WARNING: Before I get to the information on how to assemble this, I want to make an important point: I am not selling YJPS power supplies. I sell a circuit board that can be used to make a power supply. The distinction is that you made the power supply, so you must take responsibility for your own safety when building, testing, and employing it. If it fails in any way, it’s not my problem. Power supplies can kill. Be respectful of that power.
This part is the hardest one to install when there are other parts nearby, so it’s best to add it first.
Add all the resistors and test points.
Then add the small barrel and TO-92 sorts of diodes: D2, D3, D4, and D6.
You can also add the power LED (D5) if it’s going to sit down on the board. If it’s going to be up off the board on long leads, wait until you’re closer to doing the casework. Beware! The plus sign on the board for this part is on the wrong side in the v1.2 board.
Finally, add the small TO-92 transistors: Q1, Q3, and Q4.
Add the small C4, C6 and C12 capacitors. These aren’t polarized, so orientation on the board doesn’t matter.
Then, add the small electrolytics: C7 through C11. These are polarized. The positive leg is the long one, and the negative side of the cap is marked with a stripe. Look for the little plus sign on the board to match things up.
Now is a good time to add VSET, if you want an adjustable regulator. Otherwise, put a jumper here.
If you’re going to mount the fuse on the circuit board, solder its clips in place now.
Now we will set the regulated output up for a simple 2-wire connection, even if you will eventually want to use remote sensing. This will make it easier to test, below.
Add the VOUT connector, if you will use it. If you will be soldering the output wires directly to the VOUT pads, it’s better to do this later. Until then, you can use the VOUT pads as meter probe “sockets.”
Then for JRSV and JRSG, you can either add jumpers, or Molex KK connectors if you have also built shorting jumpers, like the one shown at right.
Finally, connect the JURG pads — the Jumper for the UnRegulated Ground. This connects the unregulated DC supply ground to the control section’s ground. The spacing between the jumper holes is sized to allow for a small resistor or inductor, if you want to play with decoupling these two grounds. If you’re not one to experiment, just use a simple wire jumper here. Use the thickest wire you have on hand that fits into the hole.
Add the bridge diodes (D1s), being careful to get the orientation right.
Next, add the preregulator (U1), the pass transistor (Q2), and the SCR (T1).
The best way to install the heat-sunk parts is to put a thin layer of thermal grease on the part’s tab, then bolt it to the heat sink, and insert the whole assembly into the board and solder it down.
If you have a tap set and are using the standard heat sinks, you might cut some #4-40 threads into the heat sink before you solder the assembly down; the holes in the board accept #4 screws. This is optional, with the pins on the TO-220 part sufficient for keeping the part stable under normal vibration stresses. If your screw heads are large enough, you’ll need to use a non-conducting washer for the Q2 mounting hole between the two traces.
The wire pads are big enough for 18 gauge wire.
There are two pads for AC line, marked LA and LB. The LA pad is the one to use if you will be using the on-board fuse holder. If you will be using an IEC inlet module with its own fuse holder, use the LB pad instead. Under no circumstances should you avoid using a fuse!
The AC neutral wire goes to the N pad, and the AC safety ground goes to the G pad.
If your AC inlet module uses FASTON type connectors, crimp the mating lugs onto the each of the three wires you just added for attaching to the AC inlet module. Keep the wires short. If you’re putting the power supply into a metal case, you should crimp two wires into the ground lug: one wire goes to the board, and the other goes to the case for safety reasons. If the case is aluminum, sand the inner surface around the bolt hole to remove any surface oxide and anodization to ensure good contact with the case ground wire.
The A through D pads configure the transformer’s primary side for your country’s wall voltage:
100-120V power: Put a short jumper from the A pad to the B pad on the board. Jumper the C pad to the D pad.
220-240V power: Put a short jumper from the B pad to the D pad on the board.
If you won’t be using the line filter, jumper across the filter choke. You need two wires, each one going between the two pads joined by the curly wire symbol on the board.
