filter values

[M Fowler]

Take for instance the 8uf cap in a Fender Bassman 5f6a change that to 20uf and there is a big difference probably faster response time I suspect.

[iknowjohnny]

What circuit is your amp?

The PA is pretty much jcm 800 and the preamp is 3 gain stages>cathode follower>tone stack. Theres no amp i can really say the preamp is close to. It's just a design that is common but the values and arrangement is what i ended up with after a very very long time spent on the design. So i can't really say it's this circuit or that circuit.

It's well known that lower filtering in the pre-amp/PI will result in a looser feel to the amp, and higher filtering stiffens/tightens things up.

Exactly but i find that there is an optimal value beyond which it becomes too stiff when raised and too soft/sloppy when too low. But i found that optimal value varied wildly from node to node. Hence a 100UF on the V2 node that comprises the cathode follower and last gain stage and a 20uf on the PI. Huge difference, but thats how things worked out for the best. In the past I have had all nodes in the pre thru PI at 20uf for all, 80uf for all, and 100uf for all. Nothing felt as good as this. Think about it for a second.....why would it NOT make sense that they would all require different values to get the best feel and even sound to some degree? Why would they all have to be the same for the best result? Nothing else in an amp is like that, so why would filtering be any different? Maybe this is uncharted territory that should be explored, but for the life of me i don't understand why considering it worked so well in my amp.

[Ken Moon]

I'm surprised nobody has mentioned that the cap is only one component in each stage's RC filter, with the resistor being the other.

The value of the cap doesn't have much meaning without knowing the value of the resistor - for instance, a 10k resistor and 15uF cap will provide the same filtering as a 15k resistor and a 10uF cap.

Just saying you have a 10uf or 15uF cap doesn't tell you anything - I'm guessing that most folks are assuming a set resistor value (10k?).

The resistor value is chosen first, to get the desired voltage drop. The required minimum capacitance is then calculated using the formula C = 1/(2 * pi * R).

So for a 10k dropping resistor, 16uF is plenty of capacitance, but if your dropping resistor is only 1k ohm, you'll need a 160uF cap for adequate filtering.

Adding more capacitance to an adequately-filtered amp doesn't noticeably reduce ripple, or necessarily make the amp any "stiffer" either - if you heard that effect, it could be that you switched to a different cap type.

Old amps with 1k ohm/16uF RC filters are way under-filtered - if you think the ripple adds something good to your tone, cool - but most designers of modern amps don't under-filter on purpose.

One last note - for reliability, try to use 500V caps on power supply nodes over 350V or so - I've seen a lot of new amps with close to 400V on the PI and a 450V filter cap - too close for my comfort - F&T makes reasonably-sized 500V axial filter caps, if you're looking for a source.

[PCollen]

So for a 10k dropping resistor, 16uF is plenty of capacitance, but if your dropping resistor is only 1k ohm, you'll need a 160uF cap for adequate filtering.

The TW Express uses 1K between an 80uF and 40uF cap (plate and screen supply nodes). The Komet 60 uses 1k between a 40uF and a 20uF filter cap. Are these examples of inadequate filtering ?

[Structo]

[IMG:49:35]http://i23.photobucket.com/albums/b392/ ... 3270a2.gif[/img]

[iknowjohnny]

Well, i long since have accepted the voltages i have throughout the preamp after a lot of experimentation. And there is indeed one resistor that is fairly low, but that node worked best with a average amount, IE: 60UF. The other nodes are 10k and one worked best with 20UF and the other with 100UF. So i'm not sure how that theory relates. All i know is the values are what worked best and the voltages were long ago decided on.

[Ken Moon]

I'm guessing that it was desired to keep the screen voltage high in relation to the plate voltage, hence the 1K resistor - the ripple will be high on the screens with that smallish cap, but that's generally not as much of a problem in that location as it is in sensitive preamp circuits, and because most of the ripple on the screens and plates is rejected in p-p amps as common mode.

Lots of amps use a choke in place of that 1k resistor, but this has its own pros and cons.

Notice that the Express preamp nodes do have adequate filtering, with higher resistor and capacitor values than the typical vintage amp.

[FYL]

The required minimum capacitance is then calculated using the formula C = 1/(2 * pi * R).

Yes, but the frequency part is missing: your formula is good for a 1 Hz TC only.

