Loadline suggestions for a tube preamp.

[Anthon]

I'm in a process of designing my own tube amp, and I would appreciate some input.

I picked the topology, which is similar to MILA preamp and already built a first bare bones prototype for experimenting.
Now I need to pick appropriate load lines for the stages.

Given:
Topology (not the actual schematic of course).

3 common cathode stage, first 2 using 12AX7, last one using 12AU7 in parallel. All cathode resistors fully bypassed.
1:10 input transformer, still not sure about output transformer.  Perhaps a LL5402 (4:1). + DI input for guitar/bass.
About 55-60dB of total gain is needed (in mic mode).
Volume is controlled with a gain pot (100k for now) and there is a 4 position 3 pole switch, for negative feedback control.
Goes without saying, but I would like maximum clean headroom this topology would allow (and least distortion).
Power supply is 300V max, and can be reduced for respective stages.
Grid leak resistors are 1M and grid stoppers are 270 Ohms.

The questions are:

1) How would you setup the different stages? I included blank graphs in the attachment - I would appreciate if you could draw some load lines with quiescent points, that would be a good starting point. No need to give resistor values etc - I'll calculate cathode and plate resistors myself.

2) I'm not sure how I should spread gain factors over stages? Just evenly? Or maybe I should more gain on the 12AX7 stages, and low gain on the 12AU7 stage (use it to reduce output impedance), considering 12ax7 is a higher gain tube than 12au7.

1:10 Transformer 20dB
Stage 1:  20 dB
Stage 2:  20 dB
Stage 3:  10 dB
Output transformer 4:1 -12dB

3) Output transformer: if I understand it correctly,  the main goal should be (aside from balancing the signal) to reduce the output impedance (for impedance bridging, at least 5 times less than the following device of typical 600 Ohm)? Or is there something else?

Any other remarks are welcome. Maybe changing the tubes or replacing the last stage with different configuration? (triode in parallel seems like a waste).

Thanks.

 

[ruffrecords]

The 1st stage is a compromise between headroom and noise. The transformer helps the noise performance a lot but you still need to be careful with the first stage. 20dB to 30dB is a good range. More than this and you have a headroom problem. The classic way to deal with this is to a dd a switched 20dB input pad for high signal levels. If you only want 20dB from the first stage then the 12AX7 is probably not a good choice. 30dB would be more appropriate.

The second and third stages need to be considered together as they are in a -ve feedback loop and it is the feedback that determines the gain. The gain of the two stages determine the open loop gain, the amount of distortion reduction from the -ve feedback and to a large extent the stability. You should aim for the maximum gain from the second stage and the maximum drive capability from the third which basically means getting the standing current high enough (that'd why NYD used two tubes in parallel). You could use and SRPP instead for the output stage or use them in push pull. Either way, if you want to drive a 600 ohm load then you will need two triodes. If you only want to drive a bridging load you could consider a simpler output stage and have an unbalanced output or just use a good 600:600 transformer with a high primary inductance.

Cheers

Ian

 

[PRR]

> I would like maximum clean headroom

I'd like maximum horsepower in my car.

100HP? 1,000HP? 10,000HP??

Let's get real. A sedan can't even put 1,000HP to the pavement at less than 100MPH, and we don't go that fast even on the far end of the turnpike (too many moose). I have 148HP, and rarely when I have to haul 45MPH-65MPH onto a freeway I could maybe use 200HP. But wishing for the sky is folly.

What do you need? +18dBm? +40dBm? +60dBm?

What can your next-stage input really swallow?? Most max-out not far above +20dBm; any more just distorts.

Why can't you turn-down hot inputs so they don't clip? A turned-down audio signal should be the exact same thing, only a better size.

Max power out of SE topology *mandates* an OT which handles heavy DC. Efficiency may approach 50% for ideal valves, 25% for reasonable valves. OTOH resistor-cap-transformer gives at best 6% efficiency, 4X less!

