The AM Forum

Hi Guys, Just discovered this forum, and this is my first post.
First of all, I'm 70 years old and retired as an electronics tech.
All of my work was mostly solid state circuit boards.
I have had my Tech.class license for about 3 years now.
I rebuilt a Heathkit DX-60A and it works fine, however my code skills are
terrible...
My hobby is building stuff and restoring old tube radios. I have also built 3 QRP
CW transmitters.
Now to the point, I ran across the "JUNKER" linear amplifier, It is almost complete.
However I can't seem to get it to work with the DX-60. The 813 glows nice and bright
and is getting 9.75vac to the filaments. The separate
 plate transformer is delivering
about 2300vdc. I posted a picture...I think?
When tuning up into a dummy load, it shows no plate or grid current what so ever.
I have talked to some other guys on the ARF and they think my Transformer is to small current wise
and I agree. But wouldn't you think I would get something out?
I hope someone can help me out, by the way, I just started rewinding a new power transformer,
this one is about 450ma. weighs about 15 lbs.
I was wondering if the 813 tube is bad? and how can I test it?
Thanks for any help,
Gary
KB3WYZ

After reviewing your picture, your problem seems to be that you don't have any wires connected to anything.   ;)

If you're an ARRL member, you can review the article here:
http://p1k.arrl.org/pubs_archive/63846

Gary

Congrats on your upgrade from Tech.  What did you upgrade to?

http://www.qrz.com/db/kb3wyz (http://www.qrz.com/db/kb3wyz)

You may have posted the wrong pic.  No 813, meter etc.  We'll need more specifics such as a schematic of what you have, grid drive (ma).  Are you using cutoff bias with gridleak etc.  The more pics the better but that seems to be a very early pic. the 813 socket is still in its poly bag

Recheck and repost please

Again, congrats es gl, Al

Is this a trick question??

Welcome to the Forum.  Please don't start out by trying to confuse us.  Most of us here are already pretty much confused.

Seems you put up the wrong picture as I'm not seeing any 813 or as Pete said any wires.

Anyway, too small of a xfmr would not cause you to have no plate and no grid currents.  Something else is the cause.

Fred

An 813 can be tested in circuit, or by applying the various book voltages to it and checking a few points on the curves.
Care should be used with the HV of course. The idea is to cover sitations of high current, and situations of high voltage.
I test the odd high power tubes like that and it seems good.

Just take the values from the AB1 modulator data and also use the curves:
high current test examples: eB, Eg3, Eg2, eG1, plate-mA
2500, 0, 750, -95, 25
1000, 0, 300, 0, 150
1000, 0, 300, +100, 850 (very short pulse)  and Ig1=35mA

now, it might not make the 850mA if it is worn and old, so be the judge of it.

high voltage test example, series resistances advised to limit fault current:
eB, Eg3, Eg2, eG1, plate-mA
2500, 0, 400, -300 check if it is cut off. class C data
4000, 0, 400, and bias it to cutoff - see that it is possible. Peak modulated plate voltage withstanding test. A half wave rectified ac source can be used for this.

Tube should be gettered if it is old before doing tests with high voltage.

Can't seem to reply????

I'll Try again..
http://www.w7ekb.com/glowbugs/projects/JunkerAmp.pdf
Here is the schematic.
Gary
KB3WYZ

Here is some more Pic's

Sorry I had to use up so many reply's, when I posted everything I wanted
It kept coming up ERROR.
But anyhow, I was wondering if any body would test my 813 in one of there
working projects? I would pay shipping both ways...
I got quite a bit of $$ in this project, and would hate to part it out.
Has anyone here ever built this project?
Thank's, Gary
KB3WYZ

Gary, welcome to AMfone.

First suggestion: slow down.

A couple of random thoughts to start.
Check the tube manual to see if the 813 is rated for horizontal operation - many tubes are not and some need a specific
orientation of the grids.
I suspect that your loading & plate tuning caps are on the edge as far as the voltages and power expected. So, probably a good idea to leave some extra real estate for putting in physically bigger ones.

Patrick (opcom) posted the specs from the tube manual.
Do you see PLATE current when the amp is on and keyed, no signal??
Is the Plate Meter connected, connected properly?
Do you have HV on the plate?
You can see the filaments lit.

If you do not see quiescent plate current, find out why first.
You can measure the plate current indirectly via a small resistor in series with the cathode.
Ohms law will let you convert the measured voltage across a cathode resistor to amperes.