You should aim for a low time constant, typically a decade below the lowest note to be reproduced or the lowest F3 point in the signal path, say <8 Hz for a standard guitar amp and preferably lower - <1 Hz is nice for a full-range high-fidelity amp.

Use F = 1 / (2 * Pi * R * C)

1K and 20µ give app 8 Hz, which is OK. 1K and 40µ => 4 Hz, better.

Motor-boating can be avoided by using different TCs for each node. Component dispersion is usually sufficient - a 20µ cap will measure 23µ, another one 19µ, thus the TCs will be slightly different even if the same Rs are used - but some high gain amps can only be tamed by using vastly different TCs.

Anyway, these calculations are useful when it comes to decoupling, not filtering. If you use the proper formulas for CRC or CLC filter networks, you should use a much larger cap for C2, for instance 40µ for C1 and 80µ to 120µ for C2. Yeah, I know that's counter-intuitive.

[brewdude]

Is it the value of the resistor before the cap, or the resistor after the cap that is used in this formula?

I'm surprised nobody has mentioned that the cap is only one component in each stage's RC filter, with the resistor being the other.

The value of the cap doesn't have much meaning without knowing the value of the resistor - for instance, a 10k resistor and 15uF cap will provide the same filtering as a 15k resistor and a 10uF cap.

Just saying you have a 10uf or 15uF cap doesn't tell you anything - I'm guessing that most folks are assuming a set resistor value (10k?).

The resistor value is chosen first, to get the desired voltage drop. The required minimum capacitance is then calculated using the formula C = 1/(2 * pi * R).

So for a 10k dropping resistor, 16uF is plenty of capacitance, but if your dropping resistor is only 1k ohm, you'll need a 160uF cap for adequate filtering.

Adding more capacitance to an adequately-filtered amp doesn't noticeably reduce ripple, or necessarily make the amp any "stiffer" either - if you heard that effect, it could be that you switched to a different cap type.

Old amps with 1k ohm/16uF RC filters are way under-filtered - if you think the ripple adds something good to your tone, cool - but most designers of modern amps don't under-filter on purpose.

One last note - for reliability, try to use 500V caps on power supply nodes over 350V or so - I've seen a lot of new amps with close to 400V on the PI and a 450V filter cap - too close for my comfort - F&T makes reasonably-sized 500V axial filter caps, if you're looking for a source.

[FYL]

Is it the value of the resistor before the cap, or the resistor after the cap that is used in this formula?

Before.

[Ken Moon]

brewdude, it's the resistor before the cap.

FYL, thanks for the insight - good stuff

[PCollen]

Adding more capacitance to an adequately-filtered amp doesn't noticeably reduce ripple, or necessarily make the amp any "stiffer" either - if you heard that effect, it could be that you switched to a different cap type.

.

Replacing the 16uF pre-amp node caps in a Marshall JTM45 with 32uF or 50 uf, and leaving all other things equal, noticeably changes the feel of the amp. Likewise, change the 16uF caps in a Fender 5E3 Deluxe to 32uF-50uF and see how you like it. The character of the amp changes. I'm not saying it's better or worse..that's subjective. But you can feel that the amp response has changed..this is, if you are a player familiar with the original amp.

[brewdude]

What about the previous filter nodes? How do the plate, screen & PI nodes and the resistors feedidng them effect the formula for the pre-amp node? Is there an accumulative effect?

[iknowjohnny]

You don't hear that the amp is stiffer, you feel it in the note response

exactly....just as i've been saying the whole time. My amp sounds no different than it did before, but the way it feels does sorta change the tone in your mind if you know what i mean. But in reality it's all feel. I think theres really something special in that 20uf on the PI. i used to have 20's all throughout the preamp at one time. But while i loved the squishy feel, it also made the gain stages feel too mushy. With the gain stages and cathode follower sporting big caps, the PI woth it's 20uf filter gives me back that feel but w/o the over mushiness i get with small caps everywhere else in the pre. In other words, it turns out it was apparently the PI that works well with a small cap in giving that looser feel, not the rest.

[FYL]

What about the previous filter nodes? How do the plate, screen & PI nodes and the resistors feedidng them effect the formula for the pre-amp node? Is there an accumulative effect?

Power supplies are quite complex. You may use a simplified model, Duncan-style, and only deal with two sections : filtering - aim for less than 10% ripple in PP amps - and decoupling - low TC, etc.