• WARNING: Tube/Valve amplifiers use potentially LETHAL HIGH VOLTAGES.
    Building, troubleshooting and testing of these amplifiers should only be
    performed by someone who is thoroughly familiar with
    the safety precautions around high voltages.

CLC vs CLCLC Filter Power Supply

It all depends on the current draw, really. Of course adding more C and more L is going to squash those darn ripples better. Is it needed?

What's it driving! If some nice push-pull final, then the ripple is in common to each phase, so rather significantly cancelling. Differential type amplification is like that. It is one of the reasons (apart from efficiency) why it has persisted so long, and was invented so early in the Tube Game. Even though significant (1% or more) power supply ripple might be in play, when there's no input, you cannot hear "the hum". Entirely different for single-ended amplification.

Now this is not to say that "no sound = no effect". The wavering total B+ has an effect on the instantaneous amplitude of the final-phase signal stream. This results in a kind of modulation of output, when there's output. If one where to have a single sinusoidal input to an amplifier at say 70 Hz (with 60 Hz American continent mains), then with 1% ripple, you definitely could hear a 10 Hz "beat" imposing itself. Not musical. Hence why a good design, using a combination of "whatever's cheapest and most effective to accomplish the end" is used.

Again, what's the nominal and peak current draw?

How much ripple is tolerable depends on the design of your amp and the efficiency of your speakers. With my 300B SET design and 87 dB (1 W, 1 m) efficient speakers, I found that I want the B+ ripple below 1 mV (preferably below 500 uV) to avoid audible hum at the listening position.

That's one data point...


The B+ ripple? or the ripple at the speaker terminals? The OPT divides the ripple by the stepdown ratio, so 1mV with a 5K:4 stepdown = 28uV at the speaker!
Bought those 100TL's yet? I see why you're looking at 1,100 volts - but really they like to be driven closer to 1500 for optimum performance. And your output transformer is going to have to be one crazy-big hunk of iron. 300 VA, minimum. Clearly huge. And then there's the 25 watt grid-drive requirement. 100TLs are, like many direct heated thoriated tungsten transmitting tubes ... designed to run positive grid for much of the amplification cycle. That's where the current comes from.

And your grid drive needs to be at least 275V RMS in the A/C domain. Maybe allow for musicality excursions up to 400V. The response of the output tube will be quite nonlinear though at those power levels.

If this were 1951, and I were a electrical engineer tasked with designing an amplifier using 100TL's (because of some economically compelling factor say) FIRST, I'd ask, "why single ended again?" second, "because I said so" .. then I would be working hard to design an amplifier with a LOT of gain, well above what it strictly needs, so that a bit of global negative feedback would tame the nekkid triode nonlinearity. 20 dB worth, maybe more.

And, incidentally, as is ALWAYS the case with NFB, guess what? Power supply ripple is markedly reduced further by the amount of NFB. A great result, actually.

But I'm supposing you're following a daydream - a great massive SE triode amp using one honking red-hot triode. The cost, by the time you're done (for 2 of them) is going to be ... breathtaking. And you'll still only get on the order of 75 clean watts without NFB. Well, all for the glowing bottles.



2005-08-21 10:46 pm
Yes, it does. :rolleyes:

For a start, I would suggest listening to red-plating triodes of any size for at least two hours per day... :)


Now off, chasing some hummmm... Hope I nailed it down by changing grounding scheme and twisting any wire pair that didn't manage to run away.
... show me anyone ...

I don't either. I really do prefer using combinations of technology to design and implement more ideal systems. I do. One of the more interesting ways to do this is to use combinations of FETs and tubes. (After all, no matter how they're described, "tubes" are basically giant high voltage FETs in their characteristics. Or, FETs are just tiny triodes and pentodes in their action). Even BJTs are fine, and have very powerful design characteristics that can (and I argue should) be incorporated.

