Power supply filtering question/current draw

[midwayfair]

I've got a ... thing ... that runs on a 48V power supply, and when all four circuits in it are running, it will draw up to 300mA. The highest anything in the box will draw is about 100mA, and 50mA is the lowest.

I want to filter the power supply. I have a 48V/30W power supply that can supply 650mA, so plenty of power available from the supply. It's regulated but, as a switching power supply, it needs filtering.

If I do this:

48V > 10R > 2200uF to ground > 10R > 2200uF to ground > 100mA

Ohm's Law and PIE tell me that this gives me only a 2V drop, and I need just under 5W of dissipation (though not much wiggle room there). I don't actually need so small a voltage drop for all the circuits (I'm going to drop them to 30V and 24V), but I am using this as an example.

What I can't figure out is what happens if I use the same PSU for all four PCBs like this:

48V > 10R > 2200uF to ground > 10R > 1000uF to ground > 100mA
     |> 10R > 2200uF to ground > 10R > 1000uF to ground > 24-30V reg > 75mA
     |> 10R > 2200uF to ground > 10R > 1000uF to ground > 24-30V reg > 75mA
     |> 10R > 2200uF to ground > 10R > 1000uF to ground > 24V reg > 50mA

My questions:
1) Are circuits 2-4 going to steal voltage from circuit 1? In other words, do the current draws appear in parallel to each other, or to the 48V PSU?

2) I calculated ~5W above for the 100mA circuit. If I use two 5W resistors, does that actually give me 10W of total dissipation at 20 Ohms in this case? Do I need to worry about the capacitors for purposes of determining the dissipation needed from each individual 10R, or can I still treat them as a series?

3) I was wondering if I could get away with using the same filtering line for multiple circuits. But they'll have different current requirements when in use, correct? So I should definitely keep them resistor-isolated?

[garcho]

sorry about that, I'm getting worse at reading I guess, zheesh.

[garcho]

i get 10mW going through the resistors:

48V > 10R > ... 10R > ... > 100mA

total voltage drop = 48V
circuit resistance = 48V/100mA = 480 Ohms
total series resistance = 10 + 10 + 480 = 500 Ohms
10R = 2% of total R, 480R = 96% total R
voltage drop over each resister is 0.96V, over circuit is 46.08V
4.6W for the circuit, and 10mW for each resistor.

am I way off?

should be equal voltage across all parallel circuits, right? seeing as they all return to the PS.

[midwayfair]

i get 100mW going through the resistors:

48V > 10R > ... 10R > ... > 100mA

total voltage drop = 48V
circuit resistance = 48V/100mA = 480 Ohms
total series resistance = 10 + 10 + 480 = 500 Ohms
10R = 2% of total R, 480R = 96% total R
voltage drop over each resister is 0.96V, over circuit is 46.08V
4.6W for the circuit, and 10mW for each resistor.

am I way off?

You're more likely to be right than I am. I got the same numbers for the wattage (4.6W), the voltage, and the voltage drop across the resistors (YAY I feel like I accomplished something today ... which is sad), but I must have missed a step because I didn't even see what I should be using to get to the 10mW figure. Are you saying 10Rs don't need to dissipate the total wattage of the circuit?

[garcho]

P = I * E  like you stated, but the E is referring to the voltage across the resistor, not the entire circuit. The 10Rs pass the circuit's total current - like all components in the circuit - but due to differences in resistance and voltage, dissipate different amounts of energy.

EDIT: there was a typo that i edited and fixed after you quoted me: it's 10mW (like i mention at the bottom) not 100mW.

EDIT AGAIN: man, i'm really messing this up with the typos and edits. sorry!

[garcho]

no need for parallel sources either, the whole thing has a simple equivalent.

48V, 300mA = 160 Ohms

if you put your same scheme of two 10R resistors in series with this equivalent, you get roughly 0.9W and 0.8W from R1 and R2, and 12.6W across the load. you could use two 2W resistors and skip the parallel thing. Or use four 1W 20R resistors, two in series, two in parallel. Or 8 1/2W 40R or 16 1/4W 80R, etc.

[midwayfair]

10mW just seems crazy low to me. I think I've burned up a 10R 1/4W in a delay circuit running on only 15V before, which I don't think would draw anywhere near 100mA with far less voltage involved. Not that I trust my math over yours ...

I can think of one reason I might want to stick with the parallel: I think it will keep the voltage from changing in the parallel circuits when I switch on another channel. I'll see how the layout for the filtering board shakes out; I left extra room on the PCBs for each effect in case I need to filter them individually.

