Print

[N2CJ]

I've been working on reviving an SB-220 that I picked up (minus the tubes) about five years ago and has been sitting ever since. As far as I can tell it is completely stock without any mods. It looks a little dirty from storage with some whitish oxidation on the aluminum chassis, but no signs of operational abuse. After resolving some small issues (120V cord but wired for 240v) I rejumpered for 120V and brought it up on a variac and high voltage was good (2850V), meter lights and fans came on so it seemed that both transformers were good. I installed some used Eimac 3-500z's (of unknown provenance) that purchased some years earlier and after diagnosing some dirty pins and/or tube sockets got both tubes to light and let the amp idle for an hour or so. Dummy load testing with carrier on 20M I made 500W output in CW/TUNE setting, then noticed that one of the tubes was no longer lit. I switched to antenna and made a brief QSO with good signal report, no issues reported. After cool down I recleaned the socket and cleaned the tube pins with some emery cloth and reinstalled. Both tubes were now lit.

Further testing (starting at 30W and rising to 100W exciter output) yielded the following maximum output (into my Cantenna) in both CW(TUNE) / SSB settings:

80M   900 / 1100 W
40M   900 / 1000 W
20M   800 /900 W
17M   750 / 900 W
15M    750 / 850 W
12M   700 / 850 W
10M   650 / 750 W

My budget being what it is, I am just getting the amp cleaned up and operational. I am going to install the Harback SK-220 soft-key mod,  I am not going to spring for a Harbach power supply and new tubes unless I have to. I just want to take the reasonable precautions to limit collateral damage in the event of tube or other failure. I installed back to back diodes across the meter terminals without incident. [I understand that diode protection is built in to a new metering board] I then turned my attention to the possibility of adding a glitch resistor. Not being one to blindy post without doing my homework, I found extensive discussion of glitch resistors in prior postings and on W8JI.com; however, there are a few things that I am unclear on:

1) W8JI writes (on his website) "Glitch resistors shoould always be in the high voltage line, and preferably near the filter capacitors. They should be high voltage type resistors capable of absorbing considerable energy without failing or arc. Suitable resistors are RCD 175P pulse rated resistors, or glow-bar type resistors that are physically very long. Some high power wire wound resistors will work, although generally most are not suitable." But Tom does not make specific reference to the SB-220, nor does he provide the recommended ohm value of this resistor, which other sources have anywhere from 10 to 25 ohms. Also, after extensive searches I have been having trouble finding a vendor who can supply this part.


In 2011 W8JI wrote "The single most beneficial change in the SB220 would be getting rid of the grid chokes and mica caps on the grids, and directly grounding the grids. Also a keying voltage modification and meter protection is a good idea. The rest of the changes really don't help anything, and some even make things worse." No mention of glitch resistor.

In 2010 W8JI wrote "If you wanted to protect your amplifier's expensive transformers for hard circuit faults, including tube shorts, you could install a few resistors in proper places. One place would be just after the filter capacitors. If the HV line arced to ground, a properly sized and style resistor could limit fault current to 10 amperes, giving enough time for a fuse to blow before a transformer or choke melts."

Does R-3, the .82 ohm resistor that blows in the event of a high voltage short to ground make the use of a separate glitch resistor redundant?

Also, is it prudent to replace the circuit breakers in the SB-220?

Now on to directly grounding the grids. W8JI highly recommends this and his website contains an example of 3-500z grids in a TL-922 with this modification. In his pictures Tom uses the screws which also secure the corners of the tube socket as part of the ground, which he interconnects with copper strip. I would also like to do this to avoid drilling new holes in the chassis, but I am concerned because most (but not all) of these screws have measurable resistance (up to 300 ohms) to the chassis.  Perhaps loosening the sockets and cleaning around where the screws contact the chassis can solve this. Has anyone else grounded the grids by using these screws?

[KD0REQ]

W8JI has written iirc that glitch resistors are chosen to limit the plate current to the tube max.  this way, the resistor will not let enough current flow to melt the tube.  so using the current rating in the spec sheet for one tube at that 2850 (or say 2900 for easier math) and two times the tube current because it has two tubes, plug the values into Ohm's Law (E / IxR) and see what resistance it comes up with.  I forget what he wrote for power level, something around 50 to 75 watts, and of course isolated from the chassis, because BANG!!

do the math, find the area in the amps or tubes section that discusses power levels, and order that HV power resistor.  good luck, and you'll learn something by doing the rest of the searching.  mostly by reading lots of good data about how tubes work at generating power.

