I did not read all of this,  a couple comments on tuners/ swr/ input/ etc

Internal tuners:

I once ran across a puzzle  using a Yaesu  890    into  an old amp with no tuner.    The Yaesu  (as I remember it)  has no provision for  "locking"  the tuner  settings.  I was able to tune the amp  "key down carrier"   but then while attempting to use it (or test it) using voice,  it would "walk" out of tune,  WHY?

Because   THE DRIVE  (input impedance)  of the amp CHANGES  with load and drive!!!  You can not tune  an amp on reduced power,  you can not evaluate the input impedance or matching,  or the output impedance or matching circuit(s)  under light load.

You MUST  set all this stuff up to be  correct under full output.

That is why he is using a non-inductive resistor as a dummy load across the cathode to grid to simulate the impedance of the tubes under full power, but the SWR isn't even close.

I still think I am missing something.

at this point I would be temped to turn the band switch off the 40 meter selection and measure the 40 meter slug inductance with an LCR meter and see what that inductance is.

It appears to be the only thing thing that hasn't been looked at yet.

That should tell you if the inductor is the wrong value or maybe even shorted out or opened up.

i don't think you have to u-solder it, just move the band switch off the 40 meter position and it isolates both sides of that slug.

maybe somebody did a hot switch and it opened up or created a short from an arc or maybe somebody replaced it or shortened it or something else.

If you don't get close to 1.65 uh then that is a problem.

It looks like it might be the only thing left.

I'll check resonance with the slug back in later.  The resistor is connected between the filaments and chassis ground.

The 40m caps are tied to the same ground solder lug as the coil beneath them, and all the other bands don't show this impedance issue.  I had this apart during the cleaning and tightened these back pretty good.  Also the solder lugs have little serrations like a star washer so they bite into the aluminum.  I will check continuity to ground when I get back to it but I think it's probably good.

I am not sure how to check inductance using the test equipment I have available.  Good news is the coil could be re-wound if needed, but it would be tough to get it out of the amp without taking a bunch of other stuff apart.

Oh, thought you had an LCR meter.

You can get them on amazon for about $30 measures Inductance, Capacitance and Resistance

I have the Proster BM4070 seems to work good enough, I have made coils based on math and the meter showed the correct reading and it comes with a bunch of different cables with clips and stuff

I broke down and ordered one of those LCR meters.  I figure it will make a lot of things possible building homebrew traps and filters, etc beyond just diagnosing this problem.

But I was thinking any short circuit in the coil would remove windings and lower the inductance.  This should raise the resonant frequency, not lower it.  If the coil was open at some point, that should read as off the chart SWR all the way to the edge of the Smith chart, and that's not what this is doing.

Here we see it starting near the center of the chart and heading off as the SWR rises with frequency.

Attached File


Here's a much wider sweep from 0.5 MHz to 15 MHz.  Should it have a more pronounced dip at resonance or is this what I should expect to see?

Attached File

Now looking at this plot of all the input filters together, I am thinking perhaps my test setup is wrong.  

Attached File


ETA I just realized I probably have the wrong relay contacts jumpered.

ETA2 Nope I have the correct contacts jumpered.

I think the only way to solve this, unfortunately, is going to be adjusting the slug under power, which is scary.

I'm going to have to plan that whole thing out, making sure my space is clear of clutter, distractions, and anything that might make me trip or fall into the amp.  Also will have to so this while keying the radio.  So one hand on the mic, the other on the tuning stick.

I don't know exactly what those graphs are supposed to look like, but they are supposed to be low pass filters, and while the lowest in freq one and the next highest one look like low pass filters but skewed high for the band, the others do not look like what I would expect from a low pass filter.

So there are 3 possibilities.

1. That is just the way it is because these input filters were made for tube radios that could tune 3:1 in the tank circuit.

2. Something is wrong with the test setup and 50 ohm impedance load

3. There is a failed component affecting all the filters and if so it looks like all the low pass filters have a high freq cutoff that is too high in freq.