The recommended Amveco board-mount transformers have dual secondary windings. You can configure them on the board to be in series or parallel. Neither is inherently better than the other, the right choice just depends on what you’re trying to do.
For example, say you have an Amveco 70053, which has two 15 VAC secondaries at 0.5 A each. If you put the secondaries in series, you get 30 VAC at 0.5 A out of this transformer. (Or somewhat higher voltage at lower current, up to about 37 VAC at no load.) If you put the secondaries in parallel, you get 15 VAC at 1 A instead. It’s the same 15 VA either way; you’re not getting anything for free, just trading off voltage for current, or vice versa.
To put the transformer secondaries in series, put a short, stout jumper between the W and Y pads on the board.
To put the transformer secondaries in parallel instead, connect W to X and Y to Z.
Either way, use the stoutest wire you have on hand that fits in the holes. I recommend using 18 gauge hookup wire.
If you want the line filter, add L1, C1 and the C2s. Orientation doesn’t matter for any of these parts.
Add the snubber cap, C3. It is not polarized.
Add the C5 capacitors. Pay special attention to their polarity; orient them like the other electrolytics.
The standard Amveco PCB-mount transformers can only go in the board one way. If you’re using some other type of transformer, use the markings on the board to figure out which wire goes where.
Plug the IEC cord into the power supply first, and then plug the other end into the AC outlet. This keeps you away from the power supply when it first turns on, in case something is hooked up improperly. It also keeps your fingers away from the bare wires at the rear of the AC input jack when power is applied. Power supplies can self-destruct violently, so you may want to avert your face. An exploding capacitor is...exciting.
If nothing exploded or caught fire, measure AC volts between TP1 and TP2. This is the AC output voltage of your transformer. With no load on the power supply, it will be higher than the rated output voltage of your transformer. With the Amveco 70053 (±15 V), I get around 40 V here.
Next, measure DC volts between TP3 and TP4. This is the unregulated DC voltage at the filter capacitors. You should get around 1.3× the AC voltage you measured previously. (You’d get 1.414× the AC voltage with an ideal bridge rectifier; real rectifiers have a small voltage drop across them.) You might also measure AC volts here, since this is the ripple voltage that the regulator has to remove. It should be on the order of tens of millivolts, the exact value depending on the size of your filter capacitors.
Next, measure DC volts between test points TP5 and TP6. (YJPS v1.1: TP3 and TP5 instead.) This is the output voltage of the preregulator. It should be about 2.3 V above the desired regulator output voltage. If you have a measurement preamp, you can measure AC volts here to see how much ripple the preregulator let through. This should be a very small value!
Do the same as the previous test between TP7 and TP8. (YJPS v1.1: TP3 and TP6 instead.) This is the final regulated voltage. Measuring DC voltage here, you should get the expected regulated voltage. Measuring AC voltage here with a measurement preamp, you should get almost nothing.
Now that the supply is known to function, we can work out how to connect it to the circuit it will power.
The simplest and most reliable option is to treat the YJPS like a traditional single-voltage power supply: run two wires from the VOUT pads to the board being powered. The VOUT footprint will accept either 18 ga wire or a vertical Molex KK series 2-pin male connector. (The larger type, with 0.156" (4 mm) pin spacing, which also accepts up to 18 ga wire.) The square pad, near C11, is for the positive rail, and the negative rail goes to the other pad, the one near D4.
The Molex connector is convenient, but beware that its contact resistance wipes out some of the advantage of this regulator, that being its exceedingly low output impedance. It’s better to solder wires straight to the board, if you can accept the inconvenience.
By the same token, if you choose the 2-wire output configuration, you should use the thickest wire you have on hand, up to 18 ga, and keep it as short as possible. There is literally no advantage to building a YJPS over a TREAD if you’re going to put 6 feet of 24 ga wire on its output. The long, thin wires completely wipe out the YJPS’s output impedance advantage. This would likely wipe out the YJPS’s ripple reduction advantage, too, unless the downstream circuit is running in pure class A. Even with 18 ga wire, a mere 1 mA of current variation is enough to significantly increase the ripple of the power supply over its nominal base value.
