Using LM317 as Current sink
Sy,
I think you must have looked into using the voltage regulator, LM317 as a current sink, I see nothing wrong with using it.
Did you shelf the idea? Any particular reason why?
Sy,
I think you must have looked into using the voltage regulator, LM317 as a current sink, I see nothing wrong with using it.
Did you shelf the idea? Any particular reason why?
As a current sink for what? The output stage? If so, it's a not-so-good idea. It makes the overload recovery much worse, is unsuitable for AB operation, and the LM317 is a pretty mediocre (OK, downright lousy) current sink at high frequencies. What you want there is a constant voltage sink, and one which recovers instantaneously and without incident after overload. The LEDs fit the bill perfectly.
Some people use a current sink that's heavily bypassed, but again, this is an approach that severely compromises overload recovery, not to mention passing the output signal through some big, ugly electrolytic caps.
Some people use a current sink that's heavily bypassed, but again, this is an approach that severely compromises overload recovery, not to mention passing the output signal through some big, ugly electrolytic caps.
SY,
I'm just starting out here, but this is extremely intriguing. As long as I'm still in the "paper tiger" phase, what should be my expectations with the RLD driving my 86-87 dB @ 1w, 4 ohm, small vented two-way loudspeakers?
I doubt these were optimized for a tube amp. My "reference" amp is a Marantz MA-5, which is 120W, solid state.
Thanks!
I'm just starting out here, but this is extremely intriguing. As long as I'm still in the "paper tiger" phase, what should be my expectations with the RLD driving my 86-87 dB @ 1w, 4 ohm, small vented two-way loudspeakers?
I doubt these were optimized for a tube amp. My "reference" amp is a Marantz MA-5, which is 120W, solid state.
Thanks!
The series resistance of LED is by no means negligible, besides 5-6 LEDs in series, the total resistance adds up.
Nowadays one can get high quality caps from Panasonic eg Silmic or FM series, in AC mode the series impedance to gound is negligible. I see no reason not to use bypass caps. At the end of the day, design is compact and does the job well.
Nowadays one can get high quality caps from Panasonic eg Silmic or FM series, in AC mode the series impedance to gound is negligible. I see no reason not to use bypass caps. At the end of the day, design is compact and does the job well.
Other than blocking, which is a very serious problem, and the Achilles' Heel of nearly every self-biased output stage. And the sonic problems (real and imagined) of having that big electrolytic directly in the output stage signal current path.
The impedance of the arrays is about 4-5R, pretty negligible. And common mode to boot. And flat as far out as I can measure, certainly not the case for LM317s nor big honkin' electrolytics.
The impedance of the arrays is about 4-5R, pretty negligible. And common mode to boot. And flat as far out as I can measure, certainly not the case for LM317s nor big honkin' electrolytics.
Well, I finally finished reading this thread start to finish. It took three days off and on.
I was going to try a LM317 current source for the Sven 6P1P, but now I think I'll try LEDs instead.
Thanks to all who contributed to this thread.
I was going to try a LM317 current source for the Sven 6P1P, but now I think I'll try LEDs instead.
Thanks to all who contributed to this thread.
I just measured 40 LEDs (T-10) I have from the 80s and they were all between 1.63 and 1.65V at 10mA. If I crank them up to 30mA they get up to 1.69-1.72V.
Next I guess I need to measure the dynamic impedance.
I'm looking for less than 5 Ohms, right?
Otherwise, why not configure the LM317 as a voltage source and bias the cathode at 10.5V? Eliminate the current source for a voltage source which is what the LEDs are doing any way.
I'll try all three and see for myself.
Next I guess I need to measure the dynamic impedance.
I'm looking for less than 5 Ohms, right?
Otherwise, why not configure the LM317 as a voltage source and bias the cathode at 10.5V? Eliminate the current source for a voltage source which is what the LEDs are doing any way.
I'll try all three and see for myself.
Last edited:
Won't work. An LM317 has a single NPN series pass element, so, as a voltage regulator, it can only source current, not sink it. In other words, the cathode current of the tube will reverse bias the pass device and the voltage will shoot up to whatever.
It is fine to use an LM317 here as a current regulator/sink as it will need to be bypassed with big electro anyway. This makes the HF performance of the current regulator/sink irrelevant.
But anyway, all this still looks quite pointless to me. I'd just connect the cathodes to ground and use (adjustable) fixed bias on the grids.
That is why I said adjustable. In any case, LED or voltage regulator bias at the cathode basically is fixed bias with the same issue WRT tube aging and drift / balance. You're just raising the cathode to a fixed potential above ground instead of lowering the grid to a fixed potential below ground. The former is the hard way to go about it.
Last edited:
That's correct, Glen, the bias from LEDs is fixed, but again, the issue is recovery after overload (blocking). Using conventional fixed bias, a clipping event becomes a clllliiiiiipppppiiiiinnnnngggg eeeeevvvvvveeeeennnnntttttt.
Read the references. Especially the treatment in Crowhurst and Jones. And, of course, as I point out in the text, this was experimentally verified.
I don't have Crowhurst and Jones handy, but, I just skimmed through your blog. You don't actually explain why fixed bias applied to the cathodes should improve the overload recovery there either, but I'm keeping an open mind.
The benefit of the large grid stopper resistors used is valid, but that goes for their inclusion in any conventional fixed bias circuit (cathodes grounded) as well as your LED bias circuit.
When you say "conventional fixed bias", do you mean "conventional fixed bias without the large grid stoppers"?
The benefit of the large grid stopper resistors used is valid, but that goes for their inclusion in any conventional fixed bias circuit (cathodes grounded) as well as your LED bias circuit.
When you say "conventional fixed bias", do you mean "conventional fixed bias without the large grid stoppers"?
Last edited:
Large grid stoppers definitely help conventional fixed bias as well (one more argument in favor of pentodes). If memory serves, this was a Good Trick used by ElectroVoice back in the '60s.
IMHO, clipping on a sudent "Needle in the red on the Vumeter" is not a problem by itself and the ear accomodates easily for this brief overload/distortion.
Worst is what happen just after if the peak somewhat altered the bias and until it recovers its correct value.
A cathode bypass cap may retain a wrong bias voltage for more than half a second.
Of course the link cap may possibly do the same, it shouldn't be over sized . . . il not removable !
BTW, biasing using a positive voltage source at the cathode or a negative one at the grid makes no difference for me.
Yves.
Worst is what happen just after if the peak somewhat altered the bias and until it recovers its correct value.
A cathode bypass cap may retain a wrong bias voltage for more than half a second.
Of course the link cap may possibly do the same, it shouldn't be over sized . . . il not removable !
BTW, biasing using a positive voltage source at the cathode or a negative one at the grid makes no difference for me.
Yves.