Building an adjustable high voltage bench power supply

[kward]

I need a good adjustable bench power supply to build/diagnose tube electronics. Commercial supplies are way too expensive. Heathkit type models don't source enough B+ current. I decided I'm going to attempt to design and build my own.

An older thread here discusses the design (no need to read it unless you are morbidly curious or want to see the schematic). The summary of that whole thread is: I started out thinking I was going to build a fully tubed regulated supply, but then later last year after a few determined forum members convinced me of my errant ways, I changed over to a full solid state design based on Michael Maida's approach. I've spent the last 3 to 4 months working out the design details.

A good chassis with a 5" or 6" high front panel is hard to find. (Shorter front panels won't work--not enough real estate.) The only two manufacturers I could find that do quality work were Landfall Systems and Par Metal.

I decided to go with Par Metal. Now, after a long winter's nap waiting for it to be delivered, I can finally start the assembly. It's a nice chassis, with really nice front handles, a very hefty gauge front panel, and a super thick bottom plate that can hold the weight of heavy transformers:




First things first though. I need to figure out where all the stuff goes on the front panel. Here's how I think it will be laid out (purple rectangles are the digital meters):



I'm not sure how I will get the lettering on. A proper job would have it silk screened or laser etched maybe. A trophy shop might be able to do it. Front Panel Express might also be able to do it.

I'm treading into uncharted waters with this build, and I suppose there's still a chance it could all go belly-up, but...I'm going to try anyway.

 

[Tinkerbelle]

This looks like it should be a fun build with a resulting very useful piece of test gear.

 

[chuckworkb]

 

[gadget73]

Nice. What are you thinking for displays, digital LED voltmeters ?

 

[JBL GUY]

I am in for the thread. Should be very interesting.

Will you use LCD or LED panel meters?

 

[kward]

The meters I am going to use have LCD displays. I had to direct order them from China. I looked at the LED models, even ordered one to test it and play with it, but I didn't like it as much--harder to read. The LCD meters are really bright, easy to read, react quickly, have better terminal strips for connections, and use heavier gauge wire. Overall just better built. And they were not expensive.

They also have semi decent precision given the 500V range of the supply, measuring voltage down to 1 volt accuracy. Current measurement is down to 1 mA accuracy. I wish the volt meters would measure down to 1/10 of a volt accuracy, but actually, that might bug me too much if the displays digit dance all the time, so maybe it will be okay.

Here's what they look like. The frame measures 3" x 1.75".

 

[kward]

Next problem: Inside mounting

I will have three medium sized transformers and one small transformer. I will have three breadboards: one for the negative bias supply, another for just the big power mosfets (I expect they will get hot and need breathing room), and one for the LT1084 and associated control circuits. I've also got to mount two fans for forced air cooling inside.

I decided to do all connections on/off of the breadboards with terminal blocks (similar to the one shown on the back side of the LCD meter in the previous post). I think this is a better way to go than hard wiring the boards together because if I ever need to pull out one to fix or diagnose something on it, it will come out much easier than it would if I have to unsolder a bunch of wires.

However, before I can do any of that, I need places to mount all this stuff. This chassis has a "U" shaped top and sides that come off in one peice when you pop the top off as you can see in this photo:



Par Metal sells really nice internal separators, and I did purchase one that I can use for an internal side wall to mount components on. But for the life of me I don't know why I didn't purchase two more. I really need three total: one for each side and one that runs the long dimension of the chassis to separate the power supply area that will be in the back half of the chassis from the rest of the circuits in the front half of the chassis.

I don't want to go back to Par Metal and order them now because it will probably take another 2 months to get them, so I will need to fabricate and fit some on my own. This will probably take about a week I suppose.

More later when I have those panels fabricated and installed.

 

[zandru]

K,

Me likes…….

Perhaps a (Safeguard) mechanical time clock (For reliability) so one could mind one’s time efficiently…….

My silly suggestion aside……..

Go for it. Not only is it a fun build; but It’s an essential build for all who are dead strong on benchwork……

Most impressive Kay……You are the Word……..have You heard…….

I’m looking foreword to Knock off this build, It will make for a nice gift to my Bro & freinds indeed...........

Keep up your slick workmanship…….Z......

 

[knockbill]

Nice,,, it will be a well thought out, useful tool that you couldn't buy... beats teh heck out of my Maggy 2x 5U4 + variac PS !!!

 

[kward]

I appreciate the kind words, but...it ain't working yet . And who knows what could go wrong still....

So here's the next step. Got the internal separators installed and a rough idea of how the inside parts will lay out. The divider structure is really sturdy with the left and right internal side walls having a 90 degree bend in the bottom so that screws can be used to attach them to the bottom plate. The divider structure is sturdy enough that I could easily mount the big filament toroid vertically on the front side of the middle panel and it would ride very well there. But in the end I think I will get better ventilation mounting it on the bottom plate as shown.

