New build: Princeton Reverb with mods - advice wanted!

[Enzo]

The brass plate will be a lot easier to solder to if you solder on it before you put it against the chassis. Either attach wires before installing the plate, or at least wedge something under it to bring it off the chassis for soldering.

[klooon]

I see that I have to isolate the tremolo footswitch jack from the chassis for this to work. I have seen in other 'footswitch designs' that they use a short to ground principle. How would this look like in my case?? (Excelsior tremolo)

[klooon]

The brass plate will be a lot easier to solder to if you solder on it before you put it against the chassis. Either attach wires before installing the plate, or at least wedge something under it to bring it off the chassis for soldering.

OK, thanks. I assume it's to prevent it from conducting the heat too quickly. Only, I have most things on order now, so not sure I want to go in the brass plate direction and I would like to try the 'gallictic grounding' first (that worked in my 5G9 build really fine).

[Enzo]

Why on earth do we need to leave the switch where it is? A hundred zillion Fenders kill the trem with a ground switch, why not this one too?


I must have missed the reasons the first feedback cap, C5 is coming from the source follower instead of right off the tube plate, but no matter if it works. You have a string of three caps C5,8,9 looping from the plate back to the grid of the tube triode. Isw there some reason we can;t sire C5 to C8, and use the switch to ground off either the top of R21 or R24? Then it is like any other fender insofar as grounding off the feedback loop kills the trem. As I described it, it would be a trem kill on the footswitch, not a trem enable.


I wasn't trying to steer you towards or away from the brass plate, just wanted you to know that you don't need the giant solder iron if you move the plate up off the chassis to solder to it.

[klooon]

Why on earth do we need to leave the switch where it is? A hundred zillion Fenders kill the trem with a ground switch, why not this one too?

I must have missed the reasons the first feedback cap, C5 is coming from the source follower instead of right off the tube plate, but no matter if it works. You have a string of three caps C5,8,9 looping from the plate back to the grid of the tube triode. Isw there some reason we can;t sire C5 to C8, and use the switch to ground off either the top of R21 or R24? Then it is like any other fender insofar as grounding off the feedback loop kills the trem. As I described it, it would be a trem kill on the footswitch, not a trem enable.

I am not sure how it works with the source follower etc, I just copied the Excelsior circuit... Thanks a lot for the suggestion and explanation! I have read a bit more about the Fender trem circuits now and I will add the kill switch on top of the R24. Please see below the adapted circuit:

I wasn't trying to steer you towards or away from the brass plate, just wanted you to know that you don't need the giant solder iron if you move the plate up off the chassis to solder to it.

Great - mucho appreciated! I was just trying to explain why I would choose the other route, hoping not to step on anybody's toes...

[klooon]

I looked again (and again) at Merlin's article about grounding:
(http://www.valvewizard.co.uk/Grounding.pdf)

And I updated the schematic further with a grounding scheme based on the multiple star grounding, in which the stars are daisy chained together.The voltage supply of each block and their ground returns are kept at the appropriate filter nodes. Now it's necessary to move the (negative side of the) filter cap as close as possible to such a star.

Here is the schematic (v008):

--> "inspired by" Tubeswell's schematic here: http://music-electronics-forum.com/a...-schematic.pdf

And here is the new layout (v006):


Any comments about my 'interpretation' of this grounding method are welcome!!

[klooon]

When I picked up my MM iron, I was told that probably I would fry my power tubes in fixed bias mode when the tremolo is off and better build a stock tremolo with a stronger output (e.g. lower value resistor) which should possibly do the job. I have seriously changed the schematic and layout (again...). Basically I followed/copied, the AA1165 circuit from Tubeswell (extra cap, grounding scheme) and also copied his mosfet tremolo. See the schematic and layout below.

schematic version 9:


layout version 7:


I really hope that this is the final version. I can't wait to start building. Please let me know if this is fine or not!

[tubeswell]

When I picked up my MM iron, I was told that probably I would fry my power tubes in fixed bias mode when the tremolo is off and better build a stock tremolo with a stronger output (e.g. lower value resistor) which should possibly do the job. I have seriously changed the schematic and layout (again...). Basically I followed/copied, the AA1165 circuit from Tubeswell (extra cap, grounding scheme) and also copied his mosfet tremolo. See the schematic and layout below.

schematic version 9:
[ATTACH=CONFIG]28934[/ATTACH]

layout version 7:
[ATTACH=CONFIG]28933[/ATTACH]

I really hope that this is the final version. I can't wait to start building. Please let me know if this is fine or not!

