VRDN: bipolar regulator PCB for line level ckts: ±11V to ±20V @ 1.5A with "De-Noiser"

[Mark Johnson]

A couple of diyAudio members and I perceived the need for a power supply board aimed at line level applications, such as for example Wayne's BA 2018 linestage. We felt it should include the following features:

• Rectifier diodes on board
  
• Transformer snubbers on board
  
• Usable with both dual-secondary and single-secondary transformers (AC-to-AC wall warts)
  
• True 1.5 ampere output capability, therefore: real heatsinks & big capacitors
  
• Output voltages field adjustable via multi-turn trimpots

The result is a PCB called "VRDN" - Voltage Regulator with De-Noiser technology. The DN technology was originally invented and published by Charles Wenzel, whose original article called it "Finesse" (link) . DN technology was greatly improved and expanded by diyAudio member Elvee, who named the improved circuit his "De-Noiser" (link) This regulator PCB implements De-Noiser circuitry, as shown in the schematic of Figure 1 below.

Capacitor C13, and everything to the right of C13, are the De-Noiser. Jumper block J3 allows you to easily disable the De-Noiser (jumper installed) or enable it (jumper removed). This gives the pleasant option of performing a before-and-after pair of tests, either with measuring equipment OR with human listeners. Try it one way, then try it the other. Which do YOU prefer?

The board includes two transformer secondary snubbers, namely (MOV1, C1, R1) and (MOV2, C2, R2). Metal Oxide Varistors perform double duty in these snubbers, serving as the across-the-secondary capacitance "Cx" and also acting as voltage clamps, if an AC mains spike or surge occurs. The values of snubbing resistors R1 and R2 were chosen after performing several Quasimodo tests on several different transformers, and I expect these resistor values will work well IN THIS PARTICULAR CIRCUIT, with a wide variety of different transformers. However, if you wish to perform your own Quasimodo tests on your own transformer, and derive your own optimum values for R1 and R2, please go right ahead with my blessings and good wishes. Just remember to insert your actual MOV1 component into your Quasimodo's "Cx" socket.

Figure 1 includes two bridge rectifier components, BR1 and BR2. Each of these implements a "Full Wave" rectifier, meaning it transfers energy from the AC mains to the DC filter capacitors, two times per cycle: once at the peak of the mains sine wave, and once at the trough. The circuit contains two Full Wave rectifiers, one per transformer secondary. One of them produces the positive DC voltage output, and the other produces the negative DC voltage output.

However, when conceiving this board, we envisioned that some hobbyists might wish to build a bipolar regulated supply but feed it from an AC-to-AC wall wart instead. This has the great advantage that it keeps the mains voltage completely away from the DIY chassis; instead of 115VAC or 230VAC, the wires which enter the chassis carry only 16VAC or 18VAC, which is noticeably less dangerous. An AC-to-AC wall wart also allows a smaller DIY chassis with less interior volume, since the transformer is now outside rather than inside the chassis. But an AC wall wart is just a transformer with a single secondary.

VRDN implements the conventional, textbook method for creating a bipolar DC supply from a single secondary transformer. It's called a "Half Wave" rectifier, and as the name suggests, it transfers energy from the AC mains to the DC filter capacitors, only one time per cycle: either at the peak, or the trough, of the mains sine wave. One HWR generates the positive DC voltage output, and the other HWR generates the negative DC voltage output. Exactly as is done in the Objective-2 headphone amplifier, among MANY other examples.

To use the VRDN board with a single secondary, AC-to-AC wall wart, we need to make a few very minor adjustments. They are tabulated in Figure 2 below. As you can see, the only changes are to the list of stuff-and-solder actions. No PCB tracks need to be cut, no soldermask needs to be scraped off. You simply do-not-stuff six components, add three jumper wires (each 2.5mm long) into pre-labeled locations, and stuff two discrete diodes into existing & labeled thru-holes. It's easy.

Figure 3 shows two VRDN boards. The top PCB in the photo is configured for a dual secondary transformer -- notice it has two green Euroblok connectors at the left edge of the board. One connector for each secondary. The bottom board is configured for a single secondary AC-to-AC wall wart. It has only one Euroblok connector and only one snubber at the left edge of the board. Everything else on the two boards is identical.

