Question for The Tube Experts

[Paul Marossy]

OK, well I am going to try some of y'alls suggestions and see how it goes. 

[jrc4558]

Class A AmplifierPlate Voltage ................................. 100 V
Grid No. 2 Voltage ............................ 100 V
Grid No. 1 Voltage Derived from  Cathode Bias Resistor ....................... 68 Ω
Plate Resistance (approx) ..................... 250K Ω
Transconductance .............................. 4300 µ
Plate Current ................................. 10.8 mA
Grid No. 2 Current ............. .............. 4.4 mA
Grid No. 1 Voltage for Ib ..................... 40 µa @ -20 V

I think you should lower your cathode resistor. The datasheet calls for 68 ohms. Perhaps lower the plate resistor too. It will give you higher plate voltage and definitely a better noise figure. To lower gain of a cathode biased stage either drop the cathode bypass cap, of use a 47k-47k in series for plate resistor and tap the signall off of the junction of the two. Half the signal, no loading down!

[R.G.]

I glanced through the discussion and didn't see it, so I apologize if it's already in there.

Turn down the screen voltage.

On a pentode, the screen is effectively a gain control. High screen voltage = high gain. Lower screen voltage = lower gain.

Pentodes have been used for VCAs on that basis.

[Headbuttking2]

This is kind of off topic, but just a quick question for a beginner...

Is the point of using a preamp as a pedal just to add gain to the signal?
I guess its okay to have 2 preamps then? (1 as the pedal, and then 1 in your amp)

[R.G.]

The point of using a tube preamp as a pedal is to use real tubes to make tube like sounds.

[John Lyons]

The grid stopper also affects the tone/frequency response in tubes . Tubes have what's called Miller capacitance, the grid input resistor/grid stopper along with the capacitance of the tube creates an Hif frequency roll off. I don't know all the details but you can look up Miller capacitance...

[Paul Marossy]

Turn down the screen voltage.

On a pentode, the screen is effectively a gain control. High screen voltage = high gain. Lower screen voltage = lower gain.

Pentodes have been used for VCAs on that basis.

That makes sense. Before I started tweaking stuff, I had no headroom at all - it would just have terrible sounding distortion with the gain more than 1/3 way up. So I lowered the plate & screen resistor values to make the tube move more current since it was barely, I mean barely on before. I guess I could try increasing the screen resistor some and see what effect that has.

OK, here's an update on some things I have tried so far:

1. 68K grid stopper on the input - does perceptably lower the noise floor. Not much, but it does.
2. Adding an 82K resistor in parallel with the 500K gain pot to effectively make it a 100K pot also lowered the noise level some.
3. Changing the value of the volume pot had very little effect on the signal to noise level.

Outcome: it is a lot quieter now than it was before, but there is still room for improvement.

Next on the list of things to try:

A. Smaller value cathode resistor on 12BA6
B. Lowering the voltage on the screen of the 12BA6

I'm going to keep it at because this little guy will get some nice tones from nasty crunch to nice clean sounds!  It kind of reminds me of my Octal Fatness in some ways.

[Eb7+9]

the main purpose of grid blockers is to limit current flowing into the grid circuit during positive input voltage excursions (one source of clipping) - it also helps with RF too ... you see grid blockers used mainly in high-gain distortion preamp circuits to limit compression (protecting tubes at the same time) and also at the input of the Phase-Inverter stage in tube bass amps (Traynor YBA1-mkII, YBA3, others ...) ... these don't affect noise much as you've discovered - noise is a product of device self-noise and gain scheduling ... since there isn't that much gain overall I suspect it's your pentode - lots of tube designs that employed a high gain Pentode in the front used low-noise tubes like the EF-86/6267 - often hand picked ... your circuit has nothing unusual that would make it noisy apart from the treble cap at the output ...

jc

[Paul Marossy]

your circuit has nothing unusual that would make it noisy apart from the treble cap at the output ...

Yeah, I have also considered that some of the problems might lie in the tubes themselves. I think there is still some room for improvement on this one, though. I also do know that it will only get so good. 

I will have to wait on this for a day or two as I just got an ear infection in my left ear (part of this crummy cold/flu thing I have had since Sunday), and it's 1:00AM and I can't sleep from the ringing and pressure in my ear. And I got conjunctivitis in my right eye which I got from my 2-1/2 year old son. Not a good week! Off to the doctor I will go... 

[brett]

Get well Paul.
Interesting thread.  I'd like to see more tube-based projects.  I like the sound of both 12AT7s and 12AU7s when they are "on the edge".

[Paul Marossy]

I'm trying to Brett! This darn thing is hangin' on pretty tenaciously! I'm going to see the doctor this morning to get this crap out of my system once and for all!

Anyhow, yeah, this is a fun tube project and I think it sounds great. It just needs some tweaking due to the massive gain of the 12BA6 pentode. It's a lot more useable now than it was before, but I think it can be improved on a little bit more. Hmm... I just had a thought. I wonder what a 6SJ7 would sound like in there? Hmm.... 

[R.G.]

the main purpose of grid blockers is to limit current flowing into the grid circuit during positive input voltage excursions (one source of clipping) - it also helps with RF too ...

In my experience, that is backwards.

The main purpose of grid stoppers as they are more commonly called is to prevent RF oscillation by giving the input some RF rejection. The input grid capacitance to ground and the Miller capacitance in concert with the grid stopper form a simple RF rejection.

Limiting grid current is a side effect. Unless you have a circuit that can provide significant grid current, it is a totally moot point. A 12AX7 plate driving a typical grid simply can't provide enough grid current to cause any notable damage on the grid. This might be an issue with power stage tubes, but I suspect you may have read this in some text about RF final stages where grid current and grid dissipation is an issue.

