Not SY, but you would IF the voltage drop introduced by an LED for a particular tube isn't enough to quench over-current bias point. I don't have a ready example tube, but say a tube that passes 5 ma with a 5 volt negative grid nominal bias. (Which perhaps obviously would nominally take a 1,000 Ω cathode resistor if 'done that way')
You design in 2 ea, 2.5 volt RED LEDs to give the 5 volts. All's hunky-dory until for some reason one of the LEDs goes foof, and shorts to 'wire resistance'. Now you have a 2.5 volt drop nominal instead of 5. What's the tube going to do? At first blush, it is going to multiply that 2.5 volt grid rise by gm. But of course if it nominally has an anode resistor (and not something old-fashioned /or/ fancy like an interstage transformer as load), then its plate voltage will also drop. There are a zillion great equations for where the new quiescent point will be, and who knows … maybe up to 2× the current flow, cathode-to-anode.
Is this current hogging?
Yep.
Is it bad current hogging?
Nope.
Is it going to have an audio consequence?
Yep.
What consequence?
Overall gain change.
Lower dynamic range of the stage.
Greater distortion.
Potentially out-of-spec plate power dissipation.
So.
Tho' not the doyen of today's top-shelf design, the ol' bypassed cathode resistors certainly had one darn good thing going for them: they were nearly immune to going "short" rogue. And moreover, as the tube 'above' them aged, or was replaced with something not having to-spec gm, or the like, they also adapted in voltage drop accordingly. A rather helpful characteristic.
GoatGuy
You design in 2 ea, 2.5 volt RED LEDs to give the 5 volts. All's hunky-dory until for some reason one of the LEDs goes foof, and shorts to 'wire resistance'. Now you have a 2.5 volt drop nominal instead of 5. What's the tube going to do? At first blush, it is going to multiply that 2.5 volt grid rise by gm. But of course if it nominally has an anode resistor (and not something old-fashioned /or/ fancy like an interstage transformer as load), then its plate voltage will also drop. There are a zillion great equations for where the new quiescent point will be, and who knows … maybe up to 2× the current flow, cathode-to-anode.
Is this current hogging?
Yep.
Is it bad current hogging?
Nope.
Is it going to have an audio consequence?
Yep.
What consequence?
Overall gain change.
Lower dynamic range of the stage.
Greater distortion.
Potentially out-of-spec plate power dissipation.
So.
Tho' not the doyen of today's top-shelf design, the ol' bypassed cathode resistors certainly had one darn good thing going for them: they were nearly immune to going "short" rogue. And moreover, as the tube 'above' them aged, or was replaced with something not having to-spec gm, or the like, they also adapted in voltage drop accordingly. A rather helpful characteristic.
GoatGuy
I may be missing something here; but isn't the main benefit of LED bias the low slope (AC) resistance which does away with the need for a cathode bypass capacitor? A red LED @10mA has a slope resistance of ~4R. Stringing 15-20 in series will increase that resistance to 60R-80R. Also, diode bias is "fixed" bias. None of the "compensation" of resistor cathode bias as the tube ages. I use LED bias for voltage gain stages. I just don't see the appeal for an output tube. I'm always willing to learn something new.
That's a good thought. Personally, I've been quite surprised at the uniformity of current-sharing so long as a bunch of diodes are used from the same "bag" of named-brand devices. The factory is pretty good at picking LEDs that have the same light-output for a given current flow and forward voltage.
Especially the bright blue ones.
Dunno why.
By the way, at some basic level, isn't this a job for power ZENER diodes … perhaps in a string? You know, the 5 watters.
Like http://www.mouser.com/ds/2/308/1N5333B-D-102336.pdf
It lists RZ as about 11 Ω at VZ of 33 volts.
Just saying. You're either into "flash an pizzazz" of a big bright array of LED lamps in series-parallel, or you're into a solid, maybe old-fashioned, but definitely excellent solution.
GoatGuy
As a post-script, might even more appropriately, going with a constant-current source that is bypassed with a large value capacitor retain the ultimate in what a simple resistor partially provides?
(It provides tube-ageing adaptation as well as roll-a-valve reasonable outcome for swapping tubes, or just replacing them)
In the cap-bypassed mode, it (the constant current source) acts like a variable resistance that adjusts value to match current flow. The cap bypass then turns that into a very LPF, squashing the voltage variation with signal. Average current flow remains 'constant' over hundreds of milliseconds.
Just saying.
A very little used, easily tuned MOSFET + resistor can "get 'r done". Plus a 470 μF ÷ 63 VDC low ESR Panasonic high-temperature cap. Don't forget the tiny strap on heat sink!
GoatGuy
(It provides tube-ageing adaptation as well as roll-a-valve reasonable outcome for swapping tubes, or just replacing them)
In the cap-bypassed mode, it (the constant current source) acts like a variable resistance that adjusts value to match current flow. The cap bypass then turns that into a very LPF, squashing the voltage variation with signal. Average current flow remains 'constant' over hundreds of milliseconds.
Just saying.
A very little used, easily tuned MOSFET + resistor can "get 'r done". Plus a 470 μF ÷ 63 VDC low ESR Panasonic high-temperature cap. Don't forget the tiny strap on heat sink!
GoatGuy
But you talked about LEDs in series, and of course they don't hog current - they have, by default, exactly the same current, they can't otherwise.
And light output is not related to the current sharing issue, that's irrelevant. What does impact current sharing is forward voltage versus forward current.
Jan
And light output is not related to the current sharing issue, that's irrelevant. What does impact current sharing is forward voltage versus forward current.
Jan
bold: of course, I wasn't implying otherwise.
I was however NOTING that for whatever reason, as delivered, present manufacturing bins the LED devices 'in bags' to be very consistent both in forward voltage for a given brightness, and in brightness at a given pass current. Thus (and this is the part that still produces a little bit of amazement to me), ALL the LEDs in an el-cheapo e-bay "battery powered christmas tree lights" array (ALL in parallel, with no current-leveling resistors per LED) are all observably exactly the same brightness.
Since all anodes and all cathodes are tied together, any slight variations in VFORWARD would - because of the if ≈ ebVf/kT/ nature of of semiconductor PN junctions - would have their iNOM substantially affected by tiny differences in VF. And thus brightness.
Just saying: brightness is a pretty good proxy for assessing VF uniformity.
So, that's a long winded way of rebutting your assessment of "irrelevant".
GoatGuy
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.