Utah UAT-4 PP T-55 RF Amp

[wa2fxm]

I picked up this UAT-4 at NEARFest this spring. Finding one had become quite an obsession with me and I literally almost stumbled on this one walking back to my truck. So here's what it looked like: cracked plate transformer wires, 2 beautiful glass-etched Taylor 866's with loose caps and base, rusted and locked up power switches and a way out of spec bleeder resistor. I replaced the cracked wiring and power cord, freed up the power switches and subbed a 36K bleeder. All the transformers showed reasonable resistances and both filter caps measured OK. And in the interests of historical accuracy I fired it up on the Variac the first time with those filter caps in place. One of them shorted as I got up to around 900V. And it took out one of the 3B28's I had put in there.

So I beat a hasty retreat and ordered up some perfboard and a handful of electrolytics and balancing resistors. I built a HV board to bypass the original Aerovox 2uF 2000VDC caps with 4 sets of paralleled 10uF 600VDC electrolytics with 470K balancing resistors across each set, all in series. I shoehorned it in between the plate transformer and original caps.

Never having dealt with a 2kV power supply, out in the open, on the bench here I can't tell you how hard it was to crank the the Variac up and watch the Fluke slowly increase as the power supply chassis started to hum. It took about half a dozen tries increasing a little more each time to finally get it up to 1800V. No loud noise and no smoke so I guess I'm on my way with this thing.

As always, all comments, suggestions, hand-slaps, cautions and finger-wagging greatly appreciated.

Mark - WA2FXM

[W7TFO]

I have T-55's if you need a couple.

73DG

[KD1SH]

  Mark, you are a braver man than I, for sure. I get a tad nervous even powering up a newly built home-brew supply—all components fresh from Digi-Key— with my variac.
  Sometimes I'll bring home an old decrepit looking supply or some such; look it over, and then very briefly contemplate putting it on the variac, just out of curiosity. In the end, my nerve fails me, and I decide that I'd rather go outside and juggle running chainsaws (never juggle running chainsaws indoors; carbon monoxide is dangerous) or rub my head with bacon grease and shove it down a badger hole.

[wa2fxm]

Thanks Dennis, I'm not sure how to test them yet. I will keep your offer in mind. And Bill, maybe I put too much trust in the 10A breaker in the Variac. Anyway standalone it seems to be ok. I'll stick an 8A fuse in the AC line as soon as I can find a snap-in holder. Here's what I've traced out for the power supply. First thing I wonder is how does the 1500V get up to the RF deck? The 5-pin plug carries switched AC to the RF deck for the T-55 filament transformer and then brings the 7.5V back through the 100 ohm wirewound resistor to light the 6.3V lamp on the power supply. Pin 3 is unused. I was thinking maybe a missing heavy duty spade lug on a filter capacitor with HV wire going up to the far-right ceramic standoff on the deck which is the HV input? Here's a blurry picture of the entire Utah suite hooked up together in the October 1937 issue of Short Wave & Television, page 297. It's not a great picture but I don't see anything of the kind like that.

Mark - WA2FXM

[wa2fxm]

Here's a first attempt at figuring out exactly how this UAT-4 is put together. It looks like it follows the standard push-pull circuits featured in most of the late 30's literature. In particular a 1936 ARRL Handbook 2kv push-pull amplifier which features "tap-on" link coupling which I didn't realize was a thing. I assumed all link coupling featured an inductive loop at each end but obviously here, at least on the output coil the link to the antenna tuner unit comes from taps on the output coil. The coil measures 12.5uH and with the unusually large 425pF tuning condenser the grid dip meter shows resonance on both 80 and 40 meters. It's unknown exactly what is happening on the input end because (1) there is no plug-in grid coil in the 5-pin grid coil connector (2) and there is no wiring between the ceramic rear panel input connectors and the 5-pin grid coil connector. The National connector has 3 tabs soldered into the T-55 grid circuit and 2 blank tabs which I assume would be for the input wires. There is also some serious black charing on the grid meter end of the bias wire from the grid leak resistor with residue visible on the back of the meter. The bare wire 7.5V filament CT and grid leak bias wires obviously go to ground and they both just reach to the rear chassis ground connector.

