I have been reading quite a bit of interesting discussion here and elsewhere regarding grid leak bias. It has some interesting applications such as shared cathode tubes and low signal level applications. However, since data sheets rarely give grid current information (and spice models don't usually represent it well) how does one calculate the resistor value to use with a particular tube? Is it done strictly by cut and try (stick in Xmeg resistor and check the resulting anode current)?
This is too general of a question. Lets be a little more specific.
RF grid leak for signal detection? Grid leak for output tubes? Self biased, fixed biased, at power limits, at 1/2 power dissipation? Parallel tubes need 1/2 the required grid leak resistance. Tubes that meet specs, or that do not pass the maximum grid resistor spec? Old gassy tubes?
Do you like to be conservative, or want a maxed out (or beyond) guitar amp?
RF grid leak for signal detection? Grid leak for output tubes? Self biased, fixed biased, at power limits, at 1/2 power dissipation? Parallel tubes need 1/2 the required grid leak resistance. Tubes that meet specs, or that do not pass the maximum grid resistor spec? Old gassy tubes?
Do you like to be conservative, or want a maxed out (or beyond) guitar amp?
I think I was not clear. I am talking about large grid leak resistor as a bias method in small signal circuits. Judging by some circuits I have seen the 12ax7 gets about 0.1V per Mohm but if one were to be using some other tube how do they pick the resistor to get the desired grid voltage without having data on the tubes grid current characteristics. Did the designers just find the value experimentally or possibly measure grid current at the desired operating point?
Grid leak bias only requires 1 resistor. Simple. Grid leak bias is a good way to pick up radio stations, and other interference, switcher supplies, CFC lights, etc. It is a problem if the tube gets the least bit gassy. May not be consistent from one brand new tube to another brand new tube. Requires a coupling cap from the input to the grid and grid bias resistor (imagine connecting a phono cartridge or a guitar pickup across the input that does not have a coupling cap. The signal source has much lower DC resistance than the required grid bias resistance. Simple is as Simple gets.
Self bias is well known, and easy to calculate for most small signal tubes. Requires one more resistor than grid leak bias (and a bypass cap if you want maximum gain). But it does not require an input capacitor from magnetic cartridges and guitar pickups, etc.
Self bias is well known, and easy to calculate for most small signal tubes. Requires one more resistor than grid leak bias (and a bypass cap if you want maximum gain). But it does not require an input capacitor from magnetic cartridges and guitar pickups, etc.
We know they didn't SPICE it in 1933.
The experiment takes mere minutes.
If you use ONLY gridleak bias, expect the stage to be fussy. Different idle point for every bottle. Hi-Mu triodes with large plate resistors do tend to find their own happy-point where gain is near-max and output swing is useful though not dependably generous.
There is info on this in Radiotron 4th and in Neumann/Irving. The latter has experimental(!) results on a small selection of modern tubes showing that different makers do NOT aim for the same grid I/V curves-- this is NOT a specified parameter.
Nothing to calculate; read the data sheet on the valve in question. Mullard worked it out for you with trial and error but there is a pattern most designers use; Beam Tetrode around 100k. Power Pentode similar but up to 220k. ECC83 style 1M and EF86 2M2. Tubes would be KT88 for Tetrode, 6L6 for Pentode, 12AX* for Triode and 6J32p for small signal Pentode. Increase these values and all sorts of problems occur. Trying to bias a valve with zero on the cathode and grid leak with the wrong value grid leak resistor will give problems as valves vary in values on their own without adding to their issues. Use a cathode resistor or -ve bias voltage for reliability. Many years ago, with valve based computing, I worked with a few hundred ECC81/12AT7 valves as mono stables and had no end of issues getting them to work.
