RF deck rework

[Opcom]

I pulled the RF deck out of the Tucker TX las night. I figured it would be much easier to work on it with it out on  the work bench.

Besides relocating the RF choke closer to the 4-1000 than the 20" it is now via copper strap, I want to look into the bottom compartments where the grid tuning and grid bandswitching is. There is one position, the mysterious "2726KC", where the bias is removbed from the final, very disturbing. Maybe someone let the smoke out. I am finding evidence that COL. Tucker knew how to STRAP, but was not very familiar with VHF and parasitics. He did put in many things to try and get rid of them, but I think the 20" of copper strap between the anode and the RF choke is the kind of culprit that was right in front of him and went un-noticed because he was a BC engineer where such long wires are not as important..

So far I have removed some 80 screws from the bottom plate. Maybe 40 to go. I'll post some images later of whatever secret things are in the grid tuning compartment that have not been seen in almost 60 years.

[w1vtp]

Patrick

I'm most interested in your progress.  I've been noting your commentary on the Tucker TX. It'll be neat to finally get a look inside it.  Also of interest is his use of the 304TL as modulators.  I started to collect 304's until I noticed that the prices were all over the map.  After the 3'rd one, I gave up in favor of the 813.

Love to see some good pics if the innards

Al VTP

[Opcom]

Did someone say link coupling? check out the assembly there on the bandswitch, The label says it's a B&W, but I have never seen one like this. Unfortunately, every piece of the 'plastic' that held coils together is rotted history, and this explains bias supply shorts. Would epoxy work to try and fix this?  Hot glue is definitely out. Is it toast?

About the old modulators, They were so hard to find and 4/5 of them bought from ebay cheats did not work. I had to give them up for 3-500Z's. I am OK with that, for the dissipation you can't beat the price. The only issue I am seeing is a form of amplitude distortion at crossover. At 3KV this should not happen but apparently it is common. I can worry about that later.

[Opcom]

more pics, the rotted coils can be seen. Note the red end plates on the B&W unit. Also, the coils may have been modded and positions switched around in order to put 2726KC at the midpoint (knob pointing up). This may have been a 160M coil. The 5-position B&W says 10-80.

My question is - -Was there an original 15M coil which the builder removed, and then he had to add the switch th seelct 15 or 20 when the bandswitch is in the 20M position? The move to 15 is done by shorting the last outside turn on the 20M coil to give 10 turns instead of 12 turns. This is no longer functioning as it makes no difference test-wise what position the 15/20 switch is in.

When a 5.6K resistor was put across the grid terminal to ground and the grid tuning cap meshed halfway and the exciter tuned to that resonant frequency, the SWR from the exciter to the grid circuit was:

80M - 1.8
40M - 2.9
20M - 2.5
10M - 1.7
2726 - 1.2

not bad..

[Opcom]

This shows clearly the 20+ inch long copper strap from the plate to the RF choke. Here is the 140MHz self-oscillating instability.

The schematic was drawn, took a while.  - -here it is. It is very interesting to have a bank of caps switched in for loading. There are two switches. As each is turned clockwise, more caps are put in.

I was also thinking about 160M when checking this. The grid circuit seems limited to 2MHz on the low side. The plate would go no lower than 2.04MHz with all L and C in.

My last thoughts on this were that if I could not fix the bandswitch, I could remove it and use a roller inductor and variable capacitor for continuous tuning, just like the plate circuit has.

[kc6mcw]

Interesting...

[N3DRB The Derb]

those B&W coils are always toast. The plastic ribs they used were unstable and warped over time. Some type of.
polycarbinate..... That looks large for a grid input assy. They made lower power output tanks like that as well.

[W3SLK]

Derb said:

those B&W coils are always toast. The plastic ribs they used were unstable and warped over time. Some type of.
polycarbinate

Tim is quite correct. That is how you can distingquish it of being B&W origin. The material they used for coils of that vintage seemed to be very hydrophoric. After a while, they eventually turn to dust. Those link coil turents were very common for grid coupling. Although I never saw one for 160. I do have a similar Bud coil set that does have 160 capability. I need to install it in the HN-500.

[Opcom]

The long strap to the vacuum cap is ok unless it gives you a problem resonating say like on 10 or 15 meters. However, the long plate choke connection should be eliminated. I would move the choke on the other side close to the 4-1000 and next to the resistors. Those aren’t wire wound resistors are they? If they are you should change them to carbons (non-inductive). Using wire wounds cause parasitics.

I also count only 5 bands for the grid input tuning. Maybe 10, 15, 20, 40, 80 or 15, 20, 40 80, 160, cant tell. Correction, it does look like it's link coupled, but its crazy to have 5 separate coils when a single piece of tapped B&W stock would be sufficient.

