Pentode Control Gain by Screen Grid

opacheco said:

PRR said:

> Another idea - use G3 instead of G2!

Same main problem: the plan Opacheco has shown us is a feedback amplifier. Gain is controlled by the feedback resistor ratio, not the tubes, or not until the tube is de-biased so far you might as well remove V4 and go right into V5.

I made a mistake. With the 300 ohm pots turned to maximum gain, the 200K resistor does nothing, but the 400K||1.25Meg resistors are still applying NFB. Minimum gain is still 800, 58dB, but maximum gain is 2,000 or 66dB (plus typically another 20dB in the input transformer).

Changing 200K to 20K reduces min gain to 38dB (a useful value in modern studios), but max gain is still 66dB, a long span from 38dB. With the simple linear pot it may be hard to set low gains like 40dB and 45dB. If 66dB (+20dB) is more than you could need, also change the 400K to maybe 100K. That gives min=38dB, max=54dB (58dB-74dB with input transformer).

Getting the 35dB-55dB gains you need around loud instruments, without trouble from microphonics, really requires a re-design. The tubes could be re-wired as Triode, two triodes is plenty of gain and less working parts to shake and make bad sound. The input cathode resistor could be larger, perhaps 1K with no cathode capacitor. Then the 400K could be replaced with 500K audio-taper pot.

By the way, there is no "DC feedback". Although there is DC at some points on the NFB network, the amplifier has no DC gain (5.6nFd cap blocks DC).

That's a lot of output capacitors. The 8uFd must be high voltage, so it could be expensive to use a bigger cap. Instead the NFB loop corrects any bass-loss through the 8uFd. The NFB network would leak DC from V4 cathode to output. The 24uFd blocks that, and because it is only 1V here the 24uFd can be smaller/cheaper than a high-voltage cap. But then why the 10uFd cap? Perhaps to compensate bass loss in the 24uFd cap. But only if the load is low impedance. And the cathode follower will not drive a very low load impedance. While it seems expertly designed, I don't understand some of these choices.

Click to expand...

Firts,
Thanks a lot for your analysis!!!....This unit Don't have any input transformer at all!!...Why do you thinks so?.
This unit is designed for a linear response from 10Hz to 2000Khz +/-0.5Bdb!!..Amaizing, do you?
and was used with 50Ohms load.

"The 24uFd blocks that, and because it is only 1V here the 24uFd can be smaller/cheaper than a high-voltage cap. But then why the 10uFd cap? "

I believe, the 10uf is for DC blocking from the 300Ohms resistors conected to the V4 Catode and the V4 Catode and V5 Triode Catode are DC conected by the 1.25MegaOmhs(I very little DC current).

By the way how do you make this kind of calculations gains???....do you use a specific network analysis like Feedback Amplifiers Method??

Thanks a lot again.
Opacheco.

Click to expand...

bwithwings said:

Personally I have always used a piece of paper and draw load lines to figure out bias, voltage gain etc - it is probably more long-winded than using software, but who wants to be shackled to a computer when you're doing electronics anyway 

http://www.aikenamps.com/LongTailPairDesign.htm  << this site has some very good stuff under 'Tech Info'
http://www.freewebs.com/valvewizard/pentode.html

HTH
Darren

Click to expand...

PRR said:

> Another idea - use G3 instead of G2!

Same main problem: the plan Opacheco has shown us is a feedback amplifier. Gain is controlled by the feedback resistor ratio, not the tubes, or not until the tube is de-biased so far you might as well remove V4 and go right into V5.

I made a mistake. With the 300 ohm pots turned to maximum gain, the 200K resistor does nothing, but the 400K||1.25Meg resistors are still applying NFB. Minimum gain is still 800, 58dB, but maximum gain is 2,000 or 66dB (plus typically another 20dB in the input transformer).

Changing 200K to 20K reduces min gain to 38dB (a useful value in modern studios), but max gain is still 66dB, a long span from 38dB. With the simple linear pot it may be hard to set low gains like 40dB and 45dB. If 66dB (+20dB) is more than you could need, also change the 400K to maybe 100K. That gives min=38dB, max=54dB (58dB-74dB with input transformer).

Getting the 35dB-55dB gains you need around loud instruments, without trouble from microphonics, really requires a re-design. The tubes could be re-wired as Triode, two triodes is plenty of gain and less working parts to shake and make bad sound. The input cathode resistor could be larger, perhaps 1K with no cathode capacitor. Then the 400K could be replaced with 500K audio-taper pot.

By the way, there is no "DC feedback". Although there is DC at some points on the NFB network, the amplifier has no DC gain (5.6nFd cap blocks DC).

That's a lot of output capacitors. The 8uFd must be high voltage, so it could be expensive to use a bigger cap. Instead the NFB loop corrects any bass-loss through the 8uFd. The NFB network would leak DC from V4 cathode to output. The 24uFd blocks that, and because it is only 1V here the 24uFd can be smaller/cheaper than a high-voltage cap. But then why the 10uFd cap? Perhaps to compensate bass loss in the 24uFd cap. But only if the load is low impedance. And the cathode follower will not drive a very low load impedance. While it seems expertly designed, I don't understand some of these choices.

Click to expand...