Last edited:
It's pretty much impossible to make such calculations. This method of providing grid bias is one of those "quick 'n' dirty" solutions that is best avoided. You can't calculate an exact value since there are so many competing factors that act in opposition. You have initial electron velocity that can give electrons enough energy to hit even a negative grid. This will make the grid negative. You have ionization events that make positive ions that are attracted to any negative potential. That drives grid voltages positive. How these events balance out is unpredictable even among the same production runs of the same manufacturer. It will change over the life time of the tube in question. This is why it's a method best avoided. If you don't want a capacitor bypassed cathode resistor, these days it's easy enough to substitute SS diode bias, either LEDs or series connected small signal SS diodes.
The only other possibility for sizing a DC grid return resistor is frequency and capacitance considerations. Improved high frequency performance needs smaller grid return resistors.
European AM/FM radio sets of the 1950's and well into the 60's inevitably had an EABC80 for AM/FM detection, the Triode (ecc83 like but mu of 70) in grid-leak bias driving an EL84 output. Bias R was usually 10M, resulting in 45v on the plate at 0.5mA and 200k load from 150v B+ Of course HiFi was not the goal, 10% THD considered acceptable, 5% good, below 5% over-engineered. Worked stable and reliable in mass production, millions produced, and still no problems with old and worn tubes today, those which survived ... provided the 25nF coupling caps being replaced, the old paper caps in tar-sealed paper tubes having picked up moisture over the decades. Also the vintage 10meg carbon composition resistors inevitably increase R over time, but even at 20M functionality is fine and plate v drifting upwards may even be beneficial ... as has been said earlier, "these Hi-Mu triodes with large plate resistors do tend to find their own happy-point where gain is near-max and output swing is useful though".
Last edited:
Grid leak bias is usually used for one of two purposes:
1. low hum from heater-cathode leakage, because it allows the cathode to be grounded - it requires low source impedance and low signal level to get low distortion
2. low cost - distortion not a major consideration
In most circuits 4.7M or 10M is used. You can't calculate the value because the grid current varies so much from sample to sample, and with ageing. It generally only works with high mu valves, as it cannot generate enough voltage to bias a low mu valve. You need a suitable anode resistor too, typically at the higher end of 'normal'.
1. low hum from heater-cathode leakage, because it allows the cathode to be grounded - it requires low source impedance and low signal level to get low distortion
2. low cost - distortion not a major consideration
In most circuits 4.7M or 10M is used. You can't calculate the value because the grid current varies so much from sample to sample, and with ageing. It generally only works with high mu valves, as it cannot generate enough voltage to bias a low mu valve. You need a suitable anode resistor too, typically at the higher end of 'normal'.
A 200k plate resistor from +150V could be looked at as a very bad current source.
A 10Meg resistor and a tube with 0.1uA grid leak current will produce 1V grid bias.
So is it -1V or +1V?
The 200k plate resistor controls the operating point of the tube almost as much as the grid leak resistor does. But the real operating point is controlled more by the tube, right?
I have seen guitar amplifiers with grid leak resistors at the input tube.
Does a little distortion hurt?
Does grounding the cathode reduce hum caused by filament to cathode leakage?
A triode tube was found to be Faulty in that it had 100k Ohms of filament to cathode leakage. I never saw any other tube that was within orders of magnitude of 100k Ohms (we are talking multiple Meg Ohms or more for most good tubes I know of).
Suppose that Faulty tube has a 1k self bias resistor, and a 100uF bypass capacitor.
At 60Hz, 100uF has 26 Ohms of capacitive reactance.
26/100,000 x 6.3V = 1.6mV
That is worst case of 1.6mV of 60Hz on the cathode (with a VERY Faulty tube).
If cathodes have to be bypassed to take care of hum, then Either use good tubes, Or in the case of a phono preamp or mic preamp, use DC filament power.
And if cathodes have to be bypassed, then LTP phase splitters and cathode followers can not, and do not work.
I think otherwise; just use good tubes.
A 10Meg resistor and a tube with 0.1uA grid leak current will produce 1V grid bias.
So is it -1V or +1V?
The 200k plate resistor controls the operating point of the tube almost as much as the grid leak resistor does. But the real operating point is controlled more by the tube, right?
I have seen guitar amplifiers with grid leak resistors at the input tube.