I will probably mount the negative bias and 12V supply transformers about 1/4 the way down from the top on the front side of the center divider. That should allow some breathing room for the filament toroid. The bottom plate is really beefy, so there is no chance of bottom panel sag even with the weight of the transformers. The transformers aren't that heavy anyway. I expect the finished unit will weight about 20 lbs.



I think the next step will be to do all the drilling, then get the front plate silk screened some how. I think this is going to be a bit of a slow and tedious process as well. Luckily I can remove the front, back, and bottom plates to do the drilling.

 

[JBL GUY]

Yes, like they say, what could possible go wrong?

It is looks like you are off to a good start.

Thanks for the update and the picture.

 

[kward]

Here's the CAD layout of the front panel. (Decided to go with Front Panel Express). I'm going to mull this over for a few days now before sending it in.

Front Panel Express doesn't yet have digital printing (silk screening), so all artwork will be engraved and paint filled with the colors shown (all black except for the 6.3V filament lines which are orange).

The two unmarked holes on either side of the filament section are for mounting the filament circuit breakers.

I struggled with the best location for the MAIN ON indicator and switch. As it is, the three main sections of the front panel (C- volts, B+ current, and B+ volts) are organized vertically, which seems to be the most intuitive layout I could think of. Specifically, I did not want to put the MAIN on switch and indicator in the C- column since it does not perform any function related to bias volts, even though it would look more symmetric there.

All in all, it's looking pretty good me thinks.

 

[gadget73]

I tend to be a fan of power switches on the far left. I don't really know why, but it makes sense for it to be there in my mind.

 

[JBL GUY]

It is looking good.

Front panel layout can be as simple as what works best and or what a person likes.

My preference, and it is only my preference, is for the the main power switch to be the big switch with the big indicator and usually in the center, so at most I only have to reach half way across a piece of equipment to reach the main power switch.

 

[kward]

Progress

Since last post, I have finalized the front panel layout and shipped it off to Seattle for cutting and engraving at Front Panel Express.

In mounting the transformers, the first problem was figuring out how to mount the power transformer so that I could run the wires out from the bottom without pinching them. A piece of 3/4" MDF to the rescue, with slots cut out for the wires. Even with the MDF mount, the top of the transformer is still about 1/2" below the chassis top, so it fits fine.



Here's a shot of the front of the unit without the front panel in place.



The negative bias supply transformer merits some discussion (shown at the top left of the toroid). I found it nearly impossible to find an off-the-shelf 150V - 180V AC secondary in a small form factor that can source 30 to 50 mA. I finally settled on a Hammond 262B24, which has 3 secondary windings: a 120V winding at 27 mA, and two 12V windings at 150 mA each. I connected all three secondary windings in series with additive phase so that the resultant secondary voltage (unloaded) is ~150V AC. (Loaded at 20 mA this falls to ~140V AC). This is enough to produce -90V regulated DC bias voltage at up to 30 mA output.

The left inside panel has also been removed so that I could install the main filter choke and diode bridge. The turret board is from Angela Instruments (about a 3" length cut from the original 15" length delivered). The board is mounted with three 1/2" standoffs.



I tested the diode bridge with the power transformer and a variac, and found that all is working properly so far. Next up is to mount the filter caps and fans. Then I think I can start building the power supply boards.

 

[JBL GUY]

Looking good!

Thanks for the update and pictures.

 

[stormytaylor]

Looks good. Maybe I could have saved you some time and trouble. I have in my garage packed away 3) large power supplies. I believe they go to 2000 or 3000 volts @ 250mA.

 

[phantam]

wanted voltages

I may have missed it, but what voltages and milliamps are you planning to use? Just guessing from the picture. It looks like you set up a full wave bridge and 1000 PIV diodes.
Very nice looking work!!!!!

 

[kward]

Thank you for the comments and interest. We'll see if it will actually work... (I'm sure it will).

Looks good. Maybe I could have saved you some time and trouble. I have in my garage packed away 3) large power supplies. I believe they go to 2000 or 3000 volts @ 250mA.

Well now you speak up! . I suppose you could parallel two units together to get more current output. :scratch2: For me, it's as much about the discovery and learning as it is about the final product. The learning journey has been a good one for me, so it has not been wasted effort.

I may have missed it, but what voltages and milliamps are you planning to use? Just guessing from the picture. It looks like you set up a full wave bridge and 1000 PIV diodes.
Very nice looking work!!!!!

High voltage supply: Regulated at 2V to 498V adjustable at anywhere from 0 mA to 350 mA current draw. Constant current mode settable anywhere from 10 mA to 350 mA.
Bias supply: Regulated at 0V to -90V adjustable at up to 30 mA current draw, constant current kicking in automatically at 30 mA.
Filament supply: 2x 6.3VAC at 6A, or 1x 12.6VAC at 6A, or 1x 6.3VAC at 12A, depending on the needs of the device under test.

Yes, the diodes used for rectification in the high voltage supply are UF4007 types.

I'll post the final schematic when I get it all tested and working...

 

[battradio]

Any thought to Uf4007's in series for the diodes just to have extra safety .