Who knows? Seriously, although valves with low gm can probably get away without a grid-stopper at all, it's also down to circuit layout. The only definitive answer can be given by exciting the circuit with a >10kHz square wave and looking at the cathode (if it's undecoupled) or the anode with a >100MHz oscilloscope to see if there's any oscillation. Sorry not to be more helpful, but it really is suck it and see.
Usual practice values for small signal tubes g1 stoppers are 1-10k. (Don´t ask me why, but values at the upper end of that range are often found in geetah amps, while 1k is found in HiFi amps).
For power tube g1 stoppers, its values usually is at the lower end of that range (typically 1k).
For power tube screen stoppers, typical values are in the 100-270 ohms range. The Telefunken Werkstattbuch recommends not to go much above 100 ohms, so the changing voltage drop effect due to varying screen current across that stopper will stay neglegible (and hence will keep distortion low).
Tom
For power tube g1 stoppers, its values usually is at the lower end of that range (typically 1k).
For power tube screen stoppers, typical values are in the 100-270 ohms range. The Telefunken Werkstattbuch recommends not to go much above 100 ohms, so the changing voltage drop effect due to varying screen current across that stopper will stay neglegible (and hence will keep distortion low).
Tom
dsavitsk said:
It all depends. The whole point of adding a stopper is to form a first order lowpass filter with the tube's input + Miller capacitance being the reactive element. That gives: w= 1/RC for the cutoff frequency. To find out what you need, o'scope to see what the problematic frequency is. (In my case, it wasn't an oscillation, but rather a 700KHz "clear channel" AM xmtr some 30 miles away. To keep that out, I needed a 4.7K stopper on a 6J5 with Av= 10.0. This gave a cutoff of about 50KHz, well below the 700KHz noise.
For guitar amps, you typically find much larger stoppers being used to shape the frequency response in the upper audio band. That's neither necessary nor desireable for music reproduction.
I'm afraid I have to disagree with you there. Whilst it's perfectly possible to use a grid-stopper as a CR low-pass filter, the original purpose of the grid-stopper was to lower the Q of the resonant circuit formed by stray C and stray L. Since Q = 1/R sqrt(L/C), fitting the grid-stopper resistor as close to the valve pin as possible reduces L, and hence, Q. That's why it's always recommended that the grid-stopper should be as close to the valve pin as possible, and why some people even go to the lengths of soldering SM resistors directly onto the pin.
Well I've never seen that in any old radio handbook. I've only seen them advised to limit forward grid current (power valves) and to block RF (input stage in general). Soldering it directly to the pin ensures you won't be picking up RF after the stopper.EC8010 said:
You can pretty much omit the RF blocking resistor completely if you add a capacitor from anode to grid, whose value acts with the Miller C and ouput impedance of the devide you're connecting, to produce a low pass filter.
Hi,
I can see the need for making a low pass filter on the input on apower amplifier, it is a good idea to filter out any frequency that is higher than what the amplifier can handle, it will eliminate any chance of TIM or DIM type of distortion that is produced by a feedback loop that have lower bandwidth than the input signal.
However I agree with EC8010 that the main purpose of a gridstopper is to reduce any chance of oscillation that can happen in an amplifier stage as some tubes can oscillate for certain values of input impedance when the relations between Cgp and stray inductance meet certain conditions.
However some tubes will never oscillate and I therefore think a bad idea to use gridstoppers as default as it is un-neccesary and often less than ideal to reduce bandwidth of an amplifier stage.
In my preamplifier I only use a gridstopper on the input of the line stage as I found that there was a risk of oscillations when the volume control is close to 0. In my power amplifier I use a gridstopper that make dual service as a lowpass filter so it is quite big at 4.7kohm, it give me a -3dB frequency of about 300kHz.
Regards Hans
I can see the need for making a low pass filter on the input on apower amplifier, it is a good idea to filter out any frequency that is higher than what the amplifier can handle, it will eliminate any chance of TIM or DIM type of distortion that is produced by a feedback loop that have lower bandwidth than the input signal.
However I agree with EC8010 that the main purpose of a gridstopper is to reduce any chance of oscillation that can happen in an amplifier stage as some tubes can oscillate for certain values of input impedance when the relations between Cgp and stray inductance meet certain conditions.
However some tubes will never oscillate and I therefore think a bad idea to use gridstoppers as default as it is un-neccesary and often less than ideal to reduce bandwidth of an amplifier stage.
In my preamplifier I only use a gridstopper on the input of the line stage as I found that there was a risk of oscillations when the volume control is close to 0. In my power amplifier I use a gridstopper that make dual service as a lowpass filter so it is quite big at 4.7kohm, it give me a -3dB frequency of about 300kHz.
Regards Hans
A few more words on grid stoppers. I usually use the smallest value possible for a grid stopper. I start with a few hundred ohms and raise it if necessary. I have seen a few stubborn cases where there was an oscillation that only lasted for a few cycles at about 80 MHz just as the input signal crossed through zero. As stated earlier it was occurring only at certain positions of the volume control. This seems to happen most often when RF triodes are used as small signal audio amplifiers. I found it with a storage scope while searching for a vague raspyness on some vocals. The tube was a 12AT7 this time (look it up, the 12AT7 was developed for the tuner section of an FM radio receiver). I have seen the same effect with 6BQ7, 6BK7, 6AQ8. and 5670. In the 12AT7 case I went from 150 ohms to 470 ohms.
