Tube B+ dropping resistors in tube amps

JMicro

New Member
2 issues
1 dropping ac line to 110 or 117 to make it easier on amps.
2 best practice for overall B+ voltage drop. Like when you remove a selenium rectifier.
I have seen posts on dropping ac line volts with buck transformers or zener diodes. Why is it not good just to use a dropping resistor.
I also work on Zenith FM tube radios and remove the selenium rectifier. I usually put the drop resistor after its rectified before the first cap.Then I noticed later units that used a more modern diode instead of selenium put the drop resistor before it was rectified from the factory?
On my amp I use a 5y3 .
I have another amp I built and use a 1N007 The tranny puts out too much because it was salvaged from a unit with more tubes so I used drop resistors to get it down its a pp amp it sounds OK to me
 
Mostly the reason people use buck transformers is better regulation, with a side benefit of less heat loss. Voltage drop across a resistor is a function of current flow through it, and an AB amp like most push-pull amplifiers will use more current as output power goes up. If you put an ammeter on one and listen to music you'll see it dancing around. A resistor would drop more voltage every time that happens, which moves around the tube operating points and makes things sound worse. A bucking transformer that drops say 10v will drop 10v without much care for the load on it.

Resistors in the B+ circuit have the same issue, though with a single ended amplifier the current load is pretty constant so it doesn't matter so much.
 
Yup, more resistance will cause more sag as more current is demanded from the power supply. This rule applies to class AB amps. It is not so much of an issue with class A amps because the current draw is nearly constant at any power output level in a class A amp. It's also not so much of an issue with preamps, bias circuits in power amps, or other low current circuits.

If it's a vintage unit we're talking about that was designed for 110V mains, best practice is to use a bucking transformer to reduce the voltage down on the primary side of the power transformer because all secondary windings will be high if the primary voltage is high. A good rule of thumb is to buck down to the primary voltage that will deliver 6.30VAC at the secondary filament winding(s) when the filament circuit is loaded down to its proper value (i.e., don't do that measurement without all tubes installed).
 
Also, sometimes it's necessary to lower the voltage on just the secondary high voltage winding. If you must do it this way, some people will add a choke in place of a dropping resistor, for example in between the first two filter caps. The choke removes ripple significantly better than a resistor of the same value of DC resistance of the choke. If a choke won't work (not enough room, or can't find one that fits the specs needed of the unit), then the next best practice is to place as small of a dropping resistor as possible in the circuit. I prefer to put this dropping resistor right after the rectifier but before the first filter cap . In that position, you can obtain more voltage drop with less resistance because the DC at that point in the power supply circuit has significant ripple, so you can take advantage of the voltage drop from the AC ripple current as well as from the DC current that exists simultaneously at that point. Some people forget to account for the AC ripple current and undersize the wattage rating of the resistor in that location--they do get hot. Best to use a wirewound (IMO) in that position. Usually they are sized at a few hundred ohms or less to do the job.
 
OK, this was mentioned in another thread, and frankly, I don't get why it is such a bad idea.

I am talking about exactly what kward describes. A dropping resistor between the rectifier and the 1st cap. Small value, wire-wound, plenty of wattage rating.

The filter caps should be what is actually supplying any transient load. If adding the resistor causes any change at all to the power supply's ability to provide sufficient and smooth B+, then, IMHO, the PS was under capped in the first place.

To put it another way, the Filter caps (all of them together) should be preventing the rectifier/transformer from ever seeing any transient load.

The other possibility is that you were already over driving the amp anyway (before the modification). I guess I can see it for you guitar amp guys that like to run the amps to the point they "clip", or rather don't clip, which is why you run tubes -- distort, or "get funky" or whatever you call it. What ever they do, I can see that it would entail at least some sagging of the B+.

In this case, I can see that the amp may change sound -- perhaps significantly.

But for us reasonable Hi-Fi listeners that generally prefer to keep the voice coil in the vicinity of the speaker, I don't see the problem.

/me humbly awaits rebuke/education.
 
The filter caps should be what is actually supplying any transient load. If adding the resistor causes any change at all to the power supply's ability to provide sufficient and smooth B+, then, IMHO, the PS was under capped in the first place.
The voltage on the cap is a function of the time constant of the R/C smoothing filter in which the cap participates. A period of 1 time constant will discharge the cap into a short circuit load to approx 63% of it's fully charged voltage value. If the time constant is very large the depletion rate will be slower and you'll see better "regulation." Time constants are made larger by either increasing the resistance seen right before the cap, or increasing the cap size.