The only other strangeness besides wondering how the HV gets to the RF deck from the power supply is the matter of the "outboarded" Stancor filament transformer. These Utah boxes were marketed as build-it-yourself "Add-A-Unit" kits so it's possible that the kid who built this one miscalculated his spacing needs and wound up hanging the transformer out in the air with only 2 mounting screws on the chassis. But I really don't see how any rearranging of components would allow for the transformer to fit properly on the chassis. There are no extra unused mounting holes so it's not likely this is a too-big replacement transformer that had to be shoe-horned in.

So I'm searching for a 5-pin 80 or 40 meter grid coil with center link coupling. Or at least an appropriate 5-pin plug-in mount to roll my own. Any thoughts on the HV connection, the charred bias wire and the hanging transformer?

Mark -WA2FXM

[wa2fxm]

It looks like 3-1/2 inch, 5 pin plug-in grid coils are hard to come by. I gathered a bunch of coil pieces and parts, cut and drilled a strip of Bakelite and made a test grid plug-in with a scrap piece of AirDux. And I think I've figured out where the outboarded filament transformer comes from. Originally on the power supply chassis there was most likely a dual 2.5V/7.5V filament transformer. A Utah catalog shows their 2298 transformer with dual 7.5V 6.5A CT and 2.5V 10A CT windings. When that blew up and they couldn't find another one the kid mounted a separate 7.5V transformer on the RF deck. And they used the grid input wire grommet holes to wire it up. So if I'm going to get the grid input wiring back to normal I've got to move the transformer. I don't think it's likely I'm going to find an original replacement dual filament transformer. I'm thinking of engineering some kind of bracket on the power supply chassis to stack both transformers mounted horizontally? I've got 7 inches of height to work with.

Mark - WA2FXM

[wa2fxm]

I've made a little progress with the UAT-4. I pulled the 7.5V transformer from the RF deck. On the power supply I removed one of the old Aerovox filter capacitors, drilled two holes and mounted the transformer in its place. Underneath I added an AC line fuse, some terminal strips and I tried to clean up the wiring a bit. I'm not looking forward to sliding all that high voltage crammed into 7/8" of space onto the metal cabinet. No HV yet, but I ran the filament lines to the 5-pin plug and with the two chassis connected the T-55's lit up. I wired the RF input terminals to the grid coil connector tabs and with the Siglent RF generator I was able to peak the homemade grid coil across the 40 meter band looking at the signal leaking across the T-55's to the output terminals. The plate tuning condenser also varied the signal but maxed out with full plate meshing. So I'm not really sure what I'm looking at here because it's not clear to me how the UAT-4 tunes across all bands. I get the plug-in grid coils for each band, but how does the fixed plate tank coil get tuned without any taps? The two clips are on the output which I assume is for loading the antenna? Most of the other P-P amplifier examples I've seen show plate coils which are either plug-in, or have switched taps. There is one example in the 1936 ARRL Handbook which describes shorting clips on the input and output coils for tuning 3 bands. Can anyone clarify how the UAT-4 tunes from 160 to 10 with the fixed plate tank coil?

Mark - WA2FXM

[wa2fxm]

More progress. I think. Without any documentation, and zero real live experience with 1937 amplifier technology I'm moving very slowly. Lots of trial and error. I put the plate transformer on a separate variac because 1500 volts still scares the h**l out of me. And I put together a fixed bias supply because I don't see any way the amplifier was meant to be keyed. With this lashup and a few hundred volts on the plates it squeezed out a few watts from a UAT-1 40 meter crystal input. So I guess the tubes aren't dead. Lots of tuning instability but once it's set the note sounded decent. I guess that's progress? I'm going to go back and recheck all the under chassis AC and HV connections and then mount them in the cabinet and move on from there.

Mark -WA2FXM

[W4AMV]

Hi Mark.