[WA7PRC]

Hi Glenn,

I've owned my SB-220 since the mid 1970s. A 3-500Z can fail if the filament wire-to-glass seal isn't kept cool enough (200°C maximum). The Heath cooling scheme while simple & effective needs periodic maintenance. The fan motor bearings need at least annual lubrication. Otherwise, the tube filament pins and socket contacts can overheat. An overheated tube pin manifests itself as solder that melts and drops out. The socket can also overheat and look like this:

(click for larger image)
Note the good grid contact on the left. The ends of the contacts are fully closed. Note the overheated filament contact on the right. The plating has blackened and the contact ends are not closed. The spring has relaxed from being overheated. It will not support 14.5A of filament current (there will be significant voltage drop).

Ducting of air will assure the best airflow. That means the fan blade should be centered on the panel that is cutout to accept it. Also, the panel on top (above the motor) should be in place. This is in the SB-220 manual. David K5DBX added a shroud around the fan blade and says it increases airflow and decreases fan noise. This is my version:

(click for larger image)
There are two pieces, above AND below the chassis. They're secured by the existing hardware. I added a couple of beads of RTV to hold it to the panel between the ends.

Hard-grounding the grid pins is best done with short and wide connections. The BEST way to do that on the SB-220 is with copper strips like this:

(click for larger image)
The end of the socket terminal is bent up at the point where it meets the edge of the ceramic. A soft copper strip is formed to make a short connection between the chassis and socket terminal, bent over the end of the terminal, and soldered. I used #4-40 hardware to make the connection to chassis. Yes, holes are drilled. Big deal.

You need R3. It is the grid current shunt resistor. Leave it alone. However, you can replace all the rectifiers with better (more robust) types. 1N5408 (3A/1000V) types are a good choice. Don't forget to add one for meter protection between -B and chassis. Tom W8JI shows how/where/why here: http://www.w8ji.com/fault_protection.htm.

Unless the circuit breakers are defective, leave them alone.

I replaced RFC2 (8.5 μH) with a 10Ω/20W wirewound resistor (Ohmite pn B20J10RE):

You lose a bit of X_L but it still works fine. The value isn't terribly critical but, you should use a vitreous enamel ceramic type (avoid the square 'sand resistor' types).

The 'soft key' relay driver modification you mentioned comes from Harbach Electronics. It is very popular but, still uses the slow, loud, and failure prone open relay. If you're going to replace part of the circuit, you might want to consider using hermetically sealed (vacuum) relays instead. They're MUCH faster, quieter, and more reliable. Jim W7RY developed a PCB to drive them. With that, you also get electronically regulated and switched bias. Depending on where you get the relays, expect to spend around $100 total. Jim sells on ebay as user 'radioamplifiers'.

vy 73,
Bryan WA7PRC

[W8JX]

I've owned my SB-220 since the mid 1970s. A 3-500Z can fail if the filament wire-to-glass seal isn't kept cool enough (200°C maximum). The Heath cooling scheme while simple & effective needs periodic maintenance. The fan motor bearings need at least annual lubrication. Otherwise, the tube filament pins and socket contacts can overheat. An overheated tube pin manifests itself as solder that melts and drops out. The socket can also overheat

When tube and socket was originally designed it was to have cooling air for tube applied through socket and thru a envelope around tube. When done like this there is rarely issues with pins or sockets. Heathkit and other well into modern times eliminated base applied cooling and chimneys to cut costs and simplify chassis design. This has caused issue with many pins seals and solder in pins over the years. Lacking proper cooling you want to increase airflow across tubes and sockets to reduce peak temperatures.

[KM1H]

Forget the overkill JI web page as all you really want to do is limit current until the rear breakers pop. Unlike much bigger amps the SB-220 is severely current limited by the little transformer. In that case replace RFC-2 with a 15-20 Ohm 20-25W vitreous enameled wire wound and add a .0047 6KV disc between it and RFC-1.

Carl

[WA7PRC]

I've owned my SB-220 since the mid 1970s. A 3-500Z can fail if the filament wire-to-glass seal isn't kept cool enough (200°C maximum). The Heath cooling scheme while simple & effective needs periodic maintenance. The fan motor bearings need at least annual lubrication. Otherwise, the tube filament pins and socket contacts can overheat. An overheated tube pin manifests itself as solder that melts and drops out. The socket can also overheat

When tube and socket was originally designed it was to have cooling air for tube applied through socket and thru a envelope around tube. When done like this there is rarely issues with pins or sockets. Heathkit and other well into modern times eliminated base applied cooling and chimneys to cut costs and simplify chassis design. This has caused issue with many pins seals and solder in pins over the years. Lacking proper cooling you want to increase airflow across tubes and sockets to reduce peak temperatures.

