I'm playing with an old tube PA amp, which has 12SJ7 preamp stages. Found an article (How Screen Voltage Affects Gain) that explained the importance of screen grid voltage and gain. So I thought I'd play with mine and see what it did.... Very little! The amp works well as is, just wanted to increase the gain by lowering the screen voltage. Changing the screen resistor (with a pot) while measuring the screen voltage. Well, increasing the resistance lowered the screen voltage, but reduced the signal out of the following stage (I was playing with the first 12SJ7 stage)... not what I expected. Raising the resistance from 3M to 4M reduced the voltage from 11.75V (as found) to 10.5V, and reduced the 2nd stage signal by 2/3. So I lowered the resistance, raising the screen voltage to 13V (with a pot), but got no increase. Was expecting to see some clipping (2nd stage being over-driven). What am I missing? I am more familiar with cathode-biased triodes, and this is a more complicated animal. Any help? Thanks!
P.S. The plate gets only 55VDC, but the P-P 6L6s don't seem to care. I'ts plenty loud.
P.S. The plate gets only 55VDC, but the P-P 6L6s don't seem to care. I'ts plenty loud.
Before I say what I want to, I want to first state that I'm only a beginner myself, and may very well be totally wrong in what I say.
Now, I do believe that there is a sort of sweet spot. That is, anything below, and above that voltage will only decrease the sound. This also seems to be backed up by what you've already described.
I think you're right. I did find the 3M resistor in the circuit measured 3.7M (the amp is from the 40's), so I thought I had room for improvement....I want to push this amp some (yes, I play guitar), so I will look at other ways to do it. Maybe a higher mu tube, or higher plate voltage... Thanks!
in a pentode, there are two ways to control plate current, one is by grid 1 bias voltage and another by screen grid voltage
pentode plate current and screen current ratio is about 5, but look at the datasheet for typical conditions...
voltage gain is almost gm x Rl, since the plate resistance of a pentode is very high,
Rl is // plate load resistor and next stage grid leak resistor...
gm is dependent on cathode current, and is directly proportional...
the screen grid needs to be bypassed wrt to cathode to keep voltage variations with signal zero.
pentode plate current and screen current ratio is about 5, but look at the datasheet for typical conditions...
voltage gain is almost gm x Rl, since the plate resistance of a pentode is very high,
Rl is // plate load resistor and next stage grid leak resistor...
gm is dependent on cathode current, and is directly proportional...
the screen grid needs to be bypassed wrt to cathode to keep voltage variations with signal zero.
The above link gives false information:
” A lower screen voltage enables a substantially lower grid voltage swing to create the same plate voltage swing. This is the essence of our definition of "gain." A lower screen voltage means higher voltage gain.”
The link does not mention two important factors; gm and Ra.
The gain (A) of a pentode is: A = gm x Ra
The gm of a typical sharp cut-off pentode decreases (a little) when Ug2 (screen voltage) is lowered.
If the anode resistor is kept constant and Ug2 lowered, the gain is decreased, not increased
as mentioned by the above link.
” A lower screen voltage enables a substantially lower grid voltage swing to create the same plate voltage swing. This is the essence of our definition of "gain." A lower screen voltage means higher voltage gain.”
The link does not mention two important factors; gm and Ra.
The gain (A) of a pentode is: A = gm x Ra
The gm of a typical sharp cut-off pentode decreases (a little) when Ug2 (screen voltage) is lowered.
If the anode resistor is kept constant and Ug2 lowered, the gain is decreased, not increased
as mentioned by the above link.
Unfortunately it is a bit more complicated than the article implies, as alluded to by others here.
As you decrease the screen voltage you decrease anode current. This not only effects gain, but it also shifts the bias point which in turn determines distortion.
To compensate you must also change the anode (Load) resistor.
As you decrease the screen voltage you decrease anode current. This not only effects gain, but it also shifts the bias point which in turn determines distortion.
To compensate you must also change the anode (Load) resistor.
The one good thing the article mentions is the use of pentode plate transfer plots for design. I find these more useful than the plate characteristics that are presented with V2k set to one specific (rarely two) value.
"The issue boils down to our basic definition of gain: output voltage swing divided by input voltage swing",
So far, so good. Should have stopped while they were ahead.
He even includes a sample plate xfer plot, and completely misses the point. You may never see those extremes for your particular plate load. The (0,Ia) intercepts drop with screen voltage.
"A lower screen voltage means higher voltage gain".
Nonsense: the gain equation is identical regardless of the device in question:
Av= g_m X (r_p || Rp || Rl)
Always: triode, pentode, BJT, JFET, MOSFET, IGBT, whatever. It's always the same, though the actual form of the gain equations may be different for different two port models, they can always be transformed.
The right reason for wanting to get that screen voltage as low as possible is to, ideally, keep Vpk > V2k. With some small signal pents, this is easier done than with others. The 12BY7A doesn't work right with a lowish Vpp (it was originally intended as a high gain video amp for colour TVs, and there is no lack of high voltage in a colour TV).
The reason it seems the way the article says is that decreasing the Ip decreases the g_m. The lower the Ip, the larger Rp can become. The one offsets the other, though there does seem to be a case of diminishing rewards once Rp= 100K. There have been designs that run small signal pents with Ip as low as 200uA and still produce more gain than you could get from a 12AX7 with an active plate load.
"Our thanks to Paul Reid of Rutgers University for his technical assistance with this page".
How could Paul miss this? Really?
