Dropping Resistors

[trem]

Excellent idea !! Consider it a sort of "fixed power scaling".
To boot , you will have no ripple or ripple induced hum.
The main advantage, however, is that a TO218 or TO247 MOS bolted to a heat sink will safely dissipate *lots* more than any Zener you can get.

Can you guys discuss this a bit more in detail.? I am still trying to get a rope around basic tube stuff, and I know nothing about solid state devices.
What kind of MOSFET, and where in the circuit is best to install these. I am also in the position of needing to drop quite a bit of B+ (30-40VDC) for a 2x6V6 amp. I already bought a new PT, but I would like to know the best way to implement the MOSFET solution (in series with the OT CT perhaps). I would also be happy to read any link(s) you might have if that makes it easier for you.
Thank You

[mooreamps]

How about a less brute force approach. Use a choke input filter, where the B+ hits the choke before the first cap. This will drop the voltage a LOT but you can then bring it back up with a smaller cap, like <1uF before the choke. Simply adjust that first cap to get what you want.

Makes for a very low ripple, stable supply and is easy on the rectifier.

By far the best suggestion in my opinion... Also, you don't have all that waste heat to deal with by using a FET....regardless of how well it's done...

-g

[J M Fahey]

Hi Trem, if

still trying to get a rope around basic tube stuff, and I know nothing about solid state devices.

(nothing wrong with that, we've *all* been there) I don't want to complicate your situation.
Personally, I would get another PT which fits the bill, and leave that hefty one for a later, higher power build.
When starting with a new activity, such as Electronics (or Guitar, Painting, Programming, you name it) at the beginning keeping classic/mainstream pays a lot.
If out of the question, for economic$ rea$on$, an input choke may be the best option, as suggested.
*Very* classic, will definitely drop the voltage, which in my book is something good, the sound will be sweeter, smoother, you name it.
It should cost much less than a new PT; our good Friends might suggest one.
Good luck.

[EFK]

Old Selmer TnB amps ran all of the amp B+ through a choke before the OT center tap, rather than after like Marshall/Fender/younameit amps. And it's a pretty hefty-looking choke; I doubt it was rated for full mA draw as would be "proper" (that would be bigger than the OT, probably) but it was still considerably larger than most common guitar amp chokes.

This is something I'd like to explore; does the suggestion to use a really small cap (say 1uF - 10uF) prior to the choke offer some kind of "cushion" to permit a smaller/underrated choke unit? I mean, could a typical 100-150mA at 3 or 4H (typical for 2X6L6 anyway) be used in a situation like this simply to drop voltage despite it being underrated? Certainly the rated inductance would be less at the greater current draw, but if you were using larger filter caps after the choke to finish killing the ripple, would it really matter? Or would it be more likely that you'd simply melt the choke?

[jpfamps]

If you are using a choke input you need to very careful about the AC current. This can be many times the DC current, and is higher for a lower value choke (as obviously a lower value choke has lower impedance than a higher value choke), so a 4H choke would have to be huge to avoid saturation, and hence very expensive. Possibly even more expensive than buying a more suitable transformer.....

If the choke saturates it's inductance will drop and the following filter cap will charge up to the peak AC voltage, ie what we are trying to avoid!

In my opinion choke inputs are best left to lower powered amps, or audiophiles with very deep pockets.

Other options would be to use cathode bias, or a higher impedance rectifier such as a 5U4G.

[J M Fahey]

If you are using a choke input you need to very careful about the AC current. This can be many times the DC current,

Very sorry to disagree, but a choke in series with the rectified voltage will pass current, with a DC component and an AC component.
That AC component, which causes ripple, will be much less than the DC component; a case which might be compared to ripple voltage in a capacitor being less than its DC value.
In a properly specified inductor, Impedance will be much higher than its DC resistance, that's what it's there for, or it would lose its job to a much cheaper resistor.
Won't talk about cost, because a specially wound choke might cost the same as a mass produced transformer, but that's a market size problem, not a technical one.

[tubeswell]

The voltage will drop under load, but assuming the 360-0-360 rating is for 'loaded', then you could try different rectifiers (if you have a 5V winding).

On what MWJB said, a few reverse-biased 9-15V 5W zeners in series between the HT centre tap and ground will allow you to get a reasonable B+ in a simple way, that allows you to experiment with different voltages relatively easily by subbing in different voltage zeners. Mount the zeners on a tag terminal strip so that they are away from the chassis. (If you use a bolt-on zener, it will need to be a higher power rating because of the heat it picks up from the chassis).

See pics

If you don't have a 5V winding, you can still use zeners with a SS bridge rectifier - you just put the zeners in between the bridge and the ground return (cathodes/banded-ends of the zeners pointing to ground)

[R.G.]

Let's play theory for a moment.

If the transformer puts out 360-0-360, it will put out 360*1.414 = 509V peak. A rectifier into a capacitor-input filter makes that into 509V minus the rectifier losses. For semiconductor rectifiers, that's maybe a volt, or 508V. When you load it, two things happen. First, the filter cap runs down between charging pulses, which comes out as a ripple voltage, and the resistances in the primary and secondary windings eat up some of the open circuit voltage by Ohm's law. So the peak voltage sags as well as the ripple getting larger.

