preamp design

[lowell]

So I built a nice tube prototyping board. Plank o ply w 2 preamp sockets and 2 larger octal sockets. I've wired a circuit up. 2 12ax7s. I have a single coupling cap and volume between each stage AND 68k grid stoppers. I'm having a hard time getting results to match predictions as far as scoping signal. I have stage 1 and 3 with Rk: 3.3k and stages 2 and 4 with 1k. All stages have Ra:47k. I'm trying to warm bias the stages and get asymmetrical clipping. I'd like stages 1 and 3 to compress the upgoing waveform on the plates, and stages 2 and 4 to pass signal without compressing the opposite side of the waveform, thus achieving asymmetric and even order harmonics. NOW. When inserting a 400mv 1khz signal I can get things to look right for the most part by going stage by stage with my scope on each plate and adjusting the volumes to suit. Here's my issues:

When I plug in my les Paul w 57 classics I hear distortion. This leads me to the question: what's the PEAK ac voltage of these pickups when playing aggressively? And how does one measure this?

Secondly: why does the signal before the grid stopper look different than the signal after the grid stopper (directly on the grid)? And if the Vk is -2.5v shouldn't the signal ON THE GRID not clip (grid clamp) until 2.5v?

[Merlinb]

This leads me to the question: what's the PEAK ac voltage of these pickups when playing aggressively? And how does one measure this?

Easiest way is to hook your scope up to the output of the guitar (i.e., before the grid stopper). Many humbuckers can easily give you 1Vpeak when playing aggressively.

Secondly: why does the signal before the grid stopper look different from the signal after the grid stopper (directly on the grid)? And if the Vk is -2.5v shouldn't the signal ON THE GRID not clip (grid clamp) until 2.5v?

The grid starts to conduct around Vgk=-1V, not 0V. With really hard playing it is possible you may get some mild clipping in the first stage even with -2.5V bias. The bigger you make the grid stopper, the more this is enhanced, though I wouldn't expect it to be anything but very mild clipping of occasional transients, unless the valve has a weak vacuum...

[nosaj]

Any chance you have szome pics of you board? And being a prototype board how did you decide what transformers to use and what type of circuit layout are you using?

thx,
jason

So I built a nice tube prototyping board. Plank o ply w 2 preamp sockets and 2 larger octal sockets. I've wired a circuit up. 2 12ax7s. I have a single coupling cap and volume between each stage AND 68k grid stoppers. I'm having a hard time getting results to match predictions as far as scoping signal. I have stage 1 and 3 with Rk: 3.3k and stages 2 and 4 with 1k. All stages have Ra:47k. I'm trying to warm bias the stages and get asymmetrical clipping. I'd like stages 1 and 3 to compress the upgoing waveform on the plates, and stages 2 and 4 to pass signal without compressing the opposite side of the waveform, thus achieving asymmetric and even order harmonics. NOW. When inserting a 400mv 1khz signal I can get things to look right for the most part by going stage by stage with my scope on each plate and adjusting the volumes to suit. Here's my issues:

When I plug in my les Paul w 57 classics I hear distortion. This leads me to the question: what's the PEAK ac voltage of these pickups when playing aggressively? And how does one measure this?

Secondly: why does the signal before the grid stopper look different than the signal after the grid stopper (directly on the grid)? And if the Vk is -2.5v shouldn't the signal ON THE GRID not clip (grid clamp) until 2.5v?

[loudthud]

With 3.3K on the cathode and 47K on the plate, the bias is too cold and you will be clipping on the minus half of the input waveform. What are your B+ and plate voltages?

I've see humbuckers put out 4V peak to peak if you really grind on the strings. That will clip most 12AX7 inputs unless the cathode is un-bypassed. Since the stages are biased off to one side or the other, you might see a shift in cathode voltage when a signal is being sent through the amp. Be sure to check this. Many preamp stages act differently to a guitar than they do to a steady state signal from a generator.

[lowell]

The B+ for stages 1 and 3 is ~250v and for stages 2 and 4 it's ~300v. Va for 1 and 3 is ~217v and for 2 and 4 it's ~234v. I have NO bypass caps on the cathodes.

[J M Fahey]

The scope will answer your questions.

[lowell]

JM can u be a lil more specific? Are u saying that seeing a different waveform after the grid stopper on the scope is telling me that there's grid clamp happening?

[J M Fahey]

"Different" is not quite specific
Back to the point: if you see a perfect sinewave before the grid stopper and a flattened top one after, yes, "something" is clamping it.
If all you have there is a grid, Sherlock Holmes would say : "here's the suspect".
Resistors by themselves are linear, clamp nothing.

