12AX7/12AU7 Preamp Supply Voltage Lowered

[tubeswell]

On the graph, my 12ax7 bias point is between -1v & -1.5v. The X-axis is Va=125v and the Y-axis is Ma=.55

Where does the point need to be to be considered correctly biased?

There is no 'correct' bias point. See how each grid voltage curve crosses the load line? Well, biasing in the 'middle' of where those grid curves cross the load line will be 'centre based' and the grid will take the least-clipped largest signal in that position. If the bias point shifts to the right (towards the more negative grid voltages), then a larger input signal is likely to drive the tube into cut-off because the grid can only swing -ve so far before the current stops flowing in the tube and the plate is at maximum (HT) voltage. In this scenario, it is the plate signal that is cut-off; the signal at the grid may still be swinging in a smooth sine wave.

On the other hand, if the bias point is to too far to the left along the load line (i.e.: towards the Vg=0 grid curve), then the signal on the grid will not be able to swing to a more positive voltage than 0 before grid current limiting clips the grid signal.

On the 'third' hand, if the bias point is roughly in the 'middle' of the grid curves (on the load line), then even a big enough signal at the grid can cause both plate cut-off clipping and grid current limiting clipping. i.e.: the signal could be clipped on both top and bottom.

-1.5V to -2V is a typical 'centre bias' voltage for a 12AX7 is most common scenarios. A 12AU7 mid point is more like -7 to -8V etc

In the example recorded in your spreadsheet, a 12AX7 100k load line would be about centre biased at -1V. You might find you have better gain with 150k Ra load line in that case. On the other hand a 12AU7 is looking a bit sick and might do better with a 22k* Ra load line, which should get the bias to about -5V. JM2CW

* in which case, use a 2W 22k resistor.

[Intergalactic Tourister]

In the example recorded in your spreadsheet, a 12AX7 100k load line would be about centre biased at -1V. You might find you have better gain with 150k Ra load line in that case.

Wow, completely not what I expected. Did I measure the B+ correctly at 182V. I thought I'd have lower voltage.

On the other hand a 12AU7 is looking a bit sick and might do better with a 22k* Ra load line, which should get the bias to about -5V. JM2CW

* in which case, use a 2W 22k resistor.

Yes it sounded a bit sick too. There wasn't much amplification going on with the 12au7 I noticed. That may be part of the reason it sounds like it does and why I actually like it. Or maybe the bias point (which was my first suspicion). Interesting.

On a side note I've experimented with coupling caps ranging from .0018uf to .5uf because it was too bassy. The bass knob set to 2 or 3 was already too much bass. The .0018uf to .02 seemed like they just lowered the volume and not cut out the bass.

*I'll increase to 2W 22k
*Experiment with the bias above and below its current bias point

Thanks for the great discussion and help

[Intergalactic Tourister]

On the other hand a 12AU7 is looking a bit sick and might do better with a 22k* Ra load line, which should get the bias to about -5V. JM2CW

* in which case, use a 2W 22k resistor.

The 12au7 already has a 22k Ra. Just caught that re-reading your post.

I guess you meant lower Ra since the graph put the 12au7 at about -7v with Ra at 4k.

[tubeswell]

Wow, completely not what I expected. Did I measure the B+ correctly at 182V. I thought I'd have lower voltage.

Your load line is set up using 182VDC as your HT voltage. This means that if the tube is not conducting (i.e. in cut-off with no current flowing), then the plate will be at HT voltage (i.e. 182V) with zero current. This is point at which the load line touches the X axis (182V, 0A).

Now if the tube was conducting fully and all the voltage was being dropped across the plate resistor, then the voltage at the plate would be zero, and with a 100k resistor, you would have 182V/100,000R = .00182A (or 1.8mA) current flowing through the tube. This is the point where the load line touches the Y axis (0V, 1.82mA).

