Wonder what the plate load is? I don't see how an accurate load line could be set for an EL84 with 5.5v and still see 10-15ma of idle current. That's a strange, strange load line for that tube. The operating characteristics are unpredictable at that operating point.
I dont see the point you are making.
If the stage is series transformer loaded, then the reality is nothing like the typical small signal stage.
DC operating point is just bias, not much point plotting it with a OPT of say 300 DCR
We know its steep, as long as dissipation is ok then we dont care.
The way i usually take the loadline is to mark a point at the intended anode voltage, at 0 current.
Then take the OPT reflected load to calculate the point at 0 anode volts, and the short circuit value of current (B/Z)
Then I move that line up, keeping the same gradient, to my intended bias point, e.g. 300V @ 25mA. Draw the line extending to the limits of the V axis, 0mA.
I think then, you can read off the Vg1 volts at the bias point, where X=300V and Y= 25mA, as per my example.
If I'm still not getting the gist of your question, well I give up 😕
Some valves have odd behaviours, but I havent built with EL84.
I have found with a small signal valve, I could operate with B+ at >300V, and just 0.6V grid bias, from a single diode.
Bias was about 5mA, total Pd about 0.8W and anode volts down very low, 45-55V.
I removed the diode and ran the valve with no bias, open loop, and the gain increased markedly, compared with a more proper biasing at 0.5 B+ or 0.67 B+ Volts.
To boot THD was better too, (output of a 3 stage amp), so cancellation probably had a influence
If the stage is series transformer loaded, then the reality is nothing like the typical small signal stage.
DC operating point is just bias, not much point plotting it with a OPT of say 300 DCR
We know its steep, as long as dissipation is ok then we dont care.
The way i usually take the loadline is to mark a point at the intended anode voltage, at 0 current.
Then take the OPT reflected load to calculate the point at 0 anode volts, and the short circuit value of current (B/Z)
Then I move that line up, keeping the same gradient, to my intended bias point, e.g. 300V @ 25mA. Draw the line extending to the limits of the V axis, 0mA.
I think then, you can read off the Vg1 volts at the bias point, where X=300V and Y= 25mA, as per my example.
If I'm still not getting the gist of your question, well I give up 😕
Some valves have odd behaviours, but I havent built with EL84.
I have found with a small signal valve, I could operate with B+ at >300V, and just 0.6V grid bias, from a single diode.
Bias was about 5mA, total Pd about 0.8W and anode volts down very low, 45-55V.
I removed the diode and ran the valve with no bias, open loop, and the gain increased markedly, compared with a more proper biasing at 0.5 B+ or 0.67 B+ Volts.
To boot THD was better too, (output of a 3 stage amp), so cancellation probably had a influence
Last edited:
I found the source circuit for this project. He has a 12K load resistor and 420v B+ that gives 270v on the plate with 470R Rk. This is not transformer loaded EL84. So look at a load line for EL84 @ 12K and 420v. Try to find the 5.5v G1 @ 12mA point.
How about...
Post your loadline and I'll see your point. (I guess an impossible load line.)
Post your schematics
Check you havent switched the g1 and g2 wires.
Is the tied to cathode or separately? I dont know "just off the top of my head", and I'm not going to dig out the datasheet to find out.
If separately check that connection.
Check A to g2 short somewhere.
If operation is as far out if whack as you seem to be trying to say...
Well then, theres a problem, or a bad tube, bad joint, bad cap, incorrect hook up.
Post your loadline and I'll see your point. (I guess an impossible load line.)
Post your schematics
Check you havent switched the g1 and g2 wires.
Is the tied to cathode or separately? I dont know "just off the top of my head", and I'm not going to dig out the datasheet to find out.
If separately check that connection.
Check A to g2 short somewhere.
If operation is as far out if whack as you seem to be trying to say...
Well then, theres a problem, or a bad tube, bad joint, bad cap, incorrect hook up.
I'm terribly late to the party, but it seems that the preference for LED VF is often given marvelous and life-changing reviews (by some, not all), yet in practice it is hardly better if at all from the well-bypassed cathode resistor approach.
Let's face it: all effort to bias a valve is to produce some kind of (usually) negative voltage potential on the grid, that remains relatively constant in spite of voltage variations imposed on the grid from the previous stage, or input jack.
Historically, large value capacitors were ridiculously expensive, so cathode-bias-with-bypass-cap wasn't economically feasible. When electrolytic capacitors came about, and after several novel methods were invented/discovered to create them cheaply, well … cathode bias eliminated the finicky 'C–' supplies that previously were used to create the negative grid bias. Thus life was good.
However, the audiophile community then began to question the legitimacy of cathode-resistor “biasing”, from a variety of possibly real and just as likely, imagined reasons. Maybe the 'fixed bias' approach “of Olde” might have been better?
Then along comes LEDs that can handle more than a few milliamps, and because of Physics of PN junctions, just so happens to have a strong correspondence between the color of the light emitted, and the VF forward voltage drop. A voltage drop almost completely independent on the current passing thru (not exactly, but certainly strongly so).