After all ... how many times do we have to see a "CCS loaded" or some such design here, that incorporates some (essentially internally mysterious) chip to accomplish in a near-perfect fashion, that which is easy to write down ("CCS"), but really, really really hard to implement with conventional tubes to the same degree? And what devices are really working in these mysteriously designed metablocks? Why... FETs, BJTs, diodes, and resistors, of course. Lots of 'em.

So... for this reason, I find it both sweetly compelling and remarkably abstractive to substitute a big high-power MOSFET for a 100TL. You still can get some starving transformer winder to cobble together the 10:1 stepdown audio output transformers, and you can with enough heat sinking, run them happily right in the middle of the 1200 volt band (need 2 in series, of course), dissipating 100, 200 "plate watts", and responding remarkably linearly to the MUCH LOWER drive voltage. No power input, either ... just voltage. Much sweeter design spec, and no loss of the fundamental 'stuff' that makes single-ended amplification so appealing.

Is this going over to the dark side? Nah.

Moreover, if you really want to be a tube-a-holic, then fiddle not with antique (and rather poor) 1940s triodes, but just haul out a hundred bucks, and get ahold of a pair of 4CX250s on EBay. An entirely superior tube, tetrode, capable of running in the zillion-volt range. Huge power dissipation, and a very cool look as the silver plating gradually oxidizes through browns and golds. Still need the big fat 500VA transformers though.

And lots, and lots, and lots of very-high-voltage experience, not to find yourself fried to a crisp from those 2 kV plates. Yet, what a beautiful shrine to vacuum tube technology a pair of monoblocks using these would make. Not even push-pull, just wastefully single-ended. [The problem is, of course, that you'd only want to run the unit in cold weather, as a room heater. Summer? EXTERNAL venting is a must! If the hot effluent never mixes with the room air, well, the Air Conditioner has less work to do.

And yes ... I still prefer using big heat sinks, big MOSFETS, conventional high-voltage/low-current finals. Solid-state people think its nuts (but then they angle inevitably toward more efficient AB push pulls, which isn't the point) ... but it is not. There is something remarkably interesting about what a large output inductor does to current flow, and to driving the remarkably quirky Z and Q characteristics of 2-way and 3-way speaker setups.

TO YOUR HUM question?

USE A REGULATOR. That's the sole, and right solution... not just more C's and L's. Again, our friend the MOSFET comes out for the fun ... along with a 1200 volt voltage reference... which need be nothing more than a nice resistive divider (on clamp output) series resistor (for shunt regulation), the MOSFET (for shunting), a heat sink, and a bit of medium-voltage OP-AMP offset amplification. All quite doable, and it gets your ripple down to microvolts.



2005-08-21 10:46 pm
The hum is gone :)

Without regulator *g*, just doing a proper wiring instead of the chaos it was before...


PS. I basically agree to what you wrote, my post about that dream of everyone was meant funny ;). Although, your words about the 4CX250 sounds as if you got that disease as well...


2010-06-06 8:17 pm
Please stay on topic guys.

I am simply trying to learn more about ripple voltage and its effects.

The 300mV I am getting is from peak to peak. If I understand correctly, the actual ripple voltage is measure from the center, or half of peak to peak. I I actually have 150mV ripple with CLC.

Back to my questions:
1. What are all the effects of high ripple voltage ?
2. What percentage of ripple to output voltage is good ?

According to Morgan Jones' book, 5% - 20% is ok (page 345, 4th Edition)
I find this VERY high. Maybe I am misunderstanding.


2005-08-21 10:46 pm
Back to topic.

In a push-pull EL34 setup, I have approx. 35mV p-p (13mV RMS) sinusoidal 100Hz hum on a B+ rail of 400V.

Output hum is practically inaudible, you have to put your ear into the cone of the woofer to notice it.

You have approx. 4 times the B+ voltage, so a hum figure of a few mV should be extremely good.


EDIT: Besides the hum, I noticed some low-frequency variation of the B+ voltage (timescale of seconds) with excursions as large as approx. 2-3V. Seems the choke-input supply is not very good at rejecting LF variations of the mains. Or do I disregard any other sources for that variation?
Last edited: