Thursday, May 7, 2009

SB-200 - correcting the filament voltage

Most of the SB-200 amplifiers (including mine) suffer from "high filament voltage". One theory I've heard is that when the amplifier was designed in the 60s, the power transformer was calculated for AC mains of 110V/220V and today we are using 120V/240V AC. While a bit higher plate voltage is welcomed , higher filament voltage is a bad thing! According to an EIMAC study: " A 3% increase in filament voltage above the maximum rating will result in a 50% decrease in tube life."
The factory specification for 572B heater voltage is 6.3V +/- 0.3V @ 4A. My calibrated (NIST traceable) DVM measured 6.65V AC (True RMS) for the filament voltage @ 240V AC mains. This reading is beyond the maximum allowed and it should be corrected!
I decided to bring the voltage down, just slightly less than the recommended 6.3V.
The increase in the filament voltage is of no benefit to the output of the valve. 572B is performing at full output with even less than 6.0V filament voltage! My target voltage was 6.25V - about 0.4V down. A thing to note is that the voltage of my 240V AC line does not fluctuate a lot over time - just a couple of volts. Using resistors to correct the filament voltage is the obvious solution except sometimes it is hard to find the right ones for the job! Each amplifier will require custom values for the resistors (determined by the individual transformer and maximum AC line voltage).
To drop the voltage across a resistor by 0.4V (in my case) while drawing 8A of current (2 x 4A for each valve) results in calculated total resistance of 0.05 Ohms. To complicate the matter further, this resistor should be capable of dissipating at least 4W. To preserve the circuit symmetry (the filament secondary is center tapped) I decided to use 2x 0.025 Ohms /2 W resistors - one for each side of the secondary.
Precision current-sensing power resistors could be one possibility. Another solution would be to use a coil of small gauge Teflon insulated wire - AWG #22 for instance. The small gauge will cause the wire to heat up increasing the resistance and corresponding voltage drop. (Even better - coiled on a ferrite rod, thus creating a second filament choke - more inductance equals better choking in this case). I might actually try this someday but there is not much space in the tube's socket compartment and I want this to look neat.
Instead, I made DIY resistors out of Nichrome-60 AWG #22 wire (0.65 mm), left over from my Low-Q VHF suppressors kit. This wire has very high electrical and thermal resistance. To reduce the heating and thermal stress on the short piece of wire required for each 0.025 Ohm resistor (I am using two of them in series for 0.05 ohm total resistance), I actually made each resistor out of two paralleled 0.05 Ohm "resistors".
Recap - i need a 0.05 ohm resistor but for circuit symmetry, I'll use 2 x 0.025 ohm in series (one on each end of the filament winding, just before the filament choke). To increase the power rating of each 0.025 ohm resistor on the other hand, I'll be using 2 x 0.05 resistors in parallel to form the actual resistor.

I used a little over 3cm length of Nichrome wire, bended into a "U" shape. Both ends of the wire are soldered together, creating small elongated loop. The solder points for the resistor are at each side of the elongated loop. This way, each 1.5 cm length of the wire would have to dissipate about 1W for a combined power rating of 2W per resistor.

I inserted each leg of the U-shape wire into Teflon tubing before soldering them together, forming the "resistor". This will prevent an "internal short" in the "resistor". My LCR meter shows exactly 0.025 Ohm.

Each 0.025 Ohm/2W resistor was placed in-line with the filament secondary, just before the filament choke. For soldering the Nichrome wire I used the instructions, flux (corrosive!!!) and silver solder supplied with the suppressor kit. The solder is a tin-silver alloy with high melting point and it should be used also for soldering the "DIY resistors" to the filament choke terminal strip and to the transformer leads. The heat produced by the resistors is substantial - I barley can keep my finger on the resistor (right after I shut the amplifier off (!)), so high temperature solder is needed in order to prevent the joints from failing over time. The result is as expected - filament voltage now measures exactly 6.25V @ 240V AC Mains!

I was ordering parts from Mouser for another project so I ordered a few high-power current sense resistors. Here is a more "commercial" modifications for those who don't feel like making their own resistors. The end result is the same as it is with the DIY resistors. I have the feeling that the commercial resistors heat up a bit less (obviously, larger surface area - better cooling) than the nichrome wire but it is hard to tell just by touching them with my finger.