The need to use very thick, very short output wires to get the full benefit of the YJPS is inconvenient. The YJPS gives you a couple of ways out of this uncomfortable position, all forms of remote sensing.
Above, you put jumpers across JRSV — the Jumper for the Remote Sense Voltage — and JRSG — the Jumper for the Remote Sense Ground. If you remove these jumpers, you’re left with four wire pads. You can use these instead of VOUT, putting the YJPS into the full remote sensing mode. All four wires go to the remote circuit, where the two JRSV wires connect to the load’s positive supply rail, and the two JRSG wires connect to the load’s negative supply rail. (That is, V-, not virtual ground, in the amps described elsewhere on this site.) This extends the YJPS regulator’s feedback loop out to the point of load, allowing it to correct for ripple introduced by the load, and for the VI drops in the output wires. The improvement in regulation performance can be significant.
Because the remote sensing configuration corrects for output wire resistance, the JRSV and JRSG pads do not have to be as big as the ones for VOUT. These pads will accept 22 ga wire, either soldered straight to the board or connected via the smaller sort of Molex KK connector with 0.1" pin spacing. There’s no problem using Molex connectors here, because remote sensing will correct for contact resistance, too.
One of the costs of remote sensing is that, while it can correct for the VI drops introduced by long output wires, the longer the wires the bigger you make the regulator’s feedback loop. At some point, the resulting antenna will pick up enough noise or add enough stray inductance and capacitance to the circuit to make it unstable. Remote sensing doesn’t change the fact that it’s better to use a long AC supply cord than a long DC output cable.
It can be difficult to find 4-wire connectors needed to allow the YJPS and the circuit being powered to be in separate enclosures. DIN-4 and XLR-4 connectors can be had, but they’re not as widely available as the 3-pin versions.
You can still get most of the benefit of remote sensing with a 3-pin power connector, but with a 4-conductor cable. The JRSV wires go to two of the connector pins, and both JRSG wires connect to the third. This allows the JRSV wires to continue on right to the point of load inside the circuit being powered, while the JRSG wires just go most of the way, stopping right at the power inlet to the circuit being powered.
It may be tempting to just leave JRSG jumpered, so you can use 3-conductor cable with your 3-pin connector, running the ground wire from the round pad of VOUT. Don’t do this. It means you have no remote sensing on the ground side, when adding it is so easy. 4-conductor cable, unlike a 4-pin connector, is easy to find or make.
Another word of warning: don’t use a TRS plug for this! The way they insert allows cross-connection of power and ground, a potential disaster for the circuit being powered.
Following the above scheme one step further, it’s possible to get most of the benefits of remote sensing with just a 2-pin connector, such as a common DC barrel connector. Just as above, you still run 4-conductor cable from JRSV and JRSG, with the difference being that you make the Kelvin connection for both JRSV and JRSG inside the connector.
The 3-wire alternative performs slightly better, but this version has an important advantage. Many off-the-shelf devices use 2-pin DC power plugs, typically some sort of barrel connector. This configuration lets you use remote sensing without opening up the other device and replacing its power inlet connector just so you can run the YJPS output wires straight to the point of load. Getting them close is far better than using no remote sensing at all.
You want to spend some time up front to think about casework details. Don’t go soldering power input and output wires to the board before you think about how to do the casework. Think about where the noise sources are, how power will flow, and so forth. It’s easy to end up with something that needs a lot of rework, either because it’s not physically practical to do what you half-planned to do, or because the result has a lot of noise because you did a poor job of planning the cable routing.
First, study the board layout. You can see that it carefully puts the AC hum-inducing parts way off in one corner of the board. The power path then goes around the board edge in a clockwise fashion, with the regulation increasing as you go along. The loop doesn’t quite complete, so that the DC output pads aren’t back up against the transformer. There’s some dead space between the transformer and the regulator, and the power path doubles back on itself there near the end to keep the DC output at the front of the board. All this is important to keep in mind when designing your case layout.
