Circuit Improvements for the Heath SB-220 Amplifier

{updated, html version of original manuscript for the article in 11/90 and 12/90 QST }

by Richard L. Measures, AG6K

11/09/2001

end-notes: [...]

---------------------------------------

. The Heathkit SB-220, two 3-500Z, amplifier made a notable impact in the world of amateur radio. It was the first, reasonably priced, intelligently designed, SSB-CW, HF-amplifier that was sold to the amateur radio community. Unfortunately, this amplifier kit is no longer being manufactured. The SB-220 has some excellent design features and a few design weaknesses that are easily corrected. In this article, I will discuss both topics and the cures for its weaknesses.

HV Power Supply

Before the arrival of the SB-220, there was a popular notion that a legal-limit SSB amplifier needed a heavy-duty power-supply that required two grown men to move it about. Heath engineers knew that this idea was based more on amateur radio folklore than on sound electrical engineering principles. They also knew that the average duty-cycle of a human voice was only about 15% when no carrier was present, as is the case for SSB operation. So, why build a 100% duty-cycle, AM, "lock-to-talk" power supply when one was not required? Thus, they designed a power-supply that would competently do the job that was needed. This resulted in a considerable size, weight and cost savings, which they happily passed along to the buyers of their product.

At first, some people in the ham community had negative comments about the SB-220's "wimpy" power supply. With the passage of on-the-air time, it became apparent that the power supply would do the job, and do so with a low failure-rate and with no detectable on-the-air ripple.

This was no accident. Heath engineers had wisely specified a HV-transformer design that had an exceptionally low secondary-resistance of only 12 ohms. This minimized the voltage drop under full load in the capacitor-filter fullwave-voltage-doubler rectifier circuit. Such circuits have an extremely high peak-current to average output-current ratio. So, minimizing the transformer winding resistance is essential for good voltage regulation and to minimize the I2R heat loss in the transformer's windings.

Many hams initially labeled as "inferior" the capacitor-filter fullwave-voltage-doubler rectifier circuit. They did not realize that this circuit has some advantages over the traditional, fullwave-bridge rectifier circuit. These advantages are:

Cooling

Since about half of the power that is consumed by a linear amplifier is converted into heat, another important amplifier design consideration is cooling. Most of the heat that a 3-500Z, or any other internal-anode tube, dissipates is carried away by radiation from its anode.

Here's how it works: during normal operation, the anode becomes so hot that it glows a bright orange color. The surrounding objects are relatively much cooler, so the anode looses heat to the surrounding objects by radiation-cooling. Unfortunately, the anode also heats up other parts of the 3-500Z that are sensitive to heat. This includes the critical glass-to-metal seals and the solder that is used to fasten the pins to the base of the tube. These heat-sensitive parts must be cooled with forced airflow.

The Heath engineers came up with a deceptively simple method of effectively cooling the 3-500Zs. They realized that the conventional, and very uncheap, air-system-socket / glass chimney cooling method had some serious trade-offs, such as:

1. It was difficult to force enough air through the restrictions to airflow in the "air-system" to adequately cool the filament-pins and filament-seals of the tubes.

2. The horizontal fins on the standard anode [plate]-cap coolers were obviously not designed to be cooled by the vertical airflow from the air-system chimney.

3. The restrictions to airflow with the air-system cooling method dictated that a high-pressure centrifugal blower be used. All high-pressure blowers have one thing in common: High Acoustic Noise.

What was needed was a cooling system that would quietly move high-velocity air past the hot filament-pins [1], filament-seals, the anode-seals, and the glass envelopes.

The Heath engineers knew that when horizontal air flows across vertical cylinders, such as a 3-500Z envelope and its pins, the airflow will follow the curve of the cylinders. This provides fairly uniform cooling to all areas of the cylinders, eliminating hot spots. They concluded that, with horizontal airflow, the cooling air would have a direct path the heat-sensitive parts of the tube. Horizontal airflow would also allow the anode-cap cooler's fins to lineup properly with the flow of cooling air, allowing it to function efficiently.

Since the filament-pins are below the chassis, and the filament-seals and anode-seals are above the chassis, the Heath engineers used an open-ended chassis so that a single, 15cm [6"] diameter fan blade could simultaneously blow cooling air above and below the chassis.

To position the four, hot filament-pins optimally in the under-chassis airflow, the pair of tube-sockets were mounted with the two pairs of filament-pins (#1 and #5) facing each other. This positioned the hot parts in front of the tips of the fan blades where they protruded under the chassis. [1]

The cooling system design was brilliantly simple. It was relatively quiet and it worked well. Reports of 3-500Z-pin solder melting in SB-220 amplifiers are very rare -with the exception of cases where the fan motor bearings eventually seized because they were never oiled!

On the other hand, I have heard of many 3-500Z-pin solder melting episodes in other amplifiers that used air-system chimney/centrifugal blower cooling.

One weakness in the SB-220's cooling system was the failure to reduce the heat-reflectance of the bright aluminum surfaces that were adjacent and parallel to the anodes. This would have reduced the radiant heat that was reflected back at the hot anodes. The best color for this purpose is black.[2]

This deficiency is easily corrected: After the tubes have been removed, black, liquid shoe-polish can be applied to the vertical, aluminum surfaces, that are near the anodes. The built-in felt applicator, that is attached to the bottle cap, works well for applying the black coating.

One weakness in early models was the failure to provide oil-holes for the fan motor bearings. This problem can be corrected by drilling a small hole, no more than 1/4" [3mm] deep, above the front and above the rear oilite bearings.[3] The fan should be lubricated yearly with a thin, non-gumming oil such as Hoppe's 1003.[4][5] About 2 to 3 drops of oil needs to be inserted in each hole. What isn't absorbed into the felt wicks, that surround the oilite bearings, runs out the bottom. More oil is not better, just messier. The easiest way to get the desired amount of oil in the holes is to apply the oil with a disposable insulin syringe. Each "unit" on the syringe equals about one drop of oil.

Premature Filter-Capacitor Failure

Aluminum electrolytic filter-capacitors are very sensitive to heat. For every 10 degrees C increase above room temperature, the life expectancy of the capacitor is approximately cut in half.