Example:
If an R/C power supply filter consists of 50 ohms resistance with a 100 uF filter cap, the time constant will be 100 uF * 50 Ohms = 5 mS. Thus in 5 mS, the cap will be at 63% of it's fully charged value, if the R/C circuit is drained into a short circuit load. For the cap to be considered still charged, it wound not lose more than say 2% of it's fully charged value. With a time constant of 5 mS, losing 2% of it's fully charged value would take maybe 1/20th of one time constant or 5 mS/20 = 250 uS.

As you can see, it doesn't take long to deplete the charge on the cap--much shorter in time duration than the "low frequency" transients of a audio amp. Thus hopefully you have some insight now that the typical cap size in the R/C smoothing filters in the power supply doesn't offer much regulation at the frequencies in question.

Increasing the time constant would offer better regulation. How big would the filter cap in the above example need to be to create a time constant that would deplete the cap by no more than 2% in one cycle at 20 Hz into a dead short?

20 Hz = one cycle every 50 mS. So in 50 mS we need the cap to be at 98% of it's fully charged value.

1/[(C*50 Ohms)/20] = 20 Hz, and solve for C.

C = 1/50 = 20 mF, or 20,000 uF.

So with 50 ohms resistance in this R/C filter, you'd need a filter cap size of 20,000 uF so that you had only 2% voltage depletion at 20 Hz. Granted this is a contrived example, and some hand waving to prove a point, and it is probably not as bad as I'm making it out to be because an output stage does not have zero resistance even when tubes are full on conducting during the reproduction of LF transients. In any case, you get the idea of the principles involved.

If you wanted this level of regulation (2% is quite poor in fact), the more sane approach would be to use an active regulator--and you'd probably get 10x better regulation also.
 
Audibly, poor regulation in the power supply of an AB amp tends to produce mushy bass. Basically its a function of the time constant Kward describes. The cap discharges and the supply voltage drops, reducing peak available power output with it.
 
2 issues
1 dropping ac line to 110 or 117 to make it easier on amps.
2 best practice for overall B+ voltage drop. Like when you remove a selenium rectifier.
I have seen posts on dropping ac line volts with buck transformers or zener diodes. Why is it not good just to use a dropping resistor.
I also work on Zenith FM tube radios and remove the selenium rectifier. I usually put the drop resistor after its rectified before the first cap.Then I noticed later units that used a more modern diode instead of selenium put the drop resistor before it was rectified from the factory?
On my amp I use a 5y3 .
I have another amp I built and use a 1N007 The tranny puts out too much because it was salvaged from a unit with more tubes so I used drop resistors to get it down its a pp amp it sounds OK to me
If you have a lot of B+ to drop you can also consider a choke input filter. The only downside of which is the size and the cost. Old school solution but there is a reason why the old amps from the 1930s and 40s still work, that is because they usually had chokes, and large PIO film caps in the power supply. They sound very good for how small the caps are in the PS.
 
Power supply design also plays into that. Choke input supplies give roughly 0.9x the AC voltage as DC output vs a cap input which gives 1.4x the AC voltage as DC out. Need the right type of choke to handle that, and that assumes changing the voltage that much is what you need to get done.

Really old amps also used small value capacitors just because of the technology. Prior to electrolytic caps it was all just paper caps, and a large value paper cap is physically massive. To get sufficient filtering the chokes were a necessary evil. As cap technology evolved you could get a larger value cap in a given physical space, and they also got cheaper. By the 50s caps were large enough in value that small chokes were all that was needed.
 
Really old amps also used small value capacitors just because of the technology. Prior to electrolytic caps it was all just paper caps, and a large value paper cap is physically massive. To get sufficient filtering the chokes were a necessary evil. As cap technology evolved you could get a larger value cap in a given physical space, and they also got cheaper. By the 50s caps were large enough in value that small chokes were all that was needed.

Another piece of that puzzle is rectifier limits.

The shift from chokes to capacitors happened in the late 1930s and early 1940s as PIO technology improved and the cost dropped, much of it driven by manufacturing for highly technical weapons and communication systems. The confiscated German capacitor technology circa 1945—spoils of war—created the great revolution in capacitors, as Western Electric and others began shifting to the coated kraft paper technique and the cost dramatically dropped. That was the death knell for choke filters. (Ask not for whom the power supply bell tolls, it tolls for the swinging choke.)