[garcho]

I think I've burned up a 10R 1/4W in a delay circuit running on only 15V before, which I don't think would draw anywhere near 100mA

15V 100mA = 150Ω load resistance
10Ω + 150Ω = 160Ω total resistance
10/160 = 6.25% total resistance
15V * 6.25% = 0.9375V
0.9375V * 100mA = 93.75mW, a little over 1/3 the rating of a 1/4W resistor

if you short about 100Ω out of the load, making the 10Ω about 17% of the total resistance, you would be maxing out a 1/4W resistor.

[hymenoptera]

I don't understand what all the 10r resistors are for. Is this some SMPS wizardry?

I'm used to seeing mains > PT > rectifier > filtering > regulators > more filtering, and maybe one small current limiting resistor per rail, in rack gear, consoles, etc, but this configuration is new to me.

Is it something unique to SMPS?

[midwayfair]

I don't understand what all the 10r resistors are for. Is this some SMPS wizardry?

I'm used to seeing mains > PT > rectifier > filtering > regulators > more filtering, and maybe one small current limiting resistor per rail, in rack gear, consoles, etc, but this configuration is new to me.

Is it something unique to SMPS?

There's no SMPS involved. It's a 48V DC power supply.

[hymenoptera]

...It's regulated but, as a switching power supply, it needs filtering.

Sorry, maybe I'm just not familiar with the terminology. 

I would think sufficient filtering before and after the regulator, and taking care to bypass as needed, should be plenty.

Consider the various active sections in any one circuit. If one is likely to draw a significant fraction of current you can bypass the supply with an appropriate sized cap to isolate that componant. Multiple circuits will appear no different to the power supply than those individual parts combined. The PSU doesn't care. No one circuit will steal current from another with enough filtering/bypassing and available current.

You can always mock it up, scope it, and look for ripple, oscillation, etc.

[Transmogrifox]

When considering drawing current through a resistor the most direct relationship to use for calculating power dissipation is:

Power = I*I*R

[Isquared R]

Here's how we get to these relationships:
P = VI (or PIE as you call it)

V = I*R

P = V*I = I*R*I = I² * R

OR
if V = I*R, then,
I = V/R

P = V*I = V*V/R = V²/R

You still need to make sure your resistors have a good physical size to handle large startup currents.  Remember the capacitor looks like a short circuit until it is completely charged.  Some resistors have a surge rating in their datasheet. For some you are safer to over-engineer it and use 2 Watt resistors so you know you aren't causing small amounts of damage every time you turn it on (resistance changes value if there is enough heat, even if it's not enough to make smoke)

If using several 1000's uF capacitance you may not even need the resistors.  Just connect your linear regulator circuits directly to the 48V.  The regulators will reject a lot of noise too -- probably a lot better bang for your buck than resistors and large-valued capacitors rated for this voltage.  Then you can put capacitors at lower voltages and depending how low you end up dropping the voltage in the end, could go to lower voltage ratings on the caps...which are a lot cheaper...but then maybe you already have these capacitors and it costs the same whether they sit in your drawer or your 48V supply  .

[tubegeek]

I have a 48V/30W power supply that can supply 650mA.... It's regulated...

That gives us Assumption #1: the regulator is designed OK and gives minimal voltage change between 0 current and full rated current. In other words, a "stiff" or low impedance supply.

Based on that assumption, no "stealing" would be happening. In the case of the multiple parallel connections (with series elements in each parallel circuit) you are really worried about power supply "sag," - to my mind, I think "stealing" would refer more to what happens when you tap off a power supply in different series spots (like the various filter nodes in a tube amp.)

1) Are circuits 2-4 going to steal voltage from circuit 1? In other words, do the current draws appear in parallel to each other, or to the 48V PSU?

They appear in parallel to each other AT the 48 V PSU.

2) I calculated ~5W above for the 100mA circuit. If I use two 5W resistors, does that actually give me 10W of total dissipation at 20 Ohms in this case? Do I need to worry about the capacitors for purposes of determining the dissipation needed from each individual 10R, or can I still treat them as a series?

Each R in series will carry the total current drawn by that subcircuit's load only, say, 100 mA in your examples. Each one will be required to dissipate .1 x .1 x 10 = .1 W. (P = I^2*R)

If we use another formula to double check, we have a voltage drop of V=I*R, V= .1 x 10 = 1 V per resistor, then P = I*V = .1 x 1 = .1W.
OK.
How did you get 5W?

3) I was wondering if I could get away with using the same filtering line for multiple circuits. But they'll have different current requirements when in use, correct? So I should definitely keep them resistor-isolated?

