Fluctuating waveform ~

[Gaz]

I was asked to add an effects loop to a 70's Twin Reverb. I realized the mix point of the dry and verb signals is about perfect for effects send level, so decided to just add a buffer, and use the following stage as a return stage.

I added a simple solid state buffer that I've used before but noticed something strange when I bring up a test signal on the scope: The signal comes up then bobs up and down before settling. This happens everytime I change the gain. I've never seen this before and I'm not sure what's going on! I'm guessing it has to do with the 1uF cap charging up? I really don't know, but am guessing it's something obvious

Gaz

[uneumann]

I was asked to add an effects loop to a 70's Twin Reverb. I realized the mix point of the dry and verb signals is about perfect for effects send level, so decided to just add a buffer, and use the following stage as a return stage.

I added a simple solid state buffer that I've used before but noticed something strange when I bring up a test signal on the scope: The signal comes up then bobs up and down before settling. This happens everytime I change the gain. I've never seen this before and I'm not sure what's going on! I'm guessing it has to do with the 1uF cap charging up? I really don't know, but am guessing it's something obvious

Gaz - The idea is reasonable, but the buffer leaves room for improvement. Using an enhancement mosfet and 2 10M bias resistors means that you just added 50% of the power supply noise to your signal. I suspect you're seeing that as the bobbing. The power supply sags as you turn up the signal and you could also see AC line fluctuations. Anyway, the better solution (IMO) is to use a depletion device (like an LND150) in a follower configuration (like a triode tube). That way the power supply rejection is much higher. I suspect it will reduce hum and bobbing and act more as an ideal follower/buffer.

[Gaz]

Thanks for the reply! I may try the other follower out of curiosity, however I have used this buffer a lot before with no noise or issue (I actually don't have any noise at all right now). The power supply is very well filtered, and perhaps that's why.

Since posting I did resolve the problem, but I'm not sure why, and maybe you could help me figure it out. Wht I did was add another .022uF coupling cap and 1M resistor to ground before the grid, and now the 'bobbing' wave form is gone! I'd love to know why this worked...

[uneumann]

Thanks for the reply! I may try the other follower out of curiosity, however I have used this buffer a lot before with no noise or issue (I actually don't have any noise at all right now). The power supply is very well filtered, and perhaps that's why.

Since posting I did resolve the problem, but I'm not sure why, and maybe you could help me figure it out. What I did was add another .022uF coupling cap and 1M resistor to ground before the grid, and now the 'bobbing' wave form is gone! I'd love to know why this worked...

Sure - a well filtered supply helps. As for the added cap+resistor, you added a high pass filter that removes the low freq bobbing. The 1uf+1M has a roll off below 1Hz. The .022uf+1M rolls off ~50x higher. See if you can still observe the bobbing on one side of the .022uf cap, but not on the other.

[66 Kicks]

It doesn't matter if you use depletion or enhancement mode MOSFET's. The only difference is that one will idle with the Gate at a little lower voltage than the Source and the other will idle with the Gate at a little higher voltage than the Source.

There is no reason to idle the follower at such a high voltage because we're not dealing with a very big signal. If you use a lower idle voltage, the voltage divider for the Gate will reduce power supply noise at the same time. 10M resistors just looks like an invitation to pick up extraneous noise to me so I would spend a tenth of a mA in order to use smaller resistors.

I don't see anything inherent in your FX Loop that would cause the observed bobbing and settling.

I have attached a Twin FX Loop for your consideration.

[uneumann]

It doesn't matter if you use depletion or enhancement mode MOSFET's. The only difference is that one will idle with the Gate at a little lower voltage than the Source and the other will idle with the Gate at a little higher voltage than the Source.

There is no reason to idle the follower at such a high voltage because we're not dealing with a very big signal. If you use a lower idle voltage, the voltage divider for the Gate will reduce power supply noise at the same time. 10M resistors just looks like an invitation to pick up extraneous noise to me so I would spend a tenth of a mA in order to use smaller resistors.

I don't see anything inherent in your FX Loop that would cause the observed bobbing and settling.

I have attached a Twin FX Loop for your consideration.

Yes - In practice, an enhancement mode devices may work and people do use them in similar circuits - but I think it's misleading to say that there is no difference or that it doesn't matter. In your circuit, you've improved on the PS rejection by using 2M and 240k as the bias divider. That means you're only introducing about 10% of the power supply noise into the signal. That's still far more PS noise than a depletion device (LND150) follower circuit. You also point out quite correctly that high value bias resistors may also pick up noise. The LND150 approach does away with both of these problems. It's simply a better circuit for this purpose and that's worth pointing out to anyone that's considering a buffer for an FX jack (or any other purpose). Why choose the less ideal solution? Do you really want to add noise to your signal?

