Power Supply filter design question

[vibrovox7reverb]

I have a question about the design of the power supply of my Vibrolux Reverb. Instead increasing the value from the two 70 uf caps in series to two 100 uf caps in series, why not do a 40 uf cap and then a small resistor and another 40 uf cap as the mains filter? They would be rated for 600 volts of course (my amps DC is about 430 at the mains). Would this give better filtering and tighter low end than even the two 220 uf caps in series? I know this is not a new idea, but I'm wondering if it's safe enough, and out of curiosity if it would give better filtering than just one filter stage that had higher capacitance than would the two separate stages combined.

[J M Fahey]

Where would you buy 600V 40uF caps?
Just thinking aloud.
And no, your hypotethical 40uF cap is way smaller than the 110uF equivalent you´ll get with 2x220uF caps in series.

[vibrovox7reverb]

I saw some available at Weber's site. Maybe not the best caps, but the others in my amps have been fine so far. Not sure how large the 40 uf 600v will be though.

Anyway, I meant to say that I will have two 40 uf caps, one before the OT centertap, and one connected to the OT centertap with a small resistor between them. I only have room for 5 e caps in the cap can, and I'm trying not to exceed the 5AR4's capacitance rating. I was asking about how the filtering would be with 2 sections of 40 uf as opposed to one section of 110 uf.

Either way I'll have more capacitance than I really need, but since I'm attempting to tighten up things in my amp while still retaining the rectifier tube (not necessarily trying to speed things up) I prefer not to use the 2x220 uf series setup, but if it will tighten up the amp a significant amount then I may just use it and switch to a copper cap.

[R.G.]

I did some simulation of the various cases. They all give a similar amount of ripple, except when the "small resistor" between the 40uF caps is either very big or very small. Very small gives you the expected result for 80uF, and very big gives vanishing small ripple - but also loses you the V = I*R drop you'd expect from the current that passes through it.

That is, it gives better filtering, but worse DC drop.

What you're fighting of course is the pulse current from huge first filter caps versus making AC ripple smaller. If you have a tube rectifier, this is a big deal. Solid state diodes make it not matter. Putting an inductor between the first and second caps gives you low DC drop and also smaller ripple. But then you have to put in an inductor.

[vibrovox7reverb]

I did some simulation of the various cases. They all give a similar amount of ripple, except when the "small resistor" between the 40uF caps is either very big or very small. Very small gives you the expected result for 80uF, and very big gives vanishing small ripple - but also loses you the V = I*R drop you'd expect from the current that passes through it.

That is, it gives better filtering, but worse DC drop.

What you're fighting of course is the pulse current from huge first filter caps versus making AC ripple smaller. If you have a tube rectifier, this is a big deal. Solid state diodes make it not matter. Putting an inductor between the first and second caps gives you low DC drop and also smaller ripple. But then you have to put in an inductor.

Hmm, I could drop SOME DC but I'd rather not lose too much or I'm not going to tighten up the amp at all lol. Also I don't want to fork over the dough, nor do I have the room to get a choke able to handle the current requirements of the power tubes. Hmm...if I don't have enough spare voltage to use the size resistor I need does that mean that I'll end up creating to much pulse current for my rectifier tubes despite my attempts not to? *&^%

I guess the standard totem stack of two 100uf caps will work unless I'm gonna go solid state, which I'm now contemplating as I want to get more tightness anyway. Thanks for the help RG, I'm pretty green at this stuff so every bit of advice I get is invaluable.

[Chuck H]

If you don't have Duncan PSUD2 you should download it here:

PSUD2

I ran some rough simulations and it looks to me like a 40u-100R-40u pi filter and a diode rectifier will give the same voltage and sag as the totem 100u caps and a GZ34 rectifier, BUT, the pi filter/diode arrangement will have less than one third the ripple of the totem/tube arrangement. In fact, the 40u-100R-40u pi filter is better than a totem pair of 220u caps. The pi filter also has the added benefit of lower ESR. The drawback to the pi filter is that it does require a resistor, and therefor sag is induced. If you plan to stick with the tube rectifier using the pi filter arrangement will ADD sag.

Being as my simulations were done to achieve the same voltage, and sag was the same too, I can't say there is any advantage to the pi filter for tightening bass response. But used with a diode rectifier it has much better ripple rejection and lower esr.

[vibrovox7reverb]

Thanks for the download link Chuck!

I had planned on keeping the rectifier tube solely for the benefit of it's warm up time after reading about the horrors of standby switches. However, it may be worth a try to go solid state if I can reduce the ripple that much. The amp's level of ripple is not bad now (totem stack of for 50uf as of now), but I guess it never hurt to try, and I can alway just keep the volume on my amp turned down until it's on, or just use the standby as I've never experienced a standby switch related problem before anyway.

I'm gonna use the program to see what I can do to increase the voltage and decrease sag a bit as well, because it may be what I'm looking for anyway. Just gonna have to experiment.

Thanks again for the link, and for running the sim for me...thanks to both you and RG for doing that. I really appreciate it

[Chuck H]

Welcome. You may have noticed at the site already, but there's a tone stack simulator there as well... Very cool tools. To get reasonably accurate PSUD2 readings you basically need to program in the antire power supply complete with current taps for the expected loads.

[vibrovox7reverb]

I didn't notice the Tone Stack Simulator actually, thanks for pointing it out. Lol I've been having a bit a trouble figuring out exactly how everything works on this, but I'll get it down. Thanks for the tip about the loads, btw do I need to enter the load for each stage, or just one load with the cumulative value at the end? I figure the load for each stage, but just want to check to make sure I'm not messing up results.

[Chuck H]

You do need to insert a current tap for each stage. For the load at the end of the chain I usually replace the resistor with a "constant current source" as the current load for that stage. None of my amps (and none that I know of) have a load resistor at the end of the HV rail.

[vibrovox7reverb]

You do need to insert a current tap for each stage. For the load at the end of the chain I usually replace the resistor with a "constant current source" as the current load for that stage. None of my amps (and none that I know of) have a load resistor at the end of the HV rail.

Ahh, ok thanks. I downloaded the tone stack simulator...I had no idea that Fender tone stacks behaved that way! So weird! I wish there was one for the brownface Vibroverb tone stack, as that's the tone stack I currently have in my amp, and I'd like to see how to tweak it to get the best response.

[Chuck H]

The "Fender" tone stack is the AB763 model. It's almost the same as yours with the exception of the 70k tap on the treble pot and the bleed cap off the bass and mid side of the stack. What exactly these two differences will make I can't say. But in broad strokes it means your stack probably has a sharper knee in the upper mid cut. Just set the mid pot on the simulator at 6.8k and leave it. That will simulate the fixed resistor in your stack.

[J M Fahey]

Software simulators are excellent design tools, and huge timesavers too, but you can still do it the old way: actually build the circuit you are interested into, play a few chords, and tweak some values until pleasant to your ear.
State of the art design.
And I´m not kidding.
Let´s not have the tree hide the forest.

[Chuck H]

Oh, absolutely. It means a lot less to the eyes than to the ears anyway.