Basic Unregulated Power supply - designing

[maxwall]

Looking to take a power transformer ( pri 120Vac) that outputs 6.0vac secondary unloaded at 4 amps and supply
DC voltage to several tube heaters for a guitar amplifier design. During the design of this
prototype power transformer I had it built to output 5.6-5.7 VCT which is what it supplies when loaded. I
went with this odd secondary because I was having trouble with elevated secondary voltages due to
elevated mains voltages ( 123vac mains ) here in central California. As a example, 123vac mains would usually
put a normal 6.3vac secondary center tapped heater voltage around 6.7-6.9 vac loaded so I asked that the secondaries tap be loweredto compensate for the elevated mains voltage which was the source of the problem. Long story short, I now have a
transformer with a secondary tap that is too low and as a result I'm seeing 5.6-5.7vac on the secondary heater tap.
This heater winding has a center tap to keep noise low since a floating ac voltage usually induces hum into the overall
sound of the amplifier. The design goal now is to use a half/full wave rectifier with this center tapped secondary to produce
a voltage closer in a range of 6.0 - 6.3 voltage dc loaded.  I'm expecting the new DC voltage scheme will help the amp run even quieter than one powered form AC voltage, to what extent I cannot be certain. The total heater/filament load in this amplifier is 2.7 amps (rounded to 3 amps in the last equation below), the transformer has a maximum capacity of 4 amps, about 30% reserve on the sideline for this design.​

First:
I was hoping to use a half or full wave rectifier to output a dc voltage that will get close to 6.3vdc. I know that the diodes will drop .7v (large block sized 5-6 amp diode or similiar) each but I'm not sure if my overall dc output after the filter capacitor will increase voltage to a point that approaches 6.3vdc. My intention is to boost the voltage a little using dc rather than the normal ac filament heater voltage. The AC is low as I mentioned earlier and I want to try rectified filtered DC instead. Unregulated for now.

Secondly:
I have found thru my calculations that the filter capacitor needed to meet a ripple of 2.5% and load regulation of 5% looks too large
can someone veryify my calculations  ?

Diode drop for rectifier
D2 & D4 = .7v * 2 or 1.414

calculation to find min ac voltage of seconday tap to drive  a 6.3V load. Using diode drop in above calculation 1.414 and place in calculation below.
Vsec = [6.3v + 1.414] /1.414
Vsec = 7.714 / 1.414
Vsec = 5.45 vac , nearly spot on with my existing secondary ac heater tap.


6.3v(target heater voltage) x.025(desired ripple 2.5%) = .158 Vms

Vripple = .158(Vrms) x 2.828 ( peak to peak value or 1.414*2=2.828)
Vripple = .442

Charging pulses on 60hz mains power in my case = 1/(2*60hz)
charging pulses = 1/120 or .008333(t)

C1 lets calculate the filter capacitor for this power supply running 6.3 vdc at 3 amps by plugging in the correct numbers into equation below.
C1 (uF) = [ ( IL * t )  / Vrip ] X 1000000(10 to power of 6)
C1 = [ ( 3 X 0.00833 ) / 0.442 ] x 1000000
C1 = [.025/.442] x 1000000
C1= .0525 x 1000000
C1 = 52,500 uf (in this case min 16 voltage rating since its seeing only 6.3vdc)

 

[pstamler]

My calculations suggest that if your transformer has a nominal output of 5.6Vac at 120V line voltage, 4A draw, and you use a 20,000µF filter capacitor and actually draw 2.7A, you should get a nominal DC output voltage of 6.31Vdc; when the line is 10% low you'll get 5.35Vdc, and when it's 10% high you'll get 7.23Vdc. That assumes two diode drops of 0.65V each.

Seems doable.

Peace,
Paul

 

[maxwall]

Paul,

Thanks for checking this. However I am concerned that only using one capacitor with no resistors will create
ripple noise in the amplifier. Will I need a Resistor / capacitor network of some kind to tame ripple noise in this basic
design ? I see that many supplies do have them in the early stages in the B+ section. Maybe for a filament it wont
matter.

 

[pstamler]

I think you'll have about 0.4V of ripple with a 20,000µF capacitor. Normally I'd suggest a second stage of filtering for sure, but your transformer's voltage is so low that there isn't enough voltage to compensate for the voltage drop an RC filter would create. If the ripple proves too high, you might think about an LC second stage of filtering, but again there's the DC resistance of the inductor which will drop voltage.

Why not go out and get a 6.3V @ 4A transformer from Allied Electronics? This being a guitar amp, you could probably just run the tube filaments from the 6.3V AC and, if you're careful with wire routing, be fine. If there is a hum problem, floating the filament winding up to +75Vdc should solve it. (A voltage divider on the plate supply will give you the +75V; just connect it to the heater winding's center-tap.)

Peace,
Paul

 

[Samuel Groner]

At 4A 0.65/0.7 V for the diode drop seems somewhat optimistic for standard parts--I'd use 1 V. You could get Schottky diodes with somewhat lower forward drop.

And more important: your mains won't be a neat sine-wave; the peak output will be reduced by say 10% because of waveform clipping.

Samuel

 

[PRR]

> voltage around 6.7-6.9 vac
> DC ...quieter than ...AC ..., to what extent I cannot be certain.

Oh what a tangled web you got into.

