Why use electrolytic in this Fender reverb circuit?

[TUBEDUDE]

RG. So what’s the deal with non polarized electrolytics ?

Polar electros oxidize one of two foils, the other being left bare. The bare aluminum foil serves only to make good contact to all of the electrolyte, which is the actual conductor on one side of the oxide. The oxidized foil is the other conductor, and the oxide layer is the insulation.

For non-polar, they oxidize both foils. That way no matter which way the voltage is, one foil is protected by a properly-biased oxide layer, and the other is running right at the edge of conduction. With an alternating signal, neither oxide layer gets abused very much.

Ideally, non polar should not be used with a DC bias, as the smallish current through the "incorrect" oxide layer lets it degrade, and eventually you have a (mostly) polarized cap. But they will last a long time anyway.

We can fake the both-sides-polarized by hooking two polarized ones back to back. You get simuiar electrochemical action, but use two caps to do it. The oxide insulated layer is the nominally positive one in a polarized cap, so hooking the two negative leads together is a near approximation of an NP.

Just another reason to avoid electrolytics where space permits. Unless you manufacture on a large scale on thin margins..

[R.G.]

Just another reason to avoid electrolytics where space permits. Unless you manufacture on a large scale on thin margins..

I agree. It just might be possible to replace them all. I typed in a lot of stuff, and realized that its better as a separate post. See the thread on the Immortal Amplifier.

[TUBEDUDE]

Just another reason to avoid electrolytics where space permits. Unless you manufacture on a large scale on thin margins..

I agree. It just might be possible to replace them all. I typed in a lot of stuff, and realized that its better as a separate post. See the thread on the Immortal Amplifier.

I normally use all metalized film in the power supply and cathode caps. The one cap I have trouble with is the bias filter.

[R.G.]

In re: film for bias supply
I did some tinkering with what a reasonable cost-benefit range is for films in the bias supply, using a Fender Twin bias supply as a starting point.

A quiet bias voltage is a general Good Thing. Sure, in theory, ripple on the bias supply is cancelled out by the push-pull nature of the typical output stage, but this works only to the degree that the AC gains (not DC bias gain) of the two output tubes are matched, and only works to some extent near zero signal. I used a regulated bias source in the Workhorse amps, and this made for a very quiet bias supply. I think this contributed to the nearly no hum in them. But regulation is a pain, costs parts and design time, and may be hard to hack into existing amps. I looked instead for what I could do with film caps to make a quiet power supply that didn't use an electrolytic filter cap on the bias.

In the pictures that follow, the red flat-line trace is the DC bias voltage at the bias filter cap; it's not very interesting, as it just shows a flat DC - which is good!
The sawtooth red trace is the AC voltage across the first bias filter cap. It keeps the same sawtooth shape, but varies in size with the first filter capacitance.
The green trace is the AC voltage from the top of the wiper pot to ground. It's shape varies as filtering from top of wiper pot to ground goes from none to some capacitance.
The blue trace is the AC ripple at the wiper pot. The wiper stays fixed at 50% for all the variations.

1. Stock Fender Twin Bias Circuit
Generates -65V on bias filter cap, -51V at middle of bias pot.
AC ripple at bias cap is 291mV, and 231mV at wiper of bias pot

Fender Twin Bias Circuit.png

2. Added 2.2uF/100+V film cap from top of bias pot to ground lowers Vbias ripple to 61mV at the wiper. A suitable 2.2uF cap at Mouser is 667-ECQ-E1225JF, $1.04.

Fender Twin Bias Circuit w 2.2uf.png

3. Adding a second 2.2uF at the pot wiper drops ripple to 12.9mV.

Fender Twin Bias Circuit w 2.2-2.2uf.png

4. Trying to go all film, no electro: 10uF first cap, 10uF second cap gives 139mV at the pot wiper. A suitable 10uF film cap at Mouser is 667-ECQ-E1106KFZ, $2.75 in single quantity.