But you said all the bands work under power except 40 meters, so maybe that fact points to an issue with the test load / impedance

Looking at your pic of the 2 resistors in place, they look like common carbon film resistors.

Common everyday carbon firm resistors are inductive. 100 ohms may not be very inductive, but

They will introduce a different impedance at different RF frequencies, which is why the procedure specifies using non-inductive resistors to simulate the load even at 100 ohms.

I think this very much may be your test issue and it is influencing your graphs of SWR.

Since they are parallel the over all inductance could be less but since they are shrink wrapped together up against each other they could also be coupling increasing the inductance of each

@Jambalaya

Thabks for the input and the offer.

I know I have some SMA sockets lying around.  If I connected one in place of the resistors and used the 50 ohm calibration standard from my Nano VNA, that should suffice, right?

ETA I found this on the W8JI website (emphasis mine):

He goes on to say:
Attached File


Reading it this way, it sounds like there should be no less than 4 separate 100 ohm, non-inductive resistors, each installed from one of the filament (cathode) pins to one of the grid pins (which is ground, in this amp, since it has grounded grids).

The antenna analyzer would then be looking at 25 ohms through the input filter, right?

Yes I think you are right. I don't know how we missed that. On the posts I went and crossed out I saw that but got sucked into it needing to be 50 ohms and then crossed it all out. With 2 tubes, is indeed 25 ohms with grounded grids.

Each pair of 100 ohm resistors is 50 ohms with grounded grids, but with the original not grounded grids, the input circuit was seeing some other impedance because of the components that go from grid to ground.  But the amp works with grounded grids and I dont see anybody online talking about changing or even adjusting the tuning circuits after grounding the grids.

So yep, you are right, you need 2 sets of 50 ohms from filament to ground which gives you 25 ohms.

So I was right, but decided I was wrong and crossed it out on page 3. LOL I missed it by that -{}-  much

So 2 things, that and needing to use non-inductive resistors.  The filament is supposed to have a resistance of only an ohm or two, so essentially a short.

That obviously isn't going to fix the problem but maybe give you proper graphs to see what is going on and maybe you just need to adjust the slug properly. You did see the SWR change under power from 3.5 to 2.0 when you replaced the first cap, maybe that fixed the problem and you just need to adjust the slug and could not get the right spot with the VNA because the load was wrong.

Fingers crossed

ETA: the pi low pass filter is a low pass filter, but it is also the same layout as a T impedance matching circuit ( like a T layout for a tuner ) so it does both jobs, low pass filter and impedance match from the 50 ohm radio output to the 25 ohm input to the tubes. Obviously if you make the tubes input 50 ohms, it will mess up the match.

That and the inductive carbon film resistors changing impedance with frequency.

So I did a little reading on these very small carbon film resistors.  Supposedly the inductance is negligible, almost like that of a bare wire of the same length.

It can't hurt to try them just to see if it yields a drastically different plot and can be tuned to resonance on 40m.  The plot with my 50 ohm load calibration standard was insignificantly different from the one using the little resistors.

The other thought I had is in W8JI's last comment, he says "per tube", which would mean using four 200 ohm resistors (and back at square one with 50 Ohms).   It's also possible he meant "per filament pin", as in the first quote.  

It would have been clearer if he had specified a resistor value rather than impedance under load.  

Another thing I tried that I forgot to mention is that I saw a video where the bandswitch detents can get out of time with the wafer positions.  I did some sweeps with the band switch held several degrees to the left and right to see if it plotted differently, and it did not, so I can at least eliminate another potential cause.

So I tried four 100 ohm resistors just to see what would happen.  I got the following plot, which seems to put the resonant frequency even lower.  If 50 ohms was the wrong resistance, and 25 was correct, it's expect the resonant frequency to move up, not down, right?  

Unless the added inductance from the resistors is that significant, but the resistors would add parallel inductance in this arrangement, and the slug coil is in series.


