6SL7 to 6SN7 substitution in power amp?

[dmitriy167]

Hi,

I decided to create a new thread to deal with the lack of gain that my 6L6 PP amp exhibits. At full volume from my CD player, I'm only getting four or five watts, but I should be getting 30 per channel with my 6L6WXT+ tubes.
The input tubes are 6SL7's and I was wondering if I can make a substitution to the (higher gain) 6SN7. What resistors would I need to replace, and what would be their approximate new values?

The schematic is attached.

[Giaime]

6SN7 hasn't higher gain than 6SL7. In fact, it's just the opposite

6SN7 gain = 20

6SL7 gain = 70

I think that you'll have to decrease a bit negative feedback to have some more gain. Try to make R15 4.7kohm.

Or better, build an active linestage

Or even better use a CCS in place of R3.

[kevinkr]

Looking at the schematic it is pretty clear that closed loop gain is right around 10dB, which is not really enough for most sources.

I don't really recommend eliminating the feedback because even though this is a UL amplifier, I suspect output impedance will be pretty high.

Going with a CCS will give you a significant increase in loop gain, but based on the existing plate load resistance possibly not enough to justify the trouble. See below - comment on R4.

There is another potential problem with this design and that relates to how much input headroom is present in the first stage, I'll bet it is not much, and the amplifier is only linear even at lower powers because of the presence of feedback at the input stage cathode.

The 6SL7 has mu of about 70, the 6SN7 a mu of about 20. The 6SL7 arguably doesn't have much gain margin in this circuit, with 10dB of feedback applied I doubt that the overall feedback margin is much more than 6dB.

Bypassing R4 with a large high quality electrolytic of about 470uF will help by removing the degenerative effect of R4 on the stage's gain and also by not nearly doubling rp as R4 is currently doing.

I thought initially that adding an external line stage might be just the ticket, but now I doubt the input stage can handle much more than a volt or two rms without serious linearity issues.

Changing from 6L6 to EL34 and decreasing R12 to about 330 ohms will net you a little more headroom and some more gain as well, no other changes required.

All of these changes would allow you to increase the feedback resistor to about 3.9K which should allow you to achieve about 20W without clipping the input stage.

Actually the Russian text clearly spells out that this is a 20W design, and given the plate to plate impedance of the output transformer (5K) and the very limited drive voltage available I think this is just about right.

[gingertube]

I think you can reduce feedback without significant problems.

My very first impression on looking at the circuit is that R4 and R5 have had their values swapped
That is:
R4 should be 1K and R5 should be 100R - that would give a large increase in closed loop gain.

Try it and see - check for loose/boomy/muddy bass which would indicate that feedback has been reduced too far.

If so - "mess" with the R4 and R5 values some more keeping R4 + R5 = 1k1 (e.g. you could try)
R4 = 820R and R5 = 270R - a bit more feedback
R4 = 680R and R5 = 390R - a bit more again
R4 = 560R and R5 = 560R - and some more again

If you are happy with the bass end and therefore have feedback about right but still need some more gain then do as Kevinkr suggests and bypass R4 with a 470uF electrolytic (only needs to be a 6.3V rated part).

Cheers,
Ian

[kevinkr]

No, actually the configuration as shown is correct, there is insufficient loop gain for 20dB closed loop gain which is roughly what you would get..

I have used the 6SL7 in a lot of designs and in this configuration the bias voltage will be about 0.5V (estimated) so without feedback this stage will be highly non-linear at anything above a few hundred mVrms.

[kevinkr]

The thirty minute limit has expired for editing my previous post:
What I meant to say is that swapping those resistors would result in a theoretical closed loop gain of about 26dB, however there is insufficient available loop gain for this to be the case. The implementation as shown is correct.

Kevin