Max power out of parallel 12AU7 is a couple Watts; use 3K load. THD will be high. The usual compromise is 2X Rp, or for 2 units 6K. As this will give +33dBm(!), or 34Vrms(!!), it is probably excessive. 2nd-order THD will decline as load rises, so 10K seems a sufficient load.

> and least distortion

Again... is 0.001% too high? Is 1% too low? If you remove "all" the distortion from a tube amp, is it still a "tube amp"? With point-oh THD chips at $10, you need to balance what you *really* want.

 

[Anthon]

> I would like maximum clean headroom

I'd like maximum horsepower in my car.

100HP? 1,000HP? 10,000HP??

Let's get real. A sedan can't even put 1,000HP to the pavement at less than 100MPH, and we don't go that fast even on the far end of the turnpike (too many moose). I have 148HP, and rarely when I have to haul 45MPH-65MPH onto a freeway I could maybe use 200HP. But wishing for the sky is folly.

What do you need? +18dBm? +40dBm? +60dBm?

What can your next-stage input really swallow?? Most max-out not far above +20dBm; any more just distorts.

Why can't you turn-down hot inputs so they don't clip? A turned-down audio signal should be the exact same thing, only a better size.

Max power out of SE topology *mandates* an OT which handles heavy DC. Efficiency may approach 50% for ideal valves, 25% for reasonable valves. OTOH resistor-cap-transformer gives at best 6% efficiency, 4X less!

Max power out of parallel 12AU7 is a couple Watts; use 3K load. THD will be high. The usual compromise is 2X Rp, or for 2 units 6K. As this will give +33dBm(!), or 34Vrms(!!), it is probably excessive. 2nd-order THD will decline as load rises, so 10K seems a sufficient load.

> and least distortion

Again... is 0.001% too high? Is 1% too low? If you remove "all" the distortion from a tube amp, is it still a "tube amp"? With point-oh THD chips at $10, you need to balance what you *really* want.

Maybe I explained it the wrong way...
I really would like to experiment with different values and judge it by ear and see how different values sound , and I was looking just for a good place to start. Because figures don't tell me much :-\
By 'maximum clean headroom' I mean a preamp that performs like a ''real world professional recording preamp'' should. (what ever that means to you, personally).

Then like you said, 20dBm would be good if that's what the most devices can take (I was thinking about about 24 dBu)
I can't turn down the input, because my RME UCX doesn't have input potentiometers  ;D (which has max input of 19dBu).


About distortion: again, what I mean is just a reasonable level of distortion expected from a good quality preamp, as a starting point. I expect a certain level of fidelity (what I mean by 'least amount of distortion'), but more 2nd order harmonic distortion will probably give nicer 'tube sound', otherwise what is the point of using tubes.
Because this is my first attempt at designing a tube circuit, I can't give a figure.
I trust your judgement on what is good ;D

Thanks, I'll use values you have suggested as a starting point for the output stage.

 

[CJ]

just build a V72 or V76,

 

[DerEber]

Don't listen to him  ;D

 

[PRR]

> I can't turn down the input, because my RME

Not the next-stage! Turn down the preamp!

I think there is a between-stages pot in your plan? If you are overloading in the first stage, have a pad. (Big condenser on a mad drummer, hiss is not your main problem.)

As for load-lines... tubes like higher impedance and transformers do NOT. Hi-Z windings need hyper-fine wire (expensive to wind) to get bass inductance up, leakages inductance tends to want to be a proportion of bass inductance, capacitance hardly changes, so you have a resonance at the top of the audio band.

But one point want to be provisionally "decided" very early on. Will the OT carry the last-stage's DC current or not? If not, power output is way down but the OT design (cost) is much lower. And of course we can't even start to draw a loadline until we know this.

As CJ says, plagiarism is a respectable path. Especially since ready-made parts will be FAR cheaper and quicker than custom design parts.

