Matchless Cathode Bias El34 resistor choice

[johndandry]

So here is another question on my Cathode Bias El34 project.
I've been researching different EL34 Cathode biased amps and noticed that
the 1k screen resistors in the Matchless stuff are rated for 12watts. Are they that big because cathode bias puts more screen current than fixed or is it just over sized? Would a 5watts be ok or should I go for at least 10watt.
Also, grid resistors into the EL34s, both the Selmer Zodiac 30/Bad Cat-Hot Cat/ and the Chieftain have large grid resistors, 10k/22k. The Cubman has 1.5k ala standard El34 grids (although Hiwatts have 22k as well). Thoughts?

[Reeltarded]

Tasty food for thought: http://www.aikenamps.com/InputRes.htm

I dunno why big screens unless they are dropping a bunch for a reason?

[gingertube]

Cathode Bias EL34 at B+= of 375V

Ig2 idle = 12.5mA each tube
Ig2 at max power = 19.5mA each tube.

For 1K screen resistors (Power = I squared R) = 380mW so I would be using a 2W metal film.

I use a COMMON 470 Ohm screen resistor (Mullard HiFi trick) rated at 5 Watts.

Cathode Bias Resistor(s)
Ik at idle = 75mA each tube
Ik at max power = 95mA each tube so the cathode bias resistors (270 Ohm each tube) need to dissipate 2.436 Watts - use 5W wire wound as a minimum.
I use separate 270 Ohms resistors (one each tube) rather than a COMMON 130 Ohm cathode bias resistor. The 270 Ohms I use are 10 Watt but that is only because I did'nt have 5W jobbies in my parts bin.

If your B+ is much higher than that 375V then power dissipation will go up a bit but the above are "ballpark" values.

Grid Stops Resistors
EL34/6CA7 in pentode mode the Miller capacitance at grid1 (to 0V) will be probably around 50pF but let us assume it is 100pF for calcs.

The -3dB HF roll off point introduced by the grid stop resistor will be:
1K5 grid stop and 100pF => approx 1MHz , I don't think 1K5 is big enough to guarantee no parasitic oscillation particularly for a high gm tube like the EL34.

So how big can you go.
R x C = the "timeconstant"
Take 1 over the timeconstant then divide by 2 pi and that gives you the -3dB HF roll off frequency.

10K and 100pF => 159kHz , thats OK
47K and 100pF => 33.8kHz, thats OK too

NOTE: Once you start using grid stops of greater than say 47K then you need to worry about stability if you have global feedback around the power amp. The grid stop/Miller capacitance may well become your dominant pole and if so any other stability components which were put in to introduce a dominant pole somewhere else should be removed.

We almost certainly over estimated the Miller Capacitance and Guitar Amp Output trannies are usually 3dB down by around 15kHz so I would be comfortable with grid stops up to about that 47K value for amps with global feedback.

For a zero feedback amp take the grid stops pretty much as high as you like, certainly 100K would still be OK.

I use 15K on my home brew cathode biased 6CA7 amp which does have global feedback. Having just re-read what I said above I might actually try higher values for the grid stop to make sure this is the dominant pole.

WHY? Because high value grid stops bring another big benefit . They severely limit grid current in overdrive which will eliminate or at least greatly reduce blocking distortion.

Cheers,
Ian

[Reeltarded]

Yeah, that is exactly what I was going to say next. haha

Thanks Ian, you always teach a good lesson!

[martin manning]

Thanks Ian, good post! A commercial example using large grid stops (100k) is the Marshall 4001 Studio 15, with 2x 6V6 output section. It does have GNFB to the PI in the usual way but there is no presence control, so feedback of higher frequencies is always reduced. Note also it has a combination of fixed and cathode bias.

[Firestorm]

Thanks Ian, good post! A commercial example using large grid stops (100k) is the Marshall 4001 Studio 15, with 2x 6V6 output section. It does have GNFB to the PI in the usual way but there is no presence control, so feedback of higher frequencies is always reduced. Note also it has a combination of fixed and cathode bias.

That combined fixed/cathode bias is usually associated with the least-loved Fender models, but there's nothing really wrong with it and it can provide real benefits if done well.

[gingertube]

Thanks Guys,
Just be a bit careful with my post - the actual -3dB HF roll off will be due to the Miller Capacitance and the Grid Stop Resitor PLUS the output impedance of the Phase Splitter (Assume about 50K typically). If (in an amp with feedback) you make this the dominant pole by using a large grid stop then you will need to modify it if you ever want to use Ultralinear or Triode Mode in which case the Miller Capacitance will be much higher.
Cheers,
Ian