Before you send the tube anywhere (which could break it) make 100% certain that you have wired everything
properly.

MAKE 200% CERTAIN THAT THERE IS NO HV PRESENT ANYWHERE AND THAT THE PS IS UNPLUGGED WHEN WORKING INSIDE THE CHASSIS ONCE IT HAS EVER BEEN PLUGGED IN. PERIOD.

Just looking at your pictures was wondering if you are grounding the center tap of the filament trany in transmit??An yes the 813 will fine laying down as long as the plate structure is in a vertical plane as you have it......
Your layout of parts may need to change some.....And pin 3 needs to have a shorter ground ....I would ground it right at the mounting screw closes to the pin .....

Looking at the schematic (had to open it with Reader 9.0 to see the schematic, tnx Adobe!) you can put a 1 or 10 ohm resistor between the CT of the filament xfmr and ground and measure the vdrop across it to determine plate current.

I assume you can directly measure with a HV probe the B+?
In which case you should monitor the B+ no load, and key down, no input signal (no RF). It ought not drop much at all
with quiescent idle current. That would be the first set of tests.

Double check ur wiring, make sure you have the inputs and outputs connected?

Btw, what is the little cube relay doing??

                 _-_-

PS, if ur rewinding the power iron, suggest you change it so that the secondary can use a simple full wave, not a doubler.
Much better.

Hi Gary,

What they said.

To measure plate current, you can also break the filament transformer CT and place a 500 ma meter directly in series with this lead to see plate current.  Later, once you determine what plate voltage you will be using, you can add a string of diodes or a zener in series with the fil CT to get the proper idling current.

Under full load, you may have problems getting a high enough plate voltage and good regulation with the voltage doubler method using a small "TV-type" transformer. Though, one 813 in linear service is a pretty mild load.  

An 813 in linear service can go up to about 3500 volts. I know some guys who have run 4KV. But  at 2000 volts, you may be running out of voltage headroom before the tube puts out all the power it should.   This is not to be confused with the lower ~2KV voltages that are normally run in class C plate modulated service.  I would look around for a more robust plate transformer that can do  1800V-2200V using capacitor input into a bridge diode at 250-400 ma. This will give you what you need for peak performance.

Why do I worry so much about HV plate regulation? It's because poor regulation will produce poor IMD and splatter. Right now in GG, the grid and screens are tied to ground, so they are rock solid. The only thing susceptible to modulation current sag is the plate supply. Get that covered and you have a solid, clean amplifier.  

Be sure to load the stage heavily, thus, peak C1 and C2 for max power out and then unmesh C2 a little to produce about 7% less power out. This is the most linear way to run the amplifier.


This may then require you to replace the plate tuning cap with a higher voltage rated unit, but the rest of the circuit may be fine.  A big Variac on the high voltage supply is a good idea to use for testing and later finding the amplifier maximum limits.  Do you have a big Variac?

Also, replace the red HV thin wire lead that runs from the 813 plate to the pi-network plate coupling cap. This is an RF path and needs to be of copper strap or at least copper tubing.  Grounded grid is a pretty stable config and will probably be OK as-is below 40M, but might as well do it right.

If you find the amplifier has instability problems, especially on the higher bands, consider shielding the filament and 813 input circuitry from the  plate circuitry. Right now they can see each other, though in GG you may be OK.

As the guys have suggested, the first thang is to get plate idling current and go from there. That is an interesting metering circuit the way Lew is able to monitor plate current, voltage and grid current all on one meter. Hopefully the meter is still working and it is wired correctly.

There's a few more suggestions, but get it working first and we'll chime in later on.  Be super careful with the HV, OM.

Tom, K1JJ

Thanks for your support and Ideas.. First of all, the filament transformer that I
rewound don't have a CT...Could this be the problem?
I would like to stick to the original plan as much as possible.
I have already started to rewind a bigger Transformer, the primary is done, and
the wire for the secondarys is on order.
Here is how I have been trying to make it work. I tune up the DX-60 on 40m to 7.120
and get the best SWR on the tuner going to the Ant. The output of the exciter, DX-60
to imput of the amp, the amp output to input of the ant. tuner.
The cube you see is a relay which bypasses the amp, and when in the transmit mode
it switches it through the amplifier. Going into a dummy load, with key down I get no output
at all through the amp. When I bypass the amp, the output is just fine.
I don't like testing like this because it's going to take out the final on the DX-60.
I get nothing for plate or grid current, I do get a constant 2300vdc on the plate of the 813, key up or down.
I do not have a HV probe.
I hope you guys can read the project write up..I sent the link to.
Thanks for your kind help and patience.
Gary
KB3WYZ

Gary the tube will not work unless the cathode is at ground which is the fil on an 813......No center tap then try a 2.5 mh choke from one side of the fil to ground an you should start drawing some current then...

