Tried to make a gain stage, made crummy tremolo... Help?

[_ej_]

You guys have probably seen some of my posts working on a version of a Mojotone NC3015. I've stripped it down to just the brilliant channel with a single input and that's all working nicely and sounds excellent.

Since I had a triode left over from the from normal channel I'm trying to incorporate a "cascade" switch. Essentially adding the gain stage from the original normal channel in front of the Brilliant with a few voicing changes. I got it all wired up tonight and seems like a made a crummy tremolo (see video below)

I'll pull it apart and double check my wiring but I did go through it before firing the amp up and it seemed to match my schematic. Can you guys take a look at it and see if it's a schematic issue?

Im an absolute novice and just playing around so I wouldn't be shocked if I screwed something up.

[Esquirefreak]

Try separating V1a/b cathodes and give them each a separate C/R.

/Max

[_ej_]

Try separating V1a/b cathodes and give them each a separate C/R.

/Max

I was wondering if it might be that. Will do

[pdf64]

Also 3 gain stages in cascade being supplied by the same HT node isn't good practice, that type of effect can be the result.

And with cascaded gain stages, grid stoppers mounted at the socket lug for the grid, and shielded cables up to them, become very beneficial.

Also grid stoppers at clipping stages will help to mitigate bias shift / blocking distortion (farting out).

Great info on the topic https://www.aikenamps.com/index.php/grid-resistors-why-are-they-used
https://www.aikenamps.com/index.php/what-is-blocking-distortion

[_ej_]

Also 3 gain stages in cascade being supplied by the same HT node isn't good practice, that type of effect can be the result.

And with cascaded gain stages, grid stoppers mounted at the socket lug for the grid, and shielded cables up to them, become very beneficial.

Also grid stoppers at clipping stages will help to mitigate bias shift / blocking distortion (farting out).

Great info on the topic https://www.aikenamps.com/index.php/grid-resistors-why-are-they-used
https://www.aikenamps.com/index.php/what-is-blocking-distortion

Thanks I'll read through that over lunch and make some tweaks.

What would be the solution to the node issue? Grab and a cap resistor and add a node?

Also, this only tangentially related but what does the value of the resistors in a voltage divider effect? Since the math is all about the ratio not the actual value what is the difference between a pair of 100k ohm resistors cutting your voltage by half and a pair of 1ohm?

[_ej_]

So after reading that article I added some grid stoppers and removed the bypass cap from the "cascade" gain stage.

It seems to work a lot better. I'd actually like to drop the gain of this boost stage a little bit, would reducing the plate voltage be useful in that? Or would I be better with a more aggressive voltage divider?

The tone is really thick so I may drop the value of the coupling cap. Also with the gain stage engaged there is a massive background hiss, but I'm guessing that's due to the shitty amazon resistors I'm using. I need to order some higher quality ones.

Here's a clip, first bit is just the brilliant channel then halfway through I turn on the addition gain stage.
https://soundcloud.com/ejendres/nc3015-1

[pdf64]

So after reading that article I added some grid stoppers and removed the bypass cap from the "cascade" gain stage.

It seems to work a lot better. I'd actually like to drop the gain of this boost stage a little bit, would reducing the plate voltage be useful in that? Or would I be better with a more aggressive voltage divider?

The tone is really thick so I may drop the value of the coupling cap. Also with the gain stage engaged there is a massive background hiss, but I'm guessing that's due to the shitty amazon resistors I'm using. I need to order some higher quality ones.

Here's a clip, first bit is just the brilliant channel then halfway through I turn on the addition gain stage.
https://soundcloud.com/ejendres/nc3015-1

A 470k grid stopper on an input stage seems too high, it'll tend to be hissy and lose treble.
To mitigate those issues, 33k might be seen as a useful upper limit.
470k might be used on a clipping stage, eg the last common cathode stage prior to a master volume.

Also, to minimise liability to hum and noise, input stages might benefit from a fully bypassed cathode.

It's undesirable to leave V1b grid floating when it's switched out of the signal path.

[tubeswell]

I'd actually like to drop the gain of this boost stage a little bit, would reducing the plate voltage be useful in that? Or would I be better with a more aggressive voltage divider?

Reducing the resistance of the plate resistor to 100k or 50k will decrease gain. As will changing the plate resistor into split load configuration (like there is in a 6G15 preamp), as will putting a voltage divider elsewhere in the load

[pdf64]

 ... What would be the solution to the node issue? Grab and a cap resistor and add a node?