The electrolytic filter-capacitors in the SB-220 are subjected to a high degree of heat during normal operation. The major source of this heat is from the eight, 30k ohm , voltage equalizer resistors that are adjacent to the eight filter-capacitors. During transmit, a minor source of capacitor heating, results from the 60Hz ripple-current flowing through the capacitor's internal resistance (a.k.a. ESR/equivalent series R).

The capacitor heating problem is compounded because cooling air does not reach the capacitors due to the molded plastic capacitor holders that surround and insulate each capacitor. In some cases, the heat present will partially melt the ends of the capacitor holders that are nearest to the 30k ohm resistors.

The heat that is dissipated can be reduced by approx. 70% if 100k ohm 2W or 3W, <5% film resistors are used to replace the original 30k ohm resistors. This simple modification will greatly prolong the life of the (8) electrolytic filter-capacitors.

Other resistance values may be used, up to roughly 150k ohm , provided that the resistors can withstand the voltage that is applied to them and the resistances are within +/- 5% of each other. I do not recommend using 2W carbon-composition resistors for this application. They do not maintain their rated resistance tolerance with age. Note: Increasing the equalizer resistances will also increase the capacitor bleed-down time after the amplifier is shut off.

Since this amplifier has a shorting HV-interlock, that grounds the HV positive when the perforated cover is removed, it is advisable to wait until the voltmeter indicates nearly 0V before allowing the interlock to short the HV to the chassis. . Here's why: When the HV-positive is shorted to ground, the stored energy in the filter-capacitors is applied directly to the grid-current meter shunt resistor, R3 {0.82 ohm }, which is the only HV-negative path to chassis. The peak discharge current can be substantial.

For example, if the filter-capacitors are at the 100V level when the interlock shorts, the peak-current through R3 is: I = E/R = 100V/0.82 ohm approx. 100A (not counting the ESR in the filter capacitors). If a substantial voltage exists in the filter-capacitors when the interlock shorts, R3 can be literally blown away by the discharge current-pulse. If the multimeter happens to be in the grid-current position, the meter can also be crispy-crittered.[6]

It is for this reason that I removed the interlocks from both of my Heathkit amplifiers. Another consideration was that the interlock does not prevent the operator from contacting the potentially fatal voltage from the electric-mains when the amplifier is plugged in and switched off. The interlock will not prevent the operator from being electrocuted. In other words, the safety-interlock does not make the amplifier safe.

Another advantage of removing the interlock is that it allows the perforated cover to be removed while optimizing the tuned-input circuits. This allows better access to the tuned input circuits.

There is no safe substitute for pulling the electric-mains plug before putting fingers inside any amplifier.

Intermittent Meter Readings

There are at least two problems that can cause intermittent meter readings in the SB-220. If this occurs only on the voltmeter, the most likely problem is with the (3) 4.7M ohm , 1W voltmeter multiplier resistors [R6, R7, and R8]. These resistors, which are rated at 350v maximum per-unit, are subjected to approx. 1000V-per-unit in the Heathkit circuit. This can lead to resistor deterioration, which causes fluctuations and/or inaccuracies in the 0-3500V meter indication.[7] The abused resistors can be replaced with modern, 2W flameproof spiral-film resistors that were designed to handle this voltage.

The other source of trouble lies inside the meters. Here's why: Different metals are used for the various parts of the meter. These parts, which conduct current to the meter armature, are fastened together with screws. Over a prolonged period of time, moisture in the air causes electrolysis to take place at the junctions of the dissimilar metals. This increases the resistance at the junctions, which causes intermittent meter indications.

This problem can be corrected by prying off the meter face, carefully removing the meter scale, and applying small dabs of conductive-paint to all of the dissimilar metal junctions that carry the current to the armature.

The conductive-paint can be thinned with acetone to facilitate penetration into the narrow areas between the parts. The conductive paint should be allowed to dry 15-minutes before the plastic meter faces are replaced.

Transceiver Relay Contact Failure

During receive, the voltage across the Antenna Relay Jack rises to about +115V. A bypass-capacitor, C52, is connected in parallel with this jack, so the capacitor charges up to 115V during receive. During transmit, the transceiver's relay (if one is used) places a short-circuit across this jack and the fully-charged C52. The SB-220 relay coil current is only about 0.025A, but the peak discharge current, produced by placing a direct short on the charged capacitor, can be surprisingly large. This action is like an electric spotwelder. In time, the contacts in the transceiver relay can become pitted and fail to make contact, or become spotwelded together, causing the amplifier to key-up continuously.

This problem can be corrected by placing a 100 ohm to 200 ohm , 1/2W resistor [to limit I] in series with the center pin [blue wire] on the Antenna Relay Jack. C52 must be connected to the blue-wire end of the resistor. See Diagram 1.

The voltage lost in this resistor will = I x R = 200 ohm x approx. 0.025A = 5V, which is insignificant to the 110VDC relay coil.

Filament Circuit

The most popular, published modification for the SB-220 has been filament inrush-current limiting. It is true that a large number of 3-500Zs in SB-220s suffered from filament-to-grid shorts, so some people began to theorize that excessive filament inrush current was the villain.

Another theory was that the filament-to-grid shorts were caused by a "manufacturing defect". Neither theory turned out to be true, even though Eimac®, as usual, generously made good on the tube warranties.

Curiously, none of the authors, who wrote the SB-220 inrush-current limiting articles, ever bothered to measure the actual filament inrush-current. So, I decided to measure it with my trusty old HP-1706A oscilloscope. After all, my name is Measures, so why not?

Here's what I found: The maximum, peak inrush-current through the 3-500Z filaments in an SB-220 is only 60% of what Eimac® allows. This esoteric feat was accomplished by the use of a special current-limiting core in the Heathkit filament-transformer. The core is similar to the cores used in current-limiting neon-sign transformers. Externally, this core appears to be substantially different than the core used in the HV-transformer.

About a year after I measured the inrush-current, the fact that the SB-220 used a current-limiting filament-transformer, was mentioned in QST Magazine.

