On a sunny day (Wed, 23 Nov 2011 19:27:04 -0500) it happened Jamie wrote in :
I still cannot decrypt that, you are saying that at zero load there is no current, and then what? In that case one input of the the current limiter comparator will be zero, and the other input will be set by Vout. Any Vout will flip that comparator to off, and the PWM stops.
On a sunny day (Wed, 23 Nov 2011 20:58:44 -0800 (PST)) it happened Tim Williams wrote in :
OK, you may be right, but I can assure you that current transformers have no problem with a DC bias, as long as it does not saturate the transformer. I think the input regulation specs, and the efficiency for this design that now is at 69% at nominal input, speak for themselves. Only the AC component of the triangle wave form is used (passed by the CT). the actual value is highly irrelevant, you can think of the current transformer's winding ratio + your losses as part of the open loop gain of the system. The voltage feedback sees perhaps a slightly changing open loop gain, but that is why it is there, the feedback, to keep the output stable.
You can test this by reducing the load resistor on the CT by half, the effect on the output is neglible. Plenty of open loop gain. In fact I just grabbed a resistor, thought it could be a bit lower, and added a second in parallel. Most non-critical thing you can imagine. Just like a shunt value, higher more open loop gain. Just do not make it too close to zero, or leave it open ;-) Same for the CT winding ratio. The PIC comparator has an offset around 10 mV IIRC, and its gain is huge, so 100 mV on the CT output is already plenty if not too much.
Personally I am very happy with this circuit, as it is a zillion times better than the 'Joule Thieve', that one would not even start with a capacitive load, leading to destruction of the switching transistor due to overheating. You need some introspection. :-)
Updated the diagram for the 'on' and 'off' touch button MOSFET added.
I added some info on that mysterious current transformer: primary 1 turn (loop through). secondary 11 turns. Load resistor 5 Ohm. No idea what the core material is, but I have a lot of those, for $$$$$$$$$$$$$$$$$$$++ a piece I am willing to part from these, but I could tell you where I got those from for much less. To put an end to speculations about waveforms that I do not understand, here is the voltage over the 5 Ohm load resistor of the current transformer versus the MOSFET drain voltage, externally triggered from the gate drive:
You can see that when the MOSFET is ON, so the rectangular waveform low, the current in the drain *sorry in the inductor* almost linearly rises with time, until it reaches the PIC comparator's reference voltage causing the gate drive to reverse, and the MOSFET goes off, and its drain voltage goes high.
The little spikes on the triangular wave are mostly scope probe pick up as it was sitting against the 33 uH inductor.
It is altogether a nice clean switch with just a little bit of RF on the start of the ON waveform, considering also the rather 'free' layout, then this is REALLY good:
This is it folks, hope it is of use to you. Those who want to prove it does not work, I am sure I will hear about it :-)
Don't be such a putz, it only serves to deepen the wound.
I don't criticize any one here, except maybe SLow-Man, I merely try to assist in observations I make with experience of my own. Unlike a few un-named here, I try not to second guess before sticking my neck out. If you find it difficult to understand my explanations of things I point out, so be it. We all can't be mind readers how ever, I find that those that really know their business do not seem to have any issues with it.
And btw John, I don't recall seeing much of anything out of you that impressed me, if at all.
Jamie
Give it up. You'll save more hair :)
Jamie
That's because you don't recall.
I never consider an SMPS even marginally tested, let alone done, until I've done some line and load variation and step response testing.
For instance, how does it respond under short-circuit conditions? Ok, so a perfect short takes over from the MOSFET more or less, then the supply gives out. But what about a marginal short? A load with a voltage higher than the battery but lower than nominal output. That will load it, saturating the voltage feedback, and commanding whatever maximum PWM is. Current feedback with no DC reference might as well be disconnected to save that pesky transformer!
Tim
-- Deep Friar: a very philosophical monk. Website: http://webpages.charter.net/dawill/tmoranwms
On a sunny day (Fri, 25 Nov 2011 19:16:06 -0600) it happened "Tim Williams" wrote in :
Tim ???? A perfect short in this sort of setup without the battery switch as in the last diagram, would ALWAYS give a high current due to the inductor-diode path.
In the MOSFET battery switch configuration the battery is switched off
I have tested with 200 % overload even in the design without battery switch. When output drops due to overload, below some point even the comparator reference drops, and in some limited way there is fold back, but this is irrelevant irre elephant well whatever, the battery switch will have done its work.
Sure, that is the whole idea of a regulator. But nothing saturates, as the current limit prevents that, to a point.
Wrong, you do not understand this architecture. It is very much the same as ones I previously discussed with Jamie, if you short the current transformer secondary then the whole things turns into a hysteretic voltage controlled regulator (on /off control).
In the latest incarnation I have added some more cools stuff. You can now set the output voltage via RS232, I moved the MOSFET battery switch to the + (no longer in between batteries), and I used PIC outputs to generate a 10V DC control for that battery switch MOSFET gate now the battery switch MOSFET only drops 30 mV at double load. Next thing I will remove the 2 resistors that are now low voltage cutoff sense to a PIC comparator, and use the ADC, and allow low voltage cutout to be specified via RS232 too for different battery types. All is saved in EEPROM of course. Oh, and did I mention it displays battery voltage and output voltage via RS232 too? So it is becoming as close as perfect as it can be, good enough to incorporate it into my nice gamma scintillators. I did a NiMH battery lifetime test yesterday, and I can run just 10 hours from 2 NiMH, with LCD back light on. Not bad. And output stays constant within a few mV over the full battery range, It did cut off at exactly 1.99 V on the Chinese multimeter.
So, perfecting the design, better than perfect is not possible. Sure it has more components than your joule thingy, but is is way more accurate, reliable, foolproof, and most of all I designed it, as opposite to buying a stupid module from ebay that does not even have a RS232 interface, and probably uses obscure parts. Hey :-)
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.