Hello good people,
I am wondering about a thing -
I have seen several examples of a medium capacitor connected in parallel with an expensive and good quality one in order to achieve better sound quality.
Can't I do the same with resistors?
If, for example, I connect an Amtrans and a Wishay Z-foil in parallel, I will in theory achieve the sound quality of both components.
Of course, both resistors must be the same value and twice the size of the desired one. Like 2K Amtrans AMRG parallel with 2K Wishay Z-foil = 1K Amtrans/Wishay as gridstop resistor.
I can't find any discussions on this topic.
Can you follow me and is there anyone here who has tried bypass resistors in this way or am I completely nuts?
Pft.
Henrik Vingborg
I am wondering about a thing -
I have seen several examples of a medium capacitor connected in parallel with an expensive and good quality one in order to achieve better sound quality.
Can't I do the same with resistors?
If, for example, I connect an Amtrans and a Wishay Z-foil in parallel, I will in theory achieve the sound quality of both components.
Of course, both resistors must be the same value and twice the size of the desired one. Like 2K Amtrans AMRG parallel with 2K Wishay Z-foil = 1K Amtrans/Wishay as gridstop resistor.
I can't find any discussions on this topic.
Can you follow me and is there anyone here who has tried bypass resistors in this way or am I completely nuts?
Pft.
Henrik Vingborg
You don't want a lot of junk hanging off the grid. Just a low inductance, physically small resistor,
connected as close as possible to the grid pin.
connected as close as possible to the grid pin.
I tested several different resistors Dac I/V stage and found that cheap metalfilm resistor sound best.
I think there's more of a problem that can appear as a result of paralleling different components together than the desired result. In general, all components are not "pure" as the schematic would expect them to be. All components will exhibit some inductance from the connecting leads, any ferrous material in the construction of a component can make it susceptible to magnetic interference. Paralleling them together can create additional "tank" circuits which can result in some alteration in the perceived sonic change, some more of a state of mind.
In my humble view, I'd rather buy the best components (at values that fit the design criteria) I can afford and optimize the circuit the design as best as possible and not muck about with tacking additional components on top of others hoping for a minor miracle... again, just my $0.02.
In my humble view, I'd rather buy the best components (at values that fit the design criteria) I can afford and optimize the circuit the design as best as possible and not muck about with tacking additional components on top of others hoping for a minor miracle... again, just my $0.02.
1. Two identical resistors in parallel allows you to:
Create a resistance that might not be a standard RETMA resistance value
Doubles the total maximum power dissipation.
These are fairly obvious results.
2. The only time I remember a special effect of two parallel resistors, was at microwave frequencies.
The construction of the microwave circuit was called "Stripline".
There is a ground plane on the backside of the insulating substrate, that covers the whole backside.
Two small 100 Ohm resistors were connected in parallel (50 Ohms).
The width of the pair of the resistors, side by side, was exactly the same as the width of the Stripline 50 Ohm transmission line.
Also, the inductance of the short leads of the resistors is in parallel, which reduces the parasitic inductance. That provides better high frequency [microwave] operation.
The 2 resistors job was to properly terminate the 50 Ohm Stripline to ground.
The width of the Stripline transmission line controls the impedance of the line, versus the ground plane on the backside.
The width of the two resistors controls the impedance of the resistance, versus the ground plane on the back side.
The impedance of the Stripline transmission line, and the impedance of the 2 resistors was maintained at 50 Ohms.
As I said, that is a special case to use 2 resistors in parallel.
3. Audio amplifier circuits are not microwave circuits. If they are, you have a problem to solve.
Create a resistance that might not be a standard RETMA resistance value
Doubles the total maximum power dissipation.
These are fairly obvious results.
2. The only time I remember a special effect of two parallel resistors, was at microwave frequencies.
The construction of the microwave circuit was called "Stripline".
There is a ground plane on the backside of the insulating substrate, that covers the whole backside.
Two small 100 Ohm resistors were connected in parallel (50 Ohms).
The width of the pair of the resistors, side by side, was exactly the same as the width of the Stripline 50 Ohm transmission line.
Also, the inductance of the short leads of the resistors is in parallel, which reduces the parasitic inductance. That provides better high frequency [microwave] operation.
The 2 resistors job was to properly terminate the 50 Ohm Stripline to ground.
The width of the Stripline transmission line controls the impedance of the line, versus the ground plane on the backside.
The width of the two resistors controls the impedance of the resistance, versus the ground plane on the back side.
The impedance of the Stripline transmission line, and the impedance of the 2 resistors was maintained at 50 Ohms.
As I said, that is a special case to use 2 resistors in parallel.
3. Audio amplifier circuits are not microwave circuits. If they are, you have a problem to solve.
Hi Henrik,
If you are a cheepskate (like many of us in DIY), and have a lot of higher value decent quality 1/4 watt metal film resistors, you can have fun using them in parallel to get exact values you may need (just like 6A3sUMMER said above).