Attached File


I'm going to try 100 ohms across the output of the coax before the cathode connections.  When I tried 50 ohms there it didn't make a big difference.  Just as an experiment I want to see if doubling the resistance brings the resonant point up in frequency.

So interestingly, 100 ohms total yields the following plot.

Attached File


This one is resonant just above 40 meters, and I could move the slug in and get that resonance closer.  

Makes me wonder if he meant 100 Ohms for a pair of 3-500Z under load.

Here's a plot of 40, 20 and 15 meters with a 100 Ohm load.

Attached File


Whatever the problem is, it must be related to the impedance changing as the drive power increases.  It's just weird that it starts out with a reasonable match at 5 watts of drive and goes over 3:1 at 15 watts.   The input circuit should have a mismatch at low power and get better as drive increases if it is optimized for full power.

In any case, I think the combined tube impedance under load is probably closer to 100 Ohms than 25 or 50 based on these experiments.

Inspect the tuning ferrite in the 40 meter coil for cracks or being broken.

I've had it out of the coil completely and aside from having some of the sticky wax on it, it appears to be intact.

Maybe try swapping it with one of the others?  Last night I read through everything you have tried to get the input SWR to come down on 40 and other than your tuning method, you have tried everything I would have.  The frustrating part is that this is a simple circuit.  If I think of something else I'll post it up.

The actual tube specs say: first number is 2500 VDC and the second number is 3500VDC

Driving Impedance .....................................................................100             115 Ohm

https://pentalabs.com/wp-content/uploads/2019/12/3-500zg.pdf


so it appears the way he wrote 100 ohms from each cathode pin to each grid pin per tube is wrong, because that would result in 50 ohm impedance, not between 100 and 115. ohms.  He probably meant each cathode to grid, not each cathode pin to grid because in reality the cathode filament is essentially a short circuit.


But that puts us back to 50 ohm load with 2 tubes which is what you had been using.

I am back to something is wrong with the slug

I haven't.  The other bands are okay and I don't want to mess with them.  

That said, given that removing the slug moves the resonance up the band, and it's apparent that I need less inductance rather than more, I don't think the slug is the problem.

I just wish I could see a series of plots of an actual amp under load in 5W increments to see what it should look like.  W8JI mentions the tube impedance is dynamic.  

Looking at the various plots using three different dummy loads at 25, 50 and 100 ohms, the plots are all so broad banded that it doesn't look like the tuning slug position makes a significant difference.

In order for the slug to have an effect on tuning, I should be seeing a sharp filter skirt that moves with the slug, and I'm not seeing that in these plots.  This leads me to believe my test load just isn't working.

But looking at the underside of the amp there just isn't anything between the tube sockets and the cable leading to the band switch.  Same for the input cable.  There's nothing that could cause this behavior on 40 meters and leave the other bands unaffected.

I do have that LCR meter on order.  I will check the coil impedance and also check the resistors I used for inductance.  

I suppose it's possible that my replacement mica capacitors are bad too, but that's unlikely.  

One of the few things I haven't tried yet is a different radio.

This guy posted input circuit plots for his known good SB-220 for 80m and 40m using the 50 ohm input load resistor for the tube input.

https://groups.io/g/HeathKit-HFAmps/topic/85627857#msg16103

I think he is sweeping up to 30 or 60Mhz so it is kind of hard to see because the images are so small.

But it sure does look like his 80m and 40 m filters have a much higher Q than yours.

I will see if I can dig out my amp and find some resistors and do a sweep or at least get the SWR on the inputs.

Hope this helps

So I emailed W8JI last night and this was his response this morning.  It clarifies that it should be a net 50 ohm impedance seen by the input network.


So that helps with the dummy load arrangement.  I think the reason he wants 4 resistors is to be sure to include the filaments and any associated impedance effects they may have.

He answered some questions for me via email when I was fixing my AL-80B. Nice guy.

Yes that's a considerably steeper slope on his filter skirt.  

If you can get a plot of your 40m coil that would be great, but I'll feel awful if something happens to your amp.  I try to stick with "If it ain't broke, don't fix it", but perfectionism usually wins out and I meddle anyway.