If you hang all the circuits onto one set of filters, then the voltage drop through the filter will depend on how much current is being drawn at that moment - if  Load A, B, C, & D are connected, or just A & B, etc.

If you give each load a separate filter then the maximum voltage drop when all loads are active will be smaller and will not depend on how many loads are connected (because the regulated power supply will take care of making sure your input to the filter is consistent.)

The caps act as open circuits for DC so we can ignore them completely in this DC analysis.
They act as short circuits for high-enough frequencies of AC and so they will filter the noise and ripple.

Think about the difference between these two drawings and especially what happens to the bottom one when some of the 460 ohm resistors are switched in and out.

[PRR]

> as a switching power supply, it needs filtering.

Yes, perhaps; but not against 120Hz crap, against many-KHz crap.

Which means 1,000uFd (a typical 1200Hz-crap filter) is probably too large. The impedance of an ideal 1,000uFd at 20KHz is <0.01 Ohms. The impedance of an inch of can and a couple inches of lead will be more than that.

Take a hasty figuring. 10 Ohms series and 1 Ohm shunt gives 10:1 filtering which aint bad. 1 Ohm at 20KHz is closer to 10uFd! And a 10uFd will be much smaller and can fit closer to the points being filtered.

Looking at the same reactance chart, I see that a 100uH coil gives the same 10 Ohms series impedance. Can you find a 100uH 100mA coil with less than 10 Ohms DC resistance? If so, and if cheap, you get lower DC drop, also even better filtering above 20KHz.

It would be good to know your "thing". Some things just don't mind supersonic (typical SMPS) crap. Some things just freak-out. Some freakers can be tranquilized with a little insight.
__________________________________________

> I need just under 5W of dissipation

And learn from your Uncles Transmogrifox and the other guy.

1) HOW is the 10 Ohm resistor gonna know about the 48V?? It can only know about the 100mA through it, the 1V across it, the 100 milli-Watts in it.

1a) If the "10r" resistor felt 48V and 100mA(=0.1A), then it must be a 48V/0.1A= 480 Ohm resistor. So which is it? Always work your I=E/R V=I*R R=E/I problems all ways to check for non-sense.

2) If you did have a 4.8 Watt actual dissipation, a "5W" part is NOT good for more than bench-tests. Life can be a thousand hours, which means it will not work right at some Critical Gig next year. Double the (properly) calculated dissipation and round-up when you buy the resistor.

3) If you have 10 Ohms from 48V feeding a cap, at turn-on the resistor DOES feel the full 48V, until the cap charges up. 1,000uFd takes a "long time" to charge-up through 10r. Only 0.01 seconds to 30V (18V across R), but 0.01 seconds is enough to fully-cook thin film and small bulk-carbon resistors. 10uFd at least lessens the cook-time.

[midwayfair]

Okay, lots of really good points over night.

--

>It would be good to know your "thing". Some things just don't mind supersonic (typical SMPS) crap. Some things just freak-out. Some freakers can be tranquilized with a little insight.

Before I do: I wanted to estimate a worst-case scenario, and 100mA was a nice round number that I knew was higher than actual usage. The current draw is still higher than I'm used to with pedals, and usually nothing bad comes from overpreparing, right? I also wanted to keep my question more general because clearly I don't have a handle on this part. I've never had to learn it for 9V pedal components.

So, sorry for being vague. The best I can do is a block description at the moment, because I'm still tweaking some things. It's just four line drivers for my interface (and for my PA that has a couple line in/no pramp channels and no phantom power), meant to run a couple mics and a couple instruments. The basic circuit is similar to:

(1) FET gain stage > output Buffer like the ROG Omega (but the buffer is taken right from the drain) > (2) [Optional passive tone stack] (3) a tweaked Hamptone preamp stage (this: http://home.comcast.net/~markfuksman/hamptone.jpg). The gain stages are biased at the gate/base so that it can accept any (voltage-appropriate) a BJT or JFET. The Hamtone article gives 40mA as the recommended supply amperage for the module. That buffer in it seems to draw a HUGE amount of current, enough to make me worry about the heat coming from some smaller transistors, and when I was running it from an LT1054 for 24V, it dropped the voltage by 5 volts, and the LT1054 supplies 100mA! Changing out the buffer after gain stage 1 to a simple emitter follower, and/or using a BJT gain stage drops the mA drawn.