Hi klooon,

yep that's what i was talking about - works for me. You can't use a stock PR cap can with a 'galactic' grounding system like that because the grounds in the cap can are all tied together.

The source-follower is a buffer stage between the output impedance of the LFO triode's plate and the input impedance of the 6V6s grids. I did this in my amp because I like the intensity of bias-vary trem when its buffered. You can use a cathode follower (like a 5G9) if you have a spare triode, but in this amp there's no spare triode.

In terms of the footswitch, I ran a wire from the galactic ground point for the trem to the ground lug on the jack, but you could leave it out. Its probably 'better', in terms of avoiding ground loops, to isolate the jack sleeve/ground from the chassis if you do this. Over to you to experiment.

In terms of the comment about frying your output tubes in fixed bias mode, I'm guessing whoever it was who said that was commenting on a version of your schematic which had a DC path to the output tube grids that bypassed the bias voltage source?

[klooon]

Hi klooon,

yep that's what i was talking about - works for me. You can't use a stock PR cap can with a 'galactic' grounding system like that because the grounds in the cap can are all tied together.

Great, then I will go for it. I see that you have taken the B+ to the reverb driver (1/2 12AX7 plate) from "cap 1" (10uF) stage instead of the "cap 2" (20uF). Did you have a particular (practical) reason for this? The grounds of the reverb driver go to another point (20uF: cap 2). I thought it makes sense to put it to the "cap 2" stage and is also practical?

The source-follower is a buffer stage between the output impedance of the LFO triode's plate and the input impedance of the 6V6s grids. I did this in my amp because I like the intensity of bias-vary trem when its buffered. You can use a cathode follower (like a 5G9) if you have a spare triode, but in this amp there's no spare triode.

Can't wait to hear it ;-)

In terms of the footswitch, I ran a wire from the galactic ground point for the trem to the ground lug on the jack, but you could leave it out. Its probably 'better', in terms of avoiding ground loops, to isolate the jack sleeve/ground from the chassis if you do this. Over to you to experiment.

OK, then I will put an isolated RCA jack in to try! Same goes then for the reverb RCA's and speaker jacks.

In terms of the comment about frying your output tubes in fixed bias mode, I'm guessing whoever it was who said that was commenting on a version of your schematic which had a DC path to the output tube grids that bypassed the bias voltage source?

Yes, it was the former version (e.g. schematic v008) with the two 470k's that go to ground in fixed bypass mode. Does that make sense?

In my new/last layout/schematic, I probably need to play with the value for the strength of the trem. I think I will start with the 470k you used and trying also a 220k or maybe lower... What is the function of the 220k to ground you used (I also see you omitted a 1M to ground)? Also something to change?

[loudthud]

The arrangement of the speaker jacks puzzles me. What are you trying to accomplish? It looks like when you insert a mono plug it shorts the OT tap to ground and connects the feedback to the speaker.

The resistor on pin 7 of the PI should be 470K. It doesn't need to be 2W unless you just want to use a big resistor for it's fat lead wires.

[Tom Phillips]

I'll add this additional comment about the feedback.
There is no need to change the feedback connection as you change speaker impedance. When you have multiple output impedance taps you can leave the feedback connected to just one of the taps. In your case you can just connect it to the 8 Ohm tap since that is the one for which the component values in the feedback circuit where designed. The feedback will still function correctly even if the speaker load is connected to one of the other secondary taps. You can simplify your design by using mono non-shorting jacks for the speaker output.

[tubeswell]

Great, then I will go for it. I see that you have taken the B+ to the reverb driver (1/2 12AX7 plate) from "cap 1" (10uF) stage instead of the "cap 2" (20uF). Did you have a particular (practical) reason for this? The grounds of the reverb driver go to another point (20uF: cap 2). I thought it makes sense to put it to the "cap 2" stage and is also practical?

