• 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.

HV smoothing capacitors


2001-08-06 3:29 pm
Hi everybody,

I'm about to start my very first tube amp using a s.e. EL34. I'll use a four-diode rectifier for the HV (about 400V), now I wonder: what is the correct capacitance value for the smoothing capacitors before and after a 20H inductor ? In some diagrams I've seen combinations of up to 1000uF/450V(electrolitics) while in other just 25uF MKP capacitors. What is your opinion ?

Thanks in advance for your replies,



2001-11-27 8:10 pm
Since you are doing a Single-ended design, your current draw will be (nearly) constant, so you actually dont need much capacitance. Only push-pull amps need thousands of uF.
The first rule of thumb is to put about 1/3 of the capacitance in front of the choke and 2/3 behind it. If you are using tube diodes, you shouldn't have more than 40 or 50 uF in front of the first cap, or you will burn out your rectifier tube. If you are using Solid State diodes, you can theoreticaly put in as much capacitance as you want, but in reality, you have to worry about in-rush current when you first turn on the amp. Typically the first cap dominates the ammount of inrush current you see. This can stress your transformer, dim lights in the neighborhood (ok, your house) for a moment, and even burn out the diodes. You can put in a NTC thermistor (a resistor that heats up during a surge and increases its resistance to counter the surge. Once the surge stops, its resistance goes back to near zero).
My recommendation to you is to buy some nice moderate sized oil-filled caps (AC motor start and motor run) or some of the Solen polypropelyne caps rather than huge electrolytics. The amp will be quieter and sound "nicer". Plus you will never have to replace them, vs. electrolytics which eventualy dry out (yes, some electrolytics do seem to last forever, but it is the exception, rather than the rule).
As a recommendation, try a 47uF in front of the choke, followed by two 100uF. As the other fellow said, get the PS software, and see if this combo reduces ripple enough for you.

Good luck.
James J.
My amps use a Pi filter. The first leg is comprised of a fairly heavy bank of electrolytics, bypassed with two of the 47uF 600V Solens, followed by the inductor. The final leg is <i>all</i> film cap--twenty 47uF Solens, totalling 940uF. Note that that is per channel (monobloc). It cost a few pennies, but the sound is beyond anything I've heard commercially. I can't blame CJ, ARC, et. al., as the retail price they'd have to charge for a unit with a capacitor bank like that would be high indeed, but for DIY, it's a doable proposition.



2001-11-27 8:10 pm
If you do decide to put in hundreds of uF of capacitance (and I'll say it again, you just DON'T need it for a single ended amp), break it up into several CL sections. Several C-L-C-L-C with smaller caps between the inductors is much more effective than one inductor surrounded by 300 pounds of electrolytic capacitors.
Since you're building a SE amp, you are probably more concerned with the subtle nuances of the music than earth-shattering bass (don't get me wrong, I like bass). Oil and poly- caps will inject much less noise into your signal than electrolytics. Bypassing electrolytics with a smaller plastic cap will improve the transient response of the caps, but it will do nothing to remove the thermal noise coming from the higher ESR of the electrolytics.
There are occasions that require huge quantities of capacitance, but you'd need a class AB push-pull amp with banks of paralleled high-power output tubes (EL-509, 6550) passing several amps of current, producing 100+ watts to take advantage of it all. A flea powered SE amp, producing a sweet 8 watts will need a tiny fraction of it.

I have built the KT88 kit from world audio and all's well with the sound. Infact it's very powerful in the bass section as written in the review; the high's are sweet (did some caps bypassing to improve it) and is very transparent. Problem is, the hum is pretty loud (sounds more like buzzing) and is audible from about 3 feet away from the speaker. See link to the circuit.

There are 3 rectifiers in the circuit, hence my question is, which power supply section would you recommend to have the C-L-C circuit ? It is worth to have it in all 3 sections or just the high voltage section ? What would be a starting choke value ?

Did you remember to ground the filament supply? That will give you a hum.
Check grounds in general.
The front end is, relatively speaking, more sensitive to power supply ripple. Check ripple voltages on all rails, but bear in mind that the front end ripple is most important.
My inductor is on the rail for the output, but that's because the rails for the first and second stages are regulated, which trumps inductors for flattening out ripple.
You can calculate how much inductance you'll need to reduce the ripple to an arbitrary percentage, but that doesn't tell you how sensitive the circuitry will be to the remaing ripple. (Differentials, for instance, can be quite tolerant of racket on the rail.) These days, you don't have a wide selection of inductors/chokes at the corner electronics shop. Just buy the biggest one you can get your hands on (it's like capacitance--there's a minimum you'll need to get the job done, but more doesn't hurt) keeping in mind the current rating of the part.
The choke I used was a Stancor C-2708 (.32H 600mA). They have a couple of other values that might do as well or better for your circuit. They have the added benefit of not being very expensive.



2001-11-27 8:10 pm
I've built a Mu-stage pre-amp. Because the upper triode sits at about 1/2 B+, I've had to raise the filliment supply (DC, regulated) to about 1/2 B+ too, to avoid exceeding the heater-to-cathode ratings of the tube (12AU7). I did this by attaching one of the heater rails to a pre-existing voltage divider in my B+ circuit.
I have noise in this circuit that I am trying to track down. I've seen some people who float their filliment supplies simply by hooking the center-tap of the filliment winding to ground via a capacitor.
Have you tried either of these methods? Is one quieter?


James J.
I wouldn't use a cap to ground a filament supply, as it will block DC, which is the very thing you're trying to lock down.
When I come across a circuit that needs a filament supply like you're describing, I just throw in a small transformer (one of these dinky things you can buy for a couple of bucks), and leave the rest of the filament supply grounded to the "normal" ground. That way, you can ground the "upper" filament supply anywhere you want.
If you've got DC on your filaments, it sounds as though you've done the right thing, there. Attaching either side of a DC filament supply to a voltage divider sounds like a good plan to me. I doubt that's your problem. If you were using AC, then it would be something to consider, though. If you've got a center tap on your filament winding, I'd ground there. Otherwise, either side will do.