Also, this only tangentially related but what does the value of the resistors in a voltage divider effect? Since the math is all about the ratio not the actual value what is the difference between a pair of 100k ohm resistors cutting your voltage by half and a pair of 1ohm?

Yes, add another RC HT dropper and decoupler stage, to create a new 'XE' HT node, to supply V1b.

However I don't think it's a good idea to think of this as a potential divider, as a valve stage might be better thought of as having a constant current draw.
Obviously that's not absolutely accurate, especially so with resistively derived self cathode bias. Anode / screen grid voltage will play into the resulting cathode current.

So, if a new XE HT node is desired (bearing in mind it seems to work fine without that), what values for resistor dropper R and capacitor decoupler C?
Firstly, how much voltage needs to be dropped / what's the minimum or maximum tolerable voltage drop / what node supply voltage is desired?
Secondly, the nodes' RC values both ways need to achieve a full bypass down below the lowest signal frequency, preferably into the subsonics, eg corner frequency of maybe 2Hz.

[_ej_]

So after reading that article I added some grid stoppers and removed the bypass cap from the "cascade" gain stage.

It seems to work a lot better. I'd actually like to drop the gain of this boost stage a little bit, would reducing the plate voltage be useful in that? Or would I be better with a more aggressive voltage divider?

The tone is really thick so I may drop the value of the coupling cap. Also with the gain stage engaged there is a massive background hiss, but I'm guessing that's due to the shitty amazon resistors I'm using. I need to order some higher quality ones.

Here's a clip, first bit is just the brilliant channel then halfway through I turn on the addition gain stage.
https://soundcloud.com/ejendres/nc3015-1

A 470k grid stopper on an input stage seems too high, it'll tend to be hissy and lose treble.
To mitigate those issues, 33k might be seen as a useful upper limit.
470k might be used on a clipping stage, eg the last common cathode stage prior to a master volume.

Also, to minimise liability to hum and noise, input stages might benefit from a fully bypassed cathode.

It's undesirable to leave V1b grid floating when it's switched out of the signal path.

So better to ground the grid than ground the output of that stage?

[pdf64]

Just to qualify that, it's bad practice to let grids or cap plates float even just momentarily, eg during changeover switching.
Because their V DC can fly off, causing a thump / click etc when the switch contacts close and bring it back down to earth.
High value resistors, eg 10M, are typically used to pull / tether these things down to their intended DC level, typically 0V common.

[_ej_]

Just to qualify that, it's bad practice to let grids or cap plates float even just momentarily, eg during changeover switching.
Because their V DC can fly off, causing a thump / click etc when the switch contacts close and bring it back down to earth.
High value resistors, eg 10M, are typically used to pull / tether these things down to their intended DC level, typically 0V common.

Can you explain that further? The large resistor would be connected to ground?

Is the same as the 1meg to ground on all the switching jacks?

[_ej_]

So did some further tweaks. See schematic below. Still having a small issue. As it's wired currently the tone and gain is right where I want it on the initial attack but as a chord fades rings out the signal begins to fade in and out.

Do you think this is an impedance issue?

I tried a similar configuration with the cathode fully bypassed (with a 1k resistor and 0.68uf cap) and a 100k at the plate. I didn't notice any issues with it fading in and out but it had way way too much gain.

Any thoughts? I can always restore the wiring that worked and then make the voltage divider more aggressive but the tube socket is easier to access so I was trying to just tweak that wiring.

[tubeswell]

as a chord fades rings out the signal begins to fade in and out.

Is this a power chord with everything dimed? (I.e., are you driving the signal to the max?) If so this could be a bit of sag and bloom from a slightly underrated power supply- which some guys like.
If not, then a sound byte might help us understand how you’re hearing this.

[pdf64]

Can you explain that further? The large resistor would be connected to ground?

Is the same as the 1meg to ground on all the switching jacks?

Kinda, I avoid using the term 'ground', as it's rather abstract, the circuit's nominal 0V / common node is what it usually refers to. A connection to planet earth is generally immaterial to the circuit's operation. 

Anyway R15 and 36 are acting as pull down resistors https://el34world.com/charts/Schematics/files/Fender/Fender_pro_tube.pdf

... as a chord fades rings out the signal begins to fade in and out.
...

A recording demonstrating the issue may be helpful :)

[_ej_]

Can you explain that further? The large resistor would be connected to ground?

Is the same as the 1meg to ground on all the switching jacks?

Kinda, I avoid using the term 'ground', as it's rather abstract, the circuit's nominal 0V / common node is what it usually refers to. A connection to planet earth is generally immaterial to the circuit's operation. 

