Merlinb, what are you talking about? Show me please on a graph of forward bias of a LED.
https://webench.ti.com/appinfo/webench/led/Rd.pdf
Dynamic resistance is proportional to angle of inclination of the tangent to the curve. As you may see, voltage drop with current increases slower than decreases dynamic resistance. That means, for minimum distortions we have to bias it up to currents where dynamic resistance is lower and more linear. With additional bias current, variable part of the current decreases in respect to the constant part, that decreases distortions. If you are afraid of increased bias voltage, just use a different LED.
https://webench.ti.com/appinfo/webench/led/Rd.pdf
Dynamic resistance is proportional to angle of inclination of the tangent to the curve. As you may see, voltage drop with current increases slower than decreases dynamic resistance. That means, for minimum distortions we have to bias it up to currents where dynamic resistance is lower and more linear. With additional bias current, variable part of the current decreases in respect to the constant part, that decreases distortions. If you are afraid of increased bias voltage, just use a different LED.
Last edited:
kylej₁₀₅₀;5415586 said:
Yes, it is true. The use of a higher VF forward voltage LED isn't adaptive to tube aging, tube manufacturing tolerances; it is however more-or-less fixed.
The (RK || CK) (resistance in parallel with capacitive bypass) has the advantage of also being a nearly "fixed" supply with suitably large C, and still adaptive. The best of all worlds (at least as far as I'm concerned) is the servo-set negative grid bias supply, cued off of anode setpoint design. If your B+ is say 200 V, your anode spec is 130 V for some anode resistor RA, having a servo with low-pass filter (2nd order) is a good way to set the grid bias. It becomes a fixed operating point for all tube substitutions and tube aging.
But its hard.
Hence why we don't see it much.
GoatGuy
PS: Actually it is not all that hard. Its called a constant current supply, and is pretty easy to do. For low current (less than 25 mA), a simple capacitor-bypassed JFET with small resistor in series with its source and gate-attached-to-ground does the trick. The "problem" (only for an equipment manufacturer) is there is a lot of JFET operating point variability. In practice, you use a trimpot to find your setpoint, turn everything off, then substitute the nearest ''standard'' E48 resistor in its place.
You can substitute more advanced chips, but little is gained in the end. The capacitor bypass gives an unfettered low impedance AC signal path sourcing capability. Allowing … the tube … to amplify! But this method is a anode setpoint proxy method. Actual anode setpoint servo has other subtle advantages.
Last edited:
In your own way, you're exposing the soft underbelly of LED 'biasing', I think. They're spectacularly nonlinear devices. But … to all fairness … they're also PN junction diodes. Which the Shockley model unified…
I ≈ IS • ( exp( VD / n VT ) - 1 )
Which can be rotated to
ln( 1 + I / IS ) = VD / nVT
Anyway, you could substitute any string of diodes whose summed forward voltage drop is what you're designing for. LEDs aren't themselves particular better at it than other types. They just conveniently glow brightly at the entirely coincident nominal valve current flow quiescent operating points.
GoatGuy
LED has higher bandgap energy, so only one is needed instead of 2-4 silicone dioses.
2.2.5 Temperature dependence of the energy bandgap
2.2.5 Temperature dependence of the energy bandgap
But rk increases faster still. Do the measurements. This is precisely why I addressed it in my book -it's one of those bits of received wisdom that people like to declare outright without actually having tested it.
The principle is independent of LED type; triode distortion decreases with decreasing bias voltage. Whatever LED you choose, if you current boost it you will get fractionally worse distortion than if you didn't current boost it. Unless maybe you think you have an extraordinarily linear triode and an extraordinarily non-linear LED combo, which is unlikely.
Last edited:
I made some tests with the attached (actual, no sim.) circuit. The LED was "Yellow" having some 2.0 V threshold voltage.
All essential numbers can be seen from the drawing.
I also studied the effect of bypassing of the LED, but there is no effect to THD, just a negligible increase of gain.
All essential numbers can be seen from the drawing.
I also studied the effect of bypassing of the LED, but there is no effect to THD, just a negligible increase of gain.
Attachments
Last edited:
Pick a tube type.
Use a current source as the plate load.
Calculate the RC self bias, R = Vdesired/Current.
Done!
(works for the life of the tube).
(gives low distortion).
(they will all have the same bias voltage, or else one tube is bad).
Use a current source as the plate load.
Calculate the RC self bias, R = Vdesired/Current.
Done!
(works for the life of the tube).
(gives low distortion).
(they will all have the same bias voltage, or else one tube is bad).
