problem getting RF power into a coil

As my first experience working with RF power I am trying to turn on a quartz torch using 13.56 MHZ, 2.5 Kw OEM-28B RF genrator using a Ashern L50 automatcher and a 3 turn copper coil (1/8 inch OD tube and about 3 cm diameter coil similar to the one used for ICP-mass). I have conneted everything and gradually increased RF power to 2000W. I can see on the automatcher power goes into the copper coil with about 2000V peak-to-peak. This means about 1 Amp current is flowing inside the coil. I checked returned power and it is almost zero which means matching is OK but this 1 amp current is not enough to fire the argon gas inside the torch (I also have used spark arc for starting plasma). I have very weak plasma close to outer quartz tube but usually much stronger plasma should be built in the center of the torch which this is not happening. What I am suspecting is some problems cause current not to flow inside the coil as much it should flow. peak to peak voltage is too high and current is too low despite zero returned power. I have connected the coil ground to automatcher chassis.
1- What might be the poblem here?
2- How much coil connections quality can affect current flow?
3- Any other idea how can I mak peak to peak voltage to decrease and current inside the coil to increase?
4- Am I in the right track?

I guess it all depends on how much you trust the auto-matcher readout! If it says there is very little reflected power, then the incident power has to be going somewhere. That means it is either going into the coil, or is heating up the electrical connections to the coil. If you look at the coil, does it look like someone took a welding torch to the connections between the wires and the coil? If so, clean them off well and reconnect tightly.

It has been a long while since I worked with one of these types of setups, but we used to use a flourescent bulb near the coil to see just how much power was radiating. The brighter the bulb, the more power going to it. I have used smaller neon bulbs to check microwave field strengths inside of a chamber--not sure if they would work down at 13 MHz, but it might be worth a try. You can put the neon bulb at the end of a non conducting stick (like fiberglass) and probe around.

I would also make sure there is no arcing going on anywhere near the coil, as that will consume all the power. Look for black carbon arc tracks leading away from the coil.

The voltage and power calculation seems wrong. The coil current can be expected to be mostly reactive for a correctly matched inductive coupler. If the generator is actually sourcing 2 kW power, the V*I product will be much higher than 2000.

On the generator I went up to 2000W and checked returned power is almost zero on the generator. What is the correct current and wattage inside the coil?

On the automatcher I can see peak to peak voltage is 2000 V. No trace of arc and no extra heat on the contacts. Everything seems very normal except there is no stronge plasma as I was expected. This type of coil I used usually used in ICP mass instruments but in ICP mass frequency is least 27 MHZ. I can see after a few minutes both generator and coil cooling waters geting warm. I have separate 10 liter tanks closed loop water cooling recirculating circuits. One cools Rf generator and the other one cools coil and automatcher. What I suspect is connecting the coil return end to the automatcher chassis. I think I should connect coil return end to the ground end of the coil inside the automatcher. I can't think of anything else. Automatcher also shows both load capacitor and tune capacitor are at their ends if this helps to diagnose the problem.
I have attached the photo of the turned on torch showing coil connection to automatcher output (ground connection is not shown) and two leads of spark generating device went inside of the quartz torch from the front.

I don't know, where the generator power is absorbed, possibly in the coil or partly in the automatcher cicrcuit (if the VSWR measurement is correct).

Generally, you should consider, that the H field of the coil doesn't directly exitate the plasma rather than the induced E field. If I understand right, it's considerably lower for same coil current at 13.56 MHz than at 27 MHz. So you have to apply either a higher current or use a larger coil for 13.56 MHz operation.

At the moment I have zero reflected power. Is that possible to have impedance matching in two points? I mean one with high voltage+low current and one with low voltage+high current?

If so I might be able to match impedance with tap increase or decrease of the coil inside the match box and see if I can get another matching point with higher current?