6G2 effect loop issue

[Leevi]

I'm trying to add an effect loop to a 6G2 based circuit and met an issue with the NFB.The NFB is controlable with pot and it works fine if the loop is OFF. When I activate the loop it seems that the NFB control works in an opposite way. With the minimum setting (56K=biggest NFB signal) the circuit starts to oscillate.If I cut the NFB or turn the pot fully open (i.e. max NFB resistor) the oscillation disappears. It seems that the NFB cannot be used in that setup at all. It has probably something to do with a positive feedback? I have checked and tested the plate wires that they are correctly wired.
Please check the circuit.
EDIT: The RETURN level control is not needed because there is already a master volume in the same node. I'm planning to move it to SEND.
/Leevi

[Lauri]

You have the effects loop inside NFB loop. When you activate the loop phase flips 180 degrees and the feedback becomes positive.

[Leevi]

You have the effects loop inside NFB loop. When you activate the loop phase flips 180 degrees and the feedback becomes positive.

Yes, that was my assumption too. A simple solution is to cut the nfb when the loop is activated or do you have any other ideas?
/Leevi

[Lauri]

I would move the effects loop so it's before the NFB loop.

[glass54]

Hi Leevi
Would you consider using a LTP Phase inverter (Uses 3 full 12a*7 tubes, could use a 12AT7 for PI) and your problems are over?? 
Proven PA design  BUT I do realise it becomes a slightly different animal to a 6G7 PA.
Feed Master Vol to C22 (10nF) at PI input?
Regards
Mirek

[Leevi]

Hi Leevi
Would you consider using a LTP Phase inverter (Uses 3 full 12a*7 tubes, could use a 12AT7 for PI) and your problems are over?? 
Proven PA design  BUT I do realise it becomes a slightly different animal to a 6G7 PA.
Feed Master Vol to C22 (10nF) at PI input?
Regards
Mirek

One option of course but in my case I should add an extra tube. I still have to add a bias tremolo which is not drawn in the schematic I attached.
/Leevi

[pdf64]

It's just an irredeemably awful design, sorry.
I suggest to put it down to experience and start over.
Starting with putting the switch and fuse in the live feed.

[Leevi]

Starting with putting the switch and fuse in the live feed.

What switch and fuse do you mean?
/Leevi

[pdf64]

Starting with putting the switch and fuse in the live feed.

What switch and fuse do you mean?
/Leevi

230V AC 50Hz mains power.

[Leevi]

Starting with putting the switch and fuse in the live feed.

What switch and fuse do you mean?
/Leevi

230V AC 50Hz mains power.

There are both switch and fuse. I noticed that one couping cap is missing from the first preamp tube which is a drawing error.Anaway the whole schematic is fast made just to illustrate the issue I took up and I will update it.If you have concrete suggestions for improvements I will gladly accept them.
/Leevi

[kagliostro]

Ciao Risto

Only something to give a try, not sure you'll be satisfied

Why dont try to connect (using a switch) the NFB to the cathode of the recovery stage of the FXLoop when it is in use, when you donìt use it connect it to V1b (as from the original plan), help to arrange the NFB level to the cathode of the return tube of the FXLoop consider to add also there a trimmer or pot, the 5K resistor will be always connected to the OT



Or, if you prefer you can put the switch before the NFB pot and use only a pot (this way you have to arrange for each configuration when you switch)



Franco

[Leevi]

Only something to give a try, not sure you'll be satisfied

Why dont try to connect (using a switch) the NFB to the cathode of the recovery stage of the FXLoop when it is in use, when you donìt use it connect it to V1b (as from the original plan), help to arrange the NFB level to the cathode of the return tube of the FXLoop consider to add also there a trimmer or pot, the 5K resistor will be always connected to the OT

Hi Franco and thank you for the tip
I used a DPST switch and made some tests. Your proposal works if the cathode bypass cap is removed.The cap removes the NFB of the stage so affect of the NFB control pot is lost.
/Leevi

[kagliostro]

Give a look on how is arranged the NFB connection to the  cathode on the Princeton Reverb AA1164

May be that way you can solve the problem ?

Franco

[pdf64]

Starting with putting the switch and fuse in the live feed.

What switch and fuse do you mean?
/Leevi

230V AC 50Hz mains power.

There are both switch and fuse. I noticed that one couping cap is missing from the first preamp tube which is a drawing error.Anaway the whole schematic is fast made just to illustrate the issue I took up and I will update it.If you have concrete suggestions for improvements I will gladly accept them.
/Leevi

You're in Finland, so non polarised mains connectors, shuko maybe?
Aren't double pole power switches required?

What is the point of the cathode follower, why have you put it in there, what benefits is it providing?

[Leevi]

You're in Finland, so non polarised mains connectors, shuko maybe?

Yes, non-polarised and shuko

Aren't double pole power switches required?

Finland is mainly following standards used in EU.It's up to devices which kind of switches can be used.I prefer double pole switches.

What is the point of the cathode follower, why have you put it in there, what benefits is it providing?

I'm just following the good practises how the buffered effect loops have been designed before.
This particular implementation is based on Merlins FX-loop.
https://el34world.com/Forum/index.php?topic=22726.0
/Leevi

[Leevi]

Give a look on how is arranged the NFB connection to the  cathode on the Princeton Reverb AA1164

May be that way you can solve the problem ?