Figure 4 is a detailed picture of the input circuitry, on a PCB configured for two secondaries. Again notice the two Euroblok connectors.

Figure 5 is a detailed picture of the same region, this time on a PCB configured for a single secondary AC-to-AC wall wart. You can see that the bridge rectifiers are not stuffed, the (MOV2, C2, R2) snubber is not stuffed, and the Euroblok connector is not stuffed. You can also see that the three jumper wires "JOR1, JOR2, JOR3" have been stuffed -- look for pink arrows on the photo. Discrete Schottky diodes (5 amperes, 100 volts rated) have been stuffed -- look for gold stars. One diode is stuffed in holes "A" and "C", the other diode is stuffed in holes "L" and "M". Exactly as required by the options table shown in Figure 2.

The Jumper Options are shown on the schematic as 22 megohm resistors. This is a limitation of the CAD system and, specifically, a limitation of the CAD system USER, namely me. In truth, in real life, on your real PCB, JOR1 is either a zero ohm resistor (i.e. a piece of 22AWG hookup wire with white insulation) or else it is an INFINITY ohm resistor (i.e. an open circuit; nothing stuffed). Same goes for JOR2 and JOR3. Completely disregard "22 megohms" on the schematic. Ignore it. Pay it no attention. Simply remember that JOR1-3 are either opens or shorts. Opens or shorts. Repeat after me: opens or shorts. Look at the photos in Figures 4 and 5. Those options are either not-stuffed-at-all (open) or else stuffed with hookup wire (shorts). You can see them right in the photographs. Repeat after me: opens or shorts.

Figure 6 is a photograph of the output side of the PCB. Individual wire-to-board connections are provided for supply and ground, for each of two stereo channels. Also visible in this photograph are the vertically mounted resistors. This PCB uses the vertical orientation exclusively. Finally if you look very closely, you can see the amber colored Kapton tape on the back of the heatsinks.

TIPS FOR BUILDERS

Please carefully study the top silkscreen text layer when stuffing and soldering the 1N4004 diodes and the 1N4148 diodes. There is a little "A" next to each hole where an Anode is soldered. You'll find that on every 1N4004 and 1N4148, the cathode (band end) of the diode is the short wire which is stuffed straight thru the PCB with no bends; the cathode is the "butt end" of the vertical diode. The anode wire is the one which is bent into a 180 degree U-Turn, and is exposed for easy probing. The anode wire goes into the PCB hole with the little letter "A" on silkscreen.

First stuff and solder all resistors and axial diodes. Second, stuff and solder the MOVs and the bridge rectifiers (if using!!). Third, stuff and solder the Euroblok connectors and the two pin headers J3 and J4. Fourth, stuff and solder the small (non 2200uF) capacitors, the trimmer pots, and the transistors. Fifth, stuff the regulators-on-heatsinks (next paragraph).

After you bolt the LM317 and LM337 to their heatsinks, including the Sil-Pad insulator and the insulated shoulder washer / bushing, I recommend that you test-fit one of them into the PCB. You will discover that your unmodified assemby doesn't quite fit! You'll need to modify it. Bend the leads of the TO-220, backwards (towards the heatsink) about 1-2 millimeters. Try this. Aha! Now it fits. Put a piece of Kapton insulating tape on the back of each heatsink, over the TO-220 mounting bolt. The two heatsinks are positioned back-to-back and we don't want them to come into electrical contact. Kapton tape prevents that.

Stuff the regulators-on-heatsinks so the heatsink edges are flush with the PCB, but only solder the IC legs for now. Do not solder the heatsink mounting pins yet. Instead, turn the board right side up and look at the heatsinks placed back to back. You want to see a nice wide air gap between the two heatsinks, about 3-6 millimeters wide. When building my boards, I folded a sheet of paper in half, and half again, and again, etc., until I had a bundle 5mm thick. Then I jammed this paper bundle down between the two heatsinks to create a definite gap between them. With the paper in place, enforcing a nice gap, THEN I soldered the four heatsink mounting pins in place. After the solder joints cooled, I removed the paper bundle and voila! A nice, permanent, air gap between the two heatsinks.

Finally, stuff and solder the twelve 2200 microfarad capacitors.

