An Essential Mod to the SB-220 and SB-221
The SB-220 and SB-221 amps are subject to rather serious damage when either of its 3-500Z tubes fail in such a way that the tube develops a grid-to-cathode short. Unfortunately, this failure mode seems to be pretty common.
What happens is that this grid-to-cathode short causes excessive current to flow from the bias transformer, which is also the filament transformer, through the tube, through the grid choke to ground. The result is that both the transformer and the choke meltdown and the transformer fails. New ones are available from Harbaugh Electronics, but they are not cheap at $175. And it is a pain in the ass to change out. It's just better not to melt it down.
Here's a photo of a healthy grid choke (in the middle of the image).
And here's what a grid choke looks like after it has melted down following a 3-500Z grid-to-cathode short.
This is a minor tragedy because all this carnage is not at all necessary. The bias circuit, which is only one of several that could have been designed into the amp, makes this damage inevitable. Simply changing the bias scheme, which is insanely easy to do, renders a grid-to-cathode short harmless. The tube will still be shot, but the rest of the amp will be fine.
The "stock" bias scheme is shown below in Figure 1.
The circuit works by applying 120 volts or so to the cathodes when the amp is in Standby (RX) mode. In this mode, the T/R relay is in the normally closed position, which directly applies 120 volts to the center tap of the filament transformer. There is no resistor in this circuit to limit current when the tube shorts the cathode to ground through the grid. In my humble opinion, this is a fairly dumb design since the bias current that is drawn when the tube is cutoff, as it is in the RX mode, is such that a resistor could easily have been included without dropping the bias voltage enough to cause any problem. It might also have been fused. But it wasn't, and the result has produce an entire product line for Harbaugh: selling replacement bias/filament transformers. The bias transformer and the filament transformers are not actually separate transformers, but rather different secondary windings on the same transformer.
In the TX mode, the T/R relay switches so the wiper is connected to the normally open contact. This connects the zener diode to the cathode and biases the tubes for Class AB2 operation.
Rich Measures, AG6K, has suggested a mod that prevents this problem. It is called "resistive bias" and is described succinctly on his web page. I have expanded that explanation here and I've included some photos of the mod. Unfortunately, I waited until a tube shorted and burned up the transformer before I got serious about modifying the circuit. The bias circuit modified as Rich suggests is shown below in Figure 2.
This circuit works by cutting off the tubes in RX mode by connecting the cathode to the zener diode (essentially to ground for the purposes of this mode) through the 100k resistor. Even though the tubes are cutoff, a small current still flows from the plate to the cathode through the resistor and to ground through the zener diode. This small current of about 1 ma is enough to cause the zener diode to conduct and develop about 25 volts (according to Rich) across the resistor. This 25 volts is enough to keep the tube cutoff. It is a good arrangement because it is "adaptive" or "automatic". If the tube starts to conduct, the plate to cathode current will increase. This increased current causes a greater drop across the resistor, which increases the positive cathode bias (which amounts to negative grid bias) and tends to cutoff the tube. So the tube will find its own bias level and, if the resistor is large enough, the tube will remain cutoff regardless of any tube-to-tube or circuit-to-circuit variations. As mentioned earlier, this is called "resistive bias", whereas the original circuit used "voltage bias".
This mod may seem a bit strange because the bias supply is no longer even used. This might lead one to think the Heath engineers could have biased the tubes this way and deleted the transformer altogether. However, even though I didn't show it in this schematic, the bias supply also powers the T/R relay. So the transformer can't really go away entirely. However, had I been one of the Heath engineers, I would have chosen this bias scheme and then used a much lower voltage for the T/R relay. But there may have been other tradeoffs too, so I'll give them the benefit of the doubt. It would actually be great to hear from some of them if they read this.
This mod is fairly easy to make. One disconnects the 120 volts from the T/R relay NC contact and moves one end of the resistor to the NO contact.
This is what the wiring looks like before making the mod. The red wire coming in from the upper right side is the DC bias voltage - about 120 volts. Notice that it goes to the NC contact of the relay and to the coil of the relay.
