Voltage Doubler Power Supplies

[AlexHempel]

Playing around with the doublers on PSUDII using Hammond iron I am a bit intrigued. With some of Hammond's high-current iron you can get crazy low secondary resistances and the choke has an easy job in a doubler. Far less than a henry looks highly effective at ripple reduction. More caps needed, but the voltage ratings are lower, which is nice. Never had much interest in doublers but I might try one in a new build I'm working on that requires really good regulation (sweep tubes pure pentode).

 

[TriodeLuvr]

Why then wouldnt Marantz, McIntosh, and HK have used Graetz bridge FW silicon rectifiers instead of doublers? Marantz doubled them in series in doublers in the Model 8 amps for extra margin. Many other period examples from the early period of silicon rectifiers used doublers.

I can't tell you why those companies selected doublers or why so few companies continue to do so. I can only tell you that diodes used in a full-wave doubler do not experience higher PRV than those in a FW bridge or FW CT. A careful "manual" analysis of the three circuits will confirm this, or SPICE can be used.

Jack

 

[trobbins]

Stick with ss diodes in PSUD2 for the doubler configuration, as there is an inconsistent response for valve diode models that Duncan is looking at - just in case you were lured to a valve diode configuration for some reason.

 

[AlexHempel]

As much as I love valve rectifiers, I wouldn't stick them into a supply designed for good regulation! But thanks for the tip -- good to know.

 

[TriodeLuvr]

More caps needed, but the voltage ratings are lower, which is nice.

There are two reasons doubler caps must be so much larger than those in other configurations. First (and most obvious), the two caps are in series. This halves the effective filter capacitance from B+ to ground. Second, although the ripple frequency at the output of the supply is twice the line frequency, the ripple current in each doubler cap is at line frequency. This halves the smoothing effectiveness of the two caps. I believe this is the reason Marantz used a small choke and a reservoir cap following the doubler.

Jack

 

[gadget73]

Lots of them just had a single cap to ground across the entire thing after the doubler caps. Not as good as a choke and a cap, but better than nothing. Talking ancient times so doubler caps would often be something like 150uf each, and the cap across the whole works might be 50uf.


as for why companies may have used a doubler rather than a bridge, it was probably cheaper. Less copper in the transformer and half as many diodes.

 

[stanko]

Lots of them just had a single cap to ground across the entire thing after the doubler caps. Not as good as a choke and a cap, but better than nothing. Talking ancient times so doubler caps would often be something like 150uf each, and the cap across the whole works might be 50uf.


as for why companies may have used a doubler rather than a bridge, it was probably cheaper. Less copper in the transformer and half as many diodes.

It's also easier to wind a transformer for half of any given voltage. Less reliance on the varnish insulation, as well as less worry about arcing to the iron. This means one can wind a layer closer to the end of the core. In those days, wire was sometimes referred as either "single dipped" or "double dipped", raising the insulation value of the varnish coating. Also less paper insulation was needed, making for a more compact core and a smaller transformer.

 

[AlexHempel]

Yeah, I think @TriodeLuvr was onto something regarding the performance merits of the doubler due to lower turns ratio. As to why it's not used today, my guess would be the following: doesn't make a ton of sense with tube rectifiers (which people like), which leads to lower demand/market and fewer transformers meant for doublers, as well as less general knowledge and discussion of doubler designs/merits. Also, high voltage caps are pretty expensive while diodes cost nothing, so now the doubler is going to cost significantly more than a bridge even if it regulates a bit better. Not many people are building high power pure pentode amps where it's worth the extra capacitors and hassle of a doubler.

 

[Pio1980]

I can't tell you why those companies selected doublers or why so few companies continue to do so. I can only tell you that diodes used in a full-wave doubler do not experience higher PRV than those in a FW bridge or FW CT. A careful "manual" analysis of the three circuits will confirm this, or SPICE can be used.

Jack

Caution, perhaps?
Multiplier stacking also allowed higher voltage outputs than elyticap limits allowed, as in amateur radio transmitters.

 

[GordonW]

It's also easier to wind a transformer for half of any given voltage. Less reliance on the varnish insulation, as well as less worry about arcing to the iron. This means one can wind a layer closer to the end of the core. In those days, wire was sometimes referred as either "single dipped" or "double dipped", raising the insulation value of the varnish coating. Also less paper insulation was needed, making for a more compact core and a smaller transformer.

And, generally, transformers couple better, the closer they are to a 1:1 winding ratio.

In many cases, a doubler transformer could be less than 1:1.5 ratio. (that would be about 170v from 115V). That's pretty easy to wind for high efficiency.

Regards,
Gordon.

 

[gadget73]

Efficiency is also probably why you don't see it much today. Solid state things run much lower voltages and often split rail supplies. Doesn't really make a lot of sense to do a doubler to get those voltages when the winding ratio is less than 1:1 already.

Lots of inertia and "this is how it was done 70 years ago" type stuff going on with tube equipment so there may simply not be any real demand for transformers that work in that config.