Franco

It helps a bit but the affect of the NFB control pot is not that big as when leaving the bypass cap totally away.
/Leevi

[kagliostro]

What happens if you connect the NFB to the 16ohm secondary instead of the 8ohm and change the pot and resistor with a lower value adopting different values for the standard and the FXLoop ?


Franco

[Leevi]

What happens if you connect the NFB to the 16ohm secondary instead of the 8ohm and change the pot and resistor with a lower value adopting different values for the standard and the FXLoop ?


Franco

That will certainly change the situation. I got the tremolo wired and I'm very satisfied with the current setup and I'm not any more eager to do some extra mods/tests.Anyway thank you for the proposal. Maybe in the next project;)
/Leevi

[pdf64]

Finland is mainly following standards used in EU.It's up to devices which kind of switches can be used.

Are you sure about that? I suspect that it may be set by the device's risk catagory, and that equipment like a guitar amp, with metalwork exposed to the user, will be required to have that metalwork earthed, obviously, but also to have both incoming mains feeds to be isolated by the device's power switch, as either could be live.

I'm just following the good practises how the buffered effect loops have been designed before.
This particular implementation is based on Merlins FX-loop.
https://el34world.com/Forum/index.php?topic=22726.0
/Leevi

But you've made a change to that design, adding a 10k build out resistor.
Hence the signal impedance at the send node will be 10k.
The signal impedance at the CF grid is 18k, and could easily be halved if eg the 1M was halved to 470k and the 18k to 9k1.
So the 10k build out resistor has been rendered the CF pointless.

[Leevi]

Finland is mainly following standards used in EU.It's up to devices which kind of switches can be used.

Are you sure about that? I suspect that it may be set by the device's risk catagory, and that equipment like a guitar amp, with metalwork exposed to the user, will be required to have that metalwork earthed, obviously, but also to have both incoming mains feeds to be isolated by the device's power switch, as either could be live.

I'm not 100% sure but there are lot of different devices on the market that have single pole switch even new devices. As I said it's depending on the device and its security classification. This is not an issue at all if you always use double pole switch.

I'm just following the good practises how the buffered effect loops have been designed before.
This particular implementation is based on Merlins FX-loop.
https://el34world.com/Forum/index.php?topic=22726.0
/Leevi

But you've made a change to that design, adding a 10k build out resistor.
Hence the signal impedance at the send node will be 10k.
The signal impedance at the CF grid is 18k, and could easily be halved if eg the 1M was halved to 470k and the 18k to 9k1.
So the 10k build out resistor has been rendered the CF pointless.

I have done changes after I posted the preliminar schematic.Among others I dropped the 1M resistor to 100K and the 18K to 15K.Then the fixed resitors on the SEND have been replaced with a 1M pot.
/Leevi

[pdf64]

I have done changes after I posted the preliminar schematic.Among others I dropped the 1M resirtor to 100K and the 18K to 15K.Then the fixed resitors on the SEND have been replaced with a 1M pot.
/Leevi

Both those changes seem ill conceived to me, sorry.
The increased signal level at the CF grid may cause grid current clipping on the upper lobes.
The 1M send pot will massively increase output impedance, unless it's set to maximum.
Set midway, it'll be about 250k.
There's good reason why the schematic 2deaf posted used a 10k pot there.

[kagliostro]

If I remember well Mr. Merlin, the output will be way better if of low resistance, so the 10K were good to me

Franco

[Leevi]

I have done changes after I posted the preliminar schematic.Among others I dropped the 1M resirtor to 100K and the 18K to 15K.Then the fixed resitors on the SEND have been replaced with a 1M pot.
/Leevi

Both those changes seem ill conceived to me, sorry.
The increased signal level at the CF grid may cause grid current clipping on the upper lobes.
The 1M send pot will massively increase output impedance, unless it's set to maximum.
Set midway, it'll be about 250k.
There's good reason why the schematic 2deaf posted used a 10k pot there.

So far no clipping has occured with those values. Have to do still some tests. The signal level goes easily down in the loop circuit. For instance if you remove the bypass cap you have to compensate it on the send side, otherwise the loop is a damper.


/Leevi

[Leevi]

I have done changes after I posted the preliminar schematic.Among others I dropped the 1M resirtor to 100K and the 18K to 15K.Then the fixed resitors on the SEND have been replaced with a 1M pot.
/Leevi

Both those changes seem ill conceived to me, sorry.
The increased signal level at the CF grid may cause grid current clipping on the upper lobes.
The 1M send pot will massively increase output impedance, unless it's set to maximum.
Set midway, it'll be about 250k.
There's good reason why the schematic 2deaf posted used a 10k pot there.

I made some tests on the SEND side:
Voltage divider 470K/10K does not work at all. I cannot reach the loop bypass level even if the 1M pot is fully open.I tried with 220K/10K and the signal level is about the same as the bypass signal if the 1M pot is fully open.10K pot would not help here
The earlier setup (100K/15K) the pot setting 2 o'clock corresponded about the bypassed signal level.And rest worked as boost. Now the question is why to have the level pot at all.
/Leevi