Dual rectifiers and filter cap limits

[65ampman]

Hey guys, I'm new around here, or new again as I lost my password and couldn't access my old email address.

I build and repair amps and have just finished a 5E8-A low power tweed twin build for a client. It turned out great and I have no issues to troubleshoot but I am curious about how a second rectifier effects the maximum recommended 1st stage filter cap value. Does a second rectifier tube increase, decrease, or not affect this value? Fender choose to use a single 16uf cap for the first stage and then 2 16uf in parallel after the choke. If tighter response was the intention of the 2nd rectifier, whey did they then choose to use such a small value filter cap? The 5E6 bassman also used this circuit.

Thanks in advance!

[Enzo]

Two rectifiers will rectify twice the current. I don't think "tighter response" was on their minds so much as just running out of current. Maybe that is what tighter means.

They chose a small filter cap because caps were expensive then. That is why two 16s were cheaper than one 32 or 40. And only an 8uf later on in the chain.

[65ampman]

Yes, I know that smaller caps are/were cheaper but Fender made plenty of amps before and after this model with 2x 20uf for a 1st filter stage. I am aware that smaller caps are used down the line in most cases but they must have used this small value for another reason. Why not 2x16uf(or larger) first and then a single 16uf after the choke as in most other tube amps(Fender and most other brands)? I do believe tighter response was the goal as this was originally a bass amp and needed more headroom and current to supply the extra power needed for low frequencies.

BTW, I've tried this amp with a single 5U4GB and a single 5AR4 as well and the dual 5U4GB sounds best. Tighter low end but with a musical sounding amount of sag or compression. They were on to something here. I suspect that the 5AR4 rectifier did not exist at the time these dual rectifier models were designed because they went to a single 5AR4 shortly afterwards.
twin_5e8a_schem.pdf

[Enzo]

This amp is relatively unusual in that it does not draw from the first filter node for the output stage. Your smallish first cap precedes a choke, before the B+ first node. Note the difference between this and 5F8A, where they did double filter caps and drew output stage power right off the first cap stage.

[Mick Bailey]

Having a smaller cap before the choke has a limiting factor on B+ with this configuration. I have an amp that uses 3uf (that's right) in this position to bring the voltage down in an amp where the voltage would otherwise be too high.

Interestingly, the 5e8a circuit has no means of ensuring equal power sharing between the rectifiers, and just as with any other parallel circuit, one tube can hog more current than the other.

[65ampman]

Enzo, It's unusual for fender but not for other early tube amps. But this is a perfect example for my original question as the later model 5F8-A uses a single 5AR4, not 2x 5U5GB as in the 5E8-A. Does the maximum recommended capacitance change when using two rectifiers? Did they lower the 1st cap value because of the additional rectifier or as Mick Bailey suggested, is it just to reduce the overall B+?

[dstrat]

Max plate voltage for 6l6g is 360 v dc.might of been part of it.

[R.G.]

Does the maximum recommended capacitance change when using two rectifiers?

I would say that this is the wrong way to ask the question. The reason there is a maximum capacitance specification is because there is a maximum peak current that the rectifier tube can support, and going over that peak current degrades the electron emitting cathode.

In a power rectifier circuit, the cap charges up to close to the peak of the AC voltage it's being fed, and then the diode turns off as the cap holds the newly-charged peak voltage and the incoming AC decreases from the peak. The cap runs down for the time until the next rising AC voltage half-cycle is high enough to turn on the rectifier again. In a steady stage, a capacitor-input rectifier circuit feeds the first cap a recharge current for a short time before the peak of the AC half-wave. This peak-and-run-down voltage on the cap is where the ripple voltage comes from. See Power Supply Basics for some pictures of what is happening in AC to DC rectifier circuits.

