Help needed for 18v phono stage PS hum
I need an 18-24v supply for a jfet phono stage. I've run it on two 9v batteries and it works well, but it uses them up quickly. So I want to build a linear PS for the amp. I have built a lot of electronics with PS but I'm not a designer and I'm having trouble wrapping my head around this problem. Amp is cased separate from any PS.
I tried a simply CRC/Pi style supply, using a 12.6-0-12.6v filament transformer I had lying around. Transformer is 2 diode full-wave rectified (giving me the 18v output I want) with center tap connected to the PS ground.
Rectifier output goes into a 470uF cap, then 10R resistor, then into three 2200uF caps in parallel to ground. This then goes to a DC jack (wall-wart style) that then feeds the amp. Thus far the PS is NOT cased at all (it's breadboarded for now). As a result, nothing is connected to the mains ground (well, it is through the neutral at the box, of course).
The amp PCB has a ground plane which connects to the turntable ground lug which is isolated from the chassis. The power inlet negative is connected to the chassis and provides the only chassis connection to ground (turntable inputs and ground are on the front of the amp, power inlet and outputs on the back).
Hum is pretty bad, though the sound quality of the amp is pretty good - better, I think, than the batteries.
What have I done wrong? Thanks for any input. I'm a willing student.
It's possible that your design is OK but its physical construction is not great. You want to connect the output ground (to the phono stage) after all of the capacitor connections -- see figure.
The physical wires that conect the center tap and the capacitor minus leads, are not perfect 0.0000 ohm superconductors. There are small (milliohm) resistances in the ground leads; I've drawn then as resistors between "ground A" and "ground B" and "ground C" et cetera. Capacitor ripple currents, drawn as curved arrows, flow through these resistors, so there will be ripple voltages produced. You don't want to pass these along to the phono stage, so you want to physically connect the "output ground" to point D, the very last point in the ground chain.
Maybe you could draw a similar "how it's actually constructed" diagram for the power supply you have now, and show how the center tap is connected to C1, C2, C3, C4, and the output ground.
Thanks, Transistor. How great you drew this - sorry I didn't provide it (and yes, that's exactly what I've done).
It's a bit hard to draw exactly as built as I'm using an old (previously used) Radio Shack prototyping board for the build which has parts going in odd directions because many of the copper pads beneath are shot. The ground is a bus in the middle of the board right next to another bus used for the + lead. But...not all cap negatives are connected directly to this bus. It's a bit of a mess, but functional and I've built like this before and not had problems.
Here's the board: http://www.radioshack.com/graphics/p...2265164reg.jpg
In answer to your question, though, the point at which the ground is pulled from the bus is just before the second cap, so what you say could be a factor. Easy enough to move that and try again.
I also realize the old board might cause some problems, but the hum was louder than I would have expected even for a crappy build. On the other hand, I've only built one other phono stage PS, and that was a bipolar supply for a Hagerman Bugle that was based largely on the PS schematic from Hagerman.
If I can get the hum down to a reasonable level, I'll re-build it better. Any value in throwing in an LM317 for regulation? I have a bunch of them around and considered it but wanted something simple to test the waters.
PS - Oh, and for now I HAVEN'T twisted the transformer leads (duh, forgot to mention that) b/c everything is just floating and it untwists as soon as I let it go. I'll screw it all down to a board and twist the leads as soon as I get to work on it again.
OK, I took a little time to work on this and pulled the amp ground from after the last cap as you suggested, twisted the transformer wires, then found the following:
1. The ground bus trace was broken and the last cap was not connected, so while I got output from my first build, it was with no help from it.
2. After fixing the ground bus, I got output of 18.5v or so using all 4 caps, with an AC component of 0.001v (according to my digital vom). AC had been more like 0.005v when I had only the 470uF first cap and the 2200uF second cap (other two caps were added later and I didn't measure AC component before moving the ground per your suggestion). I do have an old "boatanchor" scope that I'd hoped to use to measure the ripple but couldn't figure out how to get a good measurement from it.
3. Plug it into my phono amp, and I'm still getting hum just as bad as it was before.
4. Power "umbilical" is just two twisted wires (maybe 20-22ga), no shielding, and about 5 inches long, terminated with a DC jack.
5. Battery supply has a small amount of hum compared to the linear supply but it's very noticeable at full volume on my 2W SE tube amp (also DIY) which is otherwise VERY quiet into my 92dB +/- speakers.
Is there any problem having the amp case grounded at the power inlet? Should I be doing anything else with that?
Is there an earth in the system ?
I would connect zero volts to earth at your star ground point.
I have had similar problems when connecting an MP3 player to an amp and found it needed earthing.
On the other hand I had a cd player that didn't like being earthed !
I just tried connecting the PS floating ground (tied to the transformer secondary's center tap with no other connections) to the mains ground leg, but that had no effect at all on the hum. It's still breadboarded so no official star ground - no case for the testing period. Transformer primary just has connection to mains hot (load) and neutral.
Unless that JFET phono amp resembles a discrete op-amp, it just may not be very good at rejecting hum and noise from the power supply. Try the LM317 regulator.
One way to evaluate how quiet the power rail is just listen to it. Use a smallish battery-powered amp with a capacitively coupled input, like a headphone amp, or a small musical instrument practice amp. Or just the oscilloscope with an AC coupled input.
If the transformer is an annoyance, try an AC wall wart from the junk box or a thrift store.
No, I think you're right, it IS discreet jfet and single rail (not op-amp style) and isn't very good at rejecting PS noise. I was hoping brute force capacitance would solve the problem, but no luck so far. I tried one wall wart I had on hand and it was TERRIBLE but I'll probably try another. I will also next try an LM317 just for grins. I like the amp but could do without the hum.
A common method in valve amps is to use RC decoupling.
I have used this to good effect in my bipolar amps on the LTP and VAS stages.
I'm mostly a valve type guy, but I like the jfet approach and want it to tide me over until I can try Kevinkr's Muscovite tube pre. My next step for this will be to try to add an LM317 after the second cap (tuned to 17 or 18v), and add a 5R or so resistor between the 3rd and 4th caps and see what happens.
After some playing around, though, I'm finding the hum is pretty loud no matter what my configuration. Even my Yaqin MS12-B into my Marsh A200 amp gives a pretty loud hum at higher listening levels. Not sure it was that way always, but probably at least somewhat.
Commonalities are that the system is in a bay window of my living room, mostly sharing one outlet (though the TT is plugged into another outlet). There may be 3-5 inches height separating 3 levels of the audio rack (which is 2 shelves wide-6 shelves total). TT into the Yaqin with ground wire connected, then into Marsh amp results in audible hum - may not be as loud relative to the signal as the jfet preamp into the SE tube amp, but it's still pretty loud.
|All times are GMT. The time now is 12:36 AM.|
vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2015 DragonByte Technologies Ltd.
Copyright ©1999-2015 diyAudio