Notes About Capacitors
Back to Dusty Files Online
Small value capacitors :(under 4700 pF/.0047 uF/ 4.7 nF).
Polystyrene or mica capacitors are recommended as these have the overall best stability with temperature. Polystyrene has a low melting point, however, and care should be taken when soldering on them.
Most of these amplifiers have small value (between 10 pF and 1000 pF) capacitors attached somewhere in the circuit to compensate for ringing and high frequency or ultrasonic frequency response peaks or resonances. Since the output transformers you choose will very likely not be the same as the ones used originally by the designers, the value of these should be adjusted if possible to compensate for the differences. Doing this requires a voltmeter that can measure AC out to 100 kHz, a dummy load, a signal generator that can generate clean square and sine waves, and an oscilloscope. (Possibly a big SPARCstation that can do fast calculations would substitute)Values should be adjusted for the cleanest possible square and sine waves at high frequencies (5 kHz and up), and for even frequency response out as far as possible. If you do not have this equipment then the values are best left alone, it may be possible to make some adjustments "by ear" but I haven't tried it. Most feedback resistors are bypassed by a small value capacitor. This creates a time constant that should be preserved if you change the value of the resistor, ergo, the product of C X R should remain the same. Thus the value of the capacitor should be changed in inverse proportion to the value of the resistor, if you have a 1000 ohm resistor bypassed by 100 pf, and change the resistor to 2000 ohms, the value of the capacitor should change to 50 pf (47 pf is close enough).
Medium Value Capacitors (between .01 uF/10,000 pf/10 nF and 1.0 uF/1000 nF)
Generally, polypropylene, teflon, and hermetically sealed oil/paper capacitors (like Vitamin Q) are used. There is considerable debate and little general agreement over what's best.
My best suggestion is to start with the cheapest type then invest later in fancy or expensive stuff if you think it may result in an improvement. I would avoid (a) polyester (DuPont trade name Mylar) due to reported ringing problems at high frequency, (b) paper/wax or paper/mylar capacitors, paper is doesn't have as high a DC resistance (tho it's plenty high, but not nearly as much as plastic) to begin with and since these haven't been made in years, what's on the shelf now is probably really bad,(c) polycarbonate, which makes a great insulator with a very low dissipation factor, but it's also hydrophyllic, (read: absorbs water) meaning in high humidity enviroments (read: you fellows in Britain and south/northwest USA) the value and dissipation factor will change over time, and lastly (e) electrolytic, which is inherently a little leaky, this isn't a factor in filter circuits where one end is connected to ground and proportionately the leakage current is very small, but it's too much in tube coupling capacitor applications.
Vitamin Q type oil capacitors are subject of heated debates, with our customer response running from ecstatic to badly disappointed. I'd say stay away unless either you know that's what you want or you can buy 'em cheap. The combination of the coupling capacitor and the grid resistor of the following section create a time constant that determines the low frequency cutoff or rolloff point. For example, a .047 capacitor with a 220K resistor results in a rolloff beginning at about 40 Hz. Doubling the value of the capacitor or the resistor cuts the rolloff point in half, halving the value of either doubles the rolloff point.
Generally, the higher the cutoff point the more overall feedback that may be applied before the phase shift in the output transformer causes the amplifier to "motorboat" (oscillate at a low frequency). Back in the '50's when amp designers often tried to increase feedback as much as possible to jack up damping-factor specs (which was sort of a fad for awhile) coupling capacitors as low as 2200 pF/.0022 uF/2,2 nF were used in some instances. Trying to increase the value without backing off the feedback results in motorboating in these designs.