Step 1: The first step in the mod is to remove the bias voltage from the NC relay contact but keep the relay coil powered up. Here's what it looks like after that has been done. The red wire is the bias voltage, which is now connected only to the relay coil. It is no longer used at all to bias the tubes. Notice the 100k resistor is loose on the end that was connected to the NC contact. I cleaned up the NC contact by removing the solder with solder wick and then cleaned away the residual resin with alcohol.
Step 2: Connect the loose end of the 100k resistor to the NO contact of the relay. The black shielded wire coming in from the right is connected to the NO contact. This wire goes to the 5 volt zener diode. The fellow who wired up this particular amp cut the shield and outer insulation fairly close, so I had to be careful not to melt the insulation. But I was able to get in there with my trusty temperature controlled soldering iron. Here's how it looks after the "resistive bias" mod has been made.
There is another way to implement "voltage bias" that won't blow up when one of the tubes develops a grid-to-cathode short. The circuit is shown below in Figure 3.
This circuit biases the tubes into cutoff in RX mode by pulling the cathode up toward the 120 volt bias supply voltage through the 100k resistor. There will be some voltage drop across the resistor, but the cathode will still be close to 100 volts and firmly in cutoff. When the cathode shorts to ground through the grid, the current is limited to about 1 ma and the transformer will not melt down. Voile.
When the T/R relay is switched into TX mode, the zener is connected to the center tap of the filament transformer, taking the cathode to + 5 volts and biasing the amp as intended for class AB2 operation.
I haven't tested this bias circuit because I don't have a replacement bias transformer yet. When I get the new transformer and test this circuit, I will report the results here. It is an easy mod to make: just connect the loosened end of the resistor to the bias voltage where it connects to the relay coil instead of to the NO contact. I don't currently see any particular advantage to it over the "resistive bias" scheme, but I'm curious and intend to play with it.
Everybody who owns an SB-220 or an SB-221 needs to modify the bias scheme. I recommend you do it before a tube shorts and melts down the bias/filament transformer and the grid choke. It appears to be almost inevitable that a 3-500Z will develop a grid-to-cathode short.
The issue of why the tubes short is a whole other topic, and one that has been energetically, and rather famously, debated. The debate has been primarily between Rich Measures AG6K and Tom Rauch W8JI. Rich thinks the shorts are caused by VHF parasitic oscillations while Tom thinks that is nuts and the tubes short because they have gotten gassy and/or are operated improperly.
My amp did not have any parasitic suppression mods (which Rich promotes and Harbaugh sells), so I can't rule out, at least on that basis alone, parasitics as a possible cause. The amp did not exhibit any arcing behavior. Even when it failed, I was not aware of any arcing in or out of the tube. It just started to smell hot and I noticed it wasn't working any more.
However, this amp was new to me and had been used less than an hour before it melted down. I don't have any idea how long it had been since the tubes were previously turned on. Tom Rauch states that glass tubes in-gas over time if they don't get hot every so often. He recommends being sure to use a glass power tube for a few hours every month or so to drive out any air that has leaked in through the seals. I don't have any assurance that the tubes in this amp had been treated in that way. In fact, I suspect the tubes had not been fired up for quite some time. So in-gassing may well explain why this tube shorted. In fact, based on the evidence (as skimpy as it is), that seems the most likely explanation.
The end result is: I don't know what happened.
I plan to cook my remaining 3-500Zs at a low plate voltage with some grid current, as recommended by Tom Rauch. I will also install the parasitic suppressors as recommended by Rich Measures. I have no reason to doubt Rich's finding that the SB-220 tank circuit shows a high Q at 110 Mhz, although I haven't tried to duplicate it. I haven't thought about this much, and I haven't done any real analysis or measurement, but it seems that such a relatively high Q at 110 Mhz could indicate that there is a pole near the abscissa, which means it's close to the left half plane - which is, or course, unstable territory. High Q circuits "ring". Could a bit of noise or something like switching transients cause enough ringing to create an arc and then a short? Tom thinks not. But reducing the Q at 110 Mhz doesn't seem like a totally dumb thing to do. And many who have made this mod swear by it. So, in the absence of any actual analysis on my part, I figure I will do what both of these gentlemen recommend.
Remains to be seen if I will ever muster the energy to actually investigate any of this myself. And, even if I do, there is no guarantee I would add any value to the discussion.
Thanks, Tom and Rich, for your work on this issue. Too bad I was too lazy to act on your recommendations before I melted the amp!