Power supply schematic? please critique.

[mike_mccue]

Hi,
I'm trying to build a 2 channel microphone preamp. I posted a schematic for a power supply a while back and there were many suggestions for revisions etc. It took me a while to understand all the suggestions.

I'm still not sure what the correct wattage ratings are for several resistors. I've marked them in red. Can someone advise on that?

I'm a bit unsure about the R23. I mainly do large easy to see point to point type wiring... working with the LM317 is new to me... I plan to order a few of those devices and experiment with breadboard until I have it figured out. With R23, should I use a small trim pot, a larger pot with solder lugs and some gooey glue to lock the shaft, a group of resistors to get the exact rating.

Thanks for any help. I'm eager to order the remaining parts that I need and start a physical layout, The past few attempts have been stalled because I find I need to learn something and then I start over. :-)

best regards,
mike

[Merlinb]

Originally Posted by mike_mccue

Hi,

I'm still not sure what the correct wattage ratings are for several resistors. I've marked them in red. Can someone advise on that?

R23 will dissipate nearly 0.3W, so a 0.5W resistor is in order.
R22 can be a 1/4W

It is impossible to find the wattage ratings for R20/21 without knowing how much current will be demanded. However, since it's just a preamp 0.5W should be fine for both (that would allow up to 20mA total current, which is probably way more than you need, I would guess?)

I'm not sure what R24 is even there for? 2W is also much larger than necessary. You may want to make C04 larger, but C05 smaller. Say, 470u and 10u.

I'm a bit unsure about the R23.

You don't actually need precisely 48V for phantom power. 48V is the max, mic's will still work fine at much lower voltages.
A standard value of 8.2k for R23 will give you 46V out, which is fine.

You may want to put a 33V 0.5W zener (or similar) diode between input and output of the LM317, to ensure its voltage ratings aren't exceeded at start-up/switch-off.

[tedmich]

Originally Posted by Merlinb

R23 will dissipate nearly 0.3W, so a 0.5W resistor is in order.
R22 can be a 1/4W

It is impossible to find the wattage ratings for R20/21 without knowing how much current will be demanded. However, since it's just a preamp 0.5W should be fine for both (that would allow up to 20mA total current, which is probably way more than you need, I would guess?)

I'm not sure what R24 is even there for? 2W is also much larger than necessary. You may want to make C04 larger, but C05 smaller. Say, 470u and 10u.


You don't actually need precisely 48V for phantom power. 48V is the max, mic's will still work fine at much lower voltages.
A standard value of 8.2k for R23 will give you 46V out, which is fine.

You may want to put a 33V 0.5W zener (or similar) diode between input and output of the LM317, to ensure its voltage ratings aren't exceeded at start-up/switch-off.

looks pretty good, I'd throw fuses on all the secondaries though.For the LM317AHV I'd up the initial cap to 1500-2000uf and then with 10uf/10uf tantalums from both Out/Adj pins to ground, you'll have ~80dB ripple rejection which should be plenty. The adjustment resistors can be tiny, 1/4-1/8w and the 220 can even be a smd straight on the leads!

Unless I had the caps already I'd be tempted to go with all 100uf ELs on the HV rail, since nv ripple on B+02 might be overkill; I like CDE 381LX series, with 12000h lifetime higher voltages just derate it a little

[Enzo]

R25 is a mystery to me. Why is it there? The lamp (I assume?) appears to be across 120VAC. SO either there is a 120v lamp there and the resistor is for dimming it a little? That seems uneccesary. Or the lamp is neon and it needs a resistor to limit current. If that is so it needs to be more like 220k or something, not 1k. or if it is a commercail neon indicator, which has the resistor internal, then teh extra 1k is pointless.

[mike_mccue]

Thank you all very much.

re: Enzo,
I think a friend had suggested a LED and I have been planning on pilot light like on a fender lamp. I'll use a 120vAC lamp and I will just remove R25. Thanks.

re: Merlin,

Thank you for the advice on the wattage. I've learned the LM317 layout from Mr Steve Connor and it's all new to me. R24 was specifically suggested. I think I took the capacitor values from Steve's webpage. It seems like you are thinking about this in detail.
It has been observed that this schematic is over designed... and I accept that critique and often times I share that opinion.
I keep adding safety features... features that I don't think many of my commercially made amps have. Do you think the 33V 0.5W zener diodes are more like very best practice or is it a routine addition? I think the fuse and R24 were meant to protect the LM317 at start up. I get confused easily.

re: Tedmich,
As I mentioned, the LM317 is new to me. I had originally planned to use a discrete component solution. I thought the LM317 was supposed to lock down most of the ripple and voltage all by itself. I think I understand your suggestions. As I suggested above, I'm wondering when I have gone beyond practical needs? Can you advise?