Fun project. I would calculate the plate to plate load Z required for class C. Then terminate the plate to plate connection in that R termination. Then if you use a SWR bridge or the small vna, check the match to the desired balanced termination. If that were 300 ohms or 450 ohms, then you will need to make a simple 300 or 450 ohm to 50 ohm balun. That’s required since the bridge or vna is unbalanced. I suspect, the tapping of the coil instead of link coupled achieves the Z transformation as an autotransformer. That is the ratio of the 2-tap locations to the total turns. Validate that. This would be fun to put into spice and check it out. It would take 4 or 5 perfect mutual coupled inductors and run it in say LTspice. I did the same for link coupled and went through the math to convince myself how it worked.

[KD6VXI]

Put a 60 to 100 watt incandescent in series with the variac.

Now you won't have catastrophic faults when powering up the first time.  Even if a wiring error makes a dead short the most power that can be drawn is limited by the light bulbs wattage rating.

--Shane
WP2ASS / ex KD6VXI

[Jim, W5JO]

Put a 60 to 100 watt incandescent in series with the variac.
--Shane
WP2ASS / ex KD6VXI

Those bulbs are getting hard to find.  I have to order them, there are none locally.

[wa2fxm]

you will need to make a simple 300 or 450 ohm to 50 ohm balun. That’s required since the bridge or vna is unbalanced. I suspect, the tapping of the coil instead of link coupled achieves the Z transformation as an autotransformer.

I have a 4:1 balun on the output feeding a 50 ohm Waters dummyload/wattmeter. The tapping of the plate coil is indeed for Z transformation. I discovered that the 2 clips were unbalanced. The left one was forward 1 winding from center tap and the right one was clipped right on the center tap. I moved the right one back one winding and immediately got a bump in power output. And where I was previously getting muddled neutralization results I now had definite, clean nulls from both capacitors. Initially with 900 volts on the Variac at keydown I was getting 80mA @ 740 volts and 42 watts output. That's 71% efficiency. But there was lots of interaction between the UAT-1 output and the UAT-4 grid control. So I went back and rechecked the UAT-1 and saw that the plate coil was wound for balanced output with a center tapped B+ and the link windings also in the center. The UAT-1 final is a single ended pair of parallel 6L6's. The only piece of original Utah documentation I have is the coil winding specs for the UAT-1 and they show the link wound about 1/4 inch below the B+ tap which is on the 3rd winding from the bottom of the tank coil. I rustled up a new fairly sloppy coil with the link at the bottom, and direct out of the exciter was only a few watts. I then jiggered the 3 link windings up over the B+ tap and that bumped the output to a more normal 20 watts. When I put it back on the amplifier again I was getting 85mA @ 620 volts with 43 watts out. A jump to 82% efficiency. It would appear the exciter output coil construction is critical. Moving it in and out of the cabinet a few times the windings move slightly and that seems to effect the amplifier output. I think I will spend some time going back over the UAT-1 and take a look at the 3 homebrew coils that came with it.

Mark - WA2FXM

[KD6VXI]

Put a 60 to 100 watt incandescent in series with the variac.
--Shane
WP2ASS / ex KD6VXI

Those bulbs are getting hard to find.  I have to order them, there are none locally.

Agreed.  But, worth having a couple different wattage in hand.

If used with a variac they last forever too.  No Inrush.

Even here on island they are pretty much nonexistent, save for the electric supply house.  And when those are gone the light rep said their will be no more.  Have to order from China.

--Shane
WP2ASS / ex KD6VXI

[W4AMV]

Based on your good efficiency numbers about 15 k plate to plate Z. You can get there with either link coupling or tapped. I assume your efficiency values do not include the balun loss. Achieving this plate to plate Z takes care as the Q of the tank needs to be decent and tuning might be touchy. The link math is neat and straight forward if you desire to switch over to link. Either method should be validated on the bench before applying power. I would check on the exciter grid drive to the push pull PA and make sure that the grid I or the peak to peak excitation voltage is in line with the 6L6 requirements running PP.

[KC2ZFA]

the uat1 appears to have a single ended/balanced tank PA…the bottom of the tank (the end of the plate coil not connected to the tube plates) is usually employed in neutralization of the PA (info on this in the pre-70s manuals). The PA coils then must have center linking. If the center link is fixed you have no loading (hence power out) control. It is recommended that such center-linked coils have a variable link…look up the JVL series of B&W coils.