 

[Anthon]

Yes, there is an -20dB input pad and a potentiometer between stage 1 and 2.
Also negative feedback that can be controlled by the 4 position switch.

For carrying DC current, I would like to use a capacitor.

I could build an existing design, but my goal is to learn more about tube circuits.
I already built a 2 channel MILA, and I'm quite pleased with the results.
But what do you mean by custom design parts? Surely you can make a new design, using standard parts?

 

[PRR]

> goal is to learn more about tube circuits.

You know some already, more than is shown in your sketch.

Flesh it out and see where you are at.

Loadlines: on triodes for gain, aim for 2X-5X Rp. But for the hi-Mu 12AX7, this will lead to values that are not flat to 20KHz (why?). You will probably do best by plagiarizing the Fender values: 100K rp and 1.5K rk. For a power stage with a low-Mu triode, aim as low as 2X Rp.

The operating point along that line will suggest itself once drawn. Sorta in the middle of the possible signal swing.

 

[CJ]

start with stage 1 then go thru RDH4 and apply every formula you can find for a tube stage,

this is not as hard as it sounds it you leave out negative feedback,

once the negative feedback is applied, you are in for about 6 months worth of calculations,

AC, DC, phase shift, output impedance, fun fun fun!

that V72 was tweaked for about 30 years by nuclear phd's, looks simple, but every part interacts with the other, you will have a hard time beating those design teams,

 

[ruffrecords]

+1 to RDH. There is some very sound advice in there on choosing operating points.

Cheers

Ian

 

[mjrippe]

RDH4 = Radiotron Designer's Handbook, 4th Edition.  You can download a massive PDF copy from www.tubebooks.org

 

[Anthon]

> goal is to learn more about tube circuits.

You know some already, more than is shown in your sketch.

Flesh it out and see where you are at.

Loadlines: on triodes for gain, aim for 2X-5X Rp. But for the hi-Mu 12AX7, this will lead to values that are not flat to 20KHz (why?). You will probably do best by plagiarizing the Fender values: 100K rp and 1.5K rk. For a power stage with a low-Mu triode, aim as low as 2X Rp.

The operating point along that line will suggest itself once drawn. Sorta in the middle of the possible signal swing.

I didn't realize more information was needed. So here is a more detailed schematic of the topology I was thinking about. (not all values are correct, and there is a pad before the transformer).

Just 3 common cathode stages in a row.
There is no feedback loop, but it will probably go from output after 4.7 uF capacitor to the cathode of stage 2. Or before the capacitor, with an other just for the feedback loop. I'll just experiment with different setups, and see what sounds best.

As for the bypass capacitors, I guess I want them fully bypassed as a starting point.  I know how it will affect the frequency response and gain in theory, but this doesn't tell me how it will actually sound.

Okay, I'll use the values you have suggested for the plate resistors, and it shouldn't be hard to find appropriate cathode resistors.
I guess the frequency response won't be linear with high gain tubes because of the Miller effect forming a low pass filter? (Voltage gain being in the equation).

About OT:  should I pick one based on the impedance (calculate output impedance of the last stage, then pick appropriate OT ratio) or is there something else? Wattage should not be a problem, I presume?

I want to have balanced output impedance around 75 Ohm.
Most the information I find on internet is about transformers for speakers.

RDH4 is just too heavy to read

 

[PRR]

> the following device of typical 600 Ohm

600 ohm inputs have been rare for decades.

"Everybody" has 10K inputs. The 12AU7s can do 10K fine. 4:1 step-down is 12 dB gain left on the table. (However this may not be bad.)

> 1:10 Transformer 20dB
> Stage 1:  20 dB
> Stage 2:  20 dB
> Stage 3:  10 dB
> Output transformer 4:1 -12dB

The first 12AX7 stage is clearly gain near 40, 32dB. The second one has lighter load, gain 50, 34 dB.

Parallel 12AU7 with 10K DC load and 10K external load looks like gain of 10, 20dB.