I wonder if exposure to extreme temperature swings over several summers and winters would have an adverse effect on the ferrite slug?  

Another thing to consider is these tubes are for the most part unproven.  I haven't used more than 50W drive with the power supply switch in the CW position.  

Is it conceivable that the filament impedance will settle down as drive approaches 100W?  It's strange that the mismatch gets worse going from 5W drive to 15W.  And it's also strange this only happens on 40m.  But maybe the impedance settles down again as you approach 100W.  The trouble is my transceiver folds back over 3:1 so I can't even tune the amp.  

So a minor update.  I had to go out of town on business and have had a lot going in lately, bought a UTV and have been playing with that, etc.

W8JI responded to my email again saying the plot looked about right for a passive plot but that the impedance was going to change under power.

Anyhow in all the fiddling with the input coil, I managed to break a wore off the inside of one of the meter lamp sockets.  These sockets are hard to find, but I found a reasonable alternative on Amazon that will just require a second wire soldered on.

This is an automotive gauge lamp and housing that fits a 5/8" hole and I think the bayonet socket will fit the OEM Heathkit bulbs.  I'll just need to solder a second wire onto the socket since these are intended to get their ground connection through the panel they are installed in.

Attached File


There was just no way to disassemble the factory sockets to get at the solder joint and repair it.  If anyone needs meter bulb sockets, these are pretty close.

I am going to make a suggestion here. Find the equivalent bulb in LED and replace the old filament bulb with it. Seen a lot of Sb200 and Sb220 where the bulb sockets have been replaced and the bulb ends up melting the back of the meter. If you have bot done it yet I would do that and add the meter protection diodes as well. If you think the sockets are rare, the meters are much harder to find. If you do find them they are like gold. You know how many of the hams are. Old 1990s rig that is beat to hell, they want $800 at a ham fest even though you could get an IC-7300 used for that price or an FT-991a. They think everything is worth more than when they were new.

Thanks, yeah I saw some 5V LED bulbs with bayonet sockets that I think are intended for use in pinball machines that people have been using in the meters.  

I can order those along with the diodes and install it all when I get a chance.

ETA another thing that might make a difference is the tubes themselves.  Maybe a different set of tubes might show a different impedance under load and there isn't really anything wrong with the amp or input circuit.

Well I got the meter protection diodes ordered, along with some LED bulbs.  I replaced one of the meter bulb sockets today and the new socket placed the factory bulb a lot closer to the back of the meter face.  

Keeping in mind the info about the meters melting after changing the sockets, I went ahead and got the LED replacements that will hopefully run cooler.  They were advertised as suitable for many applications, including "old radios".

Do you mind posting the bulb part number?  A complete SB 220 rebuild is in the plans for this winter here.

Not sure the part number.  Just China stuff from Amazon.

Link

Attached File



ETA it's a 15 pack so if you want I can just mail you a couple.  I'm never going to use 15 of them.

Thank you!  I appreciate the offer, but I ordered a package for myself, I have 2 SB220's sitting here that could use them and will likely go through several more SB's over the winter.

Good deal.  I hope they work out.  Mine are supposed to be delivered tomorrow.  My concerns are if I need to change the polarity of the sockets.  I haven't looked at the schematic but it's possible the bulbs run on AC direct from the transformer. If that's the case they will only be lit half the time and might appear dimmer.  Lots of easy fixes for that though.

Many of these bulbs have polarity diodes in them. The lamp bulbs on most radios and amplifiers run off the filament transformer. In fact it is one of the processes for diagnosing them if the filaments are not burning. You check the meter lamps to see if you have filament voltage. I have an LED in my SB200 but it needs to be replaced. My new JW miller tuner uses 2 #47 bulbs as well so I think I need to oder a pack of them as well.

Well, looks like the LEDs work with no further modifications.  I might want to change out the other socket so they match.  It's definitely a noticeably bluer color temperature.

Attached File

Looks good.