The very large caps in the high voltage are a legacy from mine and a former mentors original design goal of making a simple tube mic preamp with a overbuilt power supply. I've already purchased the large caps so I'd like to use them.

Thanks again everyone.

I've updated and uploaded the latest schematic.

mike

[Merlinb]

Originally Posted by mike_mccue

Do you think the 33V 0.5W zener diodes are more like very best practice or is it a routine addition? I think the fuse and R24 were meant to protect the LM317 at start up. I get confused easily.

Standard practice.
Like most 3-terminal regulators the LM317 is only rated for 35V maximum voltage across itself (it might be a little higher in some models).
Since you're designing a 47V supply your input voltage will certainly be more than 35V, and at switch on the LM317 will have to withstand it!
Adding a zener in parallel prevents the voltage across the LM from exceeding its ratings, by providing a short-cut path for charging the output capacitor. At switch-off it also ensures the output cap can discharge if it needs to.

[mike_mccue]

I should have been more specific while I was writing... the schematic actually specs a high voltage version of the 317. It's a Fairchild LM317HV.

I had been thinking the LM317 was the simpler, cleaner way to go.

If I add the zener and the tantalum caps as suggested above it by tedmich it seems like the layout will be more complicated than I had imagined.

I'm rethinking my original plan to mimic a bias supply and use discrete components to build a 48v supply.

I like to build little eyelet boards and I was already stressing working with the LM317 hole thru components. I imagined I was going to burn them up.

best regards,
mike

[tedmich]

Originally Posted by Merlinb

Standard practice.
Like most 3-terminal regulators the LM317 is only rated for 35V maximum voltage across itself (it might be a little higher in some models).
Since you're designing a 47V supply your input voltage will certainly be more than 35V, and at switch on the LM317 will have to withstand it!
Adding a zener in parallel prevents the voltage across the LM from exceeding its ratings, by providing a short-cut path for charging the output capacitor. At switch-off it also ensures the output cap can discharge if it needs to.

LM317HV like he has listed (and LM117HV) are rated to 57VDC

[tedmich]

Originally Posted by mike_mccue

I should have been more specific while I was writing... the schematic actually specs a high voltage version of the 317. It's a Fairchild LM317HV.

I had been thinking the LM317 was the simpler, cleaner way to go.

If I add the zener and the tantalum caps as suggested above it by tedmich it seems like the layout will be more complicated than I had imagined.

I'm rethinking my original plan to mimic a bias supply and use discrete components to build a 48v supply.

I like to build little eyelet boards and I was already stressing working with the LM317 hole thru components. I imagined I was going to burn them up.

best regards,
mike

LM317 (or better still LM338!) work very well, and just require a little filtering to give huge ripple rejection ->nV!!

Just follow the National design guides. Another potential shock (pun!) is the heat sink tab on the TO-220 is common to V out! ie it LIVE and requires an insulating pad between it and the (completely required) heatsink.

[jrfrond]

It looks good to me, but you really DO NOT need a tube rectifier in a preamp. It's not a guitar amp, and you won't be drawing enough current to cause sag, so to me, it's pointless. At that point, you can bump-up the first filter cap before the choke to 80uF as well. What is of paramount importance is a good, stiff, clean power supply. You may also want to consider DC heaters, just to keep it as quiet as possible.

If, by chance, the idea of the tube rectifier was to provide warm-up time for the preamp tubes and minimize cathode-stripping, that function can be done with a series inrush current limiter in the AC line input.

[mike_mccue]

"It's not a guitar amp, and you won't be drawing enough current to cause sag, so to me, it's pointless."

Thanks John,
Ok, I have to admit... it's an indulgence.

I really need to start making something... and you guys have kept me running in circles. All the suggestions are 100% appreciated... but as a hobbyist it gets overwhelming when trying to maintain focus.

I'd like to stay with the tube rectifier for the sake of indulgence. I'm still open to replacing the 500uf caps for something smaller but I already have them sitting here.

I've heard of DC heaters supplies... I'll go read about them.

I went and read about the LM317 full layout on datasheets and understand tedmich's above suggestions now.