 

[Tom Bavis]

Out of curiosity, I looked in the 1961 Radio Shack Industrial catalog. 600PIV diodes, rated for 600 mA were $6-8 each; a pair of 60 uF 250V caps were $3 vs. a single 30 uF @ 500V at $1.50. 1000PIV and higher were available, but low current or much more expensive. Surge current ratings were much lower than modern rectifiers. 600 PIV is not adequate for a 300V+ supply, so four would be needed for a doubler, eight for a bridge. Transformer cost might be the same for the same VA rating, so doubler made sense. (BTW, 5U4GB was $1.05 in 1961)

 

[GordonW]

Yeah, I think @TriodeLuvr was onto something regarding the performance merits of the doubler due to lower turns ratio. As to why it's not used today, my guess would be the following: doesn't make a ton of sense with tube rectifiers (which people like), which leads to lower demand/market and fewer transformers meant for doublers, as well as less general knowledge and discussion of doubler designs/merits. Also, high voltage caps are pretty expensive while diodes cost nothing, so now the doubler is going to cost significantly more than a bridge even if it regulates a bit better. Not many people are building high power pure pentode amps where it's worth the extra capacitors and hassle of a doubler.

Yeah, I do note that most high-volume tube amp makers, seem to be using bridge-rectified supplies nowadays. Easy, and inexpensive parts costs, for basically the same performance as a doubler. Once you're using the entire secondary on each half of the conduction cycle, you've pretty much picked all the low-hanging fruit. Going from there to a doubler, isn't that beneficial, anymore.

Regards,
Gordon.

 

[trainbuftony]

My most recent build I used a solid state "full wave" I believe it's called rectifier.
It consisted of 2x 1n4007 diodes from the b+ winding to the first cap.
Just like the tube would have done.
I measured 4v of drop on b+ between idle and full 20 watts.
On a nearly 400v supply, that is nearly 1% regulation. If you told me a Voltage doubler could get you to 1% I'd certainly believe you because in theory it should have the lowest power supply impedence of all.

 

[TriodeLuvr]

I measured 4v of drop on b+ between idle and full 20 watts.

The voltage drop seen in practice is also a function of the difference in no load/full load current draw. Many push-pull tube amps draw only a little more current at full output, because the tubes idle close to Class A. This can create the appearance of substantially better power supply regulation. Actually, there's no reason any of the PS configurations can't regulate equally well, it's just a matter of component values and quality.

Jack

 

[AlexHempel]

It strikes me that in these regulation discussions it probably makes more sense to discuss supply equivalent resistance than percentage regulation, because percentage regulation is highly dependent on your output stage bias. A class A amplifier achieves near-perfect "regulation" by simply not varying the output stage current draw. In contrast, not even the most ridiculous supply is going to hit 1% regulation if your load current varies from like 50 ma at idle to 300-400 ma at full load, as you might see in a parallel pp amp making 100+ watts.

 

[AlexHempel]

After playing around with doublers and bridge supplies this morning using Hammond transformer specs, the doubler configs did have very slightly better regulation due to insanely low DCRs for some suitable transformers (such as a 500 VA isolation transformer with a .4 ohm DCR secondary), but not really worth the extra expensive caps for a few volts difference. However, one benefit of the doubler is the ability to use weird low voltage transformers that can be cheap and useful for other reasons. For instance, the isolation transformer mentioned above has a bunch of taps and I could use one to build a separate high-current unregulated screen supply.

 

[trainbuftony]

So the answer to this is to run em class A.
I can live with that.
Also, the main factor that goes into me choosing a power transformer is cost...
Do I have one that will work?
Great, cost is zero. If I don't have one, that puts me at the point of having a clean slate to design from. In that case I'd strongly consider a Voltage doubler.
Since my builds usually take the form of "what can I build with this iron I've been saving" I have not had the opportunity to do much experimentation with them. Eventually, in theory, my supply of organ transformers, scope transformers and various other useful transformers will run out and I'll be in the market to actually purchase them. Also, the sun will eventually burn out and life on earth will cease. Neither of these things is likely to happen any time soon.

 

[gadget73]

Another handy bonus from voltage doublers when dealing with pentodes, you have a convenient point for the screen supply, tapped out of the middle of the supply. Its going to be 1/2 the B+ which may not be the proper voltage but if that happens to work out its a simple way of powering them without also having regulation issues or needing anything even remotely complex. Probably going to need a choke though since its a half wave rectifier at that point which means a fair bit of ripple. Not super hard to filter that though, screens don't need a huge amount of power so the choke doesn't need to be a massive thing to get a decently quiet screen supply.

 

[TriodeLuvr]

Another handy bonus from voltage doublers when dealing with pentodes, you have a convenient point for the screen supply, tapped out of the middle of the supply.

That point doesn't provide sufficient voltage for screens in many designs, and I think it might be problematic to draw the requisite current if it did. However, I find it useful as a "half voltage" source for the resistive divider that elevates the filaments.

However, one benefit of the doubler is the ability to use weird low voltage transformers that can be cheap and useful for other reasons. For instance, the isolation transformer mentioned above has a bunch of taps and I could use one to build a separate high-current unregulated screen supply.

I don't know of any readily available OTS transformers suitable for use with a doubler to provide, say, +450V for a pair of 6L6GC. This needs about 180V RMS with the necessary VA rating. I'm building with mostly vintage HK iron at the moment, but when those are gone, I'll probably have to fall back on FW CT designs.

Jack