For sure it is a link coupled 5-bander. I just found out the bandswitch there in the grid circuit used to have a 15m circuit when bought new, and the OM removed it and put in a a 160M coil for 2726, thus the odd arrangement for 15-20 switching. 2726 was an old frequency for Texas Guard. There is a picture of the bandswitch on page 30 of the advertisement section in the '62 ARRL handbook, although it has no other data that has to be it. It is a "B&W 75 watt turret". The split cap is 100uF per section.

The plate circuit resistors are all "KOOLOHM" 10 Ohm "NIT" 10 watts. 4 in parallel go from the plate to the junction of the pi-network and the DC blocking cap, and another bank of 4 more run from that point to the 20" B+ strap there. In parallel with each bank is a small 1W resistor with some 2 turns of wire on it. One of those is cracked (burned). I have to check to see what value and if they are non-inducztive or not. I have some more checking to do. The 20" of copper strap from the pate to the RF choke - -maybe it is OK, maybe not, but I already decided to move the RF choke closer to the plate just on general suspicions.

The plate coupling cap is five Centralab type 503 500pF/30KV doorknobs in parallel for a combined current rating of 8A@14MHz.
The plate choke is a Raypar RL-100 which is rated 96uH/800mA.

Also - an error in my schematic (besides not showing those resistors) is the neutralizing voltage divider cap that goes to GND is 25pF to match the grid capacitance of the 4-1000 more closely.



I guess there is nothing to do now but remove the assy. and see if it can be salvaged. This is why I asked if epoxy resin would do the job.

[Opcom]

Today the MB-40 network was put in as the grid tank. This is a link-coupled tuned circuit as was suggested. Without drilling more and more holes in the front of the rig, I put it where the bandswitch was. It was a PITA because this puts its mounting away from the chassis and even so I had to use a driveshaft salvaged from a dead scope nd cut to fit.

I got a couple technical questions about this please.

1.) note the phasing of the driveshaft u-joints. It feels OK but with the velvet vernier turning it, I can't tell. Is this correct for not binding?

2.) The u-joints have a little slop in them. Is there anyway to remove backlash, such as with a sping, or should I just understand it and live with it?

3. The tuner is mounted in there as it will be, and aside from it looking like hambone did it, I got to wonder if there is a better way to ground the capacitor frame. The skinny wire there seems long for a ground and it just looks wrong. Would a 1/4" strap take care of this?


The whole TX is full of minor JSments anyway but as long as they work properly and are reliable, it won't make much difference. Reliable JSments. I re-used the ceramic mounts from the old tuning capacitor. One of the ceramic mounts holding the tuner is 0.1" shorter than the rest. This is how it came in the rig originally and I guess the OM just crankes the screws down till it worked. I shimmed with two washers and used a caliper.

Otherwise it was a lot of fun to work on that rig again. I even pulled out a NOS TEK CRT for a 545 to see if it will fit under the RF dech chassis for use as a mod monitor. It will, and with a 10KV accelerating voltage on the 5" CRT, there will be no squinting to see the mod scope. I want to work on that next once the grid tuning is taken care of.

thanks,
PJ

[W7TFO]

'Driveshaft' looks good, just set them up like one on your car.

Nice MB-40, BTW...

73 DG

[WA1GFZ]

People who build stable rigs do not put an inductance in series with the plate tuning cap. The only way it may work is if you connect the inductor directly to the plate cap not the coupling cap end of the strap.  Then the strap inductance transforms the plate Z a bit higher which can also be an issue. A crappy layout will get you on the low bands but above 40 you will have bad things happen. The Strap in series with the plate cap sets up a series tuned VHF tank.
Learned that the hard way a long time ago.

[Opcom]

I do not understand what picture is being referred to. I am not the builder, instead I found this in a scrap yard and have been putting it back together.

The image a few posts above where the plate parasitic suppression resistors and plate choke are shown, or the images just above here where the MB-40 tank assy is shown. That is for the grid.

If you mean the pic right here in this post where the long 3/8" wide strap is between the blocking caps and the plate choke, that has been ripped out as it was the accursed cause of a problem the original 1957 builder never was able to fix. 140MHz hog wild BTW. The plate choke has been moved right next to the blocking caps, and is now mouned upside down from the ceiling so that its RF end is 1" from the blocking caps instead of 20".

If you mean the thick 1" strap between the vacuum variable and the variable inductor is unacceptable, then it is a completely screwed design and the OM is probably spining in his grave.

[W3SLK]

I have a BUD tank turret set that covers 160-10M. I need to put it in the HN-500 to drive it easier.