Last edited:
With one minor(?) exception: current sources are potentially suspect as active loads, when loads are capacitive. But I worry too much.
In the scale of things, (resistive → inductive → constant current) is the pecking order. Each successor to the right offers higher A/C signal impedance than the previous one. All of them “home in” on an operating point that suits the tube and sometimes the designer. The use of a resistor is great in that its cheap and unless made from leopard spots and hummingbird feathers, it'll be durable for decades. The inductor or choke the same, except it is the size of a toddler's fist and costs way more than ten bucks. The constant current doohickey will also “do the job”, but it will sport wildly different VA depending on the life of the tube, the maker, luck, and the VGK bias voltage.
Just saying.
Your mileage may vary.
GoatGuy
I have successfully used chokes for current sources in cathode coupled invertors.
You have to find one that has enough inductance and low enough distributed capacitance to make it work well over a wide frequency range.
Pick the right inductor DCR, and you usually can insert a resistor between the choke and the cathodes to get the right self bias voltage. Then the resistor takes care of the distributed capacitance that might cause any oscillation. But too high of a resistance does not take care of the choke's self resonance.
Choke current sources are expensive, take up real estate, and sprays magnetic fields.
Trade offs, trade offs. trade offs.
Oops, did I forget about the plate load current sources in single ended amplifiers . . . the primary of the output transformer? I have used those too.
This thread was about using an LED (either for simplicity, or for a specific bias voltage, or both).
That is why I presented the (active) plate current source, and simple RC self bias network.
It gives a constant known bias voltage.
I regularly check plate voltages in my amplifiers that use this configuration, so I know two things:
1. The plate voltage is in the range that I designed the tube to work properly in terms of swing, distortion, dissipation, etc.
2. The tube is still viable, and has not "weekend" therefore needing replacement.
I have used LM317, LM317T, NPN, N-Channel, Chokes, and IXYS current sources. They all have their tradeoffs.
For an active current source, a resistor can often be used in series, to reduce the possibility of instability; the same goes for the other parts and topology of the current source.
A good feature of this forum is that so many do pay attention to details, and have good knowledge and experience, and share it. That really pays off in correct and desired performance.
You have to find one that has enough inductance and low enough distributed capacitance to make it work well over a wide frequency range.
Pick the right inductor DCR, and you usually can insert a resistor between the choke and the cathodes to get the right self bias voltage. Then the resistor takes care of the distributed capacitance that might cause any oscillation. But too high of a resistance does not take care of the choke's self resonance.
Choke current sources are expensive, take up real estate, and sprays magnetic fields.
Trade offs, trade offs. trade offs.
Oops, did I forget about the plate load current sources in single ended amplifiers . . . the primary of the output transformer? I have used those too.
This thread was about using an LED (either for simplicity, or for a specific bias voltage, or both).
That is why I presented the (active) plate current source, and simple RC self bias network.
It gives a constant known bias voltage.
I regularly check plate voltages in my amplifiers that use this configuration, so I know two things:
1. The plate voltage is in the range that I designed the tube to work properly in terms of swing, distortion, dissipation, etc.
2. The tube is still viable, and has not "weekend" therefore needing replacement.
I have used LM317, LM317T, NPN, N-Channel, Chokes, and IXYS current sources. They all have their tradeoffs.
For an active current source, a resistor can often be used in series, to reduce the possibility of instability; the same goes for the other parts and topology of the current source.
A good feature of this forum is that so many do pay attention to details, and have good knowledge and experience, and share it. That really pays off in correct and desired performance.
Last edited:
I bow 🙂 to the competition. Bottom lines are hard to come by, but they are still possible to derive. In this case, per DF96 - watch capacitive loads on (cathode / emitter / source) followers, and having adequate resistive (or adaptive active loading) to drain the capacitive load in lockstep to the signal.
With regard(s) to your use of both constant, and not-so-constant current sources (and the thread's "LED bias" ideas), most methods can be made to work. LED is an excellent, low cost, and conveniently “self-identifying as working” component. With again — only coincidentally, but also conveniently — a forward voltage drop sufficient, and a working current range conveniently enough to serve purpose as a near-constant-voltage cathode bump supply.
They do not need to be bypassed with large-value capacitors because the gains are so minimal. In short “they work”. The only real “problem” is that they're not standardized components for this purpose, things that you can purchase by binned VF ratings. They're “borrowed” devices, in that way. To wit: you sometimes need chains of them to get sufficient voltage drop. Or to purchase special kinds that are higher-rated in allowable current flow thru them, for larger tubes. Again, not purpose built for this application devices.
