Using R+C instead of LED.

[garfo]

Good day,

Was wondering if anybody has tried to change the LED with typical R+C and what results has he experienced.

Many thanks,

garfo.

[Paul Birkeland]

You can replace each LED with a resistor, and you will reduce the overall gain of the preamp, but it will otherwise still work.  Adding a large capacitor across the resistor will restore the gain.  (A pair of 470 Ohm resistors are acceptable)

The LED offers the best of both worlds.  It gives us the proper bias voltage that we want, with very low AC impedance (like a capacitor), but it sounds way better than any capacitor we have ever tried. 

-PB

[garfo]

Thanks a lot. Do we know how much current does the ecc82 absorb ?

[Paul Joppa]

Approximately 3.8mA.

[garfo]

Ok, so we shd actually use a lower value than 470R for the cathode resistor. Nevertheless will start with 470R tantalum/1W and 1pct and move from there. What about shunt caps ? Is 150uF ok ?

[Paul Birkeland]

A 1W resistor is physically way too big for the job.  1/100 of a Watt is enough power for the resistor, go small so you can get them to fit!  Additionally, when you put a big cap across them, the cap is providing the low impedance path to signal currents. 

For what it's worth, the LED will sound a lot better than any resistor/cap combination, so I wouldn't spend too much money on parts that you may well end up throwing away.

-PB

[garfo]

If LED biasing is that performant, why is it so rarely used in SE applications ?

[Grainger49]

Good question, but Bottlehead uses it often.  In their upgrades they are replaced with constant current sources.  But your question would be better aimed at the companies who don't use LED biasing.

[garfo]

Unfo Mr Kondo has passed away, otherwise I would have asked him... 

[garfo]

In anycase it appears that given 3.8mA absorbed by the tube, the correct value for R should be 390R.  This will give more or less the 1.5VDC tension. Ie 390x0.0038=1.482V, owise 470R will give 1.78V.

Am I wrong somewhere ?

[Paul Birkeland]

If LED biasing is that performant, why is it so rarely used in SE applications ?

Solid state parts are often "scary" to potential SE customers.  It also costs a lot more than just using a resistor/capacitor combination. 

Other technology in our amps that you won't find in SE amps:

1.  Shunt regulated power supplies.
2.  Parallel feed output topology.
3.  Constant current source loads.
4.  Cree high voltage Schottky diodes

etc. etc.

The biasing diodes give us exactly the bias we want with low dynamic impedance across the audio band (not necessarily the case with a cap).  We could save some money by going with a resistor and capacitor instead, but it just doesn't sound as good.

[garfo]

Money not an issue really, a good cap and a resistor can easly cost twice the led. For your guidance, here in Eu, we can easily buy a pair of HLMP6000 for 50c (sensibly less when buyong more than 1 piece). A good capacitor, like Evox Rifa for example, can go upto 10 dollars.

[Paul Birkeland]

Money not an issue really [...] A good capacitor, like Evox Rifa for example, can go up to 10 dollars.

I think money is exactly the issue.  We seem to notice sometimes that folks will throw money into parts for their kits that may actually hurt performance. (I did this a lot when I was first building kits)

I'll just reiterate that in our 20+ years of designing and building many different tube circuits, no resistor/capacitor combination has come close to the venerable HLMP-6000.  (Even Blackgate caps and Tantalum resistors)  You will also notice that in our least expensive tube preamp (the Quickie), we have compromised for costs and used a resistor and capacitor per channel for biasing.

[garfo]

Thanks Paul. Is the HLMP6k the only LED suitable for biasing Crack's first stage ?

[Paul Birkeland]

It's the best one we have found (it maintains its characteristic voltage at many currents, and it's quiet).

There was a builder some time ago who converted to a 5687 driver tube, and that required a different LED (for a different bias voltage).