Grid currents are small, but through a large resistance the voltage drop can be large enough to change the voltage on the grid to a more positive value which will shift the bias point to pass more current. That can lead to more grid current, more bias shift, more current and the OP of the tube destabilized. Manufacturers list maximum values for grid leak resistor in the spec sheets.
Grid leak resistor. I'm using EML AD1 and Pete Millet's uniamp front end. EML lists grid leak resistor for AD1 as 300k. But that's for SE cathode bias operation. I'm using it in push pull fixed bias operation. EML says that AD1 will work in any schematic that is designed for 2A3. 2A3 in this situation would be 50k. That seems a good place to start.
Gird stopper is 1k CC.
Thanks guys, Kevin
Gird stopper is 1k CC.
Thanks guys, Kevin
Vous parlez français? 🙂
http://frank.pocnet.net/sheets/145/a/AD1.pdf
p.s.
2A3(Cga): 16.5pF, mu:4.2
AD1(Cga): 23pF, mu:4
Cin = Cgk + Cga(1 + A)
Compute Cin (without Cgk): AD1 has 40% larger input capacitance, thus HF rolloff would be greater (at similar grid leak and grid stopper resistance).
Last edited:
When the tube comes very hot, 1st grid can emit electrons (which are attracted by g2 or plate) an so 1st grid becomes minus negative, i.e. more positive. Plate current increase, and so do heat... further and further... This can cause tube destruction, and even overheating of the output or/and main transfos.
In order to minimize this effect, 1st grid resistor value must be low. Very good tubes have also 1st grid plated with gold.
In order to minimize this effect, 1st grid resistor value must be low. Very good tubes have also 1st grid plated with gold.
I agree with P.lacombe. An example is the KT88. But there is a time effect.
The cathode material tends to evaporate and settle at the grid, minute amounts, but very noticable. So when you have a grid leak of e.g. 220 K to the negative bias, the bias current will creap up in a number of month or years. The tube becomes hotter, more grid emission, lower bias etc and thermal run away, destruction.
The tubes can be restored for a big part by putting the grid at a posuve adjustable supply, no other voltages applied exept filament. Run up the voltage and keep close watch at the grid. The grid will become red hot (take care, don'tapply more voltage) and the kathode material evaporates from the grid. In minute or so, the grid emissioon will be diminished very considarable, sometimes a factor of 10.
You can use the tube again, but lower the grid leak. For KT88 I recommend 50K max. Especially for fixed bias without a kathode resistor (wich is no good idea)
I used this tube quite extensively.
I do not agree with miller capacitance effects, the AC impedance will be determined by the impedance of the driver connected that should be equal or better, low compared to the gridleak. And the stopper even less, that resistor is very small compared to the impedances and should only stop RF oscillations.
The cathode material tends to evaporate and settle at the grid, minute amounts, but very noticable. So when you have a grid leak of e.g. 220 K to the negative bias, the bias current will creap up in a number of month or years. The tube becomes hotter, more grid emission, lower bias etc and thermal run away, destruction.
The tubes can be restored for a big part by putting the grid at a posuve adjustable supply, no other voltages applied exept filament. Run up the voltage and keep close watch at the grid. The grid will become red hot (take care, don'tapply more voltage) and the kathode material evaporates from the grid. In minute or so, the grid emissioon will be diminished very considarable, sometimes a factor of 10.
You can use the tube again, but lower the grid leak. For KT88 I recommend 50K max. Especially for fixed bias without a kathode resistor (wich is no good idea)
I used this tube quite extensively.
I do not agree with miller capacitance effects, the AC impedance will be determined by the impedance of the driver connected that should be equal or better, low compared to the gridleak. And the stopper even less, that resistor is very small compared to the impedances and should only stop RF oscillations.
1. Grid current is such that the grid tends to bias itself around -2V (which has to do with thermal energies and leakage ratios). If this is acceptable (which is pretty common for small signal tubes like 12AX7), then it's not a big deal.
2. If you need a voltage very different from this (typical of tubes biased at higher currents, say, closer to zero bias, or of power tubes which are normally biased at -5 to -50V, say), then you will need to accommodate grid leakage accordingly.
3. The amount of resistance required depends upon the bias voltage, how hot the grid gets (particularly important for class 2 operation), and how close the grid wire and grid-cathode spacings are (higher precision tubes are more sensitive in general). Which also controls the transconductance and perveance, which is why, say, sweeps and frame-grid (e.g. video amp) types usually need lower values. The datasheet recommended values reflect this.
Failing to use the appropriate value, in a situation where it's needed, might not have any effect at all, but isn't guaranteed to survive the natural life of the tube, or if another perfectly average tube is plugged in in its place. Grid leakage is due to a number of factors, which vary between individual tubes and through a tube's life.
Operation at high temperature hastens the effect, which is the ultimate danger here: unstable grid bias produces a runaway condition where grid current increases, causing more wear, and more grid current, and so on, until the grid voltage stabilizes at a new value (which might be -2V or so, again, at which point your poor power tube is burning holes in its yellow-hot plate).
Tim
2. If you need a voltage very different from this (typical of tubes biased at higher currents, say, closer to zero bias, or of power tubes which are normally biased at -5 to -50V, say), then you will need to accommodate grid leakage accordingly.
3. The amount of resistance required depends upon the bias voltage, how hot the grid gets (particularly important for class 2 operation), and how close the grid wire and grid-cathode spacings are (higher precision tubes are more sensitive in general). Which also controls the transconductance and perveance, which is why, say, sweeps and frame-grid (e.g. video amp) types usually need lower values. The datasheet recommended values reflect this.
Failing to use the appropriate value, in a situation where it's needed, might not have any effect at all, but isn't guaranteed to survive the natural life of the tube, or if another perfectly average tube is plugged in in its place. Grid leakage is due to a number of factors, which vary between individual tubes and through a tube's life.
Operation at high temperature hastens the effect, which is the ultimate danger here: unstable grid bias produces a runaway condition where grid current increases, causing more wear, and more grid current, and so on, until the grid voltage stabilizes at a new value (which might be -2V or so, again, at which point your poor power tube is burning holes in its yellow-hot plate).
Tim
Thanks Tim, The amps uses AD1's from Emission Labs. Grid voltage is about -80 volts so I think a resistor is out. I'm leaning more towards a grid choke. The option pointed out to me by a DIY friend.
There are 3 currents, 2 of them are contradictory: one created by cathode emission, another created by a dirt. The first one in clean small tubes can be used to bias them; the higher is the grid leak resistor, the more negative is the voltage. It is an auto-bias. In big hot dirty tubes another current prevails, it causes increase of anode current, further heating, increase of a grid current, and may cause run-away, red anode, melted glass. For power tubes max grid leak resistor's value is often specified by the manufacturer.
That's my problem. EML only publishes grid leak resistor for SE and cathode bias. I'm running the tubes PP with fixed bais.
For fixed bias assume a max grid leak resistor value of between half and a quarter of what the datasheet specifies for cathode bias.
- Status
- Not open for further replies.