In a way, they're exactly the same as the thread of using NE–2 bulbs as voltage drops for non-capacitively connected valve interstages. Not really probably the right device for the job. Certainly not tolerance-specified for the duty. But if you're building stuff “one-off”, you can always sift thru your “parts box” and find units that happen to work.
If you find my tone a bit negative, its because of this. Happenschance. Serendipitous and/or fortuitous parts repurposing.
GoatGuy
GoatGuy,
Agreed.
I guess part of my education is showing. When the Captain of a US Naval Destroyer wants an electronic equipment repaired, and there are no correct parts on the Destroyer that is in the middle of the Pacific Ocean, I did do a repair with the parts I had.
The same goes for test equipment . . . when I did not have the test gear I needed, I either invented a new test using what I had, or somehow figured out the cause of the problem.
Agreed.
I guess part of my education is showing. When the Captain of a US Naval Destroyer wants an electronic equipment repaired, and there are no correct parts on the Destroyer that is in the middle of the Pacific Ocean, I did do a repair with the parts I had.
The same goes for test equipment . . . when I did not have the test gear I needed, I either invented a new test using what I had, or somehow figured out the cause of the problem.
Love it. Even higher respect for your life experiences, Old Salt. Its one of the reasons I like(d) the nearly bomb proof-ed-ness of the old VTVM¹. You could overload that poor pair of input jacks with voltage spikes practically breaking down the insulation, and the meter'd survive it. AND remain mostly in certification.
Well, not the ampere current settings. Those poor, poor shunt resistors. Took beatings and became smoked, warped, open, fused, stinky, afire, dead … just like that. Only one twist of the knob away.
I too have had a life's worth of use-what-you-got practicum. Figuring out how to scope GHz signals with only double-digit MHz kit. Figuring out how to fix a VTVM without its schematic, at 3 AM, with something due by 7 AM. Soldering a cascade of lower-value resistors together to make a “suitable power resistor”. Creatively using series capacitors to get a lower value for RF resonance situations.
And … in (one really must admit) the ultimate in make-shift … Antenna design-and-raising. There isn't a single thing in antenna work that actually is “4 sig-figs per equations”. Nothing. The equations, the polar diagrams, the YAGI business are just hopeful hints. Your finally-cobbled-together assortment of hardware, cables, tubing and so on will do whatever it likes while you “get used to it”. But, like a tiger, it can be tamed.
Good talking to you.
GoatGuy
¹ VTVM ← Vacuum Tube Volt Meter
GoatGuy,
Thanks for sharing those stories.
I never even thought of using series capacitors to lower the series resonance. Seems 'intuitive' now (once you told me about it). Wires make great inductors at RF.
I do remember the VTVM, there are many who do not even remember the later TVM.
Antenna Design: mostly Art, Art, Art. What Science?
Raising Antennas: Who will get killed this time when something snaps?
Fixing an antenna while standing high up on an open platform that did not have guard rails . . . at least the ship was in port.
When the Destroyer got a new radar technician, you could count on the Bolometer power head being blown within the first month. The technician would forget to put the attenuator on the slotted line pickup before connecting the Bolometer. Kind of like using the 10mA scale on the VTVM to measure 1Amp.
Thanks for sharing those stories.
I never even thought of using series capacitors to lower the series resonance. Seems 'intuitive' now (once you told me about it). Wires make great inductors at RF.
I do remember the VTVM, there are many who do not even remember the later TVM.
Antenna Design: mostly Art, Art, Art. What Science?
Raising Antennas: Who will get killed this time when something snaps?
Fixing an antenna while standing high up on an open platform that did not have guard rails . . . at least the ship was in port.
When the Destroyer got a new radar technician, you could count on the Bolometer power head being blown within the first month. The technician would forget to put the attenuator on the slotted line pickup before connecting the Bolometer. Kind of like using the 10mA scale on the VTVM to measure 1Amp.
Gotcha. You assume the circuit is not re-adjusted for the higher bias voltage. In this crowd, that may be commonly done.
The mis-bias "should" be readily apparent by the change of plate voltage (about Mu times the change of LED voltage). But I know full well that many enthusiasts don't meter amplifier guts; in fact getting that info to assist debugging is sometimes futile.
That is not how I understood Merlin's assertion.
I understood the case to be added (non-tube) current flowed into the LED, to "reduce its dynamic resistance". Which it does, but also increases its voltage. Which does generally move away from lowest THD (which is usually at least bias acceptable without grid current).
Attachments
Last edited:
- Status
- Not open for further replies.