Experiences with Asian LM317K HV/LV power supply board

[doorunrun]

Advertised on eBay and AliExpress with such names "High voltage + filament Regulated power supply board for Tube preamp DIY L12-41," I opted to purchase one to power my "Mystery RCA mixer-amp" rebuild project, covered in another thread here.
PCB available, too:

The various vendors who sell the board don't offer much in the way of documentation and what is supplied can be a little hard to understand.

So far, I'm happy enough with the board, it does what it says it does. But, lacking a schematic I was feeling a little lost. With some time on my hands I did some digging and would like to share some information about it. What I like about the board is they included ventilation holes throughout the board for the heatsinks and higher wattage resistors. The I/O screw connectors and PCB are of good quality.

Cutting to the chase, I drew a schematic that's pretty darn close to what's on the board:

Drawing is done in DigiKey's Scheme-It; it needs a little refining and a double-check of the exact values on the board.

I came across a more original version of the H.V. section in this post:
https://www.diyaudio.com/community/threads/tip50-lm317-high-voltage-regulator.61831/#post-2050076
The L.V. section schematic comes from a similar product for powering filaments only.


After reading up on Maida H.V. regulation I felt like there were a few things lacking in this board. The first being not much in the way of capacitor bypassing around the LM317s; the second is the large value capacitors on the outputs.

I've installed input/output bypassing with 100nF caps to hopefully suppress noise.
I pulled the 100uF/400V cap on the board's H.V. output and using (for the moment) a 15uF/300V electrolytic in its place. On the L.V. section I'm using a 220uF/25V cap instead of the 1000uF supplied with with the board.

To build it out into a usable supply I decided to use an old Eico MX-99 Multiplex Adapter I've had for a long time and didn't think I'd ever use as a donor chassis for the project.


On the bottom side of the chassis I have two Triad toroidal transformers mounted. The HV regulator is set to provide 250V and the LV set for 6.3V.

Cheers!

 

[doorunrun]

It seems I couldn't leave well enough alone.

I've been supplying the L.V. section with ~12.6VAC with the output set to 6.3VDC. The LM317KC regulator input sees ~16VDC. That much drop in the regulator produces a lot of heat to sink.

There were some things I thought would do to help. I first wanted to get rid of the silicone pad and go the thermal grease and mica washer route. In going through the switchover process I decided I wanted to add a socket for the regulator and elevate the heatsink to allow air to circulate underneath.

The regulator's pins had been trimmed down during original installation and just barely made contact with the socket. That led me to deciding to source some spare LM317Ks with standard length pins for repairs down the road. The spares came from a seller on eBay branded with the "new" NS logo, the stylized "N."

All three of them failed on initial power up in the circuit. I think the short-circuit protection of these devices failed; the heater string for the preamp I have measures about 1.5 Ohms and like most filaments probably goes up as they warm up. Normally the heaters of the 3 tubes draw about 1A in operation.

The original LM317K TO-3 device still worked with the elevated heatsink arrangement but the shortened pins bugged me.

As a test, I installed a LM-317T, the TO-220 version, in the circuit and things got back to normal. Now the problem was proper heatsinking for the TO-220 device. In the long run I think moving away from a TO-3 device is a good idea.

With the original heatsink attached to the LM-317T and oriented vertically the heatsink gets up to about 175F and level off in free air. I can't put the perforated metal cover back on yet as the heatsink sits too high.

At this point I'm going to try moving over to the LM-338, a TO-220 device rated at 5A, rather than stick with the LM-317 series.
I also think I need to lower the regulator's input voltage to something like 10VDC.

In order to do that and continue using the transformer I have I'm adding a 4 Ohm/10W wirewound resistor in series with the transformer's secondary. That should move some of the heat dissipation away from the regulator to the the resistor.

Things are still in flux as they say, probably the best solution is to use a 7-9V transformer to start with.

And so it goes...

 

[Khron]

The spares came from a seller on eBay branded with the "new" NS logo, the stylized "N."

Well, between National Semiconductor being bought out / absorbed by Texas Instruments (later edit - back in 2011..!), and brand new LM317K's being north of $60-70 a pop...

https://www.mouser.fi/ProductDetail...317K-STEEL-NOPB?qs=EQxOuqCyzqS9XZffD%2B6BQg==

 

[doorunrun]

Getting closer to where I want this project to be.
I installed two 1.2R wirewound 5W resistors in the LV section between the rectifiers and upstream filter caps in a R-C-R-C arrangement. To make room I replaced the axial-lead caps with radial versions. Now the input voltage to the regulator is ~10.1V. The power resistors do warm up; each one dissipates about 3W.

I replaced the LM317T with a LM338T. I beefed up the heat-sinking with a combo of two that I had on-hand and now have it mounted vertically, which I think is better. With this resistor/h.s. arrangement the '338 runs much cooler.

A couple of photos showing the current board:


From another thread here I found an Antek transformer with a 200V secondary and dual 6.3V windings. I pulled the two Triads in favor of the Antek. The only issue was the Antek sits higher than the Triad due to its higher VA rating. I only had to increase the stand-off spacing by 1/4" to allow the bottom cover to be mounted.

The perforated cover of the old Eico MX-99 enclosure is a real plus for passive cooling. Too bad there aren't (m)any available with that option these days.

HNY!

 

[drtechno]

Well, between National Semiconductor being bought out / absorbed by Texas Instruments (later edit - back in 2011..!), and brand new LM317K's being north of $60-70 a pop...

https://www.mouser.fi/ProductDetail...317K-STEEL-NOPB?qs=EQxOuqCyzqS9XZffD%2B6BQg==

There is a lot of TO3 case stuff being reworked because of no lead. But this one will be available as to-220 because it really doesn't need to be in a to3 case.
I'm sure somewhere someone is sitting on a lot of them thinking they can get $60-70 each for the ones that don't know you use a To-220 version now.

 

[Cqwet Dbdfte]

Metal case semiconductors were very common, years ago as it also provided hermetic sealing, later not much of an imperative with glass passivation, making lower cost epoxy encapsulation viable. Some TO-3 cans had desiccant inside.
I always try to use the metal box as a heatsink, mounting TO220 etc parts under the PCB, also using 1/4" aluminum standoffs, for spacing and enhanced thermal transfer.
I use larger hole for mounting TO220 (fullpak) to give space for a screwdriver. One sided aluminum PCB are now low cost and can be used for circuits and as a precision machined mechanical part, sometimes sandwiched together with a regular PCB.
Hole sizes are very precise and holes can selected for later easy self tapping without a tool.
For sheet metal work a Roper punch beats drills, and a nibbling tool is very handy. The larger Roper punch can be used for punching and setting PEM thread inserts for smaller jobs.
Making a nice looking box for projects may be the greatest challenge.
There are a viable few options for low volume production if you want to spend a few more bucks.
Aluminum PCB can also be used as front panel, and artwork used for text.