The actual cause of virtually all grid-to-filament shorts in the 3-500Zs was later discovered to be a very brief, and usually very noisy, VHF parasitic-oscillation at roughly 110MHz. As will be discussed later, the large grid-current-pulse, that accompanies the parasite, was creating a large magnetic-pulse inside the 3-500Zs, which pulled the hot filament-wires off-center, causing them to touch the grid wire cage.

Another interesting feature of the SB-220 is the fact that they normally use a filament-voltage at the low end of the recommended voltage range; typically about 4.85V. This may not seem important, but, according to Eimac®, each 3% reduction in filament-voltage, provided that no drop in PEP output occurs, will double the life expectancy of the amplifier-tube. Thus, the amplifier-tubes in a SB-220 can be expected to last at least 4-times longer than the tubes in some other, much more costly and prestigious brands of two 3-500Z amplifiers.

Just because something costs more does not automatically mean that it is better.[8]

For example, the contemporary, approx. 3db more expensive, Henry Radio Co. 2KD-5 was generally considered to be a better designed, higher power output, and more rugged, amplifier than the SB-220, This amplifier typically had a filament-voltage of more than 5.90V with a line voltage of 240V. 5.90 volts clearly exceeds the maximum filament-voltage rating of 5.25V and reduces the useful emission life of the two 3-500Zs to only a few-percent of what could have been realized if the same tubes had been used in an SB-220.

Another example is the Trio-Kenwood TL-922. In this amplifier, about 5.3V is applied to the 3-500Z filaments when the amplifier is operated from 120V or 240V. This reduces the useful-emission life of the tubes by more than 75%. Of course, if your electric-mains voltage is 216V, the filament-voltage will be ideal.

Obviously, the filament-circuit in the SB-220 needs no stepstart circuit to limit inrush-current, but there is still a good reason to add a stepstart circuit to the SB-220.

Here's why: If the amplifier is started up, on the SSB mode, when powered by stiff, 240V electric-mains, the inrush-current, through the Power Switch and other components, is considerable. A stepstart circuit will eliminate this potential source of trouble.

An easy-to-build, stepstart circuit is shown in Diagram 2. In this circuit, the stepstart relay can close only when the filter-capacitors in the +110V and +3000V power supplies have charged up to about 2/3 of their normal voltages. If the stepstart relay closes before the HV reaches 2/3 of its normal potential, the 2k ohm nominal-value resistance should be increased. If the relay has trouble closing, the resistance should be decreased, which will increase the current through the relay's coil. If the circuit is functioning properly, the stepstart relay will close in about one-second, as the voltmeter passes the 2000V level. The amplifier may be operated at full-throttle 1-second after the relay closes.

The two, 20 ohm, 10W, resistors and the stepstart relay can be glued directly to the bottom of the chassis, directly under the filter-capacitor bank, using silicone-rubber adhesive. [9] The resistors should be held up off of the chassis a few mm, by the silicone-rubber. This mounting method is appropriate because drilling mounting holes in this area could harm the filter-capacitors.

Since the stepstart relay adds to the current burden on the +110V power supply, it is probably a good idea to replace the stock, halfwave rectifier [D16] with a full-wave-bridge rectifier unit. Note: The grounded red-wire, on the transformer's 80VRMS winding, must be ungrounded and connected to the input of the full-wave-bridge rectifier.

If an SB-220 is always started up on the CW/Tune mode, and then switched to the SSB mode, the inrush-current is reduced and a stepstart relay is probably not needed.

Adding A Standby Switch

Another popular modification for the SB-220 is the addition of a standby switch. A standby switch is really not necessary in this amplifier because it uses instant-on type tubes and a current-limiting filament-transformer. This transformer is gentle to the filaments, so the amplifier may be switched on or off as often as you like with no problem. The exception is: if you have just made a long RTTY or FM transmission, the glass-to-metal seals in the 3-500Zs should be allowed to cool down for about 1 minute before switching off .

Silicon Rectifier "Protection"

In the 1960s, silicon-rectifier manufacturing technology was hit and miss. There was considerable variation between individual rectifiers of the same type. The variation between rectifiers led designers to use resistor-capacitor equalizer circuits in parallel with series rectifiers.

Today, silicon-rectifier manufacturing technology has improved, so that same-type rectifiers are very uniform in their parameters. Modern silicon rectifiers do not need to be equalized. Unfortunately, old habits die a slow death and many hams are still clinging onto 1960s design techniques.

Much has been written about adding "equalizing" resistor-capacitor protection networks across the rectifiers in the SB-220's HV power supply. Unfortunately, these "protection" circuits not only do not perform as advertised, they can lead to premature rectifier failure.

Here's why: The 1/2W resistors that are typically used are rated at 250V maximum. How can a 250V resistor be trusted across a 600V or a 1000V rectifier? If anything breaks down in a series-rectifier circuit, it is like dominoes falling. One resistor failure can wipe out the remaining good parts in the series circuit.

The most frequent failure mode for HV power supply rectifiers is too much reverse-current. This problem can be eliminated if the total PIV capability of the series connected rectifiers substantially exceeds the peak voltage in the circuit.

In any series circuit, the current in all of the elements is exactly equal. The rectifiers are all in series. So, the reverse-current burden is exactly the same for each rectifier unit. How is it that something which is already exactly equal needs to be "equalized"? (see page 11-9, column 2, in the new ARRL Handbook.)

During the half-cycle application of reverse-voltage, it is important that all of the rectifiers in a series leg have similar junction capacitances. If they don't, then the reverse-voltage across the lower capacitance rectifiers will be excessive. Here's why: in a series circuit, smaller capacitors charge-up faster, and to a higher voltage, than larger capacitors.

Approximately 10nF [0.01uF or 10,000pF] of bypass capacitance across each rectifier is probably a good idea if, for example, 1A rectifiers are placed in series with 6A rectifiers. This is the case because of the wide difference in junction capacitance between 6A and 1A rectifiers.

If all of the rectifiers in a series leg are similar, they will all have similar junction capacitances, so no external capacitors, or resistors, are needed.