The bigger the cap, the less it runs down between half-cycle peaks. That is what we want the cap for, to reduce ripple on the rectified DC. But the bigger the cap and the lower the ripple voltage, the less the cap's voltage runs down between charging pulses, and therefore the shorter the time the cap gets to be charged when the peak voltage comes along to charge it. Since 100% of the charging current comes in the pulses through the rectifier at the AC peaks, if the pulses get shorter, they have to get higher to get the same amount of power through, on average.

So the bigger the cap and the lower the ripple voltage, the higher the peak currents through the rectifiers get. Add to this the peak current limit of a tube rectifier cathode, and dumb it down for a 1950s tube designer that doesn't want to do a lot of math for his rectifier circuit and doesn't have the instruments to measure peak currents in rectifier circuits, and you get the simple answer of the tube manufacturer telling you not the peak current on the rectifier tube in a circuit, but instead the biggest cap you can use in a "typical" circuit. They specified something simple that's not exactly what is going on but works well enough for the vast majority of applications.

Overall, the "maximum capacitance" spec is actually a peak rectifier repetitive current spec.

With that as background, we can get back to the question of whether the max capacitance can change with dual rectifiers. The short answer is that it depends on whether you can ensure some kind of current sharing so that one rectifier section doesn't take on more of the current pulses with a bigger cap, exceed its peak repetitive current limits, have one section die, and then have 100% of the peaks appear on the other rectifier, which shortly kills it too. Chain failure like this is common with semiconductor power stages, and ways around it are some of the first things a new EE deals with when they first start to practice on power circuits. Tubes are much foggier about this because they have a much higher internal impedance to start with, and this high impedance tends to equalize things more than with very low impedance semiconductors.

You either need a ton of math and lots of measurement data or some "try it and see". I suspect that matched rectifier tubes would be fine with bigger caps. Random rectifier tubes from different makers, and especially with heaters that run the cathodes at slightly different temperatures would start spoiling the matching that keeps current sharing from happening would make it worse.Adding external resistors in series with the rectifiers as is done with semiconductors would make current share better, and enable raising the first capacitor.

Did they lower the 1st cap value because of the additional rectifier or as Mick Bailey suggested, is it just to reduce the overall B+?

That's a tough question that may never be settled unless we a notebook or something where the original designer set out his thinking. But here is some thinking on what does happen as you tinker with rectifiers and first filter cap values.

If you make the first filter cap be tiny, say a few pF, then there is no overall smoothing at all. The "DC" the rectifiers put out is just full wave rectified AC. The DC part of this can be computed from the waveform and some integration. The peaks of this 100% ripple waveform are the peaks of the AC half waves, and the valleys go to substantially zero. As you increase the capacitance, it has the effect that stored charge in the cap starts holding up the "valley" voltage. Increase the cap a lot, and it begins to fill in the valleys. The average DC output rises as this happens, lying somewhere between the peak and the valley of the ripple voltage. A bigger cap makes the ripple voltage smaller in the direction of the peak of the wave. This increases the average voltage out of the cap.

The suggestion that an inductor after a small cap can produce a smaller DC voltage is correct - but this only works well at a constant load. For low loads, the voltage out of the cap will still drift up to the peak of the incoming DC voltage, even with an inductor, as the inductor has less and less current to use to average the voltage.

The first capacitor issues can be helped dramatically by using a second capacitor separated from the first by resistors or inductances. These series impedances after the first cap let you make the first cap small, and then make the second cap be any value you want, as the peak current issues are solved in the first cap and series impedances.

[trobbins]

If you were able to measure the power transformer winding resistances, and the choke inductance and resistance, then it is very easy to 'do the maths' using PSUD2 to closely estimate both the initial hot turn-on current stress and the continuous peak current levels for both one or two 5U4G. I'd expect that those peak current levels are no where near the valve peak current limits even when using just one 5U4GA.

If that was the case then the query 'has legs' as to what they were targeting with the 2 diode valves.

The standby switch operation is effectively the same for the diode valve peak current stress as an initial hot turn-on when the standby switch is closed.

