I feel like an idiot. grrrrr

I'm sorry about that! I guess I can get carried away sometimes.

If you just want a quick output waveform, the circuit AcHmed99 suggested works nicely. Just open up this app note, and look for Figure 28 (just Ctrl+F and type "Figure 28", it jumps right to it). They made a charge pump, but if you just leave off the charge pump itself (everything to the right of the 3843), you'll get a 50% 100kHz wave, and you can play with things from there. Sorry for my (perhaps dead-end) suggestion for getting it working, I use this chip a lot but never anything other than current mode, so it hadn't really occurred to me to do something that simple.

[sonnytiger]

I had the circuit I was briefly describing already set up days ago I was just filling in on what I already had just to let you know.

[sonnytiger]

I have set up the app note circuit just as it is shown, with V+ being ~24V and ground being the negative of the battery I am running it off of I have not put in the diodes and capacitors as shown in the app note instead putting a 1k to ground measuring the output on my scope. What I am getting on the output is a square wave, but at like a 99% duty cycle. I have triple checked my circuit and nothing seems to be different from the app note.

[sonnytiger]

I'm looking at the posts before in my other threads and will clarify a few things, it is an air core inductor, and I calculated, not measured it to have an inductance of 2.2uH (12 turns, 5 cm wide, 15cm long) seems to make sense.

[c4757p]

Yep, you're right, I just noticed that too. The app note calls for UC3844/5, which has a maximum duty cycle of 50%, so it doesn't go above that. The circuit does indeed run with a very high duty cycle with a UC3843. I'll see if I can figure out an easy way to limit that circuit to 50%.

[c4757p]

If you're not having much luck with the 3843, why not put it aside and try a 555? I'm not sure you'll get a high enough frequency with a 2.2uH coil to keep the current from being too great, but give it a try. (Much higher than 100 kHz will be difficult for other reasons as well, so we're not really going to get something "easy" here.) You may be able to push it higher than 100 kHz, anyway, you just can't depend on any old 555 doing the same. Also, see about adding a couple ICM7555 to your collection, too, and in the meantime you could even experiment with the good old two-transistor multivibrator. I've gotten them up to a couple hundred kHz with high gain transistors.

[sonnytiger]

Right I need to take note of these things argh nooby mistakes I'm sorry. I thank you for any help you can provide and if you do figure out a away, be sure to explain it to me, I don't want to just follow circuits, I want to understand them .

[sonnytiger]

Okay I will try this thanks! These ICM7555 do seem like an excellent addition, thanks for the suggestion!

[c4757p]

These ICM7555 do seem like an excellent addition, thanks for the suggestion!

They are excellent. Almost an "ideal" 555 - the input impedance is very high, so you can use huge resistors if you want to, max frequency is very high, and they use significantly less power. (Normal 555 gulps a ton of current at every edge, so you have to give it a really big decoupling capacitor.) They are expensive, though, so definitely try to use a normal 555 when possible...

[c4757p]

Right I need to take note of these things argh nooby mistakes I'm sorry.

Eh, no big deal. I use this chip all the time and I'm banging my head against the wall too, trying to figure out how to get it to do something unusual. In retrospect it might not have been the best suggestion, though it will definitely work, and if you're going to be doing a lot of stuff with PWM and coils you might find it more useful down the road. Sorry for giving you such a pain in the ass 

[c4757p]

Sadly, it looks like putting a ramp into the current sense input is about the only way to get a lower duty cycle out of the 3843. The difference between that and the 3844/45 is that toggle flip-flop that I remember you asking about is the block diagram - 3843 omits it.

[sonnytiger]

No problem, I think i will try a a 555 timer if the multivibrator circuit doesn't work out.

http://en.wikipedia.org/wiki/File:Transistor_Multivibrator.svg

using this circuit, I have used 10k pots for all resistors and 1nF caps and 2N5551 transistors.

Unfortunately I am getting a steady voltage at probe points that should be oscillating like the base pins, of about 700mV. I heard this is because of the small capacitors, is this true?

[c4757p]

Yep. Is 2N5551 all you have? Higher gain transistors should get it going - 5551 is pretty low gain. 2N2222 ought to work with 10k/1nF, and 2N5089 will definitely work (that's what I usually use for fast multivibrators).

Additionally - do you mean all resistors, or just the base (timing) resistors? The collector resistors should be about one tenth of the base resistors.

[c4757p]

Also, if you have Schottky diodes (I usually use BAT85S), you can increase the top speed by putting one from the base to the collector of each transistor. This prevents them from saturating, so they can turn off quicker. It has to be Schottky, though, because the voltage drop has to be less than that of the transistor.

Played with it a bit in SPICE and was able to get basic 2N2222 running at over a MHz (10k/68pF), but I have no idea if they would do anything near that in real life. My bench is a bit of a mess at the moment so I may or may not actually attempt this.

Follow-up: LTspice is indeed correct, a couple of El Cheapo 2N2222 Schottky-ified with BAT85S will happily do 4.2 MHz with 4.7k/47pF! Looks like you're set for whatever frequency you want.

[sonnytiger]

Okay thanks for the tips! I will try this tomorrow, but here its 1:30am now so maybe time for bed hahaha