Long ago, before they knew better, some commercial high-voltage silicon rectifier-stack manufacturers used "equalizer" R/C networks. These manufacturers stopped using "equalizers" for the same reasons that were previously outlined above. I don't know of any commercial HV-rectifier manufacturer who has not abandoned the malpractice.

Silicon Rectifier Failure

When a silicon rectifier fails from excessive reverse-current, the rectifier will short-circuit. This failure mode is very rare in SB-220's because the per-leg total rectifier PIV-rating used {= or >4200v} is more than 1000V higher than the actual PIV {approx. 3100v} in the circuit. This represents a conservative design because, during a voltage surge, the eight electrolytic filter capacitors, which are rated at 3600V-max, would likely fail before a = or >4200PIV rectifier string.

A much more common type of rectifier failure in early production SB-220s is rectifier-opening. This is caused by a defective spotweld inside the silicon rectifier. Eventually the weld breaks and the rectifier opens. The forward voltage jumps the gap at the open weld. When this happens, a 60Hz arc can usually be heard from inside the amplifier when current is being drawn from the HV supply. It is important to switch off the amplifier immediately when this noise is heard. Here's why: In a fullwave voltage-doubler rectifier circuit, there are two, series, filter capacitors.10 One capacitor charges on the positive half of the cycle. The other capacitor charges on the negative half of the cycle. The two capacitors discharge in series. When current is being drawn from the supply, and if one of the filter-capacitors is not being charged by its rectifiers, the capacitor that is being charged will force reverse-polarity current through the capacitor that is not being charged. If unchecked, reverse-current will cause electrolytic capacitors to discharge their corrosive electrolyte through their safety vents. In other words, reverse-polarity current will destroy polarized electrolytic capacitors in short order.

RL1, The Antenna and Bias Relay

Diagram .

1. The added, reverse diode across the coil absorbs the reverse-voltage spike when current stops flowing in the coil. This prolongs the life of the coil insulation and quenches the magnetic pulse from the coil. If the magnetic pulse is unchecked, it can retrigger the transceiver's VOX circuit if a nearby, high-Z dynamic microphone is being used.

2. In its stock wiring configuration, +110V is connected to terminal #2 of the relay. During receive, the relay connects this voltage to the center-tap of the filament-transformer [terminal #8], which is the DC cathode-current path to the filaments of the 3-500Zs. The positive cathode-voltage causes the tubes to cutoff during receive. The tubes cut-off because the grids are relatively 110V more negative than the cathode's electrons. Electrons are negatively charged, and since like charges repel, no electron-current flows in the tube.

A sticky problem arises if one of the tubes happens to develop a filament-to-grid short: Since each grid is DC-grounded, a shorted tube also short-circuits the +110V power supply. This power supply is powered by the unfused filament-transformer. Thus, if a filament-to-grid short occurs, and the amplifier is not switched off promptly, the filament-transformer will literally melt-down and short-out. The black tar that comes out of the overheating transformer makes an unpleasant mess inside the amplifier. This potential source of grief can be eliminated if the relay is rewired as shown in Diagram 1. The new circuit uses resistor-cutoff bias, utilizing the existing 100k ohm resistor [R27], which is rewired to relay terminal #5. The current through this resistor, during receive, is normally less than 0.25mA [P<0.007W], so the 1/2W rating is more than adequate. (In other words, it is risky to connect a power supply to the bias contacts of any grounded-grid amplifier. The TL-922 has this problem)

3. The antenna-relay is mounted on a rubber grommet. This was intended to reduce the vibration that the relay transmits to the chassis, which acts as a sounding-board. In time, the rubber in the grommet hardens. This increases the acoustic noise that the operator hears when the relay makes or breaks. This problem can be corrected by: removing the mounting screw and the grommet from the top of the chassis and applying a small dab of silicone-rubber adhesive through the hole.

After the silicone-rubber cures, a further noise reduction can be gained by installing U-shaped strips of thin flexible copper ribbon between the stiff wires that solder to terminal #s 4, 6, 7, and 9, and the relay terminals. The stiff wires should be shortened about 1cm before the U-links are soldered in. The flexible U-links act as shock-absorbers, and prevent the stiff copper wires from transmitting vibration from the relay to the chassis. This simple modification results in a substantial noise reduction.

4. During "barefoot" operation on 10m, when the amplifier is switched off, the bypass, or "straight-through"-SWR of the amplifier is much less than wonderful. This is due to the inductive reactance in the relay. The relay's inductive reactance [XL] can be cancelled by adding capacitive reactance [Xc] between terminal #9 of the relay and chassis-ground. The capacitance required is usually between 30pF and 43pF with a 1kV rating. Some, apparently conservatively rated, 500V micas seem to be able to handle the job. If a ceramic capacitor is to be used, the NP0 type is best, rated at 1 or 2 kV.

If you are a perfectionist, you may want to install a compression mica trimmer-capacitor, and adjust the trimmer for the lowest 10m bypass-SWR when the amplifier is connected to an accurate 50 ohm termination. The trimmer is removed, the value of capacitance measured on a C-meter, and a fixed capacitor of similar value installed permanently in the amplifier.

Parasites

The most costly design weakness in the SB-220 is its tendency to have an intermittent, VHF parasitic-oscillation at roughly 110MHz. This problem is far from being unique to the SB-220. I know of no model of commercial MF/HF ham band, grounded-grid amplifier that has not occasionally had a VHF parasitic-oscillation.

Some owners will take issue with this statement because many amplifiers will play for years and never exhibit any sign of a VHF parasitic-oscillation. Other amplifiers will pop, arc, or unexpectedly go bang, often eating amplifier parts. The difference between amplifiers that "take-off" and those that don't lies in the particular tubes that are installed in the amplifier. Tubes that have above-average VHF-amplification are those that are most likely to cause grief. Below-average gain tubes, or broadcast station "pulls", will probably never be able to sustain a VHF parasitic-oscillation.