For this type of 'galactic grounding' method, the ground reruns should go to the same filter cap node as the supply for that part of the circuit.

In my new/last layout/schematic, I probably need to play with the value for the strength of the trem. I think I will start with the 470k you used and trying also a 220k or maybe lower... What is the function of the 220k to ground you used (I also see you omitted a 1M to ground)? Also something to change?

The 220k between the mosfet source and the ground is the load resistor for the buffer stage. Up to you whether you try 100k or 470k here

Not quite sure what you mean by 'omitted a 1M to ground'?

[klooon]

The arrangement of the speaker jacks puzzles me. What are you trying to accomplish? It looks like when you insert a mono plug it shorts the OT tap to ground and connects the feedback to the speaker.

and

I'll add this additional comment about the feedback.
There is no need to change the feedback connection as you change speaker impedance. When you have multiple output impedance taps you can leave the feedback connected to just one of the taps. In your case you can just connect it to the 8 Ohm tap since that is the one for which the component values in the feedback circuit where designed. The feedback will still function correctly even if the speaker load is connected to one of the other secondary taps. You can simplify your design by using mono non-shorting jacks for the speaker output.

My (simple) reasoning was that I thought it would be better to use (for instance) a 16 Ohm same tap for both the 16Ohm speaker wire and for the NFB wire. I thought the OT sees the (reactive) load based on this impedance and probably will send the ‘right’ signal (e.g. in terms of the frequency response, etc.) back in the NFB loop. But if it is just putting the right ‘voltage’ in the loop (change gain?) and the amount of feedback is designed by this voltage in combination with the feedback resistors, then it’s better (and easier) to leave them on one tap (e.g. 8Ohm for what the amp is designed for). As I was planning a variable NFB pot, I thought I could have both. I have read a bit (e.g. Aiken’s article on NFB) and from what I understand (now), it’s not common to use the e.g. 16Ohm tap for NFB then, but other than the practical reason that using a different tap will modify your amount of feedback using the same resistor configuration, I could not find a good explanation. But my knowledge is too limited here, and I don’t know what could be the difference between the “16Ohm” signal compared to the “8Ohm” signal when they are fed back into the loop, other than the difference in gain. But OK, I will keep things simple and run the NFB from the 8Ohm tap and use the mono jacks!

The resistor on pin 7 of the PI should be 470K. It doesn't need to be 2W unless you just want to use a big resistor for it's fat lead wires.

OK, I will put a 2Watt 470k for the fat wires!!

[klooon]

For this type of 'galactic grounding' method, the ground reruns should go to the same filter cap node as the supply for that part of the circuit.

Please correct me if I am wrong, but in your AA1165 pdf, I see that the supply for the reverb comes from the cap1 ("green") part (to the plate of the 1/2 12AX7), but the return is through the blue part to cap 2 node to ground in your schematic.

The 220k between the mosfet source and the ground is the load resistor for the buffer stage. Up to you whether you try 100k or 470k here

OK, is this in addition to a change in the 470k? Or should I only play with a different value for the 220k value? Is then 470k giving more or less gain compared to the 220k?

Not quite sure what you mean by 'omitted a 1M to ground'?

In the Fender AA1164 schematic, there is a 1M to ground (on the left side of the tremolo part in the schematic) close to the 0.1uF that is located close to the footswitch and that goes to the grid. In your schematic there is no such 1M, but the 0.1uF goes straight to ground and nothing goes to the grid??

[tubeswell]

Please correct me if I am wrong, but in your AA1165 pdf, I see that the supply for the reverb comes from the cap1 ("green") part (to the plate of the 1/2 12AX7), but the return is through the blue part to cap 2 node to ground in your schematic.

Well spotted - the recovery stage should be going to the green ground if that is where I took the supply from, but I can't remember. In the amp I'm pretty sure I had it going to the correct return point tho. My bad.

In the Fender AA1164 schematic, there is a 1M to ground (on the left side of the tremolo part in the schematic) close to the 0.1uF that is located close to the footswitch and that goes to the grid. In your schematic there is no such 1M, but the 0.1uF goes straight to ground and nothing goes to the grid??

I have re-drawn the schematic to correct the grounding error. (Klooon - if you keep this up, you'll definitely be kept on when the going gets tough. :-)