Eimac's recommended cooling system used a pn SK410 socket and SK406 chimney:

Heathkit and many others used a simpler and less expensive ceramic socket and fan (above), with the fan blade arranged to push air past the tube envelopes AND filament pins. As measured, it has about 50°C worst-case margin. In either system, airflow needs to be maintained. Keeping the motor bearings (oilite bronze sleeves) lubricated annually is VERY important.

[W8JX]

Forget the overkill JI web page as all you really want to do is limit current until the rear breakers pop. Unlike much bigger amps the SB-220 is severely current limited by the little transformer. In that case replace RFC-2 with a 15-20 Ohm 20-25W vitreous enameled wire wound and add a .0047 6KV disc between it and RFC-1.

Carl

With a bigger higher voltage transformer it could become a bit of a beast.

[WA7PRC]

With a bigger higher voltage transformer it could become a bit of a beast.

When running high duty cycle modes such as RTTY, FM, AM, etc, that's a good thing. To a point, increased C in the HV filter can compensate a bit for dynamic load changes (peaks) when using lower D/C modes (SSB or Morse). More C is generally less costly.

[N2CJ]

These are the sockets after some gentle cleaning.

The socket can also overheat and look like this:

(click for larger image)
Note the good grid contact on the left. The ends of the contacts are fully closed. Note the overheated filament contact on the right. The plating has blackened and the contact ends are not closed. The spring has relaxed from being overheated. It will not support 14.5A of filament current (there will be significant voltage drop).

I'm not seeing any blackening. I don' think this amp ever saw any hard use.

Hard-grounding the grid pins is best done with short and wide connections. The BEST way to do that on the SB-220 is with copper strips like this:

(click for larger image)
The end of the socket terminal is bent up at the point where it meets the edge of the ceramic. A soft copper strip is formed to make a short connection between the chassis and socket terminal, bent over the end of the terminal, and soldered. I used #4-40 hardware to make the connection to chassis. Yes, holes are drilled. Big deal.

This is clearly the neatest and most elegant solution I have seen. But of course it means drilling six new holes, moving two lengths of coax and the bifilar winding out of the way.

You need R3. It is the grid current shunt resistor. Leave it alone. However, you can replace all the rectifiers with better (more robust) types. 1N5408 (3A/1000V) types are a good choice. Don't forget to add one for meter protection between -B and chassis. Tom W8JI shows how/where/why here: http://www.w8ji.com/fault_protection.htm.

I never considered removing R3, just wondered if it made the glitch redundant.

Unless the circuit breakers are defective, leave them alone.

Fair enough, but how can I know they will open when the time comes?

I replaced RFC2 (8.5 μH) with a 10Ω/20W wirewound resistor (Ohmite pn B20J10RE):

You lose a bit of X_L but it still works fine. The value isn't terribly critical but, you should use a vitreous enamel ceramic type (avoid the square 'sand resistor' types).

Forget the overkill JI web page as all you really want to do is limit current until the rear breakers pop. Unlike much bigger amps the SB-220 is severely current limited by the little transformer. In that case replace RFC-2 with a 15-20 Ohm 20-25W vitreous enameled wire wound and add a .0047 6KV disc between it and RFC-1.

Great idea, I hadn't considered RFC2 as the appropriate place for the resistor.

The 'soft key' relay driver modification you mentioned comes from Harbach Electronics. It is very popular but, still uses the slow, loud, and failure prone open relay. If you're going to replace part of the circuit, you might want to consider using hermetically sealed (vacuum) relays instead. They're MUCH faster, quieter, and more reliable. Jim W7RY developed a PCB to drive them. With that, you also get electronically regulated and switched bias. Depending on where you get the relays, expect to spend around $100 total. Jim sells on ebay as user 'radioamplifiers'.

Worth considering

[KM1H]

Eimac's recommended cooling system used a pn SK410 socket and SK406 chimney:

Only for CCS or a lot more than the 2000W PEP INPUT of the SB-220. That was a cost reduced socket that had far less back pressure than the SK-400 kluge used earlier

Heathkit and many others used a simpler and less expensive ceramic socket and fan (above), with the fan blade arranged to push air past the tube envelopes AND filament pins. As measured, it has about 50°C worst-case margin. In either system, airflow needs to be maintained. Keeping the motor bearings (oilite bronze sleeves) lubricated annually is VERY important.

EIMAC approved that sort of cooling as far back as the Johnson T Bolt which ran 4-400A's at the same 2000W PEP....with the same Johnson sockets as the SB-220. As long as base and anode seals were held well below 200 and 225C respectively all was well.