Trying to increase the value of the grid resistor to reduce the low frequency rolloff point has its limitations. All tubes draw a small amount of signal grid current which affects power tubes and low-mu triodes & beam tetrodes the worst, and high mu triodes and pentodes the least. The result is a phenomenon known as "grid-leak bias", where the tiny amount of current drawn across the signal grid resistor creates a negative voltage that changes the bias of the tube. Since the exact amount of grid current varies from tube to tube even within tubes of the same type and make, and changes with heat and actual grid voltage, weird drifting and instability can result if the grid resistor is of excessive value. The moral is to read the spec called "Max DC Grid Circuit Resistance" for your tube in the manual and stay a safe distance (say 50% of the max value) away from it. Decreasing the value of the grid resistor excessively will tend to overload the preceding stage, thus it will either not swing sufficient signal voltage or will distort excessively.
Decreasing the low frequency rolloff point by increasing the value of the coupling capacitor is the best bet up to a point, but too large of a value can cause problems, to wit:
(a) Acoustic feedback, where low speaker frequencies vibrate the amplifier, preamp, or turntable which thence go back through the amplifier, creating an oscillation. Unless you have masonry walls and concrete floors this is an outside possibility, the more gain there is, the more tubes you have, and the less stable of a foundation you have your system sitting on, the greater the chances of this occurring.
(b) normal tube drift close to zero frequency either setting up a low frequency oscillation or saturating the output transformer. Output tubes can give a slow blue blinking as a symptom. This problem is more common as the number of tubes in an amp/preamp increases.
Large Capacitors (over 1.0 uF/1000 nF) These are normally smoothing/filtering capacitors to eject AC hum from the power supply. Usually, electrolytics, oil, or large polypropylene and oil/polyproylene capacitors are used. Electrolytics are often paralleled by a polypropylene capacitor of a smaller value, because the series resistance of electrolytics tend to increase (thus the efficiency drops off) at higher frequencies. These large value capacitors are mostly passing 50 or 60 Hz hum to ground, but they also serve the function of preventing signal from passing through undesirable paths between stages in the amplifier, and these frequencies of course go out to 20 kHz. This lends a lot of credence to the polypropylene bypass idea. They also can help prevent radio frequency interference from feeding in through the power supply.
In days of yore (in this case yore is before 1970), amplifier designers, mostly due to cost constraints, calculated the "minimum decoupling" value, (meaning how small of a capacitor they could get away with) when specifying filter capacitor values. Later, it was discovered that using capacitors of a much higher value, particularly directly attached to the center-tap of the output transformer (thus the least amount of impedance between it and the output tubes), could improve perceived bass response and transient overload recovery. Thus in almost every case larger filter capacitors than those specified on the diagrams results in an improvement in sound quality. The factor you should be most aware of is that large electrolytics (or any type of large capacitor) can store a big charge, often for a long period of time after power is turned off. The result is that if you touch a charged cap you will get a BIG NASTY SHOCK, especially if you get it across your body (say your left hand is touching metal and you get the cap with your right) instead of just across your fingers. Many latter day amplifiers have capacitors as large as those in emergency room heart defibullators, which can and do literally raise the dead, up to a couple of feet off the bed! Expect to wake up on the floor, if you do wake up, after tangling with one of those. I have seen some neat arrangements with a DPDT power switch which throws a 10K 5 watt metal oxide resistor (don't use cement block resistors, which will open up) across the capacitor(s) to discharge them when you shut the amp off. Electrolytics also make a big mess if they short out due to physical damage or overvoltage. The newer ones have nice vent holes for discharge so they will not normally rupture and spew stuff all over, but I'd still feel safer with the things under a cage or cover. At the bare minimum make sure the vent hole is pointed to an enclosed area (usually the interior of the chassis).
A considerable amount of discussion has been done over using either oil, oil/polyprop, or polyproylene capacitors in place of the the usual electrolytics. There is a lot to be said for this approach, since overall performance of these are superior. The tradeoff is in size and weight. On smaller amplifiers this presents less of a problem since less capacity is required for good performance, but on larger amps these factors become a problem. On a 60 watt amp I'd like to be able to use at least 200 uf of capacity or possibly a lot more, but it's not practical to use anything but electrolytics unless there's a couple of feet of rack space handy and an extra chassis just to mount the power supply capacitors. Perhaps the best idea is to combine two varieties on one capacitor bank.
Back to Dusty Files Online