 The filament transformer in the image is a Signal 241-8-24. This is a 24vct @ 4Amp transformer. Apparently with a split bobbin design, the coupling primary to secondary allows the secondary voltage to droop to 9.75V. Does that transformer get hot?

The correct transformer in the family is a Signal 241-7-10 which is 10vct at 5 amps.

Either way your transformer has a CT. Put a 50K 2W CT to ground for standby bias, and then short CT to ground to operate. If you key the B+, then just hard ground the CT all the time.

Jim
Wd5JKO

Jim, Your right about the transformer. But I removed the 24V secondary
and rewound it to put out 10vac which under load comes out to 9.75vac.
No the transformer don't get hot.
Sure wished I would have wound it with a CT.
I think I'll try the choke idea running one side of the
Filament to ground.
Thanks, Gary
KB3WYZ

Gary,

  You can make an artificial CT with two resistors. For example, two 10 ohm resistors could be used (will dump 2.5 watts each, so use 5-10w resistors). One from each side to ground. The cathode return though will see 5 ohms to ground. At 200ma cathode current, that will add 1 volt of cathode bias. If this bias fluctuates with excitation, the amplifier IMD goes up. Dropping to 4.7 ohms each side might be better except that each one would dissipate 5 watts heat from the filament voltage.

   My suggestion would be to add the CT since your an experienced transformer rewinder.

Jim
Wd5JKO

also, pull the old incorrect label from the transformer and mark it appropriately.

It might be possible to bypass the resistors with caps that function above the 60hz freq, so that the modulation
is essentially bypassed (it looks like ground at AC).

Easiest to do this by converting the filaments to regulated DC first...

I'm going to follow this with interest since I'm about to start a build using a single 813 from the ground up though my project won't be for a linear amp but for an AM/CW rig.

I think your tune capacitor might do some arc welding.....but maybe i'm wrong.....be careful

Just for test purposes you can ground one side of your filament xfmr to see if the amp and the 813 are working.  813s are a directly heated cathode tube meaning the filament is also the cathode.  You need to ground the cathode to have the tube do anything.

You can ground through a small resistor (10 ohms) to measure the plate current by way of the voltage drop across the resistor (as others have mentioned).

Fred

Hey Gary, this is half the fun! :D

as far as I know, everyone has had to go through a similar "learning curve". And the truth of the matter is that with a very
few exceptions all of the true masters of these things are SK now. :(  We all follow behind them, in their shadows...

                                    _-_-bear

I find the prospect of messing around with 2KV for the RF and 1500 volts or so for my modulator REALLY SCARY!   I've bench-tested the power supplies and no problem, but the application is another story.  I've built a few transmitters and a lot of audio gear in my time but if nothing else, I'm building this 813 amp at a slow crawl.  Slow, SCARY crawl.  At least the guys here are patient and willing to help.

Of course there are a few slightly "off center" folks here, but they have to be because we believe in Ancient Modulation!  Oh, one more thing, when you use a linear for AM, you run into a whole 'nether world of efficiency problems provided you can get the beast working properly.

You might want to check out "Don's Tech Talks" here and the article from 2/18/2011 dealing with linears VS plate modulation.  Good luck!

Respect it, but don't be scared of it.

Another way you can get a center tap is by connecting another filament transformers secondary
across the 813 filament transformer that has a center tap. Anything above 10 volts will work. That will give you an additional voltage source from the primary of the new transformer to provide bias or some type of control voltage for a future purpose. Looks like a GG circuit and seems to be built nicely. If your going to use it on AM, the expected output will be about one third of the plate dissipation plus the drive power feed through.

No harm in running a lower voltage, especially for checking the amp out. Later if needed you can put in a bigger tune cap. The loading cap should be enough voltage rating for a single 813 at full voltage if attached to a decent load.

The drive especially G1 current will be quite higher when the tube reaches max. power at a low plate voltage, and less drive is needed at a high plate voltage. So watch you don't burn the control grid.