But the issue is more complex than merely the limited availability and high cost of capacitors in those days; one must consider the limits of a tube rectifier's cathode, limits which have over time increased. Even back in the day one could have added additional capacitors, but the rectifier could not have driven them.

I've elsewhere explained this in detail, but, in brief, the issue is the rectifier conduction angle. A capacitor only stores charge when the applied voltage, as output by the rectifier, is higher than the voltage of the charge currently stored in the capacitor. So by maintaining a larger store of charge in the capacitor the voltage is maintained, and the charging region is narrowed, so more current is drawn in a briefer region. Narrow slices of the sinewave are used, which begins to resemble a low duty cycle square wave with rounded peaks.

As the current begins to spike the rectifier cathode begins to degrade, and eventually arc in a brilliant testament that rectifier capacitance limits must not be used as optional starting points.

The rectifier designers specified a peak current (not the average current specified by the maximum current) which the rectifier cathode can tolerate without being blown apart; that's what the capacitance limit is really specifying. Few amplifier designers would have understood conduction angle—look at how little it is today understood, including for solid-state rectifiers—so the rectifier tube manufacturers distilled the specifications down to a maximum capacitance which could be honored.

This is why a tube rectifier has such low capacitance limits; it's all about instantaneous current load. Those limits were more severe with the older rectifiers, so the small capacitance was both a reflection of the poor capacitor and rectifier technology available.
 
If you have a lot of B+ to drop you can also consider a choke input filter. The only downside of which is the size and the cost. Old school solution but there is a reason why the old amps from the 1930s and 40s still work, that is because they usually had chokes, and large PIO film caps in the power supply. They sound very good for how small the caps are in the PS.

Choke supplies can have wonderful regulation. But the issue with choke filters is critical inductance. If the load drops below a certain point the filter becomes unstable and the voltage climbs to dangerous levels which can blow up capacitors and downstream components.
 
sure, rectifiers had limits. The 80 was ubiquitous in the 30s, and it was re-styled with a flashy new octal base as the 5y3. The Cunningham datasheet for the CX-380 says 4uf is typical for a cap input. Tung-Sol datasheet from 1949 for the 80 and 5y3 says 10uf. Not sure the date on the Cunningham sheet, but I'd say not later than 1930 based on the tube numbering and the S bottle pictured. Beefier rectifiers that allowed for higher inrush would also allow for use of larger filter capacitors. I'm sure the availability of larger caps prompted development of beefier tubes and the availability of those tubes drove designs that used larger caps and smaller chokes. Its a chicken and egg thing.
 
The patents being granted during the 1920s and 1930s show that capacitor development was proceeding at a rapid pace, and one not solely driven by the need to filter the output of small tube rectifiers.

Most of the rectifiers of the day were mercury, providing up to 500 Amperes as a routine practice. Arc-back was dealt with using RC filters, so capacitors certainly were needed for that purpose. Selenium rectifiers were just starting to be routinely manufactured and used in the early 1930s. The small tube rectifiers, a hundred milliAmperes or so, were used in radios and small transmitters. Most of the capacitors were mica, and the electrolytic filter capacitors were the gel type instead of the wet type. (Invented in New York, I might add, in the days when humans in NY made actual things instead of merely pushing bits around.)

The early tube rectifiers like the CX-380 note very different voltage drop for choke-input vs. capacitor-input, and the limit for the capacitance shows that the conduction angle (charging period) must have been nearly the waveform's full span to avoid cathode destruction. Remember the maximum current is per half waveform, so a big difference occurs if the current is delivered over the entire waveform (choke-input) or subset (capacitor-input). The bigger the capacitor, the smaller the time, so the greater the current.

This is why high capacitance blows up rectifiers. So does PIV stress, which is why adding solid-state rectifiers to a tube rectifier greatly extends its lifespan, but that's the subject of a different discussion.
 
Power supply design also plays into that. Choke input supplies give roughly 0.9x the AC voltage as DC output vs a cap input which gives 1.4x the AC voltage as DC out. Need the right type of choke to handle that, and that assumes changing the voltage that much is what you need to get done.