[WA1GFZ]

I wonder if you could rotate the cap 90 degrees with a right angle drive and eliminate the long strap

[Opcom]

I will have to check and see if there is room and also measure the strap length to see what the resonance might look like. A 90 and a flex shaft might do the job without making it too hard to turn. I don't want any other issues once this is put back together.

[Opcom]

I think I want to just try it first. It's been a long time out of that cabinet and probably wants to get back inside. I moved the HV connector as well to avoid running a long HV line inside the plate area. I need to do one more thing first, and that is add a modulation monitor. The last remaining "difficult" mechanical thing to do is cut the 5" hole in the front and mount up the CRT. It will sit right below the 4-1000 socket and lie front to back. I have a new in box Tektronix CRT for a 545. The shield/mount and bezel will come out of a scrapped Tek scope and it ought to fit OK. I have some measuring to do before any drilling etc.

[Opcom]

I needed my workbench back and finished up the Tucker RF deck and put it in. Since last post, I added a 1/4" wide strap between the MB-40 frame and ground. it is 1.5" long.

I am amazed how easy I can shoot the grid current off the scale with that thing with only the filament lit. But that is a worry.

This marks the first time it has been actually ready to be tested. There is no trace of instability like before.
By reading the meters, they don't act funny and take off wildly now. There is no self oscillation when I bias it for some plate current and tune through resonance on the grid and plate.

Here is what I am getting:

Eb: 2500V
Ib: 200mA
Eg2: 400V
Ig2: 30mA
Eg1: -150V
Ig1: 100mA
Po: 200W
Pdrive: 10-20W forward to the Tucker, 3:1 SWR on the driver

The stage is clearly operating class C (as far as I can tell)
The output seemed vary appropriately with drive power and bias adjustments.
The output varied linearly with plate voltage to a degree, I think it is tuned properly and the dial indications match the OM's charts closely as well.
I was able to push the grid to 100mA even with -200V on it, just by increasing the drive to 20W. That does not seem right.
Where is all the juice going? I think I dare not push the grid like that, but this thing ought to be making a whole lot more power than 200W, with 500W into the plate.

One other thing I see, that I don't understand, is the grid current meter seems to be going negative below zero when the filament is on and there are no other voltages applied to the tube. When I apply some bias, it goes to zero. Can anyone explain that?

Thanks for comments,
Patrick

[WD5JKO]

Patrick,

Sounds like big progress.

  Between your antenna project, and your Tucker project, it sure looks like a big signal will be coming from your QTH soon.

  I was wondering about that 3:1 input SWR. I bet it is high Z, and if so, maybe swamping the drive with resistance loading will help. Maybe a big non inductive resistor across the grid tuning cap, say 5K 25W is a starting point.

  So that 200w out after 500w in is a concern. Getting 40% efficiency for a class C RF amp is pretty low. I wonder if you have a AC compensated 1000X HV scope probe? I mean one of those dudes that can take at least 5KV peak. If so, run the rig on 80M, or 2723Khz, and see what kind of plate swing you have. To get good efficiency, the plate swing has to swing down to the G2 voltage, and possibly a little lower. It could be that the impedance transformation design of that pi-net is for higher voltage at lower current.
It might also be useful to look at G1 with a normal 10X 600v peak scope probe. Using these kinds of probes at RF can be a problem due to introduced capacitance from the probes, but I have found them to work OK on 160/80M.


Jim
WD5JKO

[Opcom]

Jim,

The guys doing the tower showed up again; the boss there, his back is better. The other guy is now out of the hospital. These guys are old men and know their stuff, but it makes me wonder if enough young men still enter the profession.

For sure I will try the non-inductive load across the tank. I may have a bin of Carborundums around here of 2.5K/20W. I like them but mounting is a pita as they are just tubular with no terminals. I don't know if their silvery ends can be soldered to.

2723/2726 is regrettably gone along with the ruined grid tanks. 80M is the lowest it can go.

I do not have an RF probe that can take the plate voltage. The closest thing I have is a 30KV TV set proble that is supposed to be used with a 10KOhm/Volt VOM. It works out to about 200:1. I doubt it will work accurately on 80M because the resistor inside is one of those long spiral type high-meg jobbies. The same goes for the universal slip-on 5KV proble tips I have.

I can look at the grid with probes I do have. Some older tektronix probes.

The pi network impedance issue might be the problem. I might be way off, even though it seems 'tuned', as you say the Z might be wrong.

Th PI network is made of:
Jennings UCS 9-300pF vacuum cap
Johnson 226-1 variable inductor 0-22.5uH, 27 turns
Bank of transmitting mica caps 1600pF total, variable in 100pF steps

The attachment accurately shows the RF deck now.

I hope this is enough flexibility to use 2500-3500V, and whatever current up to maybe 500mA unmodulated, inlcuding settings like 3000V@200mA so the power can be kept low-ish for 'legal limit' operation. The OM used to run 1KW DC and sometimes more acording to his son.