Ideally, you want to maintain this layout, with noisy AC coming in one end of the board and quiet DC going out the other.
This is not always practical. What if it makes more sense to have the DC output on the same panel as the AC input? This requires sending the DC output right back past those noisy AC sections. Since we can’t redesign the board so the power path is in a U shape to make this easy, we have to think very carefully about cabling.
I recommend using top-quality 4-conductor shielded copper cabling in cases like this. The shielding doesn’t have to be copper; it can be anything non-ferromagnetic, like aluminum. Connect the shield to RGND and the inner conductors to the various remote sense positions as above. (Consider remote sensing mandatory in this situation!) Then run that cable back over the unregulated DC section, as high as you can get it. In the recommended Hammond 1455N16 case, this means putting it up in the corner of the case. Make no mistake, though, this is not ideal. You are subverting the design principles I laid out above. Rethink whether you really need the DC output and AC inlet on the same end of the case before doing this. And if you must, I repeat: copper cabling, 90+% shielded, but not with steel.
Putting the YJPS and the circuit being powered in the same case can avoid the awkwardness above, if you use a large enough case.
The best layout I’ve come up with is to put the AC inlet in the back left corner of the case, and the circuit being powered up in the right front. Then the YJPS board goes along the back left part of the case, leaving the regulated DC output at the back right, near the rear of the other board, where its DC inlet probably is. The pair make a kind of L shape.
If the L configuration isn’t practical due to insufficient room in the case, you have to worry about what paralleling the two boards will do. In the worst case, it can put the circuit being powered right up against the noisy AC section on the YJPS board. The YJPS-centric way of laying the two boards in the case is to put the transformer end of the YJPS in the back left corner, say, with the DC output ending up in the front right. Then it has to go right back past the AC section to get to the back end of the board being powered, if that’s where its DC inlet is. It’s better to flip the YJPS around, with the transformer up in the front left corner, fed by an AC cable running along the left edge of the case, so that the YJPS’s DC output can be right near the other board’s DC input.
The general principle you can distill from all the above is simple: long AC wall supply wires are better than long regulated DC wires.
Another trap is soldering the DC output cable to the board early, before doing the casework, thinking that it’ll make testing easier and it has to be done anyway. Then when you go to do the casework, you’ll probably find that the DC output connector doesn’t fit through the hole in the case that was sized for the strain relief. Now you’ve got to desolder the cable from the board so you can feed this end back through the strain relief hole. Better to plan ahead, and either use Molex connectors on the board, or not try to get ahead of yourself and do a full-on DC output cable before it’s time to do the rest of the casework.
The strain relief in the parts list is for a regular 2-conductor DC power cord. It requires a 7/16" hole.
For the recommended AC input jack, you need a rectangular hole about 27 mm × 33 mm.
If you’re using a metal case, run a wire from the AC ground to a bolt on the case. If your case is aluminum, the anodization or oxidation will need to be removed around the bolt hole so the ground wire makes a solid electrical connection.
If you will be mounting the board into the bottom of your case, the mounting holes are at the corners of a 3.55" × 5.9" rectangle.
Unlike with simpler supplies, it’s important to re-test a YJPS after casing everything up and figuring out the wiring. It’s quite possible for a YJPS to work fine on the bench with all the jumpers jumpered, but fail when you try to actually use it due to the complexities of grounding, remote sensing, or load interactions. Realize that the YJPS is a wide-band circuit. Ideally, you should test the final setup with an oscilloscope and a measurement preamplifier, to get both qualitative and quantitative measurements, and to check both audio band and high frequency behavior.
This space intentionally left blank. :)
|
<< Part Selection Guide |
| Updated Tue Jan 26 2010 01:24 MST | Go back to Young-Jung Power Supply | Go to my home page |