zener regulation ?

[shooter]

In a *standard* PS PI network, would adding a zener aid in ripple rejection? 

[eleventeen]

"Under some conditions". If you picture the standard waveform produced from a full wave rectifier it would theoretically chop off voltage peaks which were higher than the zener breakover voltage. If the troughs in the input waveform were below the zener's BO voltage, those would be unaffected. So it *could* reduce ripple; but it could also produce some on/off transients that as a practical matter might be more troublesome than standard 60/120 Hz ripple. A zener in that config would prevent the filter output volts going north of the zener's BO volts.


If you showed me the schematic and asked "what's that zener for" I would NOT say "to reduce ripple", I would say to limit voltage to no more than whatever the zener's BO voltage rating is. It would leave alone, or "fail to regulate" voltage levels below the zener BO voltage.


Also, you are sort of comingling the terms "regulation" (which means maintaining constant output voltage or current level) and "ripple rejection" which means smoothing out noise and normal rectifier ripple from what is probably intended to be a smooth, level DC output. Normally, we would want to have both of those things present in a supply we call "regulated", and some parts of the ckt will do the regulating and some will do the smoothing.

[HotBluePlates]

In typical regulated power supplies, a zener is used as a reference voltage for the "real regulator circuits". In an old tube regulated power supply, the zener's circuit position is occupied by "gas discharge tubes" (like 0A2, 0C3, etc) which perform the same function of providing a reference voltage.

[shooter]

Thanks, I'm thinking about a Fil DC supply, trying to get *near 0* ripple, which rc / lc do well, my mind's eye was seeing the zener clamping anything over 6.3, AFTER the PI does it's thing.  I didn't consider the transient spikes, ( note to self, add a .1uf   )

[eleventeen]

Go ahead and experiment with that but I predict you'll find it a serious challenge.....that accomplishes little or nothing if indeed you are able to get it to work.


Why? First of all, let's assume you're talking about a Princeton Reverb = 2 x 6V6 @ .600 amps ea + 4 x 12AX7 @ .3 amps you are proposing to build a regulated 2.4 amp (minimum) supply, we had better call it 3 amps. You are going to have trouble regulating 3 amps without a real heat-sunk pass element, a 2N3055 on a good sized chunk of heatsink or something similar. 


Regulators have a characteristic where they have to "throw away" a few volts. An LM7812 3 terminal regulator need about 3 volts more than the 12 volts it is intended to put out. MORE input volts if the current draw is higher. A real, honest regulator circuit that is intended to put out 6.3 volts DC, you probably have to feed it at least 9 volts, probably more. So...you have to have a bridge rectifier (as you show) to get your 9+ volts (after filtering) but now you have only half the ampacity from your heater winding. If you have 2 qty 6.3 windings, OK, better. But now you have the 4 amps AND a higher voltage drop to get rid of. More heat sink, more careful design.


I'm not going to hold myself out as a master ckt designer but I have tried to do this, and it's not easy. It's definitely not easy if you start with a regular 6.3 winding or two of them in series. The voltages are awkward and force you to use a fat heatsink on your pass element. No TO-220 thingy (like a 7812) sticking up on a bit of sheet metal. You will need a fat heatsink, or, the chassis itself.


Meanwhile, those who have made DC circuits for heaters in tube guitar amps report very little improvement in noise. So bottom line, it's very likely to be a load of work for little gain. That doesn't mean you should not try it, because you'll learn some stuff about rectifiers.

[shooter]

you'll learn some stuff about rectifiers.

I've built a bunch of 5v 5A back in the '80s for trans/ic stuff, I've done 1 dc fil but couldn't get ripple where I wanted it.  The main ponderings here was can a zener *smooth*  - and by smooth - eliminate AC fluctuations any better than rc .   Oh, btw I can mix my metaphors with the best of them 
 

comingling the terms

All my pondering are now put to bed, 3-0-3 vac it is!
thanks guys for the schoolin

[HotBluePlates]

you'll learn some stuff about rectifiers.