I used very high value film resistors (sometimes as high as 1Meg) to trim cathode resistors sometimes. Since the value was so high, the required wattage was lower on the trimming high value resistor. Dialing in the ideal value for a non by-passed cathode resistor is a bit of an art...
However, for grid stoppers I see absolutely no advantage. I don't bother to seriously 'match' grid-stopper resistors unless the signal is from a microphone or MC phono pickup, etc.
I have a box of 1% Vishay Dale RN60D 300 Ohm resistors that I bought from Mouser during the pandemic. It still is about 1/3 full. I also use them for things like bias of CCS's etc. Sometimes I use 1k Ohm so have those too, if the need is there.
I also have some fancy non-inductive shindo resistors for grid stopping. I save those for customers who feel the need to spend extra money. I don't hear any difference whatsoever with them though. I guess my ears are not 'golden' enough.
If you are a cheepskate (like many of us in DIY), and have a lot of higher value decent quality 1/4 watt metal film resistors, you can have fun using them in parallel to get exact values you may need (just like 6A3sUMMER said above).
I used very high value film resistors (sometimes as high as 1Meg) to trim cathode resistors sometimes. Since the value was so high, the required wattage was lower on the trimming high value resistor. Dialing in the ideal value for a non by-passed cathode resistor is a bit of an art...
However, for grid stoppers I see absolutely no advantage. I don't bother to seriously 'match' grid-stopper resistors unless the signal is from a microphone or MC phono pickup, etc.
I have a box of 1% Vishay Dale RN60D 300 Ohm resistors that I bought from Mouser during the pandemic. It still is about 1/3 full. I also use them for things like bias of CCS's etc. Sometimes I use 1k Ohm so have those too, if the need is there.
I also have some fancy non-inductive shindo resistors for grid stopping. I save those for customers who feel the need to spend extra money. I don't hear any difference whatsoever with them though. I guess my ears are not 'golden' enough.
Last edited:
No current flows through a grid stop resistor so no voltage appears across it. What non-magical change can it make to the signal?
All good fortune,
Chris
All good fortune,
Chris
The grid stopper resistor (in combination with the output impedance of the stage driving it) form a low pass filter with the Miller capacitance of the tube. This controls the high end rolloff of the coupling. Chosen properly, the grid stopper should have no in-band impact on the stage coupling response. However, chosen improperly it can limit the high frequency response. It also can however slow bias excursions under moderate overdrive conditions improving dynamic response. And, if it is inside a feedback loop, it will affect the high frequency stability conditions of the amplifier.
All "non-magical" effects on the circuit. 😉
Yes, and these are all effects of its resistance alone. Flights of fancy about exotic and expensive (it has to be expensive!) grid stop resistors just proves that today we all have too much money. I say give that money to animal shelters or winos or whatever, but don't accept BS from the Interweb.
All good fortune,
Chris
All good fortune,
Chris
Most common tubes used in audio amplifiers can be made to oscillate at 30 MHz, some all the way to 300MHz.
The 807 with a plate cap, can be used at full power up to 60MHz, and at 55% power at 125MHz.
The 6L6 (essentially an 807 without the plate cap), has been used up to at least 30MHz.
The 12AT7 was designed to work up to near 300MHz.
At those frequencies, it only takes some wires (inductance), and some stray capacitance to cause an oscillation over 100MHz.
A grid stopper that causes a high frequency roll off at 3 MHz, will be the highest rolloff frequency of all the rest of the amplifier.
How many output transformers are not attenuated at 3MHz.
Phase, the output transformer will be the problem.
But the tube will probably not oscillate with that grid stopper.
$0.03
The 807 with a plate cap, can be used at full power up to 60MHz, and at 55% power at 125MHz.
The 6L6 (essentially an 807 without the plate cap), has been used up to at least 30MHz.
The 12AT7 was designed to work up to near 300MHz.
At those frequencies, it only takes some wires (inductance), and some stray capacitance to cause an oscillation over 100MHz.
A grid stopper that causes a high frequency roll off at 3 MHz, will be the highest rolloff frequency of all the rest of the amplifier.
How many output transformers are not attenuated at 3MHz.
Phase, the output transformer will be the problem.
But the tube will probably not oscillate with that grid stopper.
$0.03
I did a rolloff calculation:
60pF capacitance, 2.7k grid stopper, is -3dB at 1 MHz.
60pF capacitance, 2.7k grid stopper, is -3dB at 1 MHz.
THE only thing critical about a grid stopper is that it acts as much like a RESISTOR as practical. The value isn’t critical (but there will be an optimum range), noise isn’t critical, temperature stability isn’t critical. Low inductance is, but any old metal film type (even the 3 cent Chinese ones) are low enough. Is carbon comp better? Yes, but metal film is good enough. If you have CC’s on hand, use them but don’t go spending stupid money seeking them out.
