Ping Sy: LED for Cathode Bias

[arnoldc]

Hi Sy,

I'm reposting this-

---
Sy, I want to try LEDs. I have a citcuit (417A) that is biased via cathode resistor 100 ohms with cathode voltage of 2V. So that's around 20mA. Can you recommend an LED that can be ordered from Farnell or RS Components?
---

I found this from Farnell- HLMP-6000. The specs said:

Current, If luminous intensity:10mA
Current, If max:50mA
Depth, external:2.2mm
Depth, lead cross-section:0.5mm
LED colour presentation:Tinted diffused
Length / Height, external:2.15mm
Length, lead:11mm
Voltage, PIV max:3V

In addition, my amp drivers are 417A at 10mA and 2V. It seems to my fickle mind that this LED will work too.

Need you thought on this.

 

[Shoog]

I've been experimenting with LED bias recently myself. The noise differences between different LEDs seem marginal so I wouldn't worry to much about that. Seems better to run them at the top of there current range rather than the bottom, otherwise the result can be grainy (from my experience). Therefore your current demand of 20mA is perfect - just so long as you don't go for a low current LED.
You would be looking at Orange LED's for your two volts drop (I think green may be OK to). My experience with some orange LED's was that there voltage drop was more variable than RED LED's - I would therefore suggest getting a few extra to match up.

Try;
Farnell - TLH04400 656-550

Shoog

 

[pinkmouse]

Hi mate

SY's off travelling at the moment, but he generally recommends standard cheap LEDs, usually bought surplus for as little money as possible. From my knowledge of the circuit, that one looks fine.

 

[EC8010]

What you want is an LED that doesn't produce much light for the current through it, so look for one producing a luminous intensity of 10mCd (milliCandela) or less. A red LED will probably give a slightly lower voltage drop than you need, so orange might do the trick. As previously mentioned, FEC 656-550 might do.

 

[arnoldc]

Hi guys, thank you very much for the help.

I've been getting mixed reactions from Audio Asylum about LED vis-a-vis cathode resistor (w/ or w/out capacitor bypass) bias so I'd like to hear it for myself on amps I'm familiar with- mine

Thanks for looking it up for me Shoog. Al, thanks for pitching in for Sy, and thanks to you EC8010 for the confirmation.

Cheers!

 

[Shoog]

Have you got any old orange/green LEDs in dead equipment - that would be the easiest way of trying things out without spending a penny. Mind you don't do what I did and use LED's with built in reistors (doo).

Its good to hear that you want to try it by your own ear. Any statement by an audio freak has to be taken with a pinch of salt. The most violent protests are often from people who "go by the gospel" and have never tested an idea themselves.

Shoog

 

[rdf]

My recollectrion is SY does dynamic testing and chose a red LED presenting an AC impedance of approximately 5 ohms at the current of interest. I've been able to find LEDs good for about 7 ohms, unfortunately unmarked models purchased by the bag for about $1.

In my testing red LEDs had lower dynamic impedance than any other colour at the cost of lower bias voltage. Even among reds the impedance varied by a factor of three. Regarding noise, biased with a constant current source no LED of any colour that I tested had a noise level obviously higher than the the measurement floor of my 24-bit card - in other words more than 124 dB below 1 Vrms.

In my circuits, bypassed by a small film and the layout kept tight, I think they work extremely well if the bias point falls where you need it. I have yet to measure an unbypassed resistor that didn't display a ton of extra low-level distortion components over an LED, but that's a taste issue.

 

[SY]

Hello, checking in from Linz, en route to London. As rdf indicated, very few companies list dynamic impedance as a spec; it generally needs to be measured. And, as EC8010 says, the older and poorer light, generally the better. I bought several varieties at surplus (unmarked), set up a jig to test them, then went back and bought a pile of the ones that tested best. But really, for this application, I don't think that there will be much of a significant difference between 4 ohms and 7 ohms, so just buy cheap surplus and avoid anything that is marketed as "high intensity" or "high efficiency."

 

[zarniwoop]

I thought I was somewhat crazy,.. I'm using a blue LED on the cathode of a 6h30 input stage. Bias is right on, but I don't know if this is better than an unbypassed resistor.

 

[arnoldc]

rdf, thanks for that additional information. I plan to test bypassed and unbypassed LED in my amp. I'm a sucker for measurements, have a scope but don't know how much to make the most of it. Can you recommend a good, available (i.e. Amazon) book on audio measurements?

Sy, glad to hear from you that my application won't be that picky in the selection of tha LED. Have a nice trip.

zarniwoop, you never compared the two? Hmm, I have a headphone amp here that I plan to use the LEDs on and hopefully I can hear something different.

 

[rdf]

Hi arnoldc! Afraid I can't help you with a book recommendation, I've been doing audio measurements for so long the source of that knowledge is completely forgotten. My recommendation is to download the Rightmark Audio Analyzer (http://audio.rightmark.org/products/rmaa.shtml) and use your sound card to play around. For a free tool it's very powerful. Any decent audio card will work for signals within its limits. Unfortunately any measurements involving DC or high levels will require some interfacing to keep from taking out the card and/or computer.