SUMMARY OF FEATURES AND MIS-FEATURES

• De-Noiser circuit with jumper enable/disable
  
• Diodes on PCB
  
• Transformer Snubbers on PCB
  
• Big capacitors on PCB allowing true 1.5A output
  
• No bleeder resistors on PCB
  
• Voltage regulator ICs mounted on heatsinks for true 1.5A output
  
• Heatsink thermal resistance is 8.6 deg C per watt
  
• This is a four layer PCB, three of which are ground planes
  
• Gerber CAD files freely available, send to any fab you wish
  
• No "pilot light" LEDs on DC outputs
  
• Can use with either dual secondary trafo or AC-to-AC wall wart
  
• No SMD components; 100% thru-hole
  
• PCB unique identifier is TZPTR7 ; suffix -A means "revision A"
  
• PCB mounting holes compatible with Modushop chassis baseplate
  
• Board size 62 x 142 mm (edges) ; 50 x 130 mm (holes)

[6L6]

This looks fantastic and the "De-Noiser" is very interesting. Will try this with the Wayne's BA2018 linestage as soon as I can get it built up.

[pfarrell]

Parts ordered. Yea Mark!

[iamwhoiam]

Very nice. Have been a bit obsessed myself with this schematic for a while in the original thread Trying to optimise an onboard-transformer solution.

Have you made any noise measurements yet Mark? Presumably will need something like this: Low Noise Measurement Preamplifier

Why not go Sziklai pair (Die-Noiser)?

[spind]

Very nice--thank you very much! Could this be configured with a centre tapped transformer?

[Mark Johnson]

No measurements so far. I left out Sziklai pair so that you would put it in. By creating a new, modified board with improvements.

[Mark Johnson]

Originally Posted by spind

Could this be configured with a centre tapped transformer?

I don't think so / I don't know how. Maybe someone else does.

[pfarrell]

Some art for this thread.

Fantastic SMALL form factor PSU. Would have totally used this for my Wayne's Pre18 build if Mark had made this last year....

Some lens flare off the caps! Love it.

—I trimmed the sinks down to make them 1U friendly. 5mm standoffs under the board. Kept the little clip-on standoffs that came with the heatsinks intact, soldered them to the board.
—Easily achieved +/-18VDC. Had an Antek 50VA sitting around that I used to test with. Out of the gate I had around 17VDC on the negative side and around 16.25VDC on the positive with that transfo and no adjustments.
—Need bolts with low profile heads for the sinks—the "normal" allen ones I had around were too tall for their back to back orientation....


Zero drama bringing them both up to match. Need extra soldering iron heat on the cap joints—due to all the copper everywhere. Thin gauge FireMettal solder from the store came in handy. And apparently I got the diode orientations correct! "A" is for Anode. Was hard to see the "A"s on the board...

Mark—any benefit to Quasi-action on transfos plugged into this little guy? Could make little plug-in rigs at the secondary connections—which I did one time for my phono preamp PSU that didn't have implicit Quasi-spots.

[Mark Johnson]

Beautiful build, spectacularly photographed! Wow, congratulations on an awesome job.

(BTW, I was able to create a larger gap between the two heatsinks and therefore, the two TO-220 mounting bolt-heads, just by jamming a 5mm thick bundle of folded paper between the heatsinks before soldering their PCB mechanical hold-down pins. This created a ~4mm gap between them)

Thanks for the feedback about finding it difficult to see which end of the 1N4004 and 1N4148 diodes get stuffed in which holes. I will modify post #1 to mention it explicitly. Here is the memorable fact: The "butt end" of the vertical diode (short wire) is always the cathode (band side) of the 1N4004 and the 1N4148. The "folded over and exposed lead" (long wire) is always the anode of the 1N4004 and 1N4148. On the top silkscreen layer of text, there is also a little "A" for Anode, next to this drill hole.

Although this board already includes a 1R+2C snubber for each transformer secondary, and although I have specified values for all of the components: If you are enthusiastic to horse around with Quasimodo plus your transformer and your own snubber component values, please go right ahead ... but I recommend that you read the fourth paragraph of post #1 before you begin. Where MOV1 is first discussed.

[pfarrell]

Ah Yes! I knew I had read something somewhere about Quasi! Thanks for the heads up reminder. Sometimes I need things BOLD and CAPPED and in Big Type <--1000pt. (And repeated.) Maybe a hazard of too many incredible projects?