"The issue boils down to our basic definition of gain: output voltage swing divided by input voltage swing",
So far, so good. Should have stopped while they were ahead.
He even includes a sample plate xfer plot, and completely misses the point. You may never see those extremes for your particular plate load. The (0,Ia) intercepts drop with screen voltage.
"A lower screen voltage means higher voltage gain".
Nonsense: the gain equation is identical regardless of the device in question:
Av= g_m X (r_p || Rp || Rl)
Always: triode, pentode, BJT, JFET, MOSFET, IGBT, whatever. It's always the same, though the actual form of the gain equations may be different for different two port models, they can always be transformed.
The right reason for wanting to get that screen voltage as low as possible is to, ideally, keep Vpk > V2k. With some small signal pents, this is easier done than with others. The 12BY7A doesn't work right with a lowish Vpp (it was originally intended as a high gain video amp for colour TVs, and there is no lack of high voltage in a colour TV).
The reason it seems the way the article says is that decreasing the Ip decreases the g_m. The lower the Ip, the larger Rp can become. The one offsets the other, though there does seem to be a case of diminishing rewards once Rp= 100K. There have been designs that run small signal pents with Ip as low as 200uA and still produce more gain than you could get from a 12AX7 with an active plate load.
"Our thanks to Paul Reid of Rutgers University for his technical assistance with this page".
How could Paul miss this? Really?
How to create confusion (Lesson One):
1. Put a graph on a web page which clearly (and correctly) shows that higher screen voltage raises gain.
2. State in the text that higher screen voltage reduces gain, because 'gain' relates to possible grid voltage swing vs. full (impossible) anode voltage swing.
3. Claim that a university lecturer has given assistance with the page, but don't specify whether this help was with content or HTML coding or obtaining disc space to store the page or checking spelling or correcting grammar or . . .
1. Put a graph on a web page which clearly (and correctly) shows that higher screen voltage raises gain.
2. State in the text that higher screen voltage reduces gain, because 'gain' relates to possible grid voltage swing vs. full (impossible) anode voltage swing.
3. Claim that a university lecturer has given assistance with the page, but don't specify whether this help was with content or HTML coding or obtaining disc space to store the page or checking spelling or correcting grammar or . . .
Many amps from that era used grid leak bias. There is a large value resistor from G1 to ground, a cap in series with the signal, and the cathode is grounded. If this is the case just about any tinkering with component values will upset the delicate bias throwing off the entire circuit. So will measuring the screen or control grid with a volt meter, since the meter is at best 11 meg ohms to ground internally.
I usually convert a grid leak bias stage to cathode bias before attempting to tweak on it since grid leak bias is inherently tempermental.
As stated the screen voltage, control grid bias voltage (or cathode bias resistor), plate load resistance and load impedance (input of the next stage) ALL interact to determine the tube's gain.
For an educational exercise replace the cathode resistor, screen resistor, and plate load resistor all with pots of 2X or 3X their value. Then be prepared to spend about a day turning knobs. There will be several places that give similar gains, but different overload characteristics. Test with bare guitar, then with pedals. Try with 2 or 3 different 12SJ7's since they are each a bit different. Then for some more fun, plug in a 12SK7 and repeat. The 12SK7 has a remote cutoff characteristic, and you can find an edge where the sound is clean, and transitions to nasty with a little bump of the guitar's volume knob.
You may find a set of values that are quite clean with the 12SJ7 and transitional with the 12SK7. I used to do this with the old Champs that used a 6SJ7, but it's been nearly 50 years ago.
Try cascading the #2 and #3 pentodes in a two or three mic input type amp.
play with the screen, cathode and plate values of #2 till you reduce the signal that hits the grid on #3. It has very pleasing distortion with the volume on the #2 acting as a gain control for the overall signal. The gain and the vol of #3 must be fiddled with to find the sweet spot but worth the effort. I have two other amps, a JTM clone and a VSss-65 and prefer this amp. It has a super creamy/sparkly 6l6 clean and the overdriven sound from right around 6 and 3 on the pots is Dumbully, well to my ears anyway.
play with the screen, cathode and plate values of #2 till you reduce the signal that hits the grid on #3. It has very pleasing distortion with the volume on the #2 acting as a gain control for the overall signal. The gain and the vol of #3 must be fiddled with to find the sweet spot but worth the effort. I have two other amps, a JTM clone and a VSss-65 and prefer this amp. It has a super creamy/sparkly 6l6 clean and the overdriven sound from right around 6 and 3 on the pots is Dumbully, well to my ears anyway.
Just an update on the S/C 33...I was having a little nirvana when I wrote the above. The Strat was making the goose bumps jump. I was a little premature though, when I hooked up the EPi LP humbuckers the bass rattle nemesis reared up. So I took time this week to check my previous work and went back to a little lower screen resistor on the #1 tube in the cascade, to reduce the overall gain. I read from some more knowledgable than I, that on these tubes that this would increase gain, ?that was not my experience? The gain was reduced at least enough to get rid of the bass freq's tendency for out of control saturation.
The bottom line is that a cascaded pentode pre-amp tube can be very musical, and if a fuzz needed ,just crank the gain pot some. It does impress.
I have probably spent way too much time on this amp but I am happy with it for what it's worth.
The bottom line is that a cascaded pentode pre-amp tube can be very musical, and if a fuzz needed ,just crank the gain pot some. It does impress.
I have probably spent way too much time on this amp but I am happy with it for what it's worth.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.