Every point on the load line between the point on the Y axis and the point on the X axis, is some trade-off between plate current and plate voltage. The load line touches the 0V grid curve at the point where the plate voltage is about 55V and the plate current is 1.25mA. So 55V is effectively as low as the plate voltage can swing because the grid voltage can't go any higher than 0. So a 'centre bias' point is going to give you a plate idle voltage about midway between 55V and 182V (i.e. 118.5V on the x axis). This corresponds to about where the -1V grid curve crosses the load line (if you take a vertical line straight up from the x axis at 118.5V to intersect with the load line). The plate current here would be about 0.625mA (extending a horizontal line across to the Y axis from where the -1V grid curve hits the load line). The value of cathode resistor needed to get -1V bias point when you have 0.625mA flowing through it is -1V/0.000625A = 1,600R (or 1k6), and if you enter 1600 into the parameters, you will see the green cathode line crosses the load line at the -1V grid curve.

With the 12AX7 now biased like that, the grid can swing 1V positive (to the 0V grid curve) and the plate voltage will be 55V given a gain of 118.5V-55V = 63.5V. Now if the grid swings down by 1V (to -2V), the plate voltage goes up to about 162.5V, which is a difference of 163.5-118.5 = 45V!. So you can see that the gain at the plate is assymetric/non-linear compared to the signal at the grid. (Incidentally - the total peak signal at the plate for a 2VPP grid signal would be: 63.5+45 = 108.5VPP). But we don't measure the gain from a 2VPP grid signal - we typically measure it from a 1VPP grid signal. If you look at the parameters sheet, Merlin has provided a chart where you can read off the gain. In this case our unbypassed 12AX7/ECC83 stage has a gain of 30.6. If we chose to put a 22uF bypass cap in parallel with the 1k6 cathode resistor, we would get a gain of 61.3.

Now with a 2k2 cathode resistor, the unbypassed gain is 25.8 and the bypassed gain is 61.3. But if we then went to a 150k plate load, the unbypassed stage's gain is 34.2, and bypassing the 2k2 with 22uF gives a gain of 70.3! So we see how using 150k plate resistor with a 2k2 cathode resistor gives a decent amount of clean gain when your HT voltage is 182.

Yes it sounded a bit sick too. There wasn't much amplification going on with the 12au7 I noticed. That may be part of the reason it sounds like it does and why I actually like it. Or maybe the bias point (which was my first suspicion). Interesting.

On a side note I've experimented with coupling caps ranging from .0018uf to .5uf because it was too bassy. The bass knob set to 2 or 3 was already too much bass. The .0018uf to .02 seemed like they just lowered the volume and not cut out the bass.

*I'll increase to 2W 22k
*Experiment with the bias above and below its current bias point

Thanks for the great discussion and help

You mean decrease it to 22k. (but use a 2W resistor)

[Intergalactic Tourister]

You mean decrease it to 22k. (but use a 2W resistor)

I think we got mixed up here.

12au7
plate resistors: 100k/pin 6 and 22k/pin 1. (original layout)

I thought I would increase the 22k plate resistor to make the 12au7 sound less sickly.

But your explanation is really great. I pretty much know where I am now with my preamp.
I'll post my results and thoughts (just for reference) for anyone else following this thread.

Thank you tubeswell.

[tubeswell]

I think we got mixed up here.

12au7
plate resistors: 100k/pin 6 and 22k/pin 1. (original layout)

I thought I would increase the 22k plate resistor to make the 12au7 sound less sickly.

Both Pin 1 and Pin 6 are plate pins in a 12A_7 dual triode. 100k plate resistor on a 12AU7 gives it a terrible load line on the characteristics chart. A 22k looks a lot better (esp with a 2k2 cathode resistor). Somewhere between 22k and 47k is probably 'optimal' for the 12AU7 plate resistors with the 2k2 Rk and the 182 HT voltage. If you bypass the 2k2 with 10-25uF you will get almost a doubling of voltage gain.

[Intergalactic Tourister]

That makes perfect sense. Thanks.

I misinterpreted one of your statements. I was only using the 22k Ra in the parameters... and I wasn't even thinking about the 100k yet. Lots of info to digest.

I'll be posting some of my findings soon for others to see.

[Intergalactic Tourister]

I was thinking that Rg was the following stage's grid leak resistance (which is typically 470k to 1M, which he has marked as 'Rl').

Is Rl the grid leak resistor if analyzing the input portion of the 12au7 tube?
And if analyzing the output portion consider Rl as simply the Load of the power amplifier? Which, say, could be a typical 100K load which the preamp would see.

When Rl=0 there is no frequency response for the 12au7 output. When Rl=100,000 ohm it displays the frequency response.