A common warning sign that any amplifier is on the verge of a full-blown parasitic-oscillation is minor arcing or spitting at the tune-capacitor or the bandswitch. The full-blown parasitic-oscillation is hard to miss. It usually makes a loud bang or pop. If the amplifier is not switched off promptly after the big bang, the filament-transformer may melt down. [11]

The reasons for this parasitic-oscillation problem are the same as the reasons for the parasitic-oscillation problem in the Trio-Kenwood TL-922 amplifier, which was discussed in an article that begins on page 36 in the October 1988 issue of QST Magazine.

Briefly, the heart of the parasite problem in the SB-220 lies in the use of high VHF-Q copper-conductors between the TUNE-capacitor and the anode-connections, or "plate-caps", on the 3-500Zs. The high VHF-Q parts includes the factory-stock "parasitic-suppressors" which are made with copper wire.

This problem can be corrected by constructing low-VHF-Q parasitic-suppressors, using VHF-lossy, nickel-chromium-iron alloy ["nichrome-60"] wire and replacing the copper-braid between the DC blocking capacitor and the top of the HV RF-choke with a pair of unequal length nichrome-60 wires.12 Note: The importance of the double nichrome wire replacement for the copper braid should not be discounted. This double wire suppressor/conductor may look unconventional and silly, but it has a powerful damping effect on the VHF Q of the anode-circuit. In many amplifiers, VHF stability can not be achieved with the stock, copper braid conductor between the blocking capacitor and the top of the HV RF-choke.

(11/97 note: Another solution is to use copper wire for the suppressor inductor, double the amount of inductance, and use a 200 ohm parasitic suppressor resistor capable of dissipating 40 watts that has under 12nH of intrinsic L)

If you have access to a dipmeter, and you would like to observe the before and after effect of installing improved parasitic-suppressors, before any modifications are made, unplug the amplifier from the electric-mains and check the anode-circuit's VHF self-resonance. The optimum check point is at the left end of the copper-braid that connects the HV RF-choke to the DC blocking capacitor at the rear of the TUNE-capacitor. The resonant frequency is typically 110MHz +/- 10MHz.

The dip in factory-stock amplifiers is so sharp that it can "suck-out" the oscillator in the dipmeter. So the dipmeter must be backed away from the conductor. This indicates that the anode-circuit has a very high VHF-Q. This is not good news unless you happen to need a self-excited VHF high power-oscillator.

After installing the low VHF-Q suppressors, check the dip again. The self-resonant frequency will usually not change appreciably, but the dip should now be smooth and broad. This indicates that you were successful at substantially reducing the anode-circuit's VHF-Q, and reducing the VHF-oscillating ability of your amplifier.

During the production life of the SB-220 and its successors, Heath made two changes in the design of the amplifier that were related to its parasitic problem. One change was to increase the spacing of the TUNE-capacitor. The other change was to decrease the C of the grid-to-ground capacitors from 200pF to 115pF.

The more reactive 115pF capacitors did cancel some of the internal grid-inductance in the 3-500Zs. This increased the VHF self-resonant frequency of the grid, making the amplifier slightly more stable. Unfortunately, this fix was not a sure cure.

Judging from numerous on-the-air and telephone conversations, Heath received many complaints from SB-220 owners who reported arcing at the TUNE-capacitor. In response to these complaints, Heath increased the voltage breakdown rating of this capacitor. Increasing the spacing of the TUNE-capacitor turned out to be a serious mistake.

Here's why: The original TUNE-capacitor already had a substantial breakdown-voltage safety factor, considering that the maximum, peak (HF) RF anode-voltage in the SB-220 is approx. 2600v. The arcing was not being caused by normal HF RF peak-voltage. Instead, the arcing at the capacitor was being caused by intermittent approx. 110MHz parasitic-oscillation voltage. Increasing the voltage rating of this capacitor did stop the arcing at the capacitor, but it shifted the arcing to the output bandswitch. In other words, the parasitic-voltage was jumping across the path of least resistance. For example, if the bandswitch's contact spacing was increased to stop the bandswitch arcing, the new wide-spaced TUNE-capacitor would probably begin arcing.

Pitting on the plates of an air-variable capacitor can be cleaned up with a file, and the capacitor will be as good as new. Arcing on the fragile contacts of a bandswitch is frequently fatal to the bandswitch.

Zener Diode Replacement

One of the more common casualties during a VHF parasitic-oscillation is the Zener cathode bias diode. Because cathode-current is the combination of the anode[plate]-current and the grid-current, the Zener gets zapped by the large grid-current-pulse that nearly always accompanies a VHF parasitic-oscillation. This is the same current-pulse that is responsible for causing the vast majority of filament-to-grid shorts in the 3-500Zs. It is also responsible for blowing away the following components: R3, {0.82 ohm } the grid-current meter shunt resistor; the multimeter meter {if the multimeter switch is on the grid-current position}; the grid-to-ground 1mH RF-chokes, and the 200pF13, mica, grid-to-chassis-ground capacitors.

Three disadvantages of a Zener diode are: They are not adjustable, they don't like high-current pulses, and they are not inexpensive. A much cheaper and more rugged, step-adjustable replacement for a 5.1V, 10W Zener [ZD1] can be made with approx. 7, ordinary silicon rectifier diodes.

The replacement circuit is shown in Diagram 3. The voltage can be controlled, in approx. 0.8V steps, by adding or subtracting diodes. This allows the operator to set the zero-signal anode-current for the two 3-500Zs. The SSB-mode current should be 160mA to 200mA for best linearity.

The diodes can be mounted on a piece of perfboard and fastened in the power supply section. Note: The polarity arrows on the forward-conducting rectifiers point opposite to the arrow on the reverse-conducting Zener diode.

Full Break-in [QSK] Modification

Under 2-milliSecond,QSK can be added to the SB-220 for under $60 if you know the best places to buy the parts. The circuit is simple to construct and no exotic parts are used. The most expensive part is the high-speed, RF-output, vacuum-relay, which can be purchased new-surplus {unused}, for about $40 from Fair Radio Sales 419-227-6573

Even if you do not operate much CW-QSK, this modification is still worthwhile because it makes working SSB-VOX enjoyable. The relays are so fast and inconspicuously quiet, it's almost like talking on the telephone, or like talking in person. This is real communication.