Another use for large grid stoppers is on the signal grid of an output pentode. When fed by the typical coupling capacitor, and grid bias resistor, it is common to encounter a "blocking distortion" where the grid is driven positive and upsets the charge balance on the coupling cap. The overload "recovery time" is determined by the magnitude of charge drained off of the coupling cap by the conducting grid, and the RC time constant of the coupling cap and the grid bias resistor (often 100K or larger). The amp will distort when hit with a large transient and continue to distort a normal signal for the "recovery time" after the transient passes. I have seen guitar amps thet were modded by an "expert" that distorted badly for up to a second after a blast of a 10 db overdriven signal. This overload is common in guitar amp applications.
If the grid stopper is big (22 to 75K ohms) the grid will not be able to suck much charge out of the coupling cap, so the recovery time is very fast. The amp will softly clip the transient and continue operating without distortion as soon as the transient passes.
Why can't I do this with an output triode? The Miller capacitance will form a nice low pass filter against this resistor and kill all of your treble. This is why you see so many DC coupled DHT circuits. There are other ways to solve this problem, PowerDrive, intrestage transformers, and cathode followers are a few.
Another use for large grid stoppers is on the signal grid of an output pentode. When fed by the typical coupling capacitor, and grid bias resistor, it is common to encounter a "blocking distortion" where the grid is driven positive and upsets the charge balance on the coupling cap. The overload "recovery time" is determined by the magnitude of charge drained off of the coupling cap by the conducting grid, and the RC time constant of the coupling cap and the grid bias resistor (often 100K or larger). The amp will distort when hit with a large transient and continue to distort a normal signal for the "recovery time" after the transient passes. I have seen guitar amps thet were modded by an "expert" that distorted badly for up to a second after a blast of a 10 db overdriven signal. This overload is common in guitar amp applications.
If the grid stopper is big (22 to 75K ohms) the grid will not be able to suck much charge out of the coupling cap, so the recovery time is very fast. The amp will softly clip the transient and continue operating without distortion as soon as the transient passes.
Why can't I do this with an output triode? The Miller capacitance will form a nice low pass filter against this resistor and kill all of your treble. This is why you see so many DC coupled DHT circuits. There are other ways to solve this problem, PowerDrive, intrestage transformers, and cathode followers are a few.
In my amplifiers I get away with a few hundred ohms on the grid with DHT's. I can say that I have only used a few different types of DHT's though. Mostly 45's, 2A3's, 300B's, 211's, 845's (all low cost varieties) and a few military surplus tubes. I have found that some of the high Mu triodes (particularly the 811A) will oscillate if you give them a chance. It often helps to add a snubber on the plate cap of these tube types. I make it by wrapping about 10 turns of enamel covered wire (about 20 gauge) around a 100 ohm 1 or 2 watt CARBON COMP resistor. The wire is soldered to the resistor leads so that it is in parallel with the resistor. This is also useful on 807's and some sweep tubes.
Hi,
Stray inductance and capacitance you usually say is what you get in addition to what the tube have and is unwanted. It comes from the tube socket and from the components you connect to the tube, a capacitor with say 10cm leads probably have more inductance than the grid itself and the same goes for capacitance where the socket sometimes have more capacitance than the tube cap between grid and cathode. BTW the grid is not a true spiral in most tubes, it is made like a spiral but the turns are short circuited at 2 places along each turn, so the inducatnce is much less than it would have been if it had been a spiral, (how would a tube like ECC81 work at several 100s of MHz with that kind of inductance?)
What is important is that the combination of gain, tube capacitance, inductance and stray additions can make the grid to be seen as a negative resistance with some sort of reactance, if then you connect some components which at some frequency look like the conjugate to this negative resistnace and reactance combination you will have created an oscillator at that frequency. Adding a series resistance will give the result will be that the conditions for oscillation are no longer fulfilled and oscillation will stop, a ferrite bead can give the same result but it will be a case by case situation.
As not all tubes, (at least not with all kind of circuit values) can create a negative resistance at the grid there is no reason to add gridstoppers to all tube amplifiers as some do, it think is better to add them only if they are needed.
Regards Hans
Stray inductance and capacitance you usually say is what you get in addition to what the tube have and is unwanted. It comes from the tube socket and from the components you connect to the tube, a capacitor with say 10cm leads probably have more inductance than the grid itself and the same goes for capacitance where the socket sometimes have more capacitance than the tube cap between grid and cathode. BTW the grid is not a true spiral in most tubes, it is made like a spiral but the turns are short circuited at 2 places along each turn, so the inducatnce is much less than it would have been if it had been a spiral, (how would a tube like ECC81 work at several 100s of MHz with that kind of inductance?)
What is important is that the combination of gain, tube capacitance, inductance and stray additions can make the grid to be seen as a negative resistance with some sort of reactance, if then you connect some components which at some frequency look like the conjugate to this negative resistnace and reactance combination you will have created an oscillator at that frequency. Adding a series resistance will give the result will be that the conditions for oscillation are no longer fulfilled and oscillation will stop, a ferrite bead can give the same result but it will be a case by case situation.
As not all tubes, (at least not with all kind of circuit values) can create a negative resistance at the grid there is no reason to add gridstoppers to all tube amplifiers as some do, it think is better to add them only if they are needed.
Regards Hans
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.