Best practices for preamp stages

[Pierre]

A couple of years ago I built an amp loosely based on TW designs. The power amp was 2 x el84 driven by a rocket PI. My question is related to the preamp. The preamp consisted of an ax7 for v1a>volume pot>5879 for v2>dc coupled to v1b ax7 cathode follower>Liverpool tone stack>Power amp. I was able to get the amp fairly quiet, and it sounded great but.... I've always wondered if it was a mistake to use v1b as the cathode follower after v2. Would the larger downstream signal contribute noise to the sensitive first stage due to their proximity and having small and large signals in the same envelope? I'm dreaming up my next design with a 50 watt power section and was thinking of using a similar preamp and I don't want to repeat a bad design if in fact it is poor practice.
Thanks,
Pierre

[Malcolm Irving]

I think the issue would be the level of cross-talk between triodes in the same envelope - which I believe is capacitive.

There was an interesting series of investigations and discussion here:

http://www.diyaudio.com/forums/tubes-va ... tubes.html

[Pierre]

Thanks for that information. I'll have to read it through a couple of times to fully digest it. I'm not sure it answers my question of having signals with a large magnitude of difference in the same envelope. I suppose that if it's a concern, I may as well avoid it in future designs.

[Malcolm Irving]

Yes, looking at the thread again it includes a lot to digest!

I think the gist of it is that the signal 'leaking' into the 'other' triode of a 12ax7 is down by about 50dB at audio frequencies. This would create feedback from the higher signal voltage stage to the earlier stage. One way to look at it would to be compare that 50dB feedback signal with the (forward) gain between the two stages. Roughly speaking, if the forward gain is greater than 50dB there may be a stability problem. 50dB corresponds to a gain of 316.

[JazzGuitarGimp]

Or, in this case, since we're talking about an attenuation of -50 dB, it would 1/316

[Malcolm Irving]

Yes, I should have written ' ... compare that -50dB feedback signal ... '.

It's not a precise stability criterion (would need to allow for phase shift and use Nyquist or Bode etc. for that) but if we can make the overall loop gain less than 1 (i.e. less than 0dB) at any frequency, then we would avoid instability. If the forward path has a gain of 50dB and the feedback path has a gain of -50dB, the overall loop gain is 50-50=0dB.

One typical gain stage would be OK. (Just as well, we do that all the time!) But two typical gain stages in series would push the forward path gain above 50dB.

[Pierre]

So, the amp that I built had V1a ax7 through a 500k vol. pot into a 5879 pentode direct coupled to v1b ax7 as a cathode follower. I didn't experience any stability issues, but I tried very hard to lower the hiss level at high volume and always wondered if the design contributed to the noise level. My design goal is always to reduce complexity which is what attracted me to the trainwreck circuits. Though I hated the thought of an unused triode! I thought the amp sounded really good overall though, and I'm hoping to build on that design but with a 50 watt power amp and a parallel fx loop and some sort of relay switching. So much for reducing complexity!

[Malcolm Irving]

I would think that would be OK. I’m not familiar with the 5879, but the cathode follower has gain of less than 1 of course.
What value of grid stopper do you have on V1a? That is usually the main source of hiss.

[TUBEDUDE]

What kind of gain are you getting out of the 5879? I employ these mainly at lower gains, but with the front end driven hard as I like the distortion from them, being a true pentode and all, but you can get ridiculous gain from a pentode, and if the gain is high, noise will be also. Even if it is still stable.

[Pierre]

I'm not sure how much gain. I don't have the schematic handy, but the pentode was fairly warm biased and being fed with about 200 volt B+. The first stage has a 33k grid stopper, 220k plate resistor and red led on the cathode. It could definitely growl!

[Malcolm Irving]

You could try reducing the first grid stopper to 10k or so. The trade off here is that a lower grid stopper can reduce the hiss, but at some point you risk radio interference getting through.

[martin manning]

You could try reducing the first grid stopper to 10k or so. The trade off here is that a lower grid stopper can reduce the hiss, but at some point you risk radio interference getting through.

Don't you mean a higher grid stopper can reduce hiss and going too low might admit RF?

[Pierre]

That amp is built and gone. My nephew is using it. I'm planning the next one. I started this post to get a feel from more experienced builders on whether the topology I had used was an obvious poor practice. I'm moving on to a 40-50 watt with fixed bias because of haven't built one yet. I want to build on the design of the previous preamp, but will add an ax7 and to change the staging. Now, what to do with v1b....

[JazzGuitarGimp]

You could try reducing the first grid stopper to 10k or so. The trade off here is that a lower grid stopper can reduce the hiss, but at some point you risk radio interference getting through.

Don't you mean a higher grid stopper can reduce hiss and going too low might admit RF?

If I correctly understand thermal noise (which is questionable...) the lower the resistor's value, the less thermal noise it generates. Since the resistor in question is at the input to the first stage, whatever thermal noise it generates is subject to the highest gain the amp is capable of. It makes sense that you would want that resistor to be the lowest value possible without introducing other issues such as stability and RF admittance.

[https://en.m.wikipedia.org/wiki/Johnson–Nyquist_noise]

[JazzGuitarGimp]

Pierre, I would say as long as there are not stability issues, your design is sound. Based on what TUBEDUDE said, and in conjustion with what Merlin says about pentodes and distortion (which, in a nutshell is: they sound good when slammed by a preceding gain stage), you might try reducing the gain of the pentode stage to reduce hiss - I suspect the brunt of the 'good distortion' is a result of hitting the grid of the prentode with V1A. One other thing to consider; if I am not mistaken, hiss can also be a byproduct of HF oscillation, which ties in with Malcolm's assertion about stability. So it might be a good idea to confirm the absence of oscillation by viewing the output of the amp with an oscilloscope, if you haven't done so already.