I decided on this setup because I wanted to have as much overlap as possible between variants built on the same PCB. (The components are present get the different sounds I want.) The outputs on all four (and the inputs on the mic inputs) are transformer coupled and I'm using the Edcor XSM series because I needed something flat 20-20KHz (the WSM series drops off around 15KHz). The frequency plot for those transformers have an uptick around 50KHz: e.g. the input xformer https://www.edcorusa.com/xsm600-10k (plot is the same for 600:600).

And ... I didn't have any caps in the house as large as what I was talking about in the first post, and I also didn't have any high-wattage low-resistance resistors either. I wanted to try to understand this before I started killing components or power supplies.

(1) I'm not getting any freaking out (I imagine an op amp would probably care more, right?). (2) I know I could strap a small cap across the transformer to correct for the Xformer bump in supersonic range if needed. (3) I didn't realize that the filtering caps were only for filtering super sonic range, not for 60/120Hz!

---

>If we use another formula to double check, we have a voltage drop of V=I*R, V= .1 x 10 = 1 V per resistor, then P = I*V = .1 x 1 = .1W.
OK.
How did you get 5W?

I thought "voltage across the resistor" (the wording on Wikipedia) was the 48V minus the voltage drop across the resistor, not just the voltage drop across the resistor (it was unclear), so I was getting about 4.8W or whatever. Since I was getting a number that would imply that a 1/4W resistor was okay and in personal experience it wasn't enough, the 5W number made more sense. More on that at the bottom.

---

>In the case of the multiple parallel connections (with series elements in each parallel circuit) you are really worried about power supply "sag."

Yes, "sag" is what I meant -- the available voltage dropping when the current pulled is more than the PSU can handle. If I do drawing #2, then all the current loads draw from the lower voltage after 20R, and the voltage after the 20R changes when all four circuits are on compared with just one. I wanted to avoid that. While I'm not sure exactly how stiff the power supply I have is, I picked one that supplies more than twice what I need int he worst-case-scenario. It's definitely stiffer before the 20R than after, right?

---

>The regulators will reject a lot of noise too

I'm using a Zener regulator to drop it to 30V or 24V, like this: https://upload.wikimedia.org/wikipedia/commons/thumb/c/c3/Voltage_stabiliser_transistor%2C_IEC_symbols.svg/415px-Voltage_stabiliser_transistor%2C_IEC_symbols.svg.png

I used this regulator because needed something that could handle the 48V input and drop it to different levels without changing my PCB layout. Do those reject noise like the L78 ICs? I'm not sure. The L78xx still need caps after them, and I would have needed to drop the voltage before them anyway.

---
>Can you find a 100uH 100mA coil with less than 10 Ohms DC resistance?

Not exactly those specs, but 6.6Ohms, 300mA, 100mH:

http://www.mouser.com/ProductDetail/EPCOS-TDK/B82731T2301A020/?qs=pGJ4H8VyKtUQCUonLeVspg%3d%3d

All the inductors closer to 100mA had much much much higher resistance.

It's actually surprisingly small and the resistance is tiny. I could see putting it right after the power supply with a single cap for the whole box. With a 470uF cap following, the peak is at 22Hz and the cutoff is 23Hz.

---

Okay, both PRR and Transmogrifox mentioned the problem with "large startup currents."

The 1000uF weren't going to work with the layout anyway. They're even bigger than I expected (16mm diameter are the smallest I can find above 50V).

So what do I do about the large startup currents without making the resistors gigantic (either value or physically) and losing a ton of voltage? Does it work put a cap right at the power supply input, before any of the small resistors, so that the cap shorts the power supply at startup and charges before the voltage hits the 10R?

470uF at 63V are 10mm diameter. I can work with that. Should I go smaller? 100uF? 10uF? (That seems awfully small, 1.5Khz.)

How did you get the time it takes before the cap fills up?

[PRR]

> I didn't realize that the filtering caps were only for filtering super sonic range, not for 60/120Hz!

I am ass-uming that your "switching power supply" smooths the 120Hz ripple well.

Most general-purpose SMPSes do.

If you use an oddball, like an LED lighting "transformer", bets are off. (However even these should not have huge 120Hz ripple, because that would make the LEDs flicker.)

But to be nimble enuff to fight 120Hz they have to run at a high frequency. Audio range would whine. Also higher frequency means smaller transformer and caps and thus lower cost. So SMPSes usually run 20KHz-100KHz (or higher in a few cases.

> using a Zener regulator .... Do those reject noise

The "dynamic" impedance of a 24V Zener at 10mA is 25 Ohms. The resistor is 24V/10mA or 2,400 Ohms. 2400/25 is about a 100:1 reduction of input voltage change.

Whether that is "enough" depends how ripply your SMPS is, and how sensitive your "thing" is.