[lowell]

Makes sense. I guess my next question is can you get grid current limiting AND avoid grid "blocking" while doing so?

[exclamationmark]

Makes sense. I guess my next question is can you get grid current limiting AND avoid grid "blocking" while doing so?

Technically by sticking a huge grid stopper (most common is 470k on high gain amps) you get extreme grid current limiting. As a follow on effect, this eliminates grid blocking, because when the grid goes positive, it can't drain the capacitor through the giant grid stop.

I have a feeling this isn't the question you wanted to ask though... You probably meant softer grid current clipping, whilst avoiding blocking distortion. In that case, you will have to scrap the coupling capacitor and go direct coupled. I've made a completely direct coupled 12au7 design, using fixed bias (though you can do the same with cathode bias - probably stabler that way, but headroom gets smaller and smaller for each progressive stage - good for clipping though!). It sounds too smooth in my opinion, so I only fire it up for solo work. It makes use of a direct coupled cathode follower, which I intentionally made to compress the living daylights out of the signal, which may account for the smoothness (and in fact may have nothing to do with the whole mess being direct coupled).

If you're feeling particularly adventurous you can even use mosfets between stages.

[lowell]

thanks Mark for clarifying that does help. Actually that's exactly what I'm after, smooth compression. It seems to me the only way to get this is throughout gentle cutoff and/or biasing to where the plate curves get bunched up. I'd like to pursue this type of action on the grid's "upgoing" signal too so I don't have to waste each "in between" stage. (though I've considered inverting buffer opamps b/t stages, but what's the point if we can get the grid to act the way we want)

Are you open to sharing your direct coupled circuit? Yes I'm totally open to and actually would prefer Mosfets as they're cheaper and from what research I've done do not impart any "solid state'ness" to the sound because there's no voltage gain in a CF or voltage follower.

[exclamationmark]

The first 'option' is basically what I've used in the past. It's stupidly simple, as it's really just a voltage divider, with a bypass cap to let through the highs (it makes a shelf filter - it basically replaces cathode resistor/cap combination, as well as letting me get more gain - you can remove it if you want and just live with half the gain). The RC constant is pretty safe, and lets things work without blocking distortion. The negative bias supply in this case should be equal to slightly less than B+ (ie, +300, and -250), but I just used a C.T transformer to get symmetric rails. Basically if you have 200 volts coming out of the previous stage, and you want to apply -2 volts bias to the stage upstream, you apply -202 volts at the voltage divider. Simple.

The second option and more common approach is to simply remove the coupling capacitor and grid leak, and just stick 2 gain stages together. This works fine, except you will HAVE to use a cathode bypass capacitor (unless you want <1 gain on a 12au7). Headroom is also reduced. Using another example, if we have 150 volts coming out of the previous stage and we wanna bias the stage upstream to -2 volts, the cathode of the upstream stage will have to be sitting at 152 volts (as opposed to 2 volts in your normal ac coupled stage). This requires using a huge cathode resistor, which will make huge negative feedback when a signal is applied, thus the cathode bypass capacitor is required. Since the cathode is at 150+ volts, the anode-cathode voltage is also reduced, reducing output swing. Setting up load-lines for this is tricky, and you should probably consult merlin's book.

The third option is to use a mosfet source follower. A crucial bit of info is that the mosfet doesn't track the input voltage on it's gate perfectly - the voltage which appears at it's source can be anywhere from 1 to 5 volts more negative than the gate, depending on the mosfet and cathode resistor. The +Vs supply for the mosfet doesn't have to be high; 10 volts would probably be enough. However, the -Vs supply will need to be fairly high (or if you wanna be technical, really low). Since the capacitor inverts the signal from the previous stage, the negative voltage could peak at more than -240 volts at clipping. If your -Vs isn't up to the task, the mosfet will clip!

If I were to use mosfets on preamp tubes, I would probably get +/- 20 volt supplies (think audio SMPS supplies) and connect it to a potentiometer. I would take the centre lug as the bias output. This would allow me to adjust the bias, and basically ignore the effect of the tracking offset (remember - to get a 0 volt output on the source you need a positive voltage to be applied to the gate of the mosfet of about 1 to 5 volts). You can use the tube as fixed bias by setting the bias to whatever you want (If you were a bit crazy, you could even set the quiescent grid voltage positive!). You could add a cathode resistor, set the grid voltage to 0, and have a cathode biased stage. Or you could use a bit of both.

You will need to use grid stoppers for the mosfet coupled method, probably in the range of 100R to 2k to stop your grid from burning up. I wonder how ridiculously smooth a mosfet direct coupled, cathode biased amp, without bypass caps would be....