Fender Twin Bias Circuit 10u-10u.png

5. To save some money, we can use a 10uF first cap, 2.2uF second cap, and 2.2uF wiper cap. This combination drops the ripple to 121mV at the wiper, but costs only $4.83.

Fender Twin Bias Circuit w 10u-2.2-2.2uf.png

It’s entirely possible to pick your own poison. If you don’t need to get rid of electros, it looks like putting a 2.2uF cap from the top of the bias pot to ground will massively cut the ripple on the bias voltage wiper. This might just be a good thing for most amps with fixed bias.

If you want to go all film, you could buy one 10uF/100V film cap for the first filter cap at $2.75 (or similar, depending on your supplier) then put one or two more 2.2uF ($1.04-ish) film caps, one at the top of the wiper pot to ground, and another on the wiper of the bias pot.

[nuke]

To carry on the digression,

871-B32523R226K

Is interesting, film cap, 22µf at 63v, only $5.00 in qty 1. That's more than 10x more expensive than a good quality radial electrolytic of the same value. Also pretty big.

Then there's a whole range of capacitor types, such as polymer and polymer hybrids and so on. For a $1.70 qty 1, you can get a teeny-tiny SMD type 22µf 25v, with a 15,000 hour 105c life, and -55 to 105c operating rage. APXA250ARA220MH70G (Chemi Con)

At some point, you just have to figure out what problem you're trying to solve.

[R.G.]

That is indeed the question! What problem needs solved?

I was solving the question of how to convert the electrolytic filter cap in the bias supply over to film types, which have a hugely longer usable life without replacement. This is a problem of practicality: how to find a suitable replacement that is small enough and does not break the budget?

It's easy enough to find a 100uF/100V film capacitor. Practically, though, these are on the order of $20 to $30 or more at Mouser (my standard for can I get it and what does it cost) and physically large. I pondered the idea of tinkering with the circuit so I could use more reasonably sized and less costly film caps. So I set up a model of the Twin bias supply to see what performance it gives. The 230mV of ripple seemed unreasaonably big to me when I did that. Hold this thought; we'll come back to it.

I scanned the prices of 10uF and up film caps and tried to pick out any sweet spots in terms of uF per dollar, coming down to 10uF and 22uF as reasonable (-ish) prices. I mentally trades off paralleled 10uFs (more expensive than one 22uF), and series-parallel setups of lower voltage but smaller caps (still came out more expensive).

I wondered how much ripple I'd get with a 22uf first cap. Yep, theory is right, it gives about five times the ripple of a 100uf. Not good. So I decided to try adding filtering at the top of the bias pot, letting the 3.3K before the bias pot do some of the filtering for me. A 2.2uF film there dramatically dropped the ripple seen at the wiper of the pot. Good! As an afterthought, I put another 2.2uF on the pot wiper to ground. That was even better.

And then I got diverted into the other problem; what happens to the stock circuit if you put a 2.2uF cap to ground at the top of the bias pot and maybe one from the wiper. That dropped the wiper ripple down to the 12mV range. Good side effect stumbled upon.

So I think I solved how to not break the bank or chassis size for 100V films in the bias supply: use a 22uF or two 10uF at 100V films chosen by where you can actually buy parts and how much they cost, followed by some $0.50 2.2uF films on the bias pot itself.

As a side effect, I solved how to really, really cut the ripple on the bias supply as seen by the tube bias resistors.

This raises the question of whether it's worth doing at all. For an Immortal Amplifier setup, yes, as it's a cheaper and better way to use film than one big direct replacement for the electrolytic. For other amplifiers, maybe, maybe not. A 2.2uF 100V film cap costs about $0.50, and a 2.2uF electro costs way less. Maybe the ripple on the bias pot makes an audible difference, maybe not. Needs testing.

[nuke]

Yeah, I went to look at the 22µf 25v as a replacement for the typical cathode bypass cap we see in so many preamp stages.