I have used grid stoppers up to 100K to get the grid rolloff down into the audio range for high gain preamp distortion. I was not happy with the results, and here is why.

When a 12ax7 grid goes positive, it changes from substantially an open circuit to a nonlinear resistor in the 5K-10K region. If it is being driven from a high impedance, like a 12AX7 plate, the plate resistance of about 68K forms a voltage divider with the grid resistance and the grid voltage is clamped by its low resistance. The higher the impedance a grid is driven from, the more the grid will flat-top when it's driven positive, as there is simply not enough current available to drive it higher.

Instead, driving from a LOW impedance will pull the grid positive in spite of its change in impedance at Vgk=0, and produce a softer limiting from that nonlinearity. This is one reason that class AB2 power amp finals use transformer drive, cathode followers, and my innovation - MOSFET followers - to drive output grids. The grids can be pulled  more positive and use more of the cathode's space charge until cathode emission limits are reached. The same thing happens in miniature with triode preamp tubes.

The 100K grid stoppers did kill otherwise intractible oscillation in the very cramped box, though.

[TELEFUNKON]

Paul: try a cap from anode to grid (10p - 470p) of the pentode.

[Paul Marossy]

Paul: try a cap from anode to grid (10p - 470p) of the pentode.

Hey, now that's an idea. 

BTW, two things I've neglected to mention/ask about earlier:

1. On the screen grid voltage thing... that's a Catch 22 in this case. If I lower it, then I get this nasty sounding motorboating kind of oscillation at various settings. So, increasing the screen voltage actually helped, but at the expense of an elevated noise floor. I guess I just need to find the tipping point and work around that.

2. What is the purpose of the 0.047uF cap to ground at the junction of the screen resistor and screen grid of the 12BA6? I have seen this on other schematics, and in at least one case, increasing the size of this cap also helped to lower the amount of hum you could hear with no input.

[jrc4558]

WEll, its a power filtering can, no? Lower voltage supply for the grid off of which this cap hangs has to be filtered a bit. Hence the cap.

[R.G.]

2. What is the purpose of the 0.047uF cap to ground at the junction of the screen resistor and screen grid of the 12BA6? I have seen this on other schematics, and in at least one case, increasing the size of this cap also helped to lower the amount of hum you could hear with no input.

WEll, its a power filtering can, no? Lower voltage supply for the grid off of which this cap hangs has to be filtered a bit. Hence the cap.

The screen grid in a pentode acts as a gain modifier. In effect, it's a second input. If you let that second input have fluctuations, those fluctuations modify whatever other signal is there, even a DC condition, so the output has an amplified version of whatever is on the screen.

If what is on the screen is ripple or hum, you get ripple or hum out on the output; if it's signal, you get signal. If it's signal of the wrong phase, you get oscillation. Keeping a screen at the right place requires making sure that neither AC hum nor power supply ripple nor unwanted stray signals get into it. Hence the cap. Some hifi rigs regulate the screen voltage to ensure that it's clean and not modulating the signal.

1. On the screen grid voltage thing... that's a Catch 22 in this case. If I lower it, then I get this nasty sounding motorboating kind of oscillation at various settings. So, increasing the screen voltage actually helped, but at the expense of an elevated noise floor. I guess I just need to find the tipping point and work around that.

You're getting gain-phase oscillation by getting extraneous signal in there. I suspect that the varying output impedance of whatever you were using to raise/lower the voltage interacted with the "bypass" cap to give you oscillation. Pot impedance is lowest at either end, hence the lack of oscillation. Grounding is critical with high gain pentodes, by the way - but you've already found that out.

Paul: try a cap from anode to grid (10p - 470p) of the pentode.

That introduces a large external Miller capacitor. Done properly, this can cut HF noise. Done improperly, it cuts audio as well.

[Paul Marossy]

Paul: try a cap from anode to grid (10p - 470p) of the pentode.
That introduces a large external Miller capacitor. Done properly, this can cut HF noise. Done improperly, it cuts audio as well.

Is there a way to calc what size the cap should be so it doesn't cut too much high end off of the signal?

[jrc4558]

llok up the inter-electrode capacitance for the junction in parallel with which you will be placing the capacitor, and add the values. Then add all the capacitances between the input grid and everything else. Then calculate the corner frequency of your low pass filter, which the grid resistor will form in conjunction with the total input (internal and the cap that you will add) capacitances.

[TELEFUNKON]

oh, yes, and the µ!

better put a (variable) resistor in series with the cap (e.g. 220k+ 5M pot),
to reduce the milleristic multiplication factor.

try 47p, 100p, 220p, 470p, and 1n.

you can take the plate signal from the node C4/R6,
so you can use low voltage caps.

[R.G.]

Is there a way to calc what size the cap should be so it doesn't cut too much high end off of the signal?

Sure. The input grid stopper resistor and all of the equivalent capacitances at the grid form a simple 1 pole RC rolloff, for which the half-power frequency is 1/(2*pi*R*C)

The complexity is - what is C?

llok up the inter-electrode capacitance for the junction in parallel with which you will be placing the capacitor, and add the values. Then add all the capacitances between the input grid and everything else. Then calculate the corner frequency of your low pass filter, which the grid resistor will form in conjunction with the total input (internal and the cap that you will add) capacitances.

oh, yes, and the µ

!

And as these two were working toward, it's the sum of the Cgk, G*Cga, and G*Cgp, where Cgk is the grid to cathode capacitance, Cga is the grid to anode internal capacitance, Cgp is the external capacitance you may have added from grid to plate, and G is the voltage gain.

Feedback capacitance appears as a capacitance to ground at the grid that is multiplied by the forward voltage gain. If your tube is running Av=100, then a 100pF cap from plate to grid looks and acts like a 10,000pF cap from grid to ground.