In any case, at the amplifier level you need a way to set proper grid drive. So either a variable link at the exciter output or a variable link at the push pull amplifier grid tank. Doing the old in and out with the variable link sets the grid mA on the final amplifier. If that amplifier is properly neutralized then varying its plate tune cap through its range, with the plate HV off, should produce no variation in its grid current.

Peter

[Opcom]

I'm lucky to have 1000bulbs.com close by in Mesquite, TX. They have a good incandescent selection including 130V, rough service, and heat lamp. Also Christmas lights which are all 130V now it seems.

[K8DI]

The link math is neat and straight forward if you desire to switch over to link. 

Can you point me to this math? If you have a reference for PP RF amps in general (not just link-tuned) I'd love to have it.

Thanks,
Ed

[W4AMV]

Hi Ed.

I wrote a paper for QEX in Jan/Feb 2022. See attached reference title. I also copied the references I used. There is a spreadsheet, download from QEX,  that does the calculations for a variety of link topologies and permits a series tuned secondary link. That link can be tuned with a series C or L and provides a range of Z terminations reflected to the primary. As time permits, I will try to assemble what I have on PP.  

[W4AMV]

I am sorry and here are the references. Apologize as I do not want to clutter Marks thread.

[wa2fxm]

After playing around with the UAT-1 output coils it seems that placement of the link really does not make that much of a difference so I'm back to wrestling with the UAT-4. And I've hit a roadblock. Attempting to put the chassis' inside the cabinet the amplifier goes into oscillation with the RF deck pushed in within an inch and a half from the front panel. Move it back and let the end hang off the top shelf and everything works fine. Push it in and plate current takes off with the slightest HV application.

Thinking that the cabinet was messing with the neutralization capacitance I drilled an access hole up top for the front NC and with the chassis all the way in re-neutralized everything, but it made no difference.

When the chassis is hanging off the back end there are two items that are outside the cabinet, the thick black HV line from the HV connector to the 'Aux' plug, and the grid input twisted link (which I added) from the grid input connectors over to the grid coil. Maybe the cabinet was effecting these somehow? I rerouted the grid link wires away from the HV line, but it made no difference,

Reading the ancient 1930's manuals one suggestion for eliminating parasitics was adding a small RFC in each grid lead, which I tried to no effect. All the other suggestions include shortening wires, straightening wires, and symmetrical component placement. It would be easier for me to build a wooden cabinet. And I don't really want to be messing with Utah's original design.

The runaway plate current occurs without any input from the UAT-1. With a little fixed bias on the grids and the chassis pulled back in non-oscillation mode I can see plate current slowly rise as I increase HV, as expected. Push the chassis into the cabinet, and try it again, plate current surges as you increase HV. Any suggestions for moving forward here?

Mark - WA2FXM

[W7TFO]

eBay is a good source for incandescent bulbs...

73DG

[KB2WIG]

A bit expensive,


https://www.amazon.com/Mandala-Crafts-Rough-Service-60-Watt/dp/B098886NYZ/ref=mp_s_a_1_2_sspa?crid=CB0SDF7N78RT&dib=eyJ2IjoiMSJ9.A23qnC_pP5LkhWrlh-GE5G2TmjIfSIG22qbPIqPUVUbeU8JY2Qrqm5K19Sos-PLksw3JWuDGzurp7r5DA2GENoiFQ6AhkQ9FoDGSE7xdzobIJoazQ_ELMFN479AcwGgVTz3laNGxtWrEiHgkiFSCh_fo-EUQ8hsGrhVqNNfAO-v7IDhdqfJMAQHAOLrzXaxT2NS-zq2TvHdUKwNBEWtydA.EoVQNEeDTLGY-A60QvSRJqgeXzbsAbwRzCRixWvZGu4&dib_tag=se&keywords=100watt+incandescent+light+bulbs&qid=1708785127&sprefix=100watt+inc%2Caps%2C170&sr=8-2-spons&sp_csd=d2lkZ2V0TmFtZT1zcF9waG9uZV9zZWFyY2hfYXRm&psc=1


Lots 'o bulbs, get 'em while they're hot.

KLC