These numbers are from a couple of tuning charts.
There are huge differences in the settings.

E/I=3500VDC/286mA, 12.2K Ohms
F=3900KC  
Q=12
Xc1=500, C=82pF
L=24.6uH
Xlc2=102, C2=400pF

E/I=3000V/500mA, 6K Ohms
F=3900KC
Q=12
Xc1=250, C=163pF
L=11.2uH (Xl=274 Ohms)
Xc2=42.3, 962pF

I hope to return to this project in a few weeks.

PJ

[Opcom]

more than a few weeks..

Found the grid current problem. The meter scale is 100mA. Its resistance is 14.3 Ohms. The movement is 7MA full scale. I have no idea what the deal is here or why it was that way. I'll solder a 1 Ohm NI resistor across it. In working this, I removed the bias chassis and found one of the three parallel 7K/10W bleeders open. They all look a bit cooked. Maybe a 2500 Ohm 50W is a better idea.

[Opcom]

The bias supply is now straightened out and has a pair of 1400 Ohm 40W chassis-mount (for heat) resistors in series, and in parallel with that a 11K 20W. The bleed is about 80mA, which I hope is enough for a 4-1000 in class C.

This only brought to light another problem.

Tried loading 'er up into the dummy load again on 20M and was only able to get about 50-60% efficiency as before. It did not matter if the input was 500W or 1200W.

The loading seemed best with 650pF. Loading is by switches and the total range is "0" to 1600pF with 100pF granularity and that should be OK as far as I know.  I have never liked these switched things but I know they work.

The usual tactic of changing the roller inductor one way and the tuning cap the other in order to change the Q didn't seem to help, it looks like the best setup was arrived at.

Increasing the plate voltage from 2250 to 3000 increased the plate current to as much as 450mA and made more power but did not notably increase the efficiency. So, lots of drive, but same efficiency issue. This said I can take it off resonance and peg the 500mA plate meter. Also with the lower voltages, the screen current was higher. I was running 500V on the screen, grid driver was anywhere from 10-50mA, screen current about 50-100mA give or take.

An odor came from the grid tuning compartment. There should not be anything in there to burn, only an MB-40 tank. When this happened, the grid drive slowly dropped to almost nothing, but it came back when given a minute to cool off ? There is nothing in there that should be heating up and a ricebox on low power should not be able to abuse an MB-40.

Something that I consider strange is that if I leave the 4-1000 filament, (and the whole monster itself) turned off, and set the exciter to about 5-10 watts and tune the tank coil, there is a place where it resonates, which is the way it should be.
But, the SWR as shown on a cross-needle Daiwa meter at the output of the exciter shows a 'good' 1:1 match at that point. How can a grid circuit with no tube load on it give a match? Its not a load, or should not be a load. There's no place for the few watts to go.

I'd be interested in any opinions of this weird behavior. What it means is I have to remove the RF deck again and try this with it upside down on the bench and see what gives. If magic smoke appears without the 4-1000 in the socket, or with it there but not heated, then what..

I know the transmitter worked before because it had a reputation in the 1960's, unfortunately I do not believe in seances so I can't ask Col. Tucker about it.

[W7TFO]

Is there coax internally in the input circuit, like from a jack to the input network?

It may become a reactive part if the impedance changes without operation of the tube.

Just a thought.  Or maybe put a pair of odor eaters in it and move on.

73DG

[Opcom]

There is a piece of RG-58 coax from the input SO-239 to the link terminals on the MB-40. It is probably 20" long. The shield is grounded at the connector, and the cable runs to the two pins on the MB-40. The center conductor to one and the shield to the other. Would it make sense to ground the coaxial cable to the chassis there as well? I have always thought not to add 'ground loops' I don't think the coupling loop would care much. It might not be strictly necessary to have a shielded cable but I did not want to have RF in the filament/screen compartment. Might be next weekend to get the RF deck back out and find out more.

[WD5JKO]

Tried loading 'er up into the dummy load again on 20M and was only able to get about 50-60% efficiency as before. It did not matter if the input was 500W or 1200W.

An odor came from the grid tuning compartment. There should not be anything in there to burn, only an MB-40 tank. When this happened, the grid drive slowly dropped to almost nothing, but it came back when given a minute to cool off ? There is nothing in there that should be heating up and a ricebox on low power should not be able to abuse an MB-40.

I wonder if the low efficiency might be because the "plate current" meter might be measuring cathode current?

It will be interesting to see what got hot in the grid department. My guess is the coil from high Q where the dissipation resistance is from I^2*R loss. I wonder how much power you were driving the 4-1 with?

Jim
WD5JKO