For a choke input filter, the output is dramatically more variable. In an unloaded choke input filter, the voltage is the same as the voltage on a capacitor input filter, because the choke works on the *current* through it, not the voltage across it. If only minimal current flows, the choke might as well not be there, and the choke-capacitor collapses to just a capacitor input filter.

As DC output current is started, the choke comes into play, and starts widening the conduction angle of the rectifier with respect to the AC power line cycle as well as - er - choking off the AC components. The DC at the filter cap after the choke starts falling. AS the load increases, the DC output keeps falling until the choke is conducting all the time, 180 degrees of each AC power line cycle. At that point the DC stops falling (excepting for changes caused by wire resistance, etc.) and the DC output stays constant with increasing load until something burns out. The DC level after the choke hits continuous conduction is always (except for minor DC losses) 0.637 times the peak of the input waveform, or 0.9 times the RMS value of the input waveform.

So a choke input filter has an output of 1.414 times the peak of the input sine wave for low loads and falls to 0.637 times the peak of the input sine for larger loads. That is, the DC output drops by 55% from no load to the "critical load" that makes the choke conduct for a full cycle, then doesn't drop any more. As a practical matter it's bad design practice to try to use a choke input filter below the critical load. The DC output wavers around hugely with minor load changes. Good design for choke input filters is to design the choke and a minimum load to keep the choke conducting all the time. So-called swinging chokes have part of their magnetic path designed to un-saturate at low loads to increase the inductance to put off discontinuous conduction and widen the load range over which the choke is in continuous conduction.

The net of this is that you don't really get to choose your output voltage with a choke input filter. If you have a 360-0-360 transformer, you get out 500+Vdc with a cap input filter and 509*0.637 = 324Vdc with a choke input filter. Anything in between is dramatically dependent on the exact value of load. You might accidentally get something that acted in a way you liked, but that is what it would be - a happy accident.

Choke input filters are great *if* (1) you like the voltage they give you after you get to continuous conduction (2) you can afford the necessary choke in terms of space, availability, and money.

There are some other options.

1. Use a dropping resistor. Cheap, but voltage wavers around a lot, and has the possibility of generating a lot of heat.
2. Use a zener or an active zener. A bit more expensive, but has the virtue of letting you choose exactly what voltage you want dropped, and letting you even set it to whatever you like, with competent design. Wastes the same amount of heat as a dropping resistor at the same current/voltage.
3. Use a bucking transformer on the primary of the power transformer. Go read "Vintage Voltage Adapter" at New Page 1. This lets you dial down the *primary* voltage into the power transformer in steps, and do it in a way that has no wasted heat. However, it has the potential to mess up your heater voltages if you have an integrated power transformer with HV and heaters in the same one, so the complexity goes up.
4. Use a voltage regulator. An LM317 can be set up to directly regulate hundreds of volts; I've done it. It's only slightly more complex and expensive than an active zener. It can be "dirtied down" with a series resistor to restore sag if you like.

Which one to choose depends on what you want out, and how much trouble you're willing to go to.

[J M Fahey]

That's quite true.
Then, what did the old timers do?
Did they burst their expensive electrolytics every other day?
No, they avoided the problem by skipping it: they specified a "minimum load current", above which the choke was a faithful partner and not an enemy.
Now I don't remember it, but there *was* a formula to calculate that.
That minimum safe current was more than met by the power tubes idling current, somewhat aided by the few mA for the preamp.
I also remember "swinging chokes", which were, say, 5Hy at low current, dropping to, say, 2 Hy when loaded.
They sure saved weight (and $$$).
All this straight from the mists of time.

[EFK]

The swinging choke concept seems seriously interesting. Seriously!

[R.G.]

That's quite true.
Then, what did the old timers do?
...they specified a "minimum load current", above which the choke was a faithful partner and not an enemy.
Now I don't remember it, but there *was* a formula to calculate that.
That minimum safe current was more than met by the power tubes idling current, somewhat aided by the few mA for the preamp.

Well, that's kind of a problem. You're absolutely correct, the idling current of the power section is enough to be a reasonably sized minimum load. The problem with that approach is that tubes burn out, and get pulled out of their sockets. In tube equipment, the minimum load current drawn by the tubes is always *zero*, at least in terms of what happens to the power supply.

You can't rely on the tube currents to keep the choke current up, especially in this day of clueless junior tube Einsteins wanting to pull tubes because they heard that amp mods were great and they want a mod all their own.

When the tubes are out, or even worse when MOST of the tubes are out, the power voltage pops right back up towards the 1.414 times the input voltage instead of the 0.637 times the input that the choke makes it if there is enough current. This has the effect of killing anything that's left on the power supply that is voltage sensitive, like preamp tubes (12AX7 is rated for 300V on the plates) and any filter caps that are sized for the "normal" choke filtered voltage. Nothing like having your power supply voltage more than double because you pulled out your output tubes.

