Hi, I have been trying to dim T8 36w tubes as low as possible and have been successful since the filaments are heated and can even run at 50% on a 9w PL-S ballast with no blackening. But when run on a HF DC ballast at 10% it causes the entire tube to striate and but goes away and becomes a normal lighted tube when I turn of the filament heating..What caused this?

IMJ

It is a form of acoustic resonance inside the fill gas, what create (standing wave) bands with higher and lower pressure. As the emission strongly depend on the pressure, it create the light and dark bands.
When you switch OFF the filament, the lamp voltage changes, what cause the inverter to change it's operating frequency, so it isn't in match with the tube resonance anymore.
Moreover the lamp current start to have quite significant DC component, what tend to act against striations. That's why dimmable ballasts intentionally (usually by a resistor connected parallel to the DC blocking capacitor) give to the lamp a small DC current component. It is too small to cause electrolysis problem, but still large enough it effectively  suppress the striations at low output settings.

By the way when electrodes are heated (either externally, or by the lamp's operation), even ballasts with asymmetric current waveform (e.g. ow voltage supplied blocking oscillators; all with single transistor output stage) do not cause DC component in the lamp, so the lamp operate on AC only (assume it is connected to a transformer secondary, so the voltage have no DC component). When electrodes are heated (so very small cathode fall), for fast current changes the arc behave as quite linear resistor.
In fact what rectify high frequencies on lamps with not heated electrodes is the cathode fall, when the voltage is asymmetric.
The current flow only in the polarity, when the circuit is able to generate high enough voltage, so when the transistor is OFF. As only cathode side is heated by the ion bombardment, what mean it would heat only the side, what is cathode during the transistor OFF phase.
As a result most of single transistor ballasts without dedicated additional electrode heating stay in the rectification mode, with only the cathode side heated by the arc. This make it a bit easier to stabilize the power (during the ON state, when the voltage is given by the input and transformer ratio, so can not adopt to the lamp need, there flow no current anyway), but the lamp current have then high DC component, what could be problematic with some lamps.

When the ballast feed high enough power into the lamp, even the anode side eventually heat up, so it stop rectification. This is, why most simple battery ballasts start the lamp in two stages: First the lamp rectify, so work only on ~1/2 of the power, after some time the anode heat up, so allow the current to flow even during the transistor ON state, what mean the overall power and the light output nearly double.

The DC ballast I used to dim has 2 transistors and two capacitors connected to the secondary on the PCB in series with the output to the tube and makes a loud whining sound at full power and fortunately it's quiet when dimmed at 10%.

Then the ballast make no DC component at all, so the changing operating frequency is, what cause the difference between heated/unheated electrodes.

Now I can dim it down to 1% but there are still some signs of wear on the tube ends is it because of the electrodes are glowing for too long?

Usually dimmed lamp with auxiliary electrode heating is expected to last the same hours as on the full current. As the filament heating only is only an extreme case of dimming, I would expect they wear off in the same way, so the time when the filament power is ON count as well...
But it is hard to tell, it may differ lamp type to type.

Compare to full power the filament is not stressed by the discharge, but the whole filament is hot, so evaporate.
On full power only one spot is hot (the end of the still working emission coat closest to the filament end feeding the arc current), the rest of the filament (farther from that hot spot) remain cold, so does not evaporate, but the hot spot is stressed by the discharge, so it is slowly sputtered away.
And it is hard to say, what mechanism lead to faster overall degradation...

I suspect it's the starting used. I use a S10 to start the tube at full power and then dimmed it from there. The starter was connected in parallel and is not in series with the filaments so maybe the starting voltage was too high since the filaments is already heated. I tried with a SI51 ignitor and it could light the tube at 20% setting which I think is not as bad as with the starter at full power so I will test again with a new tube.

When filaments are already heated at start, there would be no high voltage - the developing discharge in the tube clamp it, so it won't rise too much.

How is the lamp arranged? First you wrote you used an electronic inverter to feed the arc, so what is the starter doing there?

Oh..I'm sorry I forgot to mention I have since changed the set up to ac only. The DC needed a power supply and the inverter could only run the lamp until 20% before making the noise and still have to be ac again from there. The filaments are connected to separate transformers and connected to a dimmer so I can adjust the filament heat for testing. The tube is unable to light on it's own when first turned on at all levels of dimming except at 20% when the SI51 is used and 100% when S10 starter is used. I will post a pic of the schematics... .