Anyway R15 and 36 are acting as pull down resistors https://el34world.com/charts/Schematics/files/Fender/Fender_pro_tube.pdf

... as a chord fades rings out the signal begins to fade in and out.
...

A recording demonstrating the issue may be helpful :)

I added a 1meg and it eliminated the pop, thanks for the tip.

Here's a recording. Ironically the first chord I hit after hitting the switch didn't fade at all but the next few show it pretty clearly.

[pdf64]

I'm not sure I quite identified what you're referring to, but it may be ultrasonic oscillation.
Does removing the bright cap / turning treble / cut down affect the issue?

[_ej_]

I'm not sure I quite identified what you're referring to, but it may be ultrasonic oscillation.
Does removing the bright cap / turning treble / cut down affect the issue?

Here, I had a second to make a better clip. This is the DI signal off my Hot Plate. First bit the has the added gain stage and I shut it off of the second half. You can even see in the waveform how it has a tremolo effect on the ring out. This is with all the controls at noon. Without the boost it all works fine.

I tried the extremes of the cut control and it didn't make a difference.

https://on.soundcloud.com/k79ZYqwGR7ltDjQds1

[AlNewman]

It might be worthwhile considering changing the switchable gain stage to the second stage of the amp, rather than the first stage.

[_ej_]

It might be worthwhile considering changing the switchable gain stage to the second stage of the amp, rather than the first stage.

What difference would that make?

[pdf64]

... it has a tremolo effect on the ring out. This is with all the controls at noon. Without the boost it all works fine.

I tried the extremes of the cut control and it didn't make a difference. ...

Sorry, it's not apparent to me that the beating effdct is due to the circuit causing modulation, rather than just being the beating effect of slight tuning imperfections magnified by lots of gain.

This is why single frequency test signals are used.

If you're going to build / repair / modify amps, some means to properly test their operation becomes a pressing requirement.

In the meantime, what controls exacerbate the effect, what's the worst you can get it?

Your added stage introduces a lot of gain. A compely designed, properly functioning amp should be unconditionally stable at any control setting.

[ac427v]

What a fun project! I'm happy to be one more cook adding salt to the soup.

High gain amps need to add a moderate amount of gain with each stage. It is especially important to minimize the bass frequencies being amplified/distorted. V2a is now the third stage of gain and it has nothing limiting its gain. That is too much--especially at the end of the preamp.

You can test my theory by unsoldering one end of the 25 uf cap on the cathode of V2a. If that solves your problem, take that cap out. Alternatively, you can experiment with replacing the 2a cathode cap with a smaller one that only boosts high frequencies. 1 uf might sound good.

[_ej_]

What a fun project! I'm happy to be one more cook adding salt to the soup.

High gain amps need to add a moderate amount of gain with each stage. It is especially important to minimize the bass frequencies being amplified/distorted. V2a is now the third stage of gain and it has nothing limiting its gain. That is too much--especially at the end of the preamp.

You can test my theory by unsoldering one end of the 25 uf cap on the cathode of V2a. If that solves your problem, take that cap out. Alternatively, you can experiment with replacing the 2a cathode cap with a smaller one that only boosts high frequencies. 1 uf might sound good.

That's a good idea, I'll give it a shot.

I'm really not trying to add that much gain. More along the lines of running a tube screamer out front of the amp.

I figured plexis cascade their channels so I could do the same thing.

[_ej_]

so I just tried restoring the cathode bypass wiring on the switchable gain stage. It obviously added tons more gain and it seemed to help but if you let a chord ring long enough the oscillation comes back. So it seems like once the signal drops to a certain point with the switchable stage in it started to oscillate.

I also tried the extremes of all the controls and it doesn't seem to effect the oscillation either way. To my mind this further indicates its something happening with the switchable stage wiring.

I've tried a bunch of things to try and get the amp to oscillate with the gain stage switched off and I can't make it happen.

Any other thoughts? Do you guys think making it the switchable stage the second stage instead of the input stage would help? I may just try that regardless. I'll have access to a scope and a signal generator at work this week but I wanted to get this buttoned up and cleaned up this weekend.

What a fun project! I'm happy to be one more cook adding salt to the soup.

High gain amps need to add a moderate amount of gain with each stage. It is especially important to minimize the bass frequencies being amplified/distorted. V2a is now the third stage of gain and it has nothing limiting its gain. That is too much--especially at the end of the preamp.

You can test my theory by unsoldering one end of the 25 uf cap on the cathode of V2a. If that solves your problem, take that cap out. Alternatively, you can experiment with replacing the 2a cathode cap with a smaller one that only boosts high frequencies. 1 uf might sound good.

