Low voltage 12AU7 (19J6) -- re "starving"

[jackinnj]

when I built this the THD% using the 12AU7 model I have simmed the same as the one I breadboarded. I've used LED's on the cathode for a number of SY designs but here you are really constrained by low plate voltage:

[DF96]

And your point is?

12AU7 and 6J6 are not the same, so THD simulation will not tell you much. Some of your distortion will be coming from the diode.

[Tony]

Some of your distortion will be coming from the diode

why? my 6H23 preamp so biased with a silicon diode sounds fantastic to my ears....

[SY]

You probably won't have much distortion from the diode (that's something I've never been able to measure in a wide variety of circuits), but you may very well have unexpected distortion from grid current, which usually doesn't show up in sims. Many tubes of that class start showing significant grid current at Vgk < 1V or so.

[Tony]

I realise that with the forward biased diode, you are limited to an input voltage of 0.7 volts peak if you are to avoid getting the grid in positive region, i have used a pot in the input to pad the signal......

[DF96]

Sorry, I misread the data. 6.44mV of signal across the diode is pk-pk, not pk or RMS. However, that would still give a few % of 2nd? (assuming a diode behaves something like a BJT Vbe? Vsig pk = THD 2nd % ?) Would be reduced because it is only a small part of the total signal. So 3.2mV gives 3.2% 2nd, but the diode signal is only 3.2% of the total so I make that about 0.1% 2nd due to the diode.

If the valve model includes grid current (some do, some don't) then you can check for distortion from this by adding some impedance to your signal source.

[DF96]

Originally Posted by Tony

I realise that with the forward biased diode, you are limited to an input voltage of 0.7 volts peak if you are to avoid getting the grid in positive region

No, fortunately it is better than that. Most of the signal is developed across the grid-cathode so the diode only sees a fraction.

However, it is not the grid going positive you need to worry about. As SY says, grid current can start well before that. I seem to recall that the books warn that positive grid current can start anywhere between -1.5V and -0.5V, and the threshold varies not only from valve to valve (even of the same type) but can also change with ageing. I try never to have a grid more positive than -1V.

[Tony]

i recall tubelab's experiments with the 845's those chines tubes seemed to pull grid currents even when not yet into positive grid.......

[jackinnj]

OK, I straightened out the SIM by using the Leach models and setting the coefficients such that the VI curves look like the 19J6 datasheet.

With the diode and the sections of the 19J6 in parallel, the THD% was 0.31%, with a 2K resistor per the Starving Student schema it was 0.052%

We're constrained by the 48V supply. Even more constrained by 19J6 becoming unobtanium.

[DF96]

If you are very patient 19J6 appear on ebay from time to time, but who wants to wait a year before building an amp? You could consider UCC84 (or possibly UCC85), although these are not too common either. They are European 100mA heater valves, with E (6.3V) and P (300mA) versions too. The '85 is almost the same as a 12AT7, apart from the addition of a screen and different heater. The '84 was intended for cascode, so should be OK at low anode volts.