The Hamptone may have very slight supply rejection.

Since you are throwing-away almost 24V, at 40mA, you *could* pencil a 24V/40mA= 600 Ohm resistor, and then a capacitor. Since the regulator transistor needs some headroom, and current demand is NOT sure, I'd pencil half of that, or 250 Ohm standard round-value. This will lose (0.040A^2)*240r or 0.4 Watts, use a 5 Watt to be very conservative (they are cheap). Then 10uFd after the 250r will give a lot of supersonic rejection. (A 36V rail will need re-computation.)

[tubegeek]

If you are thinking of building your own custom SMPS (I'm not clear on whether you are doing the SMPS off-the-shelf or custom?) I would recommend looking at the power conversion chapter I mentioned here:

http://www.diystompboxes.com/smfforum/index.php?topic=112122.msg1034012#msg1034012

There is stuff in there about improving Zener regs, and I have barely scratched the surface of that chapter. Just email them and they will hook you up with download info.

[midwayfair]

TubeGeek: Thanks, I'll check it out. I'm using a PSU but I might end up building one.

----

So I took some more measurements last night.

One problem here is that the current consumption of the Hamptone circuit varies depending on the supply voltage. And it's 100% related to that output buffer. Even using only one of them (the other half is the Omega) stock this is what I got at 48V:

...

Oh, actually, that fried the 100R 1/2W on the emitter of the bottom half of the buffer. And a 470R 1/4W.

FIRE! Very exciting.

Clearly more wattage is needed.

I don't have a 2W 100R, so I upped the emitter resistor to 470R 2W, things smell a lot less like burning, and took new readings:

47R on the power supply: 4.2V drop = 89mA (this is close enough to what I guestimated as my worst case scenario).
470R on the power supply: 23V drop = 49mA (and this is what I got as the consumption just running it on 24V ... which makes me feel less crazy for guessing 100mA ...)
1K on the power supply: 32V drop = 32mA
With a 680R in parallel with the 470R (only other 2W value I had around): almost 100mA exactly.
BJTs (I think they're 2N2269, but they aren't marked) in place of FETs @48V with 470R emitter on the output buffer: 47R: 1V drop, 21mA (about what I expected)

I don't have any zeners above 1/2W on hand to try the Zener regulator. I'm a little scared of fire at this point.

>you *could* pencil a 24V/40mA= 600 Ohm resistor, and then a capacitor. Since the regulator transistor needs some headroom, and current demand is NOT sure, I'd pencil half of that, or 250 Ohm standard round-value. This will lose (0.040A^2)*240r or 0.4 Watts, use a 5 Watt to be very conservative (they are cheap). Then 10uFd after the 250r will give a lot of supersonic rejection. (A 36V rail will need re-computation.)

Just to be clear: You're saying I use the 250R and 10uF BEFORE the zener regulator, right? Because the voltage drop from it not only doesn't matter but is actually beneficial and puts less strain on the regulator? And bigger cap is okay after the zener because the 250R is helping with that, right?

For the one I want to run on as close to 48V as possible -- If I want to use it with the 2SK30 FETs (which draw a lot of current even with big source and drain resistors), should I just leave off the filter resistor and rely on a bigger capacitor by itself (because anything else will fry on startup)? Should I go looking for some wirewound 10W resistors with very tiny values? Or should I say *!$^! it and use the BJTs in that one since they only draw 1/5 as much current?

[PRR]

> 89mA

How hot are the *transistors* getting?

If you can't touch them for a 10-count, increase the resistor to the lower Base. There is NO reason to be running even 20mA here. You aren't driving transcontinental phone lines.

> If I want to use ..., should I just ...? Should I go looking....

You are not the first to follow this path. Plagiarize!! Study every small-signal audio amplifier plan you can find. Most of your ??s relate to power distribution and filtering, so look at that part of the plans.

Same as you learned music. You didn't start writing songs without ever hearing music. You had a record-player, transistor radio, or iPod, you heard hundreds of musical pieces, you learned some structures, rhythms, harmonies which come up again and again because they work. Benefit from past experience of others.

[hymenoptera]

I think he's building something like this

http://i222.photobucket.com/albums/dd317/awlivoyl/SSEGM3.gif

With the 390R 2W the maximum current should be limited to no more than 1/8th watt (48v/390r=0.123A)*

I can't see how the 100r emitter resistor blew unless there's some mods done that we don't know about.

*edit: wrong wrong wrong. I'm sorry, I've been sick and was running a fever of 102 degrees when I wrote that. Please disregard! I have no idea what I was thinking there.