I'm old enough that the power supply design course I took in my second year had the design of choke input filters in it. The only good option recommended was to put in resistors to keep the minimum load current needed to keep the voltage down, and to use swinging chokes to keep this critical load current as low as possible.

I also remember "swinging chokes", which were, say, 5Hy at low current, dropping to, say, 2 Hy when loaded. They sure saved weight (and $$$).

They really did! Of course, the problem was that people were trying to figure out how to not have chokes at all, and worse yet, picky-to-design chokes with defined cutoff currents. The design of a good swinging choke is a bit trickier than the design of a good power transformer. I had to do a few paper designs of these, but I didn't like it.

Then when you get the swinging choke design done, you have to worry about actually winding and stacking them. The gap on a swinging choke is as critical as the gap on any inductor, plus you get to add to that the stacking of the magnetic shunt that bridges the gap to get the higher inductance at low currents.

You can do much the same thing with two chokes - one a high-inductance but modest current rating, and a second lower inductance but higher current rating. When the first one saturates, the second one will take over --- if you've done the necessary design work to ensure that you really do know what and when and how this happens.

The swinging choke concept seems seriously interesting. Seriously!

It absolutely is. It's a GREAT idea - that needs a somewhat obscure design procedure to get to work and either custom magnetic parts or cut-and-try DIY work because there aren't any Hy-sized swinging chokes commercially available.

Swinging chokes are a great idea - if you don't have a budget, a schedule, or a profit margin target. I don't have either of the last two on my amps, but I can't get away from that budget thing. After a few tries to get a well-made custom magnetic part, I suspect that most amp builders would find a renewed liking for an adjustable solid state whatsit and a heat sink that just lets them dial in what they want.

Oh, and if you're getting custom magnetic whatsits built, why not get the right power transformer voltage made for you instead of a fancy choke?

How about a less brute force approach. Use a choke input filter, where the B+ hits the choke before the first cap. This will drop the voltage a LOT but you can then bring it back up with a smaller cap, like <1uF before the choke. Simply adjust that first cap to get what you want.

Think about dVout/dIout for a few minutes.

Makes for a very low ripple, stable supply and is easy on the rectifier.
By far the best suggestion in my opinion... Also, you don't have all that waste heat to deal with by using a FET....regardless of how well it's done...

I actually graded the labs for a group of EE-lings who were designing choke input filters as part of a course. They all thought that the choke input filter was a GREAT idea too.

Until they had to build one. It got a lot quieter as the deadline for getting the project to work approached.

Power FETs were not available at the time, but they'd all have gone for the nasty heat issues of a power FET in an instant after they got the choke input filters working. But then they were consciously trying to learn.

[jpfamps]

Very sorry to disagree, but a choke in series with the rectified voltage will pass current, with a DC component and an AC component.
That AC component, which causes ripple, will be much less than the DC component; a case which might be compared to ripple voltage in a capacitor being less than its DC value.
In a properly specified inductor, Impedance will be much higher than its DC resistance, that's what it's there for, or it would lose its job to a much cheaper resistor.
Won't talk about cost, because a specially wound choke might cost the same as a mass produced transformer, but that's a market size problem, not a technical one.

The AC current is a significant issue with lower value chokes.

Using Duncan Monroe's power supply design software, with a 330 VAC input, a 5H choke, a 47 uF cap and a 5k load, the DC current draw is 60 ma, however the current through the choke peaks at 120 mA, ie twice the DC load!

Of course the simulator assumes that the choke won't saturate.......

[Steve Conner]

The peak current is always twice the average in a choke input filter at its critical current, because of the mathematical properties of sine waves.

The choke current is a 120Hz sine wave that averages to the DC current, and at the critical current it just dips to zero, therefore it must peak to twice the DC value.

As RG says, choke input filters are best avoided nowadays. There are two other options I could add to his seemingly exhaustive list -

A power supply with lousy regulation patched up by a SCR controlled rectifier, like Carver's Magnetic Field amps.

A resonant choke filter of the kind beloved of radio hams.

Of course these suggestions are both tongue in cheek. When I had excessive B+ to deal with, I used a MOSFET dropper.

[J M Fahey]

Well, I was talking about *real* "old guys", not "new guys" pretending to be them.
Way back then, musicians did *not* tinker with their amps, just used them to play, go figure.
I'm talking 50's and 60's players.
There was no "retro" fade either.
And designers *did*:

The only good option recommended was to put in resistors to keep the minimum load current needed to keep the voltage down,

As a perfect practical example of late 40's amplifier, look at thisBremia S30.
It has a choke-input supply; 550+550V PT secondaries (no, it's not a typo) for a 360V (full load) B+:

Look at the adjustable 20K 25W wire resistor, first point in the load chain.
To see it in real life:

For the full thread:
http://music-electronics-forum.com/t22192/
I got many such amps to repair in my early days (think late 60's).

[tubeswell]

look at thisBremia S30.

Wow! A South American tube amp - Cool post J M! 8-)

Did you have to repair lots of them because of the choke input filter? ;-)