Ah! An alternative to finicky negative bias supplies? Something as simple as a string of LEDs in the cathode-to-ground connection? Really that simple? And emits light too? Whooppeee!
I mean seriously, there long, long were Zener diodes that very easily could have filled the shoes of the LED's functionality. But they were reputed to be noisy. Nothing a little capacitor bypass couldn't tame! Still … once besmirched, always suspicious, I guess.
The thing that the cathode-resistor-with-large-bypass-capacitor has going for it is its operating point automatic adaptability. Valves' quiescent characteristics change as they age. They change as they're swapped for different brands. They change for different manufacturing 'run' reasons. Fixed biases — whatever the method — don't adjust themselves as those parameters change. So, the valves changes reflect into valve quiescent conditions that likewise change. For better or worse.
At one “end”, you have fixed voltage biases that don't vary at all.
At the “other end”, you have constant-current sources (cap bypassed)
… that act like super-duper resistors.
Between are LEDs closer to the 'one end', and ordinary resistors, closer to the 'other'.
My preference remains with the cap-bypassed resistor, as its modest valve long-term and swap-changes tracking is quite good. LED biasing is OK, but somewhat of a silly compromise, as any particular operating point can be rather hard to achieve. Same goes for Zeners, and even moreso for fixed-voltage bias supplies.
I've only 'played around' with constant-current sources a bit, but they really do seem to be remarkable solutions to the long-term-aging and valve-swap issues. Set 'em, and forget 'em, for the most part. The bypass capacitor need only be the same 'size' as what one might choose for bypassing an ideal resistor.
So, yah. Late to the party, but an old fahrt's musings.
⋅-=≡ GoatGuy ✓ ≡=-⋅
Let's face it: all effort to bias a valve is to produce some kind of (usually) negative voltage potential on the grid, that remains relatively constant in spite of voltage variations imposed on the grid from the previous stage, or input jack.
Historically, large value capacitors were ridiculously expensive, so cathode-bias-with-bypass-cap wasn't economically feasible. When electrolytic capacitors came about, and after several novel methods were invented/discovered to create them cheaply, well … cathode bias eliminated the finicky 'C–' supplies that previously were used to create the negative grid bias. Thus life was good.
However, the audiophile community then began to question the legitimacy of cathode-resistor “biasing”, from a variety of possibly real and just as likely, imagined reasons. Maybe the 'fixed bias' approach “of Olde” might have been better?
Then along comes LEDs that can handle more than a few milliamps, and because of Physics of PN junctions, just so happens to have a strong correspondence between the color of the light emitted, and the VF forward voltage drop. A voltage drop almost completely independent on the current passing thru (not exactly, but certainly strongly so).
Ah! An alternative to finicky negative bias supplies? Something as simple as a string of LEDs in the cathode-to-ground connection? Really that simple? And emits light too? Whooppeee!
I mean seriously, there long, long were Zener diodes that very easily could have filled the shoes of the LED's functionality. But they were reputed to be noisy. Nothing a little capacitor bypass couldn't tame! Still … once besmirched, always suspicious, I guess.
The thing that the cathode-resistor-with-large-bypass-capacitor has going for it is its operating point automatic adaptability. Valves' quiescent characteristics change as they age. They change as they're swapped for different brands. They change for different manufacturing 'run' reasons. Fixed biases — whatever the method — don't adjust themselves as those parameters change. So, the valves changes reflect into valve quiescent conditions that likewise change. For better or worse.
At one “end”, you have fixed voltage biases that don't vary at all.
At the “other end”, you have constant-current sources (cap bypassed)
… that act like super-duper resistors.
Between are LEDs closer to the 'one end', and ordinary resistors, closer to the 'other'.
My preference remains with the cap-bypassed resistor, as its modest valve long-term and swap-changes tracking is quite good. LED biasing is OK, but somewhat of a silly compromise, as any particular operating point can be rather hard to achieve. Same goes for Zeners, and even moreso for fixed-voltage bias supplies.
I've only 'played around' with constant-current sources a bit, but they really do seem to be remarkable solutions to the long-term-aging and valve-swap issues. Set 'em, and forget 'em, for the most part. The bypass capacitor need only be the same 'size' as what one might choose for bypassing an ideal resistor.
So, yah. Late to the party, but an old fahrt's musings.
⋅-=≡ GoatGuy ✓ ≡=-⋅
Addendum
Well,
once my "lockdown" back recovered, I hefted a few things about and had a look at few things. I did the measurements a while back and now have to recollect my experiments.
(1) The IV characteristics of a string of 3 of the LEDS I used (not in situ) - The semi-Ln plot ended up looking like two distinct trendlines, steeper at low V with a gentle turnover into a lower gradient straight line at c.a 5.4 to 5.7V. Which means my bias in situ of 5.8-6.0V would equate to 12-16mA. The dynamic gradient of the 3 diodes was found to be c.a 32 V/A, compared to the static resistance of c.a. 420 Ohm. Which according to Morgan Jones,looks like a good thing.