From a practicality perspective, just plopping yet another cheap ICC e-lytic cap on the bias supply after the bias wiper (or the voltage divider if you're too cheap to put in a pot). This is what Fender did in the ubiquitous Blues Junior, by plopping down a 100µf at 25v after the divider in the C- supply. I think it resolves hum just as well. Even a cheap ICC e-lytic will outlive the one-sided board used for the tube sockets.

[TUBEDUDE]

Multiple small caps definitely is the way to go.
I don't use 22uF as cathode resistors. I use 2uF to 5uF.
How much low end do you need?

[B Ingram]

I'm puzzled by why Fender uses a polarized electrolytic capacitor in the reverb circuit of the Blues Junior and other "blues" series amplifiers with solid state reverb.

They use a 0.47 µf 100v electrolytic. ... is there any electrical purpose for choosing an electrolytic that anyone can fathom?

A 470nF 100v electrolytic is smaller than the same cap with a film dielectric. The space available for the cap is pretty tight ("C23" in that manual). That the electrolytic is 20-25% of the cost of the film-variety is also a big bonus to the manufacturer.

[nuke]

A 470nF 100v electrolytic is smaller than the same cap with a film dielectric. The space available for the cap is pretty tight ("C23" in that manual). That the electrolytic is 20-25% of the cost of the film-variety is also a big bonus to the manufacturer.

You mean like this one I put in?

Smaller and less expensive than the 0.47 100v axial. I had to form the leads outward to fit the spacing. They're less expensive than the e-lytic, even in qty 5000.

C20-blues-junior.jpeg

[B Ingram]

A 470nF 100v electrolytic is smaller than the same cap with a film dielectric. The space available for the cap is pretty tight ("C23" in that manual). That the electrolytic is 20-25% of the cost of the film-variety is also a big bonus to the manufacturer.

You mean like this one I put in?

Smaller and less expensive than the 0.47 100v axial. I had to form the leads outward to fit the spacing. They're less expensive than the e-lytic, even in qty 5000.

Is a film cap always smaller than an electrolytic of the same voltage & capacitance? Is is the reverse usually true?

[nuke]

Is a film cap always smaller than an electrolytic of the same voltage & capacitance? Is is the reverse usually true?

In this case, yeah. And since there's literally 0-volts here, it doesn't even need to be a 100v.

But that's a 100v 0.47 film (polyester), in the same exact spot where the .47 100v electrolytic came out.

[R.G.]

In general, a film cap of the same capacitance and voltage rating will be bigger than an electro. This only gets fuzzy where the cap rating is small, and the packaging case and leads, etc. are a bigger part of the entire package than the working capacitor plates and insulator.

If it was the other way round, electrolytics would probably vanish.

[nuke]

In general, a film cap of the same capacitance and voltage rating will be bigger than an electro. This only gets fuzzy where the cap rating is small, and the packaging case and leads, etc. are a bigger part of the entire package than the working capacitor plates and insulator.

If it was the other way round, electrolytics would probably vanish.

Yeah, no argument.

In this particular instance, it looks like the 0.47uf at 100v was chosen only so it fit the same footprint as all the other small e-lytic axials in the amp. The cathode bypass caps are 47uf at 16v, 22 at 25v, 2.2 uf at 50v, all because they use the exact same board footprint.

It's kinda funny actually.

The drive side of the reverb uses a 22u at 25v, also with no biasing DC on it. The circuit is set up as an opamp high-pass filter with DC preserved, and DC in this case is no DC, just what you want driving the 800-ohm reverb tank coil that's isolated from ground.

On the recovery side, for whatever reason, they implemented the same sort of op-amp stage, with values adjusted. The tank has a grounded output, so it should average out to 0-vdc on the opamp. Admittedly, I'm a little puzzled by the whole recovery stage, not just the 0.47 capacitor. Seems like it could be simpler. But admittedly, it does seem to work well enough.

The basic layout of the solid state reverb looks like it has been copy-pasted from every version of the amp since 1992 til present day, in spite of the reverb and return getting moved around a number of times with very different drive and return and injection.