Really old amps also used small value capacitors just because of the technology. Prior to electrolytic caps it was all just paper caps, and a large value paper cap is physically massive. To get sufficient filtering the chokes were a necessary evil. As cap technology evolved you could get a larger value cap in a given physical space, and they also got cheaper. By the 50s caps were large enough in value that small chokes were all that was needed.
This can be tuned with a smallish cap (1-10uF) before the choke, it won't be all the benefits of true choke-input but it's possible to tune to a specific output voltage that's between that of a traditional cap input and a choke input.
 
Power supply design also plays into that. Choke input supplies give roughly 0.9x the AC voltage as DC output vs a cap input which gives 1.4x the AC voltage as DC out. Need the right type of choke to handle that, and that assumes changing the voltage that much is what you need to get done.

Really old amps also used small value capacitors just because of the technology. Pdrior to electrolytic caps it was all just paper caps, and a large value paper cap is physically massive. To get sufficient filtering the chokes were a necessary evil. As cap technology evolved you could get a larger value cap in a given physical space, and they also got cheaper. By the 50s caps were large enough in value that small chokes were all that was needed.
Sure, nowadays, capacitance is cheap. My point is that these old Western Electric, RCA, and other vintage "theatre amps" when rebuilt sound fantastic even with their small value capacitors. So one does not need a lot of capacitance to get a good sounding amp.
There is probably no component more reliable, when used within its specs, than a choke, since it is merely a spool of wire. As with any component, it has to be properly designed and spec'd for the job. Determining the right choke is not a big deal. I have read several ways for determining critical inductance from scientific to simple. I have been installing chokes in almost all my rebuilds and in new amps and have never had any problem with voltage problems with chokes being sub critical. In fact there is a group of diy'ers who build their PS for their SE amps, purposely using sub critical chokes. I prefer to have my chokes above the critical point.
In my view, there is only 2 negatives with chokes, its size and cost.
 
This can be tuned with a smallish cap (1-10uF) before the choke, it won't be all the benefits of true choke-input but it's possible to tune to a specific output voltage that's between that of a traditional cap input and a choke input.

I wonder what this does to the regulation though. Does it fall off a cliff and drop back to choke input voltage levels with any load?

My point is that these old Western Electric, RCA, and other vintage "theatre amps" when rebuilt sound fantastic even with their small value capacitors. So one does not need a lot of capacitance to get a good sounding amp.

I was just saying that the use of chokes was partly a matter of necessity. It certainly did and does work. Just because modern design goes for big caps and no iron doesn't mean small caps and big iron don't still work. Caps just happen to be cheaper.
 
Does it matter what the frequency of the drive signal is?

It shouldn't, though I suppose if the frequency was low enough and the signal level was high enough you might be able to measure the signal as power supply ripple if there wasn't quite enough reserve capacity in the supply. For every negative peak it will draw more from the supply and every positive peak it draws less, so on average it cancels itself out.
 
So this is great info/conversation, but we haven't answered the question yet.

What practical solution is there for contemporary higher line voltages, as well as other things such as replacing selenium rectifiers with modern diodes/bridges that result in excessive B+?

Here's what I think is not practical:

-- Replacing power transformer. -- Even off-the-shelf, let alone a custom wind -- probably costs more than some amps are worth, and if the amp is worth more, then making it non-original , and thus loose value is not the intent.

-- Assuming that there just happens to be a convenient and proper voltage for a bucking winding. While some units, lile hifi amps made from console, organ or jukebox pulls, may in fact have additional windings, but even in this case, there is no guarantee that the right voltage will be available.

-- If the unit has a cover/enclosure, it should still fit on. While a separate bucking transformer may fit in some amps, many chassis do not have that kind of extra space. Chokes may or may not fit into this category.

-- Always using a variac. No one is going to tell you that those coffee can things with a big knob on top are part of "positive living room feng shui". Anything that causes visitors make reference to "Dr. Frankenstein's lab" (assuming that such references would not already be made) is not considered "practical".

-- It shouldn't double the waste heat output of the amp. Filaments are one thing. At least they look cool, but smouldering power resistors don't.

I don't mean to have an "attitude" -- I realize this post may sound that way, but it is not the intent. I'm just seeing if there are any more elegant solutions out there.

Strings of diodes?
Some sort of modern buck converter referenced to a few volts (few 10's of volts?) below B+ with a Zener?
Modern buck with devices that are rated for high voltage?

Ideally, a solution would also help with the increase in filament voltage, too, or have a separate solution for the filaments.
 
Back
Top Bottom