For a simple HV probe that can be made in 30 minutes, figure out the input resistance of your meter. Might be 1 meg ohm on all ranges, as some digital meters are, or like an analog meter it might not be one value, but differing depending on the selected range.

Analog meters are usually rated in Ohms per Volt, that is, a 20,000 ohms per volt meter: on the 500V range it would be 10 Megs, on the 200V range it would be 4,000,000 Ohms, and on the 50V range it would be 1 Megohm, and so forth.

You can make a X10 probe for it that will be safe to use with care.
Multiply the input resistance on the selected range by 9.
Make the HV resistor string: Put 10 resistors in series to achieve this value.
Be conservative with wattage. Don't overlook that 1KV on a 1Meg resistance pulls 1mA current, and is 1 Watt dissipated.
Put the resistor string inside a PVC or glass tube for insulation. One end is the probe tip, the other end accepts the meter's probe.

So for a 1 Megohm input voltmeter on the 200V scale, putting a 9 Meg resistance in series with it makes it a 2000V meter.
And for a 20,000 Ohms per volt meter on the 500V scale, putting a 90 Meg resistor in series makes it a 5000V meter.

At this point many people may prefer to solder an alligator clip to the high voltage end of the HV resistor/probe and clip it to the 'B+ test point', and avoid handling the resistor while it is live. That's a good idea, and best not to rely on cheap PVC or unknown glass tubing for keeping such high voltage off your hand. Also note be sure the meter's GND or negative lead is very definitely grounded, otherwise the meter will float at the high voltage.

A similar design is in the picture attached. It is a 20:1 probe for use with a 1 meg input resistance meter or audio scope for checking modulated B+.
It's wired different, so disregard that, but just to show the resistors in a tube idea.

It probably would be advisable if future discussions on "breadboarding HV supplies" be continued in a separate thread. Let's keep this thread focused on the 813 Lew McCoy Linear Amplifier.

A "breadboarding HV supplies" thread can definitely stand on its own with probably very useful information especially on safety and design.

A new thread has been started: http://amfone.net/Amforum/index.php?topic=38298.new#new
W9ZSL and K1JJ threads have been moved.

Thanks Guys for all your great Ideas, I think I will rewind the Filament transformer
and I certainly will put a CT on this time!
I do have a winding machine, Homebrew of course, but it works.
This transformer isn't to hard to rewind, because it has two different bobbins
one ,the primary the other secondary. I have a good supply of #18,16, and 14 enameled
wire. So give me a few days and I'll see if I can get this puppy to come alive.
Yes, I am careful about the HV, I can see the caps bleeding off when I shut the rig down.
No I won't be using this amp for phone, just CW.
Thanks again, I'll be getting back soon.
Gary
KB3WYZ

Ok guys, Here is where I stand on the 813 Linear Amp. First off I put a 2.5uh choke
from one side of the 10v fil supply. That worked...At least I am now getting FR to the
Ant. tuner. However, with key down the plate voltage drops from 2300vc down to
about 900vdc. The set is now showing some plate and grid current. I have been told by others that
my plate transformer is way to small wattage wise. Before I hooked the choke to the filament
transformer I got jack squat nothing. At least I think this project may still have a chance.
On CW I am getting about 40W out put with the amp maxed out. Just the DX-60, with out
the amp will do about 75 W out. Oh yes, I burnt out the motor for my transformer winding
machine  :'( Guess I'll have to use a vari-speed drill. I have some iron for a bigger tranny
about 14 lbs. Now to get it rewound, any ideas are welcomed.
Gary/KB3WYZ

Gary,
What is the plate current while the HV supply drops to 900V?   When your bringing things up, its a good idea to have the proper metering in place so you can see some of the key parameters.  If the drop is being caused by the very small plate transformer, the transformer should be getting very hot in a short period of time.  Once you know the current you have to supply from the power supply you should then be able to size the transformer correctly based on the rectifier topology you pick.  I assume this is one of your first projects and no better way to learn than by doing!  Keep up the good work.

Joe-W3GMS     

A good source for a filament transformer base unit is a microwave oven transformer. 
Remove all the wire on the secondary or hv side, leaving the primary or 120 volt winding intact then rewind the secondary with about #14 or #12 neopreme covered wire to get the desired voltage for the filaments in use.  I believe the the turns ratio for most microwave transformer is one turn per
volt on the secondary, that being the case, only about 10-12 turns should do the trick for 10 volts out and don't forget to put in a center tap.