The QSK circuit was published in the January, 1994 QST article "The Nearly Perfect Amplifier". One of the features of this QSK circuit is that the electronic cathode bias switch [ECBS] is always in perfect sync with the RF-relays. In many other QSK circuits, this is not the case.

Here's why: In some other QSK circuits, the ECBS is RF-actuated. This may sound wonderful but it is not so because RF-actuation allows the 3-500Zs to be switched in and out of their linear bias region during and between softly spoken syllables. The net result is increased IMD, or splatter, as the bias wanders between linear and non linear operation during speech.

Making the RF-relays and the ECBS RF-actuated is not a solution because this would cause the RF-relays to hot switch every time they tried to make.

A properly designed QSK circuit puts the QSK-transceiver's amplifier-control line completely in charge of the amplifier. In this way, the 3-500Z bias will always be correct for linear operation any time that the RF-relays are actuated.

To ALC, Or Not To ALC?

If you are using a Swan 500 to drive your SB-220, you definitely need to use the ALC circuit in the amplifier. I also recommend that you NOT display your call sign prominently at ham conventions or swap-meets. This may help to avoid any unpleasant situations that may result in a spontaneous tar and feathering.

If your transceiver puts out less than 140W-PEP, the use of ALC with your SB-220 is of no value since it is impossible to overdrive the 3-500Zs. It is for this reason that I removed the ALC circuit from both of my Heathkit amplifiers.[14]

Operation On The 12m And 17m Bands

One of the main problems with using older design, ham band only, amplifiers on the new 12m and 17m bands is choke-fires.

Here's what happens: When a HV RF-choke is operated on or near one of its self-resonant frequencies, an extremely high RF-voltage appears on the choke. The voltage can easily exceed four-times the supply voltage. This can cause the insulation on the choke's wire to break down and ignite.

To prevent choke-fires, all operating frequencies should be more than 5% away from any of the choke's self-resonant frequencies.

The SB-220 will operate well on the 12m/24.9MHz and 17m/18.1MHz bands because its HV RF-choke, RFC-1, does not have any self-resonance below 40MHz.

If a transistor-output radio will be used to drive the amplifier, the amplifier's tuned input circuits for the 10m and 15m bands should be optimized for this dual-purpose. More on that later.

The only possible problem associated with 12m and 17m band operation is the increased current burden on the output-bandswitch.

Here's why: In order for the amplifier to tune down to the new frequencies without increasing L, the TUNE-capacitor and the LOAD-capacitor must be adjusted for about 35% more capacitance. This increases the operating Q of the output pi-network by about 18%, which increases the RF-circulating-current through the bandswitch contacts by about 18%. Since P=I2R, the increase in contact dissipation will =1.18 x 1.18 x R= 1.39 x R. If R remains constant, this equals a 39% increase in the heat dissipated by the contacts.15

This is unlikely to be a problem for normal SSB operation without speech-processing. For higher duty-cycle operation, the amplifier should be switched to the lower-voltage CW/TUNE position in order to reduce the average heat dissipation on the output bandswitch contacts.

Improving Input SWR

The tuned-input circuits in the SB-220 have a typical SWR of about 1.5:1. This is completely satisfactory as long as tube-output radios are used to drive the amplifier.

Now days, the only type of radios that are being manufactured use transistor-output. The transistors are used in an untuned, broadband-transformer circuit. This type of amplifier will develop full power only when it is working into a nearly non-reactive load whose resistance is close to the design load of 50 ohm s. Many transistor-output radios are so fussy that they will begin to power themselves down with a reactive load whose SWR is 1.15 to 1. This means that the amplifier will not receive full drive power unless it has a very low, input SWR.. On many bands, this is the case with a stock SB-220.

Fortunately, the input-SWR of this amplifier can be easily improved if we understand what we are dealing with.

The job of the tuned-input circuit is more complicated than just matching 50 ohm to the input-resistance of the amplifier-tubes.

Here's why: The instantaneous input-resistance of a grounded-grid amplifier fluctuates wildly during the positive and negative voltage swings of the sinewave input signal.

When the input cathode-voltage swings positive, the grounded-grid looks negative with respect to the cathode, and the current is completely cut-off, making the input-resistance nearly infinite.

During the negative swing in input voltage, the grid looks more positive, and a large current flows in the tube, making the input-resistance very low.

For example: a pair of 3-500Zs. When the driving voltage is peaking at negative 117v, the anode-current is at its maximum peak, and the instantaneous anode-voltage is swinging to its lowest point of approx. +250v, the total, peak cathode-current is 3.4a.16 Thus, the driving resistance at this point, Rin = 117v/3.4a = 34.5 ohm , and, incredibly, Ppeak = 117v x 3.4a = 397w.

Thus, the resistance swing is from near-infinite with positive driving voltage, all the way down to 34.5 ohm .17 The instantaneous drive power requirement varies from 0w to 397w at the positive and negative peaks of the sinewave input voltage. This is not the type of load that makes for contented transistor-output transceivers.

During the positive swing in input voltage, there is virtually no load on the driver, so the tuned-input circuit must store the energy until it is needed the most, during the negative crest in the input voltage.

Thus, the tuned-input circuit's job is to act as a flywheel/energy storage system, and a matching transformer.

Q is like the mass of a flywheel. More Q makes for a better flywheel, which does a better job of averaging the wild swings in input-resistance, giving a lower input-SWR. The tradeoff is that more Q means less bandwidth. This means that, with a high Q, the input SWR may be near-perfect at the center of the band, but too high at the band edges. Thus, a compromise must be made.

Eimac® recommends using a Q of five for the tuned-input circuits in a grounded-grid amplifier. As I will show below, the SB-220 uses a Q of only about one. This is why the stock, input-SWR of the SB-220 is less than wonderful. Other commercial amplifiers that were also designed in the era before transistor-output transceivers used a Q of one or even less than one. However a Q of five is apparently more than is really needed. A Q of two will do.