I tried slapping a 1k/0.68uF on the cathode of V2a since I had that wired up on a lead from the experiments I was doing with V1b. It definitely cleans stuff up and dropped the overall gain level but didn't change the oscillation.

Does sound nice though, I may keep it.

[pdf64]

I suggest to add another HT dropper stage XE to supply the added boost circuit.

[_ej_]

I suggest to add another HT dropper stage XE to supply the added boost circuit.

Worth a shot! Will do.

[pdf64]

eg 10k 16uF.
The dropper resistor will have less than 20V across it in normal operation, but best to use a 1W 500V rated component, preferably metal oxide, to cope with power ip conditions.

[_ej_]

eg 10k 16uF.
The dropper resistor will have less than 20V across it in normal operation, but best to use a 1W 500V rated component, preferably metal oxide, to cope with power ip conditions.

Yeah, I have some 1watt stuff leftover from a few previous builds. Even with a new node the issue is occuring. So I'm going to take it in to work tomorrow and see if I can pin point where in the circuit the oscillation starts and go from there.

I am tempted to remove the switch, divider and bright cap from the circuit just to see if any of that is causing the issue.

[AlNewman]

The reason I suggested switching the second stage is so you have a consistent input for your guitar.  In your last drawing, you're basically building a cold clipper stage at the input, in series with the "first" (high gain) stage.  There's also different things happening with the grid leaks of both stages.  If you add a stage in between 2 of the following stages, it could give you more flexibility to dial in the amount of gain the way you like.

If you are going to add a cold clipper, you could (maybe) add a switchable stage after the input stage, and build it closer to a JCM800.  These amps also had an extra filtering stage, one for each of the tubes in the 4 stage pre amp.

The difference between what you're trying to do, and what a plexi does by jumpering channels, is a plexi parallels the input stages, and you are trying to run them in series.

That being said, in my opinion, if you are looking for a tubescreamer type boost, probably the easiest way is to make a switchable bypass cap.

[_ej_]

The reason I suggested switching the second stage is so you have a consistent input for your guitar.  In your last drawing, you're basically building a cold clipper stage at the input, in series with the "first" (high gain) stage.  There's also different things happening with the grid leaks of both stages.  If you add a stage in between 2 of the following stages, it could give you more flexibility to dial in the amount of gain the way you like.

If you are going to add a cold clipper, you could (maybe) add a switchable stage after the input stage, and build it closer to a JCM800.  These amps also had an extra filtering stage, one for each of the tubes in the 4 stage pre amp.

The difference between what you're trying to do, and what a plexi does by jumpering channels, is a plexi parallels the input stages, and you are trying to run them in series.

That being said, in my opinion, if you are looking for a tubescreamer type boost, probably the easiest way is to make a switchable bypass cap.

That makes way more sense that they're in parallel.

Would wiring it up that way be a better plan? I think I could really rework it in parallel

[AlNewman]

Try it and see.

[_ej_]

Trying this wiring tonight. Anyone see any issues?

[pdf64]

V1a is missing its grid leak resistor.

What's the benefit in switching the input to one channel or the other, why not leave them both connected to the input?

I suggest to mix the channels before V2a, as per eg 5F6A and most classic 2 channel amps https://el34world.com/charts/Schematics/files/Fender/Fender_bassman_5f6a.pdf

[Esquirefreak]

I'd probably leave the cold clipper K unbypassed.

/Max

[_ej_]

V1a is missing its grid leak resistor.

What's the benefit in switching the input to one channel or the other, why not leave them both connected to the input?

I suggest to mix the channels before V2a, as per eg 5F6A and most classic 2 channel amps https://el34world.com/charts/Schematics/files/Fender/Fender_bassman_5f6a.pdf

I'm trying to switch in the second channel similar to when you jumper a Plexi. I this configuration you should have switch position 1 should just be the one channel, with the switch active it should be jumpered.

Why mix earlier?

[SEL49]

The top channel flips the signal phase twice, but the lower channel only flips once. IOW, the two channels will be out of phase at the mixing resistors. Mixing out of phase signals will subtract the two signals rather than add (reinforce) the two channels. This usually results in a thin crappy sound. Kinda like jumping the two channels on a stock AB763 amp.

[AlNewman]

https://el34world.com/charts/Schematics/files/Marshall/Marshall_marshall1959.gif

Here's one version of the plexi, you may want to follow their idea a little more closely, at least to start.

