Cathode Bias Resistor and Capacitor Values

[Aug]

Hi, I am reworking a Cathode Biased KT88 Amplifier. I have only discovered this mode of operation since picking this amplifier up. It uses 2 x KT88s Per Channel. Presently it uses a single 300 ohm resistor,bypassed with a 100uF capacitor. I have been reading various articles about configuration and it appears the general consensus is separate resistor/capacitors for each cathode is the way to go. I would think to keep the status quo, I would use a 600 ohm resistor for each cathode? But not sure about capacitor. Been advised elsewhere to increase value. Not 100% with this. Any advice would be very much appreciated. Some basic measurements I have taken. Anode to Cathode 420, voltage drop across the 300 ohm resistor 42.

Any advice greatly appreciated.

 

[Merlinb]

Estimating the gm to be about 10mA/V the internal cathode resistance will be in the region of 100 ohms. Plug this into the standard filter equation and use a pleasingly low frequency, say 5Hz:

C = 1 / (2 PI f R) = 1 / (2 * PI * 5 * 100) = 318uF

So yes, you want more than 100uF. Anything from about 220uF to 1000uF might be worth trying. Be generous with the voltage rating (e.g. use 100V caps), as the cathode voltage may climb high when the amp is first switched on and the coupling caps are still charging.

 

[DF96]

You can get away with a smaller capacitor across a common cathode resistor because the signal currents tend to cancel - in fact some designs omit the capacitor. With separate resistors/capacitors this cancellation no longer occurs so you need a big cap.

The value of the cap may have a bearing on LF stability if there is significant negative feedback.

 

[Aug]

OK, so definitely higher, but caution with too much to avoid lf instability? Is this capacitor performing a sort of smoothing effect between cathode and ground as well as decoupling?

 

[Merlinb]

Is this capacitor performing a sort of smoothing effect between cathode and ground as well as decoupling?

That's kinda what decoupling is...

 

[Aug]

OK, wasn't 100% whether it was a feedback type thing going on, similar to electrolytic in feedback path of solid state amplifiers.

 

[Merlinb]

If your amp has global feedback then the cathode bypass caps will be inside the loop and hence will affect stability (a bit). It will probably be OK if you use nice big caps, but admittedly it is very difficult to predict the effects -you're best off using a cut-and-try approach.

 

[Aug]

OK Merlinb, many thanks for your advice as well as others. I'll try a couple of values and see what results I get. Just as an aside, cathode bias seems to be one of the less used o/p stage configurations, what's your view on it?

 

[DF96]

OK, wasn't 100% whether it was a feedback type thing going on, similar to electrolytic in feedback path of solid state amplifiers.

The cathode decoupling cap prevents local feedback.

Just as an aside, cathode bias seems to be one of the less used o/p stage configurations, what's your view on it?

Cathode resistor bias is by far the most common method used in output stages and elsewhere. It is simple, cheap and reliable. It automatically adjusts for valve differences and ageing.

 

[Aug]

I must of lived a pretty sheltered tube life. Most of the kit I've come across have been UL or Triple with Cathode tied to ground or via 10 ohm resistor.

 

[Merlinb]

cathode bias seems to be one of the less used o/p stage configurations, what's your view on it?

You don't see it used much with big bottles like KT88s. Traditionally, if you're going to the expense of using big bottles, presumably it's because you want mucho power, so presumably you don't want to waste precious voltage swing by cathode biasing.

I suppose an argument could be made for doing things the other way around:
Use cathode bias with big bottles because they'll still give you 'more power than you need', cathode biasing provides better protection for those big expensive valves / transformers, and the grid leak resistance can be maximised so it doesn't tax the driver stage. Use fixed bias for small bottles because they don't need much negative voltage (e.g. easier to generate), the driver can cope with heavier grid-leak loading since it doesn't have to swing as many volts, and the output stage is more likely to be driven to clipping so polite overload recovery is important. Six of one, half a dozen of the other.

 

[DF96]

I suspect that fixed biasing for domestic audio is partly a marketing tool. It gives 'audiophiles' an extra adjustment to make, and gives them a sense of having something which lesser amplifiers lack.

 

[Aug]

OK, you guys have given me a lot of insight into bits I was struggling to understand. At the risk of completely disproving this, could I suggest in this mode of operation, the split cathode resistors for each tube acts almost like emitter resistors in a solid state amp. I'll let you know how I get on and post some pics once I've finished. Thanks again.

 

[Merlinb]

could I suggest in this mode of operation, the split cathode resistors for each tube acts almost like emitter resistors in a solid state amp.

That's exactly how they act.

 

[jan.didden]

You can get away with a smaller capacitor across a common cathode resistor because the signal currents tend to cancel -

Is that the case here? The tubes are // so wouldn't the cathode signal currents add?

Edit: no you're right - it's two tubes per channel, not per phase. I should read better ;-)

Jan

 

[DF96]

I thought we were talking about PP. I don't think the OP has said so explicitly, but I don't think he has mentioned SE either so 2xKT88 is more likely to be PP than PSE?

 

[20to20]

Estimating the gm to be about 10mA/V the internal cathode resistance will be in the region of 100 ohms. Plug this into the standard filter equation and use a pleasingly low frequency, say 5Hz:

C = 1 / (2 PI f R) = 1 / (2 * PI * 5 * 100) = 318uF

So yes, you want more than 100uF. Anything from about 220uF to 1000uF might be worth trying. Be generous with the voltage rating (e.g. use 100V caps), as the cathode voltage may climb high when the amp is first switched on and the coupling caps are still charging.

Where do you find an "internal" cathaode resistance spec., and doesn't a 600R cathode resistor fit into the equation somehow?

 

[DF96]

For a pentode the impedance seen looking into the cathode is 1/gm. 10mA/V is thus 100 ohms. Strictly speaking the cathode resistor should then be added in parallel, but as 600 is much bigger than 100 and electrolytics have big tolerances it doesn't really matter.

 

[Aug]

Hi, it's PP. I was trying to work out how much of the o/p is class a, if it's presumably a/b. Measuring the volt drop across the Cathode Resistor while driving the Amp to clipping reveals no change in consumed current. Does this indicate it's Class A? If not how can I test it's class a credentials?

 

[DF96]

If the voltage across the cathode resistor does not change with signal then this indicates Class A. A small change with signal may still indicate Class A, as second order distortion will do this. Class B would give a large change with signal, which is why Class B amps need to use fixed bias. The division into Class A, B etc. is never quite as clear in the real world as it appears in textbooks.