[tubeswell]

Is Rl the grid leak resistor if analyzing the input portion of the 12au7 tube?
And if analyzing the output portion consider Rl as simply the Load of the power amplifier? Which, say, could be a typical 100K load which the preamp would see.

When Rl=0 there is no frequency response for the 12au7 output. When Rl=100,000 ohm it displays the frequency response.

Rl is the following stage's grid leak resistor. However it also contributes to the AC load on the preceding stage (which is why you need to factor it in to calculate the gain) and it also forms an RC network in combination with the coupling cap (in an AC-coupled stage) to produce a given frequency roll-off point.

[Intergalactic Tourister]

Rl is the following stage's grid leak resistor. However it also contributes to the AC load on the preceding stage (which is why you need to factor it in to calculate the gain) and it also forms an RC network in combination with the coupling cap (in an AC-coupled stage) to produce a given frequency roll-off point.

I see.

But there is no grid leak R anywhere on the 12au7... only an output coupling cap which would connect to a power amp. Not sure what parameter to put in the Excel sheet. Is there a middle-of-the-road value to use that would typically be on a phase inverter or power tube?

[tubeswell]

I see.

But there is no grid leak R anywhere on the 12au7... only an output coupling cap which would connect to a power amp.

The output stage will have grid leak resistors.

Not sure what parameter to put in the Excel sheet. Is there a middle-of-the-road value to use that would typically be on a phase inverter or power tube?

Depends on the tube and on the type of stage. 470k to 1M is typical for a 12A_7 preamp tube. Recommended grid leak resistances for output tubes are typically somewhat less than that because of the greater potential for electrons to be stripped from the grid (because of the higher power of output tubes). Can vary between 100k to 220k to 470k

[Intergalactic Tourister]

Ok, thank you once again tubeswell.

I decreased the 12au7 plate resistor from 100k to 47k and I couldn't hear any difference really. It shift the bias from -2v to -4v.

I may lower it further (to 20k) closer to the bias spec you suggested anyway. Maybe I'll hear a difference since it will then shift the bias from -4v to -6v. Heck, I should go to about -7v or -8v just to see what happens.

But other than that I like the sound that comes out of it.

[tubeswell]

I decreased the 12au7 plate resistor from 100k to 47k and I couldn't hear any difference really.

Depends what its driving. A 12AU7 will drive a tone stack for more bandwidth, but less gain, than a 12AX7. Similarly, you would notice a difference if the 12AU7 was a cathodyne PI or LTP driving the output tubes (instead of a 12AX7). A 12AU7 would be a whole lot sweeter and 'thicker'/syrupy sounding.

[Intergalactic Tourister]

Depends what its driving. A 12AU7 will drive a tone stack for more bandwidth, but less gain, than a 12AX7.

The 12ax7 is the input and runs through the 1M volume. The 1M volume then runs to the grid of the 12au7. Then, the 12au7 plate runs through the 1M treble and 1M bass... then back into the grid of the 12au7. Then, of course, output from the plate via .022uf.

So the 12au7 only runs through the treble and bass.

Similarly, you would notice a difference if the 12AU7 was a cathodyne PI or LTP driving the output tubes (instead of a 12AX7). A 12AU7 would be a whole lot sweeter and 'thicker'/syrupy sounding.

You're right on... its thick and syrupy.

But I see your overall point. I'll lower the plate R again though... just to hear.

[Intergalactic Tourister]

I lowered the plate R to about 17k.

12au7
plate resistors: 17k/pin 6 and 22k/pin 1.

I still don't believe I heard much difference this time as well.
The thing I did noticed, however, I did not really get an increase in headroom... which I thought might happen by lowering the 12au7 plate R since the preamp is running about 50 or more volts less than it was designed to. But this was my initial inquiry.

Also, when a chord is struck the sound sometimes cuts out and fades back in ("sag" or compression?) with what I believe to sound like some "hard clipping" as well. Of course I was counting on either bias or plate resistance as the beginning approach.

If you bypass the 2k2 with 10-25uF you will get almost a doubling of voltage gain.

I did temporarily bypass the cathode resistor using a 1uf and was a dramatic increase.

But now that I know where the tubes are at bias-wise I feel adjustments made elsewhere won't be futile.