There's 2 volume controls, but you would need to have a fixed mixing resistance at your switchable stage, which you'll probably have to dial in to taste.  Or you could follow the idea of the November amp, and have 2 volumes instead of a switch.

https://el34world.com/charts/Schematics/files/Sluckey/November_01.pdf

If you try and mix the signals where you have it, I don't think it will work due to phase issues.

[_ej_]

https://el34world.com/charts/Schematics/files/Marshall/Marshall_marshall1959.gif

Here's one version of the plexi, you may want to follow their idea a little more closely, at least to start.

There's 2 volume controls, but you would need to have a fixed mixing resistance at your switchable stage, which you'll probably have to dial in to taste.  Or you could follow the idea of the November amp, and have 2 volumes instead of a switch.

https://el34world.com/charts/Schematics/files/Sluckey/November_01.pdf

If you try and mix the signals where you have it, I don't think it will work due to phase issues.

Do you think it would be better to mix the signal prior to the volume pot?

Also, in the November schematic it looks like there's a pair of 470k ohm resistors where the signal comes together.

When I bypass the parallel channel will that act as a voltage divider for the main channel? Do I need to switch those resistors out of the circuit?

[AlNewman]

If you don't mix the signal before the volume, when you switch your parallel stage you'll have no volume control in that stage.

You'll likely need some sort of mixing resistor setup to get the desired effect.

As it's drawn, depending on where you put the switch, there could be a voltage divider when the gain stage is off. With an SPST switch, I think it would be best to put it after the gain stage, maybe with a 1M to ground, and a mixing resistor to taste, with the switch between the mixing resistor and resistor to ground.  Then have another resistor directly off the coupling cap of the 1st stage, with the signals mixed directly before the volume pot.

There's probably better ways to do it, but that would be my suggestion to start.  Others have much more experience with this type of thing.

[_ej_]

If you don't mix the signal before the volume, when you switch your parallel stage you'll have no volume control in that stage.

You'll likely need some sort of mixing resistor setup to get the desired effect.

As it's drawn, depending on where you put the switch, there could be a voltage divider when the gain stage is off. With an SPST switch, I think it would be best to put it after the gain stage, maybe with a 1M to ground, and a mixing resistor to taste, with the switch between the mixing resistor and resistor to ground.  Then have another resistor directly off the coupling cap of the 1st stage, with the signals mixed directly before the volume pot.

There's probably better ways to do it, but that would be my suggestion to start.  Others have much more experience with this type of thing.

What do you think of this?

Edit: since I have a DPDT I changed things to switch out the mixing resistors. I'll be trying this tonight

[AlNewman]

Personally, I don't think that would be the ideal approach.

For one, the output of your gain stage will always be mixed in wherever the switch, and you're also adding in more attenuation with the gain disengaged.

Some of the extra peaker caps and such can be added or adjusted once you have your ratios correct. 
Anyways, you can try it, it won't hurt anything, and you'll learn as you go.  When you're modding an amp, it's all about trial and error.  Just think about the signal path and the most efficient way to obtain your goal.

As I mentioned, there are way smarter people than me who have already posted here.

[AlNewman]

Just to keep it simple, I would probably start with something like this.

[ac427v]

I suggest you use an already invented wheel, the Hoffman TMB Stout with the "Stout Optional Gain-stage Switch".
He placed the new gain-stage and DPDT switch between the first stage and the V2 Cathode Follower/Tone Stack. You would need to adapt it to your desired parts values.
This would be far more versatile if you also added another volume control.
https://el34world.com/Hoffman/files/Hoffman_Stout.pdf

[_ej_]

I suggest you use an already invented wheel, the Hoffman TMB Stout with the "Stout Optional Gain-stage Switch".
He placed the new gain-stage and DPDT switch between the first stage and the V2 Cathode Follower/Tone Stack. You would need to adapt it to your desired parts values.
This would be far more versatile if you also added another volume control.
https://el34world.com/Hoffman/files/Hoffman_Stout.pdf

That is a good call introducing the gain stage in series there is working. I'm not getting the tremolo effect (at least as far as my midnight testing allowed me to check)

But it's absolutely crushing the next stage. I need to add a fairly aggressive divider I think.

[Esquirefreak]

Add a 1M pot to figure out what value resistors you need for the divider?

/Max

[ac427v]

The Marshall JCM 800 can be a good learning tool for how to balance voltage gain, distorted tone and clean tone. RobRob has a great discussion of this and the schematic tracing the signal path from the second input through the gain stages.  https://robrobinette.com/Tube_Guitar_Amp_Overdrive.htm