Help - the pre amp I made is not working!

[DF96]

100K is the correct value for a voltage gain of unity, from the pot slider to the output. You are getting 'motorboating', which is due to a combination of LF phase shift and insufficient HT decoupling. Your feedback loop has two sources of LF phase shift: C10/R19 (a high-pass filter) and C9/R18 (a lead-lag network when combined with the cathode of V4). At low frequencies the HP filter will add 90 degrees of phase shift. The lead-lag will add some phase shift, but this will disappear again at even lower frequencies. The two together are not enough to create the conditions for instability, so there must be another source of feedback. This is the HT rail. At very low frequencies the smoothing capacitors etc. have less effect so the current drawn by V4 and V5 is varying the HT, and this may be getting back via V3 (or perhaps via V4). Does the LF instability disappear on tape input?

Things to try: better HT decoupling between the output and earlier stages, vary the value of C9 up or down, reduce the value of C10, make R17 equal to R19 in parallel with R20 (if DC conditions will allow this).

Alternatively, since you only need unity gain, scrap V4 and just have a cathode follower.

 

[cr0wl3y]

100K is the correct value for a voltage gain of unity, from the pot slider to the output. You are getting 'motorboating', which is due to a combination of LF phase shift and insufficient HT decoupling. Your feedback loop has two sources of LF phase shift: C10/R19 (a high-pass filter) and C9/R18 (a lead-lag network when combined with the cathode of V4). At low frequencies the HP filter will add 90 degrees of phase shift. The lead-lag will add some phase shift, but this will disappear again at even lower frequencies. The two together are not enough to create the conditions for instability, so there must be another source of feedback. This is the HT rail. At very low frequencies the smoothing capacitors etc. have less effect so the current drawn by V4 and V5 is varying the HT, and this may be getting back via V3 (or perhaps via V4). Does the LF instability disappear on tape input?

Things to try: better HT decoupling between the output and earlier stages, vary the value of C9 up or down, reduce the value of C10, make R17 equal to R19 in parallel with R20 (if DC conditions will allow this).

Alternatively, since you only need unity gain, scrap V4 and just have a cathode follower.

Thanks for the detailed explanation. I'll do a bit of homework to understand what this is all about and try some of the suggested solutions.

Unfortunately I don't think I can scrap V4. The original design may have intended it to be another gain stage but I need it as a summing mixer after the volume pot. The final design of this pre-amp would have everything up to the grid of V4 twice (bringing the signal from two turntables to line level), the V4 mixing the two (the 100k resistors isolate the signals from one another), the cathode follower reducing the output impedance and feedback for unity gain. Got the idea from the following:

The Tube CAD Journal: Vacuum tube mixers

Could I simply mix the two signals straight into V5? In other words, would it work if the left channel of Phono 1 and Phono 2 were connected using 100k resistors straight into the grid of V5 and scrapped everything else?

 

[DF96]

It all depends. If you want a proper virtual earth summing point then your output stage would need more open-loop gain than a 12AU7, so you are already compromising a little. Going straight to a CF output would mean that the level of one signal would depend to some extent on the pot for the other signal.

May be best to stick with what you have, and get it working OK. Given that you are taking feedback from the actual output point, C10 can be smaller than you might think reasonable as the feedback will restore lost bass. By the way, your aim should not be to merely stop the LF oscillation but go beyond this and avoid a large LF peak. A small LF peak may be unavoidable.

 

[cr0wl3y]

It all depends. If you want a proper virtual earth summing point then your output stage would need more open-loop gain than a 12AU7, so you are already compromising a little. Going straight to a CF output would mean that the level of one signal would depend to some extent on the pot for the other signal.

May be best to stick with what you have, and get it working OK. Given that you are taking feedback from the actual output point, C10 can be smaller than you might think reasonable as the feedback will restore lost bass. By the way, your aim should not be to merely stop the LF oscillation but go beyond this and avoid a large LF peak. A small LF peak may be unavoidable.

It works! Following your tip about reducing the size of C10 size I tried .22uF and the "thumping" is gone. The gain now is perfect for my purposes.

Thanks!

 

[DF96]

Good. If you get woofer flapping (especially on disc replay) then you could take C10 even smaller.

 

[cr0wl3y]

Quick question - where would you place the send/return for a "master" EQ/isolator in the pre-amp? Was thinking about just before the line stage but then worried that when switched on it would mess things up as it will be between the two channels and the valve that is supposed to work as a summing mixer. Putting it after on the other hand may screw up the low output impedance of the system.

Any thoughts? Could instead go for two separate EQ/isolators, one for each channel, placed where the "tape" input/output is (just before the volume pot). More expensive but also more versatile for mixing.

Thanks,
Nikos

 

[DF96]

Sorry, I haven't a clue what you are asking! I'm not a musician, I just know about valve hi-fi. Someone else may be able to help.