I've built a bunch of 5v 5A back in the '80s for trans/ic stuff ... can a zener *smooth*  - and by smooth - eliminate AC fluctuations ...

R-C eliminates fluctuations when voltage input drops by releasing stored energy from the cap into the load. But there's a limit to how well that works, which is why there is residual ripple (which is bigger with a heavier load). Zeners clamp a voltage from exceeding breakover voltage, but are noisy (which is why there is almost always a film cap in parallel with a zener in regulator circuits).

Anyway, heater regulators are a tough problem. Even if they only drop 3-4 volts from the applied d.c., the high current means large heat dissipation in the pass-device. There are approaches some use for unregulated d.c. heaters (total heater draw equals the current rating of the winding), and you may find some ideas in guitar amp designs which use d.c. heaters.

[vibrolax]

I made a 12V filament regulator using the LT1085 LDO and low forward drop diodes MUR-410G, I made a 12V filament regulator.
 
It's a 3A regulator, but my heatsink was sized for about 1A.
My design narrative: http://www.frontiernet.net/~jff/SonOfSVPCL/DIYSVTBassPreamplifierPowerSupplies.html
PDF of layout: http://www.frontiernet.net/~jff/SonOfSVPCL/Images/Visio-AmpegSVTPreamp_PowerSupply_26Aug2014.pdf

Using schottky diodes and even lower dropout voltage regulators that we have today, I believe it's possible to make a 6.3 VDC regulated supply from a 6.3 VAC winding.

You do have to use a lot of filtering capacitance before the regulator to minimize the ripple voltage.  It's also marginal for staying in regulation with low line voltage, say 10%.

[shooter]

I was gonna use dc in my 2nd mono-block, the 1st is near dead silent with AC, so it's probably a fools errand but then again....  amp is still in  the pencil whip stages and I have lots of erasers


 

[vibrolax]

I'd agree that regulated DC filaments are not the most significant feature for a quiet amp.  Grounding, layout, transformer orientation are probably more significant.

[eleventeen]

After having tried, fruitlessly, to do this a few times, it's my strong feeling and belief that you're about 3x better off trying for a regulated 12 volt supply out of an 18-21 volt source than trying for a 6 volt supply out of a....?v source. It is the relationship between the regulator overhead (the "throwaway" factor) and the final voltage level. In other words, let's say you have to throw away 3 volts in either case. That 3 volts is huge compared to a 6 volt output relative to a 12 volt output. The higher output volts means you're drawing 1/2 the overall current which lightens the req'ments for the pass element. Heck maybe you could get by with an LM350 (3A) or an LM-338 (5A) which would make your life much easier. Still have to heatsink the daylights out of either of them.

[vibrolax]

It wasn't difficult to get 12 vdc regulated from 12.6 VAC, including low line voltage margin. I couldn't immediately find a very low dropout reg in an easy through hole package, but I'm pretty sure I can get a diode / cap / reg design requiring maybe1.5v or less headroom.  I don't have an immediate need for such. Do you think a demonstration of what's practical with modern parts would be of interest?

[eleventeen]

With the understanding that for the most part, the effort doesn't achieve much, yes, I think lots of folks would be interested. Hopefully you could just supply and perhaps enhance whatever design effort you did to get your results. To rebuild it from scratch...I wouldn't ask you to do that.

[shooter]

the regulator overhead

I wasn't thinking regulator, just the zener - as the regulator, and caps and resistors.  my goal was to get fil ripple close to the single mV range since this will be a class A PSE which doesn't offer the same "cancellation" you get in PP.  the last amp - AC fil, is pretty close, although I'd have to open it to verify and, well, it works just fine so it stays closed,  playing LP's, mp3s, and even guitar!

[vibrolax]

With the understanding that for the most part, the effort doesn't achieve much, yes, I think lots of folks would be interested. Hopefully you could just supply and perhaps enhance whatever design effort you did to get your results. To rebuild it from scratch...I wouldn't ask you to do that.