[Helmholtz]

Yes - In practice, an enhancement mode devices may work and people do use them in similar circuits - but I think it's misleading to say that there is no difference or that it doesn't matter. In your circuit, you've improved on the PS rejection by using 2M and 240k as the bias divider. That means you're only introducing about 10% of the power supply noise into the signal. That's still far more PS noise than a depletion device (LND150) follower circuit. You also point out quite correctly that high value bias resistors may also pick up noise. The LND150 approach does away with both of these problems. It's simply a better circuit for this purpose and that's worth pointing out to anyone that's considering a buffer for an FX jack (or any other purpose). Why choose the less ideal solution? Do you really want to add noise to your signal?

Maybe you should point out that the main benefit of the depletion Fet is that it doesn't require a bias voltage divider as it allows self/auto biasing. And this improves ripple rejection.

[66 Kicks]

Maybe you should point out that the main benefit of the depletion Fet is that it doesn't require a bias voltage divider as it allows self/auto biasing. And this improves ripple rejection.

Is that what he means? If so, it's not the type of MOSFET that makes the difference, it's the circuit. As long as you use the same fixed bias, it doesn't matter if you use different MOSFET's as long as they are operating in their Safe Operating Areas.

[Helmholtz]

Is that what he means?

That's how I understood.
Auto-bias only works with devices that need negative bias.

[uneumann]

That's how I understood.
Auto-bias only works with devices that need negative bias.

Right - sorry if that point about the bias resistors was not clear. A tube triode follower is cathode biased - that type of bias works for a depletion device. An enhancement device requires a bias source (like your resistor dividers on the gate). The whole point is to NOT use those resistors. Just use an LND150 with source (cathode) bias.

Edit - just use this circuit http://www.valvewizard.co.uk/accf.html but substitute an LND150 for the triode.

[66 Kicks]

Right - sorry if that point about the bias resistors was not clear. A tube triode follower is cathode biased - that type of bias works for a depletion device. An enhancement device requires a bias source (like your resistor dividers on the gate). The whole point is to NOT use those resistors. Just use an LND150 with source (cathode) bias.

Edit - just use this circuit http://www.valvewizard.co.uk/accf.html but substitute an LND150 for the triode.

Which one? Merlin has self-biased and fixed biased on that page. A "tube triode follower" is not necessarily self biased and I think this is where the confusion comes from.

There are benefits and pitfalls to both methods. What it is that you are trying to achieve along with the existing conditions frequently dictates which one.

[uneumann]

Which one? Merlin has self-biased and fixed biased on that page. A "tube triode follower" is not necessarily self biased and I think this is where the confusion comes from.

There are benefits and pitfalls to both methods. What it is that you are trying to achieve along with the existing conditions frequently dictates which one.

The first one - right at the top. It has a bias resistor and a load resistor in the cathode circuit. That works for triodes or depletion devices like the LND150.

Since you asked about it, Merlin's figure at the bottom does use a resistor divider to bias the tube like your circuit, but it has all the problems I mentioned above. For some reason, Merlin doesn't mention those, but you can find lots of other circuit design books/sites that will point out the same pitfalls I describe. It's not hard to understand if you think of the power supply as a noise source (which it is). Then those bias resistors are mixing that noise into the signal going to the gate or grid. It's just not a good idea to do that unless you have a totally clean power supply, which is rare (never?) in the case of guitar amps.

[66 Kicks]

The first one - right at the top. It has a bias resistor and a load resistor in the cathode circuit. That works for triodes or depletion devices like the LND150.

Since you asked about it, Merlin's figure at the bottom does use a resistor divider to bias the tube like your circuit, but it has all the problems I mentioned above. For some reason, Merlin doesn't mention those, but you can find lots of other circuit design books/sites that will point out the same pitfalls I describe. It's not hard to understand if you think of the power supply as a noise source (which it is). Then those bias resistors are mixing that noise into the signal going to the gate or grid. It's just not a good idea to do that unless you have a totally clean power supply, which is rare (never?) in the case of guitar amps.

I'm not advocating for either approach, but there are some circumstances where one system is better than the other.

PSRR doesn't seem to be a real big concern for a lot of FX Loop designers. If it bugs you, it's certainly possible to design a very quiet fixed biased Source follower.

[uneumann]

I'm not advocating for either approach, but there are some circumstances where one system is better than the other.

PSRR doesn't seem to be a real big concern for a lot of FX Loop designers. If it bugs you, it's certainly possible to design a very quiet fixed biased Source follower.

It doesn't bug me at all. I know which circuit I'd use. I'm just answering your question about noise (which is what bobbing is) and pointing out why the two bias methods are different in terms of how PS noise affects them.

[66 Kicks]

It doesn't bug me at all. I know which circuit I'd use. I'm just answering your question about noise (which is what bobbing is) and pointing out why the two bias methods are different in terms of how PS noise affects them.

Yeah, you would use the self-biased one because it has the better PSRR. Obviously, PSRR does bug you. Did you see how I killed PS noise with a fixed-bias driver? Not only that, but it also has an excellent all around signal-to-noise ratio. Do you see why I spec'd an enhancement mode MOSFET?

Edit: I didn't ask any question about noise.