Thank you for your time, first of all "from load to no load" I meant when you take out the tip, in a constant current regulator the sudden change in an open circuit tends to overshoot the voltaje too much.
That's a non-issue here, since there is circuitry to detect a faulty/missing tip which turns off the RF stage then. This is true for the original as well as my circuit, just that mine automatically restarts once the fault condition is gone. Plus, the maximum possible output voltage from the DC/DC to the output stage is limited in this circuitry.
Also, as i said i'm not that convinced that the basic principle for the regulation here is only about constant current. The peak detector picks up the RF voltage right before the last inductor. But the voltage at that point also varies depending on the reflected RF energy. You can see that if you drive the RF stage with, lets say, only 9 volts or so (to avoid blowing up the FET) without the control loop, then enable it and connect/disconnect a 50 ohm dummy load. If my half-broken digital scope didn't play silly tricks on me, then the voltage at that point rises when no dummy load is connected. This must be reflected RF, since in an open circuit very little current could flow...
A H-bridge in rf is really complicated but once done it allows you control the current directly from the bridge, the shunt resistor with an integrator would show the current. Think the integrator must not to be so fast because of the thermal inertia of the tip but fast enough to not to burn the TVS is in antiparallel with the + and the - of the bridge.
Well, it still is far more compicated compared to the simple method that is used now, namely a very simply RF generator plus filter stage, where the output power is controlled by the supply voltage to it. If i would redo the circuit, i would keep the basic principle the same, but then use a beefier output FET, like the IXYS IXFH12N50F (or comparable) tthat is used in the newer Metcal supplies. Question is just if the currently used driver chip has enough juice to drive it, or if a different driver would be needed.
Forgive that mosfet, other talked about another much better, the IRFB4019. Note that is a "digital" audio amplifier, that means it controls the current in the speakers by varying the dutty cycle in a switching frequency much higher tan the audio frequency. Faster times, better efficiency. The mosfet I talked about was for the same purpose but its worse talking about speed and gate charge. They are designed for hard switching and that means a TVS is necessary(or a rc snubber).
I know how digital amplifier stages work. But they don't use such a high frequency. Usually they operate in the upper 100s of kHz to very low MHz regions. Plus, they usually also need LC filtering to restore the actual AC waveform. Very crude ones use the inductance of the actual speaker as part of that filtering, though.
In the end you would still need good filtering after the full bridge, the only difference then would be the use of said bridge instead of a single FET and transformer.
I agree with you about the toroids.
Power losses at the switcher I meant the IRF510. Someone talked about an rc snubber that's all. I've seen lots of rf stages dying because of a lack of a TVS.
Well, RF power stages usually die because of too much reflected power due to a impedance mismatch. (EDIT: That is assuming that the finals are not driven beyond their rating even with a correct load impedance.) There are only two sensible ways, IMHO, to remedy that: either match the load impedance, thus reducing the amount of reflected RF (not possible in this application), or reduce the output power accordingly to protect the RF final (which is what is done here). Slapping some TVS diodes into the circuit just covers the symptoms but does nothing to correct the actual problem.
The control loop controls the excess of the power but it has a time constant orders of magnitude higher than the response time of a TVS. TVSs only acts when the voltage goes over a value, see it as a hyperfast zener.
Yes, TVS's are fast. But the actual regulation is not that slow either. It is definitely fast enough to protect the output stage quite well. Keep in mind that the transition is not infinitely quick. As the tip heats up there is a more or less gradual change over time, as you can see in the graphs that i posted in this thread.
Heck, the only instance where there would be a real sudden change is when the tip is removed during opertaion, due to the time constant in the tip-detection circuitry. But even that is short enough to protect the RF stage well enough.
Yeah they are wrong by using solid magnet wire but it's not a significant problem, the current is not high. Litz wire can be made with thin enameled copper wire.
The heatup time would decrease with litz wire, and the efficiency would increase. If you have the time try it! enameled AWG40 is enough. I would like to compare theory with reality although skin effect is reality too.
While i did not try litz wire in this circuit, i did play around with different supply voltages to the RF stage, which in turn also effectively pumps more or less power into the tip. There is a point after which it makes very little to no difference in the heatup/heat-recovery time. Again, the fact that the inductors don't produce excessive heat shows that the losses in them are not that big. And even then an argument can be made that the core losses contribute the majority to the heat production. Initally i used a smaller ferrite toroid for the transformer, which then got rather warm. Using a bigger one helped with that, while the wire was the same. So, again, these losses are definitely bigger in the core than the wire.
Maybe you want to contact the manufacturers of such HAM transmitters, as well as Metcal, and tell them they are wrong, or ask them why they used solid wire instead litz wire? Would be interresting to hear their take on the issue
Greetings,
Chris