I want to revisit making a 48v from discrete components. I'm going to start drawing something up today.

thanks very much,
mike

[mike_mccue]

Here's is a schematic that just focuses on using the 50vAC bias tap to get 48vDC with components I may find easier to assemble.

Can anyone advise with the resistor ratings? I've been reading a few pages like this: Power Supplies that offer equations, but I added the second filter and don't know what to make of that situation.

How about other protection elements?

I learned a lot about the LM317 but I'd like to finish the original idea of working with discrete components which I was working on before I was introduced to the idea of them so that I can consider which is better suited for my DIY project.

Thanks again for all the help.

mike

[Merlinb]

Originally Posted by mike_mccue

Can anyone advise with the resistor ratings? I've been reading a few pages like this: Power Supplies that offer equations, but I added the second filter and don't know what to make of that situation.

The phantom power requires practically no current, so you really only need to spec according to the current you want to flow down the zener string. I would guess 10-20mA?

Since you have 70Vpeak input, and 48V output, voltage dropped across the resistors is 22V. To allow 10mA:

22/0.01 = 2200 ohms. So two 1k resistors would probably do (R44/45).

Total power dissipated:
22^2 / 2000 = 242mW Or 121mW in each resistor, so 1/4W resistors could be used.


However, if you're going to use a simple 10-20mA zener design then you might as well just supply the zeners directly from the B+ supply, rather than a separate half-wave rectifier. (Half wave rectifiers are not good for transformer health)

[mike_mccue]

Thank you for the reply.

I value your opinion about the half wave and the suggestion to move the connection, but I already have the transformer and purchased it because it had the 50vAC tap. Now that I know about the option I'll probably try using your suggestion on a future project.

I'd like to plan for 40ma max draw.

So is that:

22v /0.04a = 550ohms

If so I can use 2 - 270ohm resistors

22v^2/540 ohms = 0.896w or .448w per resistor.

1 watt resistors?

Do you think the Zener stack that I have is proper for the 48v output? It was suggested by an friend who never fully explained how the zeners work as a regulator.

I have been off reading what I can find... but I really haven't had an epiphany and fully realized how they maintain a constant voltage.

I see in my mouser catalog there is a zener with a 47v output... would that be suitable instead of the stack?

Is the fuse adequate for protection? Is there any extra protection necessary?

I've uploaded the revised phantom power schematic as well as the latest power supply schematic incorporating the protection diode and caps for the LM317HV.

Thanks very much,
mike

[jrfrond]

The LM317 is pretty standard in many phantom supplies. It is not complex, and I would probably stick with that design, but rather use fixed resistors instead of making adjustable. Phantom voltage is just NOT that critical.

I wouldn't use half-wave rectifiers anywhere in the circuit. If you use full-wave, you can drop your filter capacitance and lower the cost of your project.

I still maintain that the tube rectifier is a waste of time, but if it makes you feel better to use it, it's no big deal. A couple of 6A10 diodes is much cheaper. Yes, I know 6A10's are WAY overrated (6A/1KV) for even most guitar amps, but for the few pennies difference, I always go with the heaviest-duty diodes in critical supplies.

BTW- Zener diodes work as a regular diode until you reach the "avalanche" voltage, whereas they go into reverse conduction, and maintain the voltage at that point. The down side as opposed to an IC regulator is that a simple zener regulator draws current by virtue of it's design. If you didn't have a current-limiting resistor in series, it would pretty much be a dead short. IC regulators are much more efficient and precise, as well as inexpensive, which is why you don't see zener regulators being used much anymore. Personally, from a service standpoint, I don't like the reliability factor of a series zener string.

[mike_mccue]

Thank John,
I appreciate your opinion.

My concern with the LM317 is that I am not experienced working with small heat sensitive parts and small traces on hand made PCBs.
I really enjoy working with eyelet boards that I make on my workbench and working with larger easy to see stuff.

I take your opinion about the tube rectifier at face value and agree that your reasoning is correct.

best regards,
mike

[Merlinb]

Originally Posted by mike_mccue

My concern with the LM317 is that I am not experienced working with small heat sensitive parts and small traces on hand made PCBs.

The 317 would supply so little current that heat really isn't a problem- we're talking milliwatts.

[mike_mccue]

I am speaking of my solder and assembly skills.

I'm still open to using the LM317, I juts need to buy a few and attempt a build I guess.

best,
mike

[Merlinb]

Originally Posted by mike_mccue

I am speaking of my solder and assembly skills.