Q is defined as the ratio of the input-resistance of the tuned-input circuit, which is 50 ohm , divided by the reactance, in ohm , of the input-capacitor [XCin]. For example, in the SB-220, the 40m input-capacitor [C42] is 470pF. The reactance of C42 at 7.15MHz = XCin = 1÷[2 f C] = 1÷[2 x 7.15x106Hz x 470x10-12F] = 47.36 ohm , which is also written as minus j47.36 ohm . Thus, the Q = 50 ohm /47.36 ohm = 1.05.

When the Q of a tuned-input circuit is too low to start with, no amount of adjusting the inductor, L, and Cout can ever bring the SWR down to a low value.

Improving the input-SWR of an SB-220 is simple: increase the Q by decreasing the XCin in the tuned-input circuits. This means more Cin.

Since the resistance matching ratio of a tuned-input is semi-proportional to the Xc-ratio of the two capacitors, if Cin is increased, to increase Q, Cout must be increased in order to maintain the same resistance matching ratio. In this case it is 50 ohm to approx. 69 ohm .

Increasing both capacitances will lower the operating frequency of the tuned-input, so L must be decreased to bring the frequency back up to where it started. This can be accomplished by removing turns of wire from L, and by adjusting the inductor's tuning slug.

It is important to keep in mind that the matching ratio of a tuned input circuit can not be changed by adjusting the slug-tuned inductor alone. In order to arrive at the optimum SWR, Two components must be adjusted in order to change the matching ratio of a tuned input circuit.

Another factor that affects the input-SWR is the Q [=XL/R(RF)] of the inductor. Any RF-resistance [R(RF)] that appears in the inductor will adversely affect the SWR. Smaller wire has more resistance than larger wire. Thus, it is important to use an adequate wire diameter. As frequency increases, skin-effect increases, which increases the RF-resistance of the wire. To compensate for this, the wire diameter must be increased in proportion to frequency.

For example, at 1.8MHz, in a 100w tuned-input circuit, the wire diameter should be at least 0.5mm [approx. #24AWG]. At 29MHz, = or >1.3mm [approx. #16AWG] diameter wire is appropriate. You can't go wrong by choosing a larger diameter wire, unless it won't fit on the coil-form.

A Q of two is usually slightly more than optimum if you need to cover a large frequency spread with a single tuned-input circuit. Prime examples would be to cover 3.5MHz to 4MHz, and 18MHz to 21.5MHz, so that the 15m tuned-input will also cover the new 17m band. In these cases, a Q of approx. 1.5 should be tried. This also applies to a 10m tuned-input that will be used on the 12m band.

A Q of 1.5 corresponds to a reactance of about 33.3 ohm {Xcin=50 ohm /1.5=33.3 ohm } for Cin. At 3.75MHz this would require a approx. 1275pF capacitor.18 The nearest standard value is 1300pF. Of course, capacitors can be paralleled to arrive at the desired C.

Measuring the input-SWR of an amplifier is a very inexact science. For example, different models of SWR-meters will give different readings in the same circuit. Changing the distance between the SWR-meter and the amplifier may also change the indicated SWR.

Another complication is that modern transistor-output transceivers always use a set of switched, approx. 1.5-octave-per-filter, broadband output-filters. This is done so that their output signals will meet the FCC requirements for spectral purity.

At the extremes of an individual filter's bandpass, such as at 29MHz, the filter can introduce a reactance into the transmission line. This reactance can cause some peculiar results when you are trying to optimize the SWR of the tuned input circuits in an amplifier.

The best way to avoid this problem is to use a tube-type driver, such as a Trio-Kenwood TS-830S, when optimizing the tuned-input circuits. The driver must be tuned for maximum power into a known-to-be-accurate 50 ohm termination, and then not readjusted during the adjustment of the tuned-input's L and Cout. If the transceiver is re-tuned, it may introduce a reactance onto the transmission line that will affect the indicated SWR.

If the Q of the tuned input has been increased, by increasing Cin and decreasing L, Cout will need to be increased. The easiest way to find the new, higher, optimum value for Cout is by inserting a compression mica trimmer-capacitor in in parallel with the stock Cout.

With the maximum peak drive power applied, L and Cout are alternately adjusted for the best match at the center of the band. Cout can then be removed, its capacitance measured on a C-meter, and a fixed capacitor, of the same C, permanently installed in the amplifier.

If the amplifier is driven with a steady carrier [A0 , now called N0 N], considerable stress is placed on the HV power supply and the RF-compartment will become very hot. The stress and strain can be reduced if the driver is set to the CW-mode. The driver is keyed by an electronic-keyer, set to 50 to 60 words per minute. The keyer sends a steady stream of dits, which have a 50% duty-cycle. The amplifier's current-meter readings should be approximately doubled to find the actual current.

Adjusting the tuned-inputs is much easier if the front panel is removed. The meter leads need to be lengthened to facilitate doing this. A chassis-ground wire needs to be added between the panel and the amplifier chassis if you want the multimeter to function when the panel is separated from the amplifier.

It is important, not to inadvertently contact the nearby HV feedthrough insulator while you are adjusting the tuned inputs. Doing so could result in your premature appearance in................"Silent Keys".

A reasonable procedure is to use insulated tuning tools and to stand on a plastic mat, with one hand behind your back. It is also advisable to wrap some 1" [25mm] plastic electrical tape around the nuts on the HV-feedthrough insulator, before the amplifier is plugged in.

If you would prefer not to work around lethal voltage, it is possible to adjust the tuned-inputs of the amplifier without applying high voltage to the anodes of the 3-500Zs. Here's how:

1. Make the appropriate changes in the tuned input circuits.

2. Disconnect the red, secondary wire of the HV-transformer from the rectifiers. Insulate the loose wire.

3. Re-connect the amplifier to the electric-mains, key and drive the amplifier with about 5W initially.

4. Observe the grid-current meter and apply only enough drive to obtain no more than 250mA of grid-current.

5. Adjust L and Cout for the best SWR.

This method is not as accurate as the full-power adjustment method, but it is safer.