Annually oiling that fan will be sure to gum it up pretty fast which required more oiling until there was so much build it it sometimes seized; every 5-10 years depending upon use has long proved to be sufficient and a full cleaning/relube around 20 years. With modern synthetic oils it will likely be even longer. Of the several hundred Ive had in for service or 6M conversion only a few very grungy ones had a slow fan which quickly came back to life.
Im sure a few will report other cases of problems but it is not the general rule. Most SB-220 cooling problems are due to blocking the air flow or not mounting the blade or top cover properly. Simply reading the manual must have been too much effort after all the assembly work

When running high duty cycle modes such as RTTY, FM, AM, etc, that's a good thing. To a point, increased C in the HV filter can compensate a bit for dynamic load changes (peaks) when using lower D/C modes (SSB or Morse). More C is generally less costly.

With a transformer only rated at .4kVA those modes would require a considerable derating. About 330uF in place of the 200 uF originals is about the safe limit before chancing blowing the unobtainium AC power switch.
A step start is not a cure all as they can only handle so much before they blow, I wouldnt advise beyond 470 uF as now the transformer will be more stressed. A muffin fan on the xfmr will help prolong its life.

Heathkit and other well into modern times eliminated base applied cooling and chimneys to cut costs and simplify chassis design. This has caused issue with many pins seals and solder in pins over the years. Lacking proper cooling you want to increase airflow across tubes and sockets to reduce peak temperatures.

That is NOT correct. When builders and later owners actually read the manual the base seals were well cooled. Forgetting to put that small aluminum top back on the rear of the cover now eliminated the air deflection to the base....simple and effective. Amp Supply (after a few runs), Ameritron and others also provided adequate base seal cooling. My LK-500ZC has been run extremely hard in contests, DXing and quite a bit of AM and the original 1986 Eimacs are still making 1200W at 70-80W drive.

With a bigger higher voltage transformer it could become a bit of a beast.

For a little while. Dahl had an external Hypersil transformer that would boost power to about 1800W until the wimpy band switch and Tune cap arced over. Going to the later wide spaced cap just put more stress on the switch and I dont now how much of a safe cooling budget remained, especially for contesting. As a YCCC member I had plenty of repair work during and after contests.....many drove the amp up for a 1-3 hour turn around since I kept plenty of spares on hand and also had a couple of bone stock loaners

This is clearly the neatest and most elegant solution I have seen. But of course it means drilling six new holes, moving two lengths of coax and the bifilar winding out of the way.

Directly grounding the grids is another Hammy Hambone fix, pushed by JI and others, that isnt always a good thing. It increases the tube gain and can add another 150-200W output. OTOH it also brings an amp that is often very close to taking off on 10M beyond that point and creating a bit of unnecessary damage. A lot depends upon the quality of the original build as well as having the original parasitic suppressor resistors within tolerance.
Using the Rich Measures voodoo nichrome suppressors is pure idiocy as it doesnt address the cause. There are plenty of modern resistors on the market to replace those old carbon composition resistors. The Ohmite OY series is excellent.

Carl

[WA7PRC]

The socket can also overheat and look like this:

(click for larger image)
Note the good grid contact on the left. The ends of the contacts are fully closed. Note the overheated filament contact on the right. The plating has blackened and the contact ends are not closed. The spring has relaxed from being overheated. It will not support 14.5A of filament current (there will be significant voltage drop).

I'm not seeing any blackening. I don' think this amp ever saw any hard use.

Look again. A good grid contact is on the left. The fried filament contact is on the right.

Hard-grounding the grid pins is best done with short and wide connections. The BEST way to do that on the SB-220 is with copper strips like this:

(click for larger image)
The end of the socket terminal is bent up at the point where it meets the edge of the ceramic. A soft copper strip is formed to make a short connection between the chassis and socket terminal, bent over the end of the terminal, and soldered. I used #4-40 hardware to make the connection to chassis. Yes, holes are drilled. Big deal.

This is clearly the neatest and most elegant solution I have seen. But of course it means drilling six new holes, moving two lengths of coax and the bifilar winding out of the way.

It's no big deal to drill six holes. The filament choke wasn't touched.

Unless the circuit breakers are defective, leave them alone.

Fair enough, but how can I know they will open when the time comes?

At some point, you have to trust they'll open when needed. I don't recall hearing of any reliability issues with the CBs Heathkit used.

[N2CJ]

Look again. A good grid contact is on the left. The fried filament contact is on the right.

I was referring to my SB-220, not WA7PRC's contacts in the picture.

So anyway, drilling the holes and the obstructions weren't such a big deal after all, so I grounded the grids with copper strips, so here is the before and after

   

I decided to bend the copper instead of the terminals on the tube sockets, mainly because of an irrational fear of breaking one of the terminals. I think it is satisfactory.



Now I am going to order Ohmite resistors to remake the stock parasitics.