Maybe they only had the choke to use, and 2 diodes gave them just acceptable hum and sag compared to some other combination of parts.

[65ampman]

Thanks everyone! This gives me lots to think about. I am not an engineer but I do know ohms law and have been servicing and building amps for several decades now. This is the first request that I've had for a low power tweed twin and it turned out great. I ended up connecting the choke in the more traditional manner between the first and second filter caps and separating the screen supply and B+. This seems to be the more correct way to do things and ensures that the screen voltage is not higher than the plates. I am currently experimenting with filter caps values.

However, I have seen many vintage tube amps connected as in the 5E8-A schematic with the B+ taken after the choke with no adverse effect on tube life or tone as far as I could tell. Some vintage amps wired this way still have the original tubes which says something for the reliability of the circuit, correct or not.

[g1]

However, I have seen many vintage tube amps connected as in the 5E8-A schematic with the B+ taken after the choke with no adverse effect on tube life or tone as far as I could tell. Some vintage amps wired this way still have the original tubes which says something for the reliability of the circuit, correct or not.

Keep in mind that having the extra current of the power tubes running through the choke requires a much larger (like OT size), more expensive choke.
http://music-electronics-forum.com/t7038/

[65ampman]

Keep in mind that having the extra current of the power tubes running through the choke requires a much larger (like OT size), more expensive choke.
http://music-electronics-forum.com/t7038/

Wow, I see that now! Since I wired it like the 5F8-A with the B+ before the choke, I don't need that large choke but now I'm curious to hear the difference...

[pdf64]

...I ended up connecting the choke in the more traditional manner between the first and second filter caps and separating the screen supply and B+. This seems to be the more correct way to do things and ensures that the screen voltage is not higher than the plates...

Just to note that's a non-issue. Bear in mind that whenever the amp is passing signal, half the time the plate voltage will be less than the g2 voltage anyway, perhaps several hundred volts below at high signal levels.
As long as limits aren't breached, plate voltage is kinda irrelevant, the g2 voltage is what matters, as the action of g2 is to 'de-couple' plate current from plate voltage.
However, as with any vintage design that doesn't have them, current limiting resistor/s should be added, eg 470 ohm 1W, as g2 current (and hence dissipation) increases significantly when overdriven. The amp's designer almost certainly wasn't expecting the amp to get used like that, so saved a little money by omitting them.

Also bear in mind that the amp was designed for 6L6G, so its conduction angle and power supply was almost certainly specified to accommodate idle conditions suitable for 6L6G (19W plate), rather than the 6L6GC that might normally be used now, with it's 30W plate. So to operate as the designer intended, idle at ~13W, rather than 23W per 6L6.

[65ampman]

However, as with any vintage design that doesn't have them, current limiting resistor/s should be added, eg 470 ohm 1W, as g2 current (and hence dissipation) increases significantly when overdriven. The amp's designer almost certainly wasn't expecting the amp to get used like that, so saved a little money by omitting them.

Also bear in mind that the amp was designed for 6L6G, so its conduction angle and power supply was almost certainly specified to accommodate idle conditions suitable for 6L6G (19W plate), rather than the 6L6GC that might normally be used now, with it's 30W plate. So to operate as the designer intended, idle at ~13W, rather than 23W per 6L6.

I should have mentioned that I did add the 470 ohm screen resistors and also added a bias trim pot with plenty of range to accommodate NOS and new production tubes.

[g1]

Also bear in mind that the amp was designed for 6L6G, so its conduction angle and power supply was almost certainly specified to accommodate idle conditions suitable for 6L6G (19W plate), rather than the 6L6GC that might normally be used now, with it's 30W plate. So to operate as the designer intended, idle at ~13W, rather than 23W per 6L6.

added a bias trim pot with plenty of range to accommodate NOS and new production tubes.

I think the point he was making is that even with new production tubes, you should not increase idle current beyond what you would use with 19W type 6L6G. (13W idle current)