You may already realize this, but when playing with bypass caps remember the AC impedance of the LED will be in the 10 ohm range. To completely bypass the LED at audio frequencies requires a massive cap, which is a bit contrary to the point of using a LED. In my current circuits the bypass is only intended to provide a solid shunt between grid and cathode at RF frequencies because, to be honest, that's the kick I'm on at the moment. To my ear the quality of that cap is surprisingly audible, though I reserve the right to claim is wasn't after further auditioning.

 

[arnoldc]

hi rdf, I've downloaded that software quite sometime ago but have not played around with it. I'll check.

In your experienced opinon, will the LED require cap bypass on -

a) resistor loaded
b) choke loaded

single triode section, in driver duties?

 

[SY]

The 5-8 ohm impedance provides quite marginal degeneration, especially with low mu tubes. Just out of curiosity, I measured the harmonic spectrum of a couple of red LEDs in cathode circuits- well below my measurement limits. All I could see was a few millivolts of fundamental, exactly what would be predicted from the dynamic impedance, i.e., the thing looked like a 5 ohm resistor. Any bypass cap that has a significant effect at those low impedances will likely be more nonlinear than the LED.

I'm including some of these measurements in an article I've been trying to finish on a power amp design.

 

[rdf]

Can't say with certainty arnoldc since I haven't tried either yet. One circuit (with additional oddities of its own!) is a 6C45 into a 5 kohm SE OPT, the other is a 6C45 loaded by an IXYS solid state CCS in series with 10 kohm. Don't see why a small bypass wouldn't work in choke and resistor loaded circuits though. The values I use are usually well under 0.1 uf.

I don't want to claim too much science behind it as nothing I have at home measures very high frequencies. Spice simulations using available models for LEDs suggested their impedance rises with frequency above the audio band. When used as a cathode load this implies increasing cathode feedback at ultrasonic frequencies and beyond. That to me implies the potential for decreasing stability so I shunt the LED and let the tube roam free up there. Neither circuit uses grid stoppers with a very high transconductance tube and neither is at all prone to RF pickup or makes any noise, that's about all of a 'scientific' confirmation for the principle I can offer at the moment and it's not much. It's such an easy thing to try, I recommend giving both a go and find out if it matters for your circuit.

I see that SY posted while I wrote this. What he said! I'm not bypassing at audio frequencies, only ultrasonic and beyond.

 

[rdf]

Whoops, I see I wrote the LED bypass is intended to provide a "shunt between grid and cathode" a couple of posts back. That should of course be "cathode and ground". Sorry about that.

 

[SY]

I haven't done a dynamic impedance measurement over 20kHz, but it's certainly flat to there. And switching response time (using square wave excitation) is better than I can measure. In "Valve Amplifiers," Morgan Jones showed some models of LED nonlinearity but didn't measure it at useful currents (10mA). I note, though, that he didn't use a bypass cap in the LED-biased phono stage. In fairness, the signal levels there are low, much lower than for a driver stage. But in his 6SN7 distortion measurements and my similar 5692 measurements, the LEDs seemed to contribute negligable nonlinearity. Jones's advice was to keep currents high and maximize plate loads.

All that said, I certainly don't see how a high frequency bypass could hurt anything. It's the idea of a whomping electrolytic that spooks me.

 

[arnoldc]

I'm getting a bit frustrated, all my LEDs are blue

I picked one that looks like Agilent and hooked it up to my 45 amp, as expected, I'm not getting 2V bias but rather 3V spot on.

Grrrrrr

 

[rdf]

Attached is what LTSpice claims as a comparative plot of the cathode voltage for a 6C45 biased at 10 ma with a QTLP690C LED, the latter pulled at random from LT's default library pack. The solid green trace is for an unbypassed LED, the blue represents a perfect 0.1 uf cap, no inductance, leakage, etc.. Manufacturer's specs for most film caps that value show a self-resonance between 10 and 20 MHZ anyway. Possibly 'math gone wild' with no true correlation to reality but it's what got me trying bypasses.

 

[arnoldc]

Guys, I just have to post this.

I bought some LEDs from a local shop and picked the orange ones for my experiment.

I have a decent 45 amplifier, which I can say can go head to head with commercial ones like Sun Audio in its current form. My driver is an Raytheon 5842, with choke load on plate, cap coupled (Jensen) to globe 45s with grid choke, small F-475 Tamura OPTs as output.

I replaced the 100 ohm resistor and its Panasonic 470uF bypass cap and in goes the orange LED which gave me 1.9V bias. Just perfect (would have loved 2V).

WOW!

I played my usual torture CDs, The Blue Man Group The Complex and Audio CDs. WOW! Played some more, Switchfoot, and Cowboy Junkies... WOW! Played some vinyls, Sergio Mendez, The Great White, Crowded House, Clearaudio Test Disk (Classiscal), REM... WOW!

I have more bass depth and weight. The highs are cleaner. The resolution and dynamic contrast improved!

Geez! For *just* an LED??!!

I am so happy, I still have a wide grin while posting this. Thanks to Sy and to all who supported me on this.

I'll be leaving still wearing this grin... Off to a Sergio Mendez and Brasil 2006 concert. They're in town

 

[SY]

Glad that worked out, but really, all praise should go to MJ, whose wonderful books convinced me to try this in the first place.

Enjoy the concert; I spent last weekend in Vienna and managed a nice opera and a symphonic concert. It's so much better than even the best hifi ever made....