The best grid stopper resistor happens to be the cheapest; carbon film. Because carbon has a high resistivity, it doesn't need many turns to achieve the required resistance so it has low inductance, and that's what you need in a grid stopper. In all other applications, carbon film resistors are horrible because of their excess noise and non-zero voltage coefficient. But there's near-negligible current flowing through a grid stopper, and its value is low, so carbon film is ideal. Beware carbon composition because they have high shunt capacitance. Save your money and put it towards a better brand of Scotch.
Carbon film has more of a tendency to fail open than metal film, and failing open is bad news for power tubes. Where you want the lowest L is in high gain small signal stages where the oscillation frequency can be a gigahertz, and I use CF for both the stopper and the grid leak. Gives an application for all the 100k, 470k and 1 Megger‘s that have piled up over the years and don’t find as much use in SS gear. Got tons of 100 ohm 1/4 CFs for the stopper too, as I use them when I want a fusible resistor. If it goes open biasing a small signal pentode or 12AT7 it’s not the end of the world.
But what about magical silver foil and oil capacitors, hand rolled on the thighs of Japanese virgins under the light of the full moon, blessed by mystics, cryogenically burned in, hand selected by audiophile masters, and individually packaged in space age acoustically balanced audio felt boxes. Won't these capacitors make all your amps achieve unobtainium levels of orgasmic musical perfection? What about those?
Sorry. 😊 I couldn't resist.
For low level circuits like an MM or MC RIAA input amplifier you should consider the johnson noise (+ excess noise and other noise mechanisms) contribution of the grid stopper and its effect on SNR. Choosing the smallest resistor value that assures stability is helpful. (I generally use 47 - 100 ohm in such locations typically I use high transconductance triodes like 6S3P/6S4P/5842, etc.)
As someone else mentioned I have found I generally prefer modestly priced MF or thin film to exotic foil types. (There are additional noise mechanisms to consider in at least some foil type resistors.)
As someone else mentioned I have found I generally prefer modestly priced MF or thin film to exotic foil types. (There are additional noise mechanisms to consider in at least some foil type resistors.)
A thinking-out-loud generalization:
Many resistors at about 20 degrees C, have -174dBm/Hz Gaussian thermal noise across their terminals (thermally generated noise).
From 20Hz to 20kHz that is approximately 43dB more noise in the audio band
-174dBm/Hz + 43 dB = -131 dBm in 20kHz bandwidth
0 dBm is 1 milliwatt.
-131 dBm is 7.94 x 10^-14 milliwatts, and that is an extremely low total power that is spread evenly across 20kHz.
So there is even less power in a narrow portion of that audio band.
The power in a specified bandwidth is constant.
But the voltage varies according to the resistance.
There will be 1,000 times more thermal noise voltage across a 1 Meg Ohm resistor than there is across a 1k Ohm resistor.
Many resistors at about 20 degrees C, have -174dBm/Hz Gaussian thermal noise across their terminals (thermally generated noise).
From 20Hz to 20kHz that is approximately 43dB more noise in the audio band
-174dBm/Hz + 43 dB = -131 dBm in 20kHz bandwidth
0 dBm is 1 milliwatt.
-131 dBm is 7.94 x 10^-14 milliwatts, and that is an extremely low total power that is spread evenly across 20kHz.
So there is even less power in a narrow portion of that audio band.
The power in a specified bandwidth is constant.
But the voltage varies according to the resistance.
There will be 1,000 times more thermal noise voltage across a 1 Meg Ohm resistor than there is across a 1k Ohm resistor.
Last edited:
Which is why in low level application, lower input impedance are preferred. An MM cartridge would prefer around 50k load anyway, an MC even lower. So the grid bias resistor is going to be lower than a Meg there.
The theoretical Johnson noise is -174dB/Hz, but CC and MOX are higher than theoretical due to the construction. It’s like impurities and dislocations in a transistor raising its noise floor. “Low noise” resistors just bring it down to theoretical levels, same as transistors. The savior of it all is that 100 or so ohms in series with 50k doesn’t do squat, or even less into a pure capacitance of a dozen or so pF at audio, even if the noise is double or triple what it “should” be.
The theoretical Johnson noise is -174dB/Hz, but CC and MOX are higher than theoretical due to the construction. It’s like impurities and dislocations in a transistor raising its noise floor. “Low noise” resistors just bring it down to theoretical levels, same as transistors. The savior of it all is that 100 or so ohms in series with 50k doesn’t do squat, or even less into a pure capacitance of a dozen or so pF at audio, even if the noise is double or triple what it “should” be.
Are there any excess noises that are not in some way proportional to current? I don't know of any, but certainly don't know everything. If not, then grid (G1) stops, with no current, cannot generate excess noise.
All good fortune,
Chris
*Excess noise is a loose term for anything "in excess of" Johnson/thermal/Nyquist noise.
All good fortune,
Chris
*Excess noise is a loose term for anything "in excess of" Johnson/thermal/Nyquist noise.