I have one made this way running dual 813's with no problem and very little heating.

No correctly designed power supply is going to drop 60% of its no load voltage with a moderate load of a 813.  You need to do a serious review of your power supply.

Fred

Do have ratings for the B+ HV xfmr??

Pix of the PS?

 "You can make an artificial CT with two resistors." A resitor Y 10 ohms each leg.
This works very well BTW. But use carbons which have no L factor. I use a DX60b on a HB amp and ran into a problem at first using wirewounds. (would occasionally mysteriously go open)Guessed it was because they had L and I didnt want that so made them up from paralleled carbons instead. End of problem completely. Mine is a twin 811A but the problem was exactlly as you state when there was a problem in that circuit. Voltage present and correct but no current and no performance. If you are just using the DX60B you dont need the relay since you aren't listening on it.
I also use that trick in my directly heated 1929 one tube transmitters (45 , 01A , 171A)
don

Hi guys, here is the DC power supply.

The Plate transformer is about 100ma, the AC secondary is delivering 862 vac to
the rectifier board, I would estimate about 2350VDC no load.
I am going to start rewinding the Fil. tranny today CT this time!!
More later,
Gary

Yeah well, multiply 862v x 0.100, tell me what you get?

That's the number of watts you have available.

You need MORE watts than the expected power output.
Check the Bill Orr Handbook for better info on how to design and rate HV power supplies.

Best to get one at a hamfest, online, or used via Amazon. They were published for each year,
so you want one from some time in the late 60s I expect. You'd need to read up more, before
proceeding any further.

The power supply transformer ought to weigh at least 10lbs, or more. More is better.
Rule of thumb here.

                        _-_-bear

airc the Tron sez 10 Watts per pound for usual EI core transformer .... somewhat more for toroids or hypersil cores

How about this...About 13 lbs.
This is from a 1960's Zenith TV.
Gary

What is the the part number on the Zenith xfmr??  Should start with 92-.  I can cross reference the piece to a Stancor or Thordarson, from that we can find out more about what current the xfmr was good for.

Fred

Someone had a BC-610 HV/plate transformer in the For Sale section. Get it and quit messing around. TANSTAAFL applies. There ain't no such thing as a free lunch. Get a real power transformer and your problems will go away. This may seem harsh, but the power supply of an amp is like the foundation of your house. If you don't get that right, the rest doesn't matter.

I wasn't going to say anything, but I'm starting to feel guilty to see a project that is surely doomed for failure unless there are some changes.

Yes, Bear pretty well summed it up.  A voltage doubler, tripler or quadrupler  multiplies the voltage, but it also multiplies the poor regulation sag proportionally.   It's a cool ham thing to want to use what ya got, but using a small power-rated transformer AND using a multiplier too is like building your own tubes. Why bother when the results will suck?

In my opinion, a linear amplifier with good IMD figures (cleanliness) is the most important thang. A 2KV power supply that sagged even to 1600 volts under modulation (20%) will have your ham neighbors in a tizzy about splatter.

As Steve suggested, look for a husky transformer in the ham ads. Get something that weighs at least 40 pounds and preferably bigger..  There are a lot of older hams with cellars filled with iron that they will never use and don't care to move anymore. Put a few "wanted" ads out there and you will be flooded with offers.

You won't believe how solid a big transformer using  a choke input and 30 uf or more filter capacity can be. You might see only 50-100 volts of sag -  which is needed to be super clean.  Much will depend upon where you idle the amplifier vs: peak power -  and your syllabic rate on ssb or AM.

Good luck.

T

Thanks Tom, glad you told him and not me.  Voltage doublers have no place in ham xmtrs.

You're right about hams having basements full of iron they'll never use.  I have some plate xfmrs that I've had for for over 50 years, still waiting to use them in something.

Fred

Another way to make a suitable filament transformer is to use an easy to get 12V/6A Toroidal transformer.

I measured the voltage per winding, and calculated the amount of windings for 7.5V and 5V (For Taylor T40 and 100TH tube  use)

Used a knife to scratch of the laquer and soldered the leads. In the center I soldered a CT for both voltages.

I'll attache a picture to give an idea....

Martin, PA4WM

Maybe we should have a HV iron section on this forum.where we could trade plate XFMRs.  I must have a ton of iron (I think I'm just kidding).

 ;D

Fred, the numbers on this Transformer is...