They're probably not as wimpy as you think, you have to try pretty hard to solder one to death! (Unlike MOSFETs. Stupid MOSFETs)

[mike_mccue]

so here's the $10,000 question:

Do you think I can spread the legs of one to accommodate a fit with a eyelet board made with garolite and keystone eyelets?

I imagine the minimum spacing for the eyelets is wider than the standard package meant for PCB thru hole mounting.

For the past month I've been imagining having to use a breadboard or something for the phantom power ortion of the power supply circuit... and I suppose looking for excuses to avoid doing so.

best regards,
mike

[Merlinb]

If you can't spread the legs then you could just bolt it to the chassis and use wires to run to-from the board. Nothing unusual about that.

[Steve Conner]

Sure you can, I've mounted lots of TO-220 devices to tag boards and the like. The legs bend quite a bit, and you can solder on bits of wire to extend them if they don't go far enough.

[mike_mccue]

Thanks very much to both for the continued help.

How about the heatsink? There was some mention above about an insulator?
When I order a few LM317AHV will I also need to order a few heatsinks as well as insulators?

Can some one take a last look at the most recent schematic *20a.gif to see if I got the Phantom power correct finally?

Thanks again,
mike

[Merlinb]

Originally Posted by mike_mccue

Can some one take a last look at the most recent schematic *20a.gif to see if I got the Phantom power correct finally?

I can't see any problems with it. Might want a fuse in the B+ though...

[mike_mccue]

On the center tap?

I'm off to read your Valve wizard site to see what you recommend.

Also, did I miss the idea that you previously stated that heat is not a problem... does that mean I don't need a heatsink?

Thanks,
mike

[tedmich]

Originally Posted by mike_mccue

Thanks very much to both for the continued help.

How about the heatsink? There was some mention above about an insulator?
When I order a few LM317AHV will I also need to order a few heatsinks as well as insulators?

Can some one take a last look at the most recent schematic *20a.gif to see if I got the Phantom power correct finally?

Thanks again,
mike

The Lm317 MAY run without a heatsink, it has thermal shutdown built in, but a small extruded Al one and a mica pad (both available at radioshack) is cheap insurance; with any current they can get very hot. Heres one I built into a copper pipe cap


notice 20 turn trimmer, I potted it up once I was sure it worked correctly.

[mike_mccue]

Thanks tecmich,

It's helpful to see how you did that without a PCB. The mica is the electrical insulator, but it transfers heat? I guess it's sized to the standardized "package"?

re: Merlin's comment about the fuse on the HV.

I've added two fuses to the High Voltage AC before the rectifier.

I read Merlin's page that suggested that you can determine the amperage by experimenting first hand. Is there a practical way to predict a useful lowest amperage value?

I imagine a 500vAC rating is a good idea for whatever I end up with. Yes, no?

Here's the latest tube-preamp-21a.gif schematic.

best regards,
mike

[tedmich]

Originally Posted by mike_mccue

Thanks tecmich,

It's helpful to see how you did that without a PCB. The mica is the electrical insulator, but it transfers heat? I guess it's sized to the standardized "package"?

yes!
IME fuses are usually std. 250v slo blo units at about 1.5 - 2x the amps the circuit is expected to see

[mike_mccue]

Thanks. I'm working up a parts order.

best regards,
mike

[Enzo]

Here is a part that works like an LM317 but will take up to 125v directly.

TL783. I noticed it doing exactly this job in a mixer power supply.

Mouser sells them for under $2.

http://focus.ti.com/lit/ds/symlink/tl783.pdf

[tedmich]

Originally Posted by Enzo

Here is a part that works like an LM317 but will take up to 125v directly.

TL783. I noticed it doing exactly this job in a mixer power supply.

Mouser sells them for under $2.

http://focus.ti.com/lit/ds/symlink/tl783.pdf

Nice find Enzo! I have been keeping track of HV regulators, the Supertex LR8 will handle up to 450VDC but only 10mA:
www.supertex.com/pdf/misc/LR8K4PSS.pdf

And the audiophiles love Michael Maida's LM317/MOSFET regulator:
www.national.com/ms/LB/LB-47.pdf

or the Swenson regulator, fine tuned by Gary Pimm to handle 800v ~1A, its great for bench supplies and costs $15 in parts from Mouser:
Gary's web site:
PimmLabs
click on Swenson reg on left.

layout/bom photo
http://home.comcast.net/~pimm1028/we...or_hv_mini.zip