1:10 Transformer 20dB
Stage 1:  32 dB
Stage 2:  34 dB
Stage 3:  20 dB
Output transformer 1:1 0dB

106dB on tap.

"Too much" but you will use NFB to trade-off gain for distortion and impedance.

The only NFB possible in the first stage, without real hiss increase, is to switch-out the cathode cap, -6dB.

That first stage will be up to 5% THD at 50V peak output, 1Vpk at grid, 100mV at mike jack. 100mV is not a "huge signal" in this day of close-mike large condensors. Without bypass, ~~2% at 200mV from mike. And note that this stage's distortion depends on mike selection and placement, complicating the most important part of the job.

You have 20dB+32dB= 52dB in transformer and first stage. You hardly need any more gain to get a 55-60dB goal. You will surely need gain as low as 30dB before you use the pad. Then the path after the Volume control may have to work at "gain" of -12dB. You can take 20dB-30dB of loss in the volume control. You do have to ponder why you build it up then waste it away. Many "Nuclear PhD" designs don't run a wide-open 1st stage into a heavy loss pot.

{sigh} Many ways to skin cats. Your first try shouldn't be your last try. I think your proposed plan will work fine, even if not "maximum clean".

 

[kambo]

its 6SN7 based topic but very educational.

http://groupdiy.com/index.php?topic=12637.0

 

[My3gger]

Lower gain and better linearity of 6SN7/6CG7 as first two tubes could be advantage here. Cathode follower for output too, might require step down transformer to make sure it will drive a little heavier loads. W.E. 141-A has output like this, interesting design.

 

[gridcurrent]

Have you examined the data in the rear of RCA Receiving Tub e Manuals for
"Resistance-Coupled Amplifiers"?
Load lines were calculated eons ago for fully bypassed common cathode amplifier stages for various tube types.

Points of disagreement with some of our esteemed contributors:
like it or not, there is plenty of 600 ohm terminating equipment, maybe not in bedroom studios,
but certainly in money making facilities.
Don't cheap out on the design.

And pertaining the the allegeded superior linearity of a lower gain tube such as a 6SN7:
12AX7a is one of the lowest distortion tubes within it's limitations.
There is a reason it is found in so many time tested designs.

 

[kambo]

12AX7a is one of the lowest distortion tubes within it's limitations.
There is a reason it is found in so many time tested designs.

12AX7 is the worst tube i have ever used in any scale!
u have to hand pick 12AX7 for specific design... same 12AX7 may sound excellent  in one design,
and same 12AX7 would sound horrible in another design...

its just easy to grab one 12AX7...

 

[ruffrecords]

12AX7a is one of the lowest distortion tubes within it's limitations.
There is a reason it is found in so many time tested designs.

12AX7 is the worst tube i have ever used in any scale!
u have to hand pick 12AX7 for specific design... same 12AX7 may sound excellent  in one design,
and same 12AX7 would sound horrible in another design...

its just easy to grab one 12AX7...

That probably says more about the design than the 12AX7. As a first stage gain block in a preamp they are ideal. The only thing you really need to select for is microphonics but you have to do that with any tube.

Cheers

Ian

 

[CJ]

what is your input source gonna be?

vocals?

acoustic guitar?

sm57/marshall?

kick drum?

snare?

hard to get max performance for all those out of one box,

mic placement can take the place of a pot,

mic choice can take the place of the pot,

condenser mic into an optpmized 40db box can be sweet,

every preamp has there sweet spot, that is why you see no volume controls on the classics,

V72, API312, Langevin AM16, get NFB where you want it to get the best tone/noise ratio, then use the mic closet to control volume,

V76 has a ton of engineering used to get a great volume control,

want a transparent tube pre with no output transformer?  try the single stage WE 5814 or whatever it is, great for acoustic guitar and soft vocals,

no, WE417a

dead link city on that old thread, 

there is also the one bottle preamp probably stashed in the Meta somewhere,