The linear V-I plot tangents don't seem to change much between 5.7 to 6.3V
(2) Lack of EL84 characteristics on the web were a pain, so I tried as best as I could with an AVO mk2 char. meter, and one of teh Mullard EL84s I am using. There where some limitations in allowed settings, so I tried -Vg in the range 4-7V, and approximated the Vs as 150V. Without an inbetweeny valve plug tap, I had to trust that I dialled teh Vg spot on to the chalkline.
This was such fun. Much knob twiddling and reading an analogue meter. Anyway, the characteristic plots I was able to generate looked pretty pentodish, nothing weird, perhaps a suggestion of a wee kink at Va = 100V, but analogue meters and other uncertainties, I wouldn't bet for sure. At the same time, I looked the screen currents, and they were pancake flat for -Vg=5,6,7, with Is=1.6 mA for Vg=-6. The -Vg=4 V curve had a very definite gradient to it.
Back to the anode curves, I put in an operating point of -6V and 15mA and drew a line from (420, 0). This line crossed the -Vg=5,7V curves at 210 and 330V volts, suggesting (and I know this is rough) a fairly equal swing of 60V for a 1V change in Vg for this reqion. Experimental uncertainty notwithstanding, this suggests nice linearity and hopefully not too much harmonic distortion (yes I'm aware of several caveats, but you have to start somewhere with what you have to hand). Come to think of it I haven't measured sound levels vs input voltage. I reckon it will be far less than 1V.
Well ladies and gentlemen I've had my fun. I think I understand a tiny bit more about amplifiers, and pentodes still do my nut in as the is too much going on at once. I do (subjectively) think the LEDS do sound better than the 470R/ 100uF cathode bypass combo, perhaps I never gave the cap the X hours needed to settle down. Who knows.
Next idiot idea I have is to start the whole game over again with a C3m driver stage (and only 2 LEDS!). However, as I'm only about 33% through the record collection and loving it, I will wait till the summer has passed.
Regards to all who have contributed their wisdom,
Andrew
Well,
once my "lockdown" back recovered, I hefted a few things about and had a look at few things. I did the measurements a while back and now have to recollect my experiments.
(1) The IV characteristics of a string of 3 of the LEDS I used (not in situ) - The semi-Ln plot ended up looking like two distinct trendlines, steeper at low V with a gentle turnover into a lower gradient straight line at c.a 5.4 to 5.7V. Which means my bias in situ of 5.8-6.0V would equate to 12-16mA. The dynamic gradient of the 3 diodes was found to be c.a 32 V/A, compared to the static resistance of c.a. 420 Ohm. Which according to Morgan Jones,looks like a good thing.
The linear V-I plot tangents don't seem to change much between 5.7 to 6.3V
(2) Lack of EL84 characteristics on the web were a pain, so I tried as best as I could with an AVO mk2 char. meter, and one of teh Mullard EL84s I am using. There where some limitations in allowed settings, so I tried -Vg in the range 4-7V, and approximated the Vs as 150V. Without an inbetweeny valve plug tap, I had to trust that I dialled teh Vg spot on to the chalkline.
This was such fun. Much knob twiddling and reading an analogue meter. Anyway, the characteristic plots I was able to generate looked pretty pentodish, nothing weird, perhaps a suggestion of a wee kink at Va = 100V, but analogue meters and other uncertainties, I wouldn't bet for sure. At the same time, I looked the screen currents, and they were pancake flat for -Vg=5,6,7, with Is=1.6 mA for Vg=-6. The -Vg=4 V curve had a very definite gradient to it.
Back to the anode curves, I put in an operating point of -6V and 15mA and drew a line from (420, 0). This line crossed the -Vg=5,7V curves at 210 and 330V volts, suggesting (and I know this is rough) a fairly equal swing of 60V for a 1V change in Vg for this reqion. Experimental uncertainty notwithstanding, this suggests nice linearity and hopefully not too much harmonic distortion (yes I'm aware of several caveats, but you have to start somewhere with what you have to hand). Come to think of it I haven't measured sound levels vs input voltage. I reckon it will be far less than 1V.
Well ladies and gentlemen I've had my fun. I think I understand a tiny bit more about amplifiers, and pentodes still do my nut in as the is too much going on at once. I do (subjectively) think the LEDS do sound better than the 470R/ 100uF cathode bypass combo, perhaps I never gave the cap the X hours needed to settle down. Who knows.
Next idiot idea I have is to start the whole game over again with a C3m driver stage (and only 2 LEDS!). However, as I'm only about 33% through the record collection and loving it, I will wait till the summer has passed.
Regards to all who have contributed their wisdom,
Andrew
I used a triode strapped E180F, with a resistor load and two LEDs in cathode, and for me, LEDs sound better than resistor and capacitor. I used this to drive a 45 - simillar driving reguirements as for a 300B.