I attempted a "paper" design for a regulated 6.3 VDC @ 3A filament supply, using a 6.3 VAC transformer.
Using a bridge rectifier made with 1N5822 Schottky diodes (40V/3A), we lose about 1V in the 2 diode drops
Using an LT1764 regulator, the dropout voltage is less than 0.4V @ 3A
So 6.3 VAC * 1.4 - 1 - 0.4 = 7.42 VDC (ideal unfiltered)
In reality, we get crushed by 5 more factors:  ripple voltage, ripple current, ESR of the capacitors, transformer winding resistance, and low line voltage.
62,500 uF of filter capacitance gives about 0.4 V p-p of ripple, and 4A rms ripple current.

C (farads) = load current / (freq * ripple_c_pp) = 3A / (120 Hz * 0.4) = 0.0625 farad
To minimize ESR and to divide the ripple current, this can probably best be done with 10 x 6200 uF ecaps
Oops, our diodes don't have enough continuous current capacity, so we'll have to parallel them, or use ones rated for a higher continuous current
In order to minimize the transformer secondary winding resistance loss and have enough current to charge the reservoir capacitance, the transformer AC filament current rating needs to be 2X or more the desired DC 3A output.

Since it takes a boatload of capacitance and current to get a low enough ripple to stay in regulation (that's not even considering low line voltage), the PSU is going to cost a bit.   We can build an entire amp for what some of these directly heated triode hifi guys (300B's, 2A3's, etc) spend on their current regulated DC filament supplies.

So my regulated 12V supply design with twice the voltage and 1/4 the current was a piece of cake compared to this.
I still think it's possible (unless we want to have margin for low line voltage), but not very economical.

If I wanted 6.3 VDC regulated at 3A, I'd use a separate filament transformer with enough voltage margin  to allow more ripple and line voltage variation.

Thanks for encouraging me to work out the details.  This is good stuff to think about.[/s]

[jjasilli]

I wasn't thinking regulator, just the zener
But regulation reduces ripple.  See, e.g.:



http://www.instructables.com/answers/How-can-I-reduce-AC-ripple-current-in-a-power-supp/


http://www.teamwavelength.com/info/powersupply.php


(Google:  "does a regulated dc supply reduce ripple")

[shooter]

We can build an entire amp

Thanks Jon for the math, and JJ for the links, since I'm more ADD than anal retentive I think i'll just keep it AC     I was just hoping for an "easy fix"

[shooter]

I think i'll just keep it AC

Or not?   I was following the Vox AC30C2 thread, Sluckey posted the schematic and the amp uses DC fil for the preamp.  The PA section is still AC, (because of current?)
I drew up the AC to DC part of the vox below.

So maybe I *steal* once again.

[vibrolax]

It is indeed DC, but it will have ripple and it will be unregulated.
What RMS voltage you are going to get depends on what diodes you use, the filter capacitance, and the transformer losses.
You can model all that in Duncan's PSUD II.

I see there are some 100 nF film caps in there to try to suppress the diode switching noise.  How much of that you get depends on which diodes you choose.

[shooter]

it will have ripple

I got a couple carcasses laying around I was gonna experiment an see what I get.
Did I *read* the schematic right, it is 6VAC to 6VDC?  the notation off the Vox tranny has each leg labeled ~6v.
Also I couldn't find SG1 ground anywhere else besides the artificial CT, guessing SG1 is the preamp common tie point?

[vibrolax]

I'd say that sg1 is the preamp signal ground.

I have not previously encounteted the artificial center tap formed by the 220R's before the rectifier used in conjunction with the second artificial ct formed by the 3k3's.

[shooter]

I have not previously encounteted the artificial center tap

Had me baffled for awhile, I even started drawing it as the *input*, thinking Vox drew the bridge all wrong  .   I like *oddities* in schematics, usually steal a couple from here and there and glu'em together