Here are the optimum values I found for the tuned-inputs, using the full-power adjustment method:

Band Cin pF. . L -turns removed . Cout pF

80m 2 x 680 -4t . 1300pF

40m 820 . . -4t 680pF

20m 360 . . . -1t . . 270pF

15m 270 . -2t. 180pF

10m 180 . -2t approx. 130pF

Notes:

1. The 10m and 15m values are affected by lead length and placement, so some fine tuning will usually be needed.

2. The amplifier was equipped with two, 10 ohm [5 ohm net] cathode resistors [Rc] These resistors increase the input-resistance of the 3-500Zs by about 8%.

3. The ALC circuit had been removed from this amplifier. This slightly reduces the load capacitance on the output of the tuned-input circuit.

4. The capacitors are mica, rated at = or >500v, or NP0 ceramic rated at 1kV.

Other experimenters have reported finding slightly different optimum values, so the best values for your amplifier may be slightly different than those listed.

Removing The Tuned Input Circuit Inductors

The inductors are fastened to the chassis by two spring tabs in the base of each inductor. When the inductor base is pushed through its mounting hole, the spring tabs are compressed as they pass through the mounting hole. After the spring tabs have passed through the mounting hole, they spring-out and lock in the inductor base.

In order to remove an inductor, both spring tabs must be compressed. The upper spring tab can be easily compressed with a screwdriver blade. The lower spring tab is difficult to reach unless a special tool is used.

I made this tool out of 1/8" diameter piano wire, which can be purchased in 36" lengths in model airplane shops. Here's one method of making the tool: Using a bench grinder, cut off about 30cm [12"] of the wire. With a pencil-point flame from a propane torch, heat a spot on the wire, about 2cm [0.8"] from the end of the wire. When the metal is glowing red, grasp the end of the wire with pliers and bend a approx. 85 degree angle in the wire.

The long end of the wire-tool is the handle. The short end is hooked under the inductor base. When the handle is pulled straight-up, the lower spring tab will depress. It is much easier to remove turns from the inductors when the inductors have been removed from the amplifier.

Easy Cabinet Removal

The SB-220's cabinet can be easily removed as follows:

1. Use a book or a block of wood that is about 5-inches by 7 inches and more than 1-inch thick.

2. Hold the book or wood block spacer flat against the rear of the amplifier, between the air inlet screen and the right edge of the cabinet, with the approx. 5-inch dimension vertical.

3. Using your other hand, rock the amplifier over backwards on top of the spacer, until the amplifier is standing up. If the spacer is positioned properly, it will be supporting the full weight of the amplifier and no part of the cabinet will touch the table.

4. Remove the four screws with their rubber feet.

5. Slide the cabinet up and off.

Adding 160m Coverage

Unfortunately, a number of technically unsound, 160m conversions for the SB-220 have been published. Most of these conversions unnecessarily discard the original filament and/or HV RF-chokes and ignore RF-design rules. A better 160m conversion can be found in the January, 1989 issue of QST Magazine, beginning on page 23.

Conclusion

The Heathkit SB-220, and its younger cousins, the SB-221 and HL-2200, can provide many years of trouble free service. The only things they need from their owners are a few circuit-improvements, and a yearly cleaning and fan oiling. n

 

If you have any questions or comments about this article, please feel free to telephone me at 805 386 3734.

Rich, AG6K

 

[End Notes]

1. The filament-pins receive a considerable amount of heat through conduction from the approx. 75w filament. The amount of heat present requires that continuous forced-air cooling be directed at the filament-pins, even on standby.

2. The designers of the Trio-Kenwood TL-922 amplifier, whose circuit looks amazingly similar to the SB-220 circuit, exhibited some original thinking when they decided to incorporate black, low heat-reflective surfaces, adjacent to the 3-500Z anodes.

3. The fan does not need to be removed to be drilled.

4. "WD-40", "LPS", and similar products, are NOT non-gumming.

5. This oil can be purchased in stores that sell fishing reels and/or firearms. Ordinary, 10w or 20w SG-grade motor oil can also be used.

6. Meter damage can be avoided by connecting two, ordinary = or >1A, any PIV, silicon rectifiers across the terminals on each meter. The two diode arrows should point in opposite directions.

7. When the SB-220 is powered from 120.0V or 240.0V, the no-load HV is very close to 3080V.

8. For an extensive treatment of this subject see: "The Emperor's New Clothes" by Hans Christian Andersen.

9. The areas to be bonded should be degreased. After the stepstart parts are in place, the amplifier should not be disturbed for at least 24-hours to allow the silicone-rubber adhesive to cure.

10. In the SB-220, each of these two capacitors is made from four, 200uF, 450V capacitors in series. Thus, the four capacitors act as a single 50uF, 1800V capacitor.

11. To be more precise, after the big-bang, if the anode [plate]-current meter indicates a current flow, when the amplifier is not in the transmit mode, a 3-500Z filament has been pushed against its {grounded} grid. This short-circuit grounds everything that is connected to the filament circuit, including the unfused, +110V relay supply. For more info see "RL1, The Antenna and Bias Relay", #2.

12. If you would like to purchase a complete increased duty-cycle Suppressor Retrofit-Kit for the SB-220, they are available from me for $14, delivered An ad for this item appears on page 168 in the February 1991 issue of QST. If you would like to receive a 2-page information sheet and price list, send me a postcard with your address.

13. In later models, the grid-to-chassis capacitors were changed to 115pF.

14. For more information, see "Amplifier Driver Compatibility" which begins on page 17 in the April 1989 issue of QST.

15. In actual use, the metal in the contacts will run hotter because of the increased current burden. This will increase R. Thus, the contact dissipation will probably increase more than 39%.

16 At the instant of maximum peak-current(s), the peak grid-current, per tube, is approx. 0.5a and the peak anode-current, per-tube is approx. 1.2a. Thus, the peak cathode-current is 1.7a per-tube. This represents a meter-indicated anode-current of about 800mA for two 3-500Zs.

17. The average input-resistance for a pair of 3-500Zs is about double this value, or approx. 65 ohms .

18. The capacitors used should be 500V silver-mica type or = or >1000V ceramic, NP0 type.

19. These suppressors are supplied with the suppressor retrofit-kits.

20. For more information, see "Amplifier Driver Compatibility" which begins on page 17 in the April 1989 issue of QST.