#27 mu stage

[kevinkr]

Originally Posted by pieter t

Taking another close look it seems to me that the 33k resistor is not correct; it should have a value to provide the D3A with the right bias. Some 360 ohm should be good.

I just noticed that too, I suspect that the resistor should be much smaller than 360 ohms, say 82 -100 ohms. I'm running these at Vp of 140V and 12 - 14mA occurs with about 1.2V of bias.. And they are all over the place..

[pieter t]

Originally Posted by kevinkr

I just noticed that too, I suspect that the resistor should be much smaller than 360 ohms, say 82 -100 ohms.

No, because the current is split between the bottom tube and the 39k resistor.
There is some 4-5 mA of current there.
In the past I preferred to use double triodes for the bottom, one triode for input / voltage amplification, the other one to sink the surplus current of the top tube - more elegant than that 39k resistor.

[kevinkr]

Originally Posted by pieter t

No, because the current is split between the bottom tube and the 39k resistor.
There is some 4-5 mA of current there.
In the past I preferred to use double triodes for the bottom, one triode for input / voltage amplification, the other one to sink the surplus current of the top tube - more elegant than that 39k resistor.

I see your point.. The 27 should be running at around 6 - 7mA and the D3A at something like twice that.. I didn't work the numbers with the 39K in mind.. Total currents of 20mA are not a problem as long as dissipation limits are not exceeded.

[Jaap]

see also here: http://www.dhtrob.com/schemas/afbeel...kerMustage.gif

[zivhackman]

Thank you very much!
pieter, thank you for noticing that 33k res!
it is a drowing mestake that could lead me to alot of frustration!
in my calculation i got 333.333333 ohm resistor, so i translated it very wrong to a real world resistors...

You don't say what the intended application is.

it sould be a line amp

by the way,
I thought a lot about replacing the 39K resistor to a ccs.

is it a good idea?

[SemperFi]

Originally Posted by ruffrecords

If you want minimum distortion you should bypass the cathode resistor of the lower tube.

Cheers

Ian

? Isn't this local feedback, and bypassing it will increase distortion? Unless you need the gain, I thought having some unbypassed cathode resistor is better than not.

[ruffrecords]

Originally Posted by SemperFi

? Isn't this local feedback, and bypassing it will increase distortion? Unless you need the gain, I thought having some unbypassed cathode resistor is better than not.

Yes and no. It is local feedback but because it is series derived it means it increases the effective output impedance (ra) of the tube by approximately mu times the cathode resistance. In a typical 6SN7 running at about 5mA, ra is about 10K and the 1K cathode resistor to bias it effectively adds another 20K to ra making it 30K. Since the principal source of distortion in triode is non-linearities in ra, the mu follower seeks to minimise these by making the plate load as light is possible (i.e 10 or more time ra). Making ra 3 times bigger by not bypassing the cathode of the bottom triode simply means the effective load has to be three times greater to keep the distortion as low as it would be with a bypassed cathode.

As a rule, not bypassing a cathode resistor is not as effective at reducing distortion as it would appear at first sight.

Cheers

ian

[kevinkr]

Originally Posted by ruffrecords

<snip>

As a rule, not bypassing a cathode resistor is not as effective at reducing distortion as it would appear at first sight.

Cheers

ian

My experience generally correlates well with this comment.

[zivhackman]

Originally Posted by SemperFi

? Isn't this local feedback, and bypassing it will increase distortion? Unless you need the gain, I thought having some unbypassed cathode resistor is better than not.

I thought that to..
but there another reasons not to bypass Rk,
i just don't want an electrolythic in the signal pass..
after reading what allen wright wrote in his book,
I pulld out from my preamp and power amp all the Rk bypass caps and the sound was much better!
puting there a film cap will be costy and i will need some extra space..

Thank you all for the great replies!!

[SemperFi]

Originally Posted by ruffrecords

Yes and no. It is local feedback but because it is series derived it means it increases the effective output impedance (ra) of the tube by approximately mu times the cathode resistance. In a typical 6SN7 running at about 5mA, ra is about 10K and the 1K cathode resistor to bias it effectively adds another 20K to ra making it 30K. Since the principal source of distortion in triode is non-linearities in ra, the mu follower seeks to minimise these by making the plate load as light is possible (i.e 10 or more time ra). Making ra 3 times bigger by not bypassing the cathode of the bottom triode simply means the effective load has to be three times greater to keep the distortion as low as it would be with a bypassed cathode.

As a rule, not bypassing a cathode resistor is not as effective at reducing distortion as it would appear at first sight.

Cheers

ian

Interesting. I have always relied on local feedback and never used global, and been happily ignorant
I need to break out the breadboards again and do some testing, it doesnt quite make sense to me. I can see your point if the load is very easy, perhaps 1megohm and more. But in real couplings it seldom is that easy. Perhaps buffering with a CF, but then we add another tube.
Looking into the plate circuit from the output end, as the load will see it, the negative and positive going phases will be more assymetric if the cathode is bypassed. The tube being the low impedance source and the resistor/choke/ccs going to b+ being a high impedance source. Doing a quick Spice and using an exaggerated hard load (10kohms), the bypassed stage has much more distortion than the unbypassed. Even if adjusting the inputs for equal output or not.
I am not about to tell experienced folks like Kevin and Ian that you're wrong, just that I need to have a moment of brainscanning while having some local feedback myself (a cold brew) about this.