I'm not laughing at your web. I got even more tangled-up on a simple "Champ" run at odd voltages. I could not reach my B+ goal with a stock plate+heater PT, and switched to two "control" PTs, and still did not get my target, plus had live terminals all over the outside of the amp. I wuz gonna give it to someone, but too hard to make-safe (my plastic dog-house fabrication went real bad).

Just saying......

6.9V on "6.3V" heaters is not an emergency. I'd like lower, but I know the won't burn-out in many hundreds of hours. And when I did get down to dealing with it, a 0.1 ohm 10W resistor (maybe two 0.05r 5W) is cheaper than custom re-windings.

Heater burn-out is very rare even at high voltages. On power-tubes, IMHO it is a non-issue: you will always soften a screen or leak a seal long before the heater evaporates thin and fails. When you do see a failed heater on a power tube, it is just because tungsten is hard to make well, there's a thin-spot on every reel, and modern tube makers don't have incentive to do very-very well.

IMHO, well-done AC heat IS good enough for almost any guitar amp.

And simple pure AC heat gives simple pure 60Hz sine.... you can notch it in ProTools.

DC-from-AC spews harmonics of 60Hz all up the musical band. Buzzz on top of hum. You won't hear much 60Hz in a typical guitar speaker, 180Hz and 360Hz really stand out, and overtones reach above 1KHz into where the ear is most sensitive.

More points:

Only 2 or 3 stages in a guitar amp might "need" DC heat. The 6L6/EL34 certainly don't: they have 30V signals. Look where signal may be closer to 30mV: the first stage, the stage after the volume control, and the reverb recovery stage. 95% of your heater hum will be in these three stages.

That changes your DC demand from 17 Watts to 4 Watts. You already know how big a cap you need for 17W at 0.4V ripple; powering just key stages with DC reduces demand so you could use 1/4th the cap or get 1/4th the buzz.

Look at 0.65V diode drop against 6.3V load. Already you have 10% losses; probably 20%-30% because of multiple diodes and higher drops. When you start from a "DC Heat!" decision, you should re-string your heaters for higher voltage, nearer 24V-36V. That's easy with 12AX7 stages, and your 6L6 really don't need DC heat (though this does force multiple windings).

The cap-input rectifier will also cause more heating in the PT winding, though this may be negligible since total heaters is only half of total load, and you say the winding is over-speced.

Getting very low buzzz-ripple with a single cap will be critically dependent on LOW parasitic resistances around the winding-rectifier-cap loop. You will have ~~20 Ampere spikes when the rectifiers turn-on. Even milliOhm wiring resistance will turn that into hash as large as the computed/ideal ripple, and with stronger overtones at more-annoying frequencies. BTW: this is why they make 4-terminal caps: you inject the hashy spiky power at one pair, take clean power from the other pair, no resistance in common to both loops.

IMHO trying to get "DC" cleaner than well-done AC is futile until you go to more than one C. A C-R-C filter is 2-pole, a C-L-C filter is 3-pole; these knock-down the higher harmonics very much faster than a 1-C filter. Of course a C-R-C needs excess voltage to cover the R drop, a C-L-C needs a hefty choke which may not be a standard item.

You don't build your own tubes, you don't have to build your own power supply. A 12V 1A DC power supply is now a cheap commodity. Power 2 or 3 key 12AX7 from that. Some switchers throw audible hash, some are much better than others. Also 25KHz can be reduced with quite small chokes and caps, like the "hash filter" used on car radio.

I'd stop fussing and just run 6VAC all around. Then if hum is a problem, get obsessive with preamp heater wiring balance. Then if still not good, break off the key stages and a try a DC wall-wart just for them.

 

[PRR]

> 5.6-5.7vac on the secondary

Hey, you could wind-up another part-volt fairly easily.

Assuming the transformer is not enclosed (or the covers come off).

Assuming there is a little space around the winding.

A PT in this size runs about 10 or 20 turns per Volt. You want say 0.6V. That's 6 or 12 turns.

Get a few yards of #18 enamel wire, strip the ends. Thread 10 turns around the existing winding. Power-up and measure the ACV. Should be 0.5V-1.0V, something in that ballpark. Adjust the turns to your desired correction voltage and a wee bit more for loading.

Put this winding in series with your heater winding. Total voltage may either go up or down; if down, swap the leads of your booster winding.

Finish with some varnish to hold the winding. A layer of paper is good protection. Sleeve the ends of the winding: if they cross and short it effectively dead-shorts the whole PT and it may burn.

 

[maxwall]

Thanks for the replies.

I would usually run the correct 6vac voltage but I want to try to salvage this transformer since its fairly new-yeah its a fuss.
I have been running the filaments at 5.6 vac off and on , but am suspect that cathode stripping may be a issue going this
low on the filaments when they should be at 6.3vac. I'll have to build this power supply and see what the real world results will be. I can always drop down to some tubes that run on 4.7-5 vac although there are not many preamp tubes made to run at these voltages , but there are several choices to try out. As I suspected by the replies, it appears there are always some other things to take into account other than just the calculations. Its nice to get some insight into the DC side of it.

The first build I'll try a half wave rectifier (schotky types looks good) using the center tap & secondary 5.7vac leads into the large filter cap and lastly filaments run in parallel as usual.