95-1311
138236
In was a 13 pounder...
Guess that's still pretty small.
This tranny, the secondary was open.
I have never seen wound like this...Looks like the primary had a CT on it???
Thanks guys, about ready to cut bait.
Gary

I have a lot of transformers that would fit that sized amplifier, Dahl and Chicago/Stancor mostly. Let me know what you really want and I can look, and give you an option that way.

The series capacitors look to be different makes and sizes, maybe same rating. I advise against mixing electrolytic brands and ages in the same series circuit, as the leakage current of one may be far different than another. This would result in dangerous differences in transient, ripple and DC voltages across those caps. In essence, one might be exceeding its 450 volt rating during some portion of the waveform or turn on step. You have to put a stiff low enough resistor across each one to try and balance that, it ain't worth it in power loss to do that.

Zenith 95-1311 crosses to a Thordarson 26R59 and Stancor P-8353.  P-8353 is rated 265-0-265 vac at 300ma, 5vac at 6amp, 12.6vac at 6amp.  Primary 117vac 60Hz.  The xfmr is rated for cap input filter (300ma).  With a choke input filter the current can be increased 30%.  The xfmr has two 5U4 sockets built into the top plate.  Wt 12.5Lbs.

The 138236 is a manufacturer and date code.  138 indicates the xfmr was made by Chicago Xfmr Co which was bought by Stancor in the 1960s.  The 236 indicates the piece was made in the 36th week of 1962 or maybe 1952 (IIRC).  The date code didn't indicate which decade, you had to guess that. (The two 5U4 sockets in the top plate was more of a 1950s thing).  Once burnt my forearm on one of those top mounted 5U4s reaching around it to pull some other tube out.

You can add up all the VoltAmps the xfmr was designed for and put all of it into the HV winding and most old big TV xfmrs can supply more than that (easily 50% more) without getting too hot.

IMO the xfmr is still too small for any type of 813 xmtr.  You need a bigger xfmr.

One last thing,  the primary unlikely had a CT or any tap,  very often the HV winding is first on the core and the primary is over it with the filament windings last.

Fred

Thanks to everyone for your kind suggestions, The McCoy "JUNKER" Linear Amplifier
is now HISTORY...I started to part it out. I wanted to keep it all onboard like described
in the article. I have used voltage multipliers before very successfully in CW homebrew
transmitters. I guess it's just a matter of choice. I have been having health issues and my
eyes are failing me. I love to wind coils, but just can't see all that well to do a neat job.
I see another gentleman on this forum is going to build this McCoy amplifier. Sure do wish him
the best. Looks like he has the right Transformer. I'll be watching this forum for his progress.
Best wishes to all!!
Gary

Gary,
   I am sorry to hear that you are opting out of this project but I certainly understand. I was looking forward to being partners in crime on this project. 😊

Rich

I too regret that you won't pursue it. I always enjoy seeing what others build and rebuild and the 813 is one of those good ol' big glass tubes that there seems plenty of and can take a little occasional abuse, a long-lived tube.

Correctly rated transformers/parts are easily/readily and cheaply available. Do you have a local ham mentor/friend in your area who can help you? Make the effort, have a backbone, be patient, if its all really worth it to you..maybe it isn't?...is it all about instant gratification? What a weird outcome with this.
C'est la vie.

Al VE3AJM

I sincerely hope you feel better some time and might consider building it, or something like that later. I understand about health, sometimes I don't feel well and not like working on stuff. I know what you mean about eyesight as well. I have to use glasses for this and take them off for that, it is very annoying. No pressure here 8-)

One last piece of encouragement I can offer, is that you can use one plate supply for more than one amplifier, and this can be done really cheap.

My friend uses one plate supply+modulator hooked to two amps, and he only lights up the filaments on the amp on the band he is going to use.

That way the amp chassis has plenty of room, no crowding. Plate supply is in the bottom of the rack. Could be under a desk just the same  but he has all racks.

One of the cheapest $ per watt transformers around is the old BC-610 plate transformer (Stancor P-9920). They are old and heavy and sometimes people will give them away for the shipping. You can choose 2500V@350mA or 2000V@500mA on that one with the flip of a primary voltage switch. A 10-20H choke an 6-12uF oil cap and you are good on that. With a capacitor input, 3500 or 3000V is possible at lower current.

I have gotten several big transformers for free or cheap because they are the old cast-iron-bell styles. Does not bother me and I have not had one blow yet even if they are 50 years old! I get a young helper to assist mounting them.