Does a little distortion hurt?
Does grounding the cathode reduce hum caused by filament to cathode leakage?
A triode tube was found to be Faulty in that it had 100k Ohms of filament to cathode leakage. I never saw any other tube that was within orders of magnitude of 100k Ohms (we are talking multiple Meg Ohms or more for most good tubes I know of).
Suppose that Faulty tube has a 1k self bias resistor, and a 100uF bypass capacitor.
At 60Hz, 100uF has 26 Ohms of capacitive reactance.
26/100,000 x 6.3V = 1.6mV
That is worst case of 1.6mV of 60Hz on the cathode (with a VERY Faulty tube).
If cathodes have to be bypassed to take care of hum, then Either use good tubes, Or in the case of a phono preamp or mic preamp, use DC filament power.
And if cathodes have to be bypassed, then LTP phase splitters and cathode followers can not, and do not work.
I think otherwise; just use good tubes.
Last edited:
-1V. Assuming the valve is not gassy.
1.6mV could be a lot of hum. Remember that in days of old signal levels were generally lower than nowadays. I think I first saw grid leak bias used in a tape recorder design in Practical Wireless back in the 1960s, probably on an input intended for dynamic microphone. DC heaters were not commonly used then, neither were biased heaters.
LTP phase splitters generally have more signal than microphone inputs.
There is probably little need for grid leak bias now, except for low cost. It is sometimes seen in quite inappropriate situations (e.g. second stage fed from high output impedance first stage) which just shows that the designer did not know what he was doing.
Grid leak bias is intriguing because is simpler than most alternatives. As Sorento said, EABC80 and DAF96 always had a 10 Mohm grid bias resistor on their original application. I believe that they were designed and tested to work this way - the 10 Mohm value is a datasheet specification and is consistent trough manufacturers. I have tried a DAF96 with the standard 10 Mohm grid leak bias recently and it works well, but with a significant drawback. It does pick up a lot of interference from almost every digital device. I can hear my CD player servo by simply putting the CD player next to the amplifier. I will try again with better shielding.
I haven't noticed hum reduction from this arrangement alone - proper layout and grounding is still required. Even if true, I believe that it never was a priority for the manufacturers. Many radiograms and tube radios had a noticeable hum that is easily reduced with obvious and simple arrangements that mainly require a few cm of extra wire.
I haven't noticed hum reduction from this arrangement alone - proper layout and grounding is still required. Even if true, I believe that it never was a priority for the manufacturers. Many radiograms and tube radios had a noticeable hum that is easily reduced with obvious and simple arrangements that mainly require a few cm of extra wire.
I thought I was writing a "tongue in cheek" suggestion that we choose Not to use grid leak bias anymore for new design work.
What is the price of a small 1/4 watt resistor, and a modern physically small 100uF 10V bypass cap (or even a 1000uF 10V cap)?
And what are the price / performance ratios of a "cheap bad tube", versus an affordable good tube.
You should make things as simple as possible, but no simpler" . . . Einstein
What is the price of a small 1/4 watt resistor, and a modern physically small 100uF 10V bypass cap (or even a 1000uF 10V cap)?
And what are the price / performance ratios of a "cheap bad tube", versus an affordable good tube.
You should make things as simple as possible, but no simpler" . . . Einstein
I try to avoid electrolytic aluminium caps when possible, and putting a semiconductor on the cathode is not my first design choice. This is the reason why I have tried grid leak bias, but only with a tube that was designed for this.
An alternative I am considering are multilayer ceramic smd capacitors. They are now available with high capacity and low price. A few of them in parallel are smaller than a equivalent electrolytic, some key electrical parameters are better, and they don't decay over time. Has anyone already tested them as cathode bypass?
An alternative I am considering are multilayer ceramic smd capacitors. They are now available with high capacity and low price. A few of them in parallel are smaller than a equivalent electrolytic, some key electrical parameters are better, and they don't decay over time. Has anyone already tested them as cathode bypass?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.