PSU series resistor, part 2: can the resistor be too small?

[aion]

Yesterday I had a question about what happens when a PSU CLR is too large, and how the voltage drop curve changes as the resistance increases.

That's one end of the extreme, what happens when the CLR resistor goes far above the limit of practicality. What about the reverse? Are there dangers of using too-small resistors, like 10R or less?

Based only on the formulas, they would be less susceptible to everything since they barely resist the current flow. Op-amps aren't greedy like LEDs, so they don't need their current limited externally. A low CLR would not cause the current draw to increase over what the op-amp or other active devices are trying to draw.

But, I have also heard anecdotes of people having issues with ~10R PSU resistors burning up even in relatively small 9V op-amp circuits with a typical current draw, say 3 duals. It was speculated to be due to the inrush current as the filter capacitor is charged during power-on.

This would assume an otherwise correctly-wired circuit... obviously there are many other issues if we add reverse polarity or shorts to the mix.

Does the math check out on this? Could it be possible that a combination of a resistor that is too small with a capacitor that is 'too good' (meaning low ESR and high value) may cause issues?

I have always stuck with either 47R or 100R or no resistor depending on the current draw and have never experienced anything like this. But I've had it in my mind that 10R could be bad news in some cases and I am trying to understand if that actually makes sense or not.

[ElectricDruid]

It's already come up in the other thread, but the big problem with making the resistor smaller is that it's part of a lowpass filter, so if you go down from 100R to 10R, you need to bump up the cap value to compensate. So your 100u(?) cap becomes 1000u. And that's getting physically large and quite a bit more expensive. It makes more sense to use a slightly-larger resistor that costs cents rather than a big chunky and expensive cap to do the same job. But it's a balancing act - we want the smallest possible resistor to limit the volt drop, and the largest possible resistor to get the best filtering. So what value to choose?!? Maybe somewhere in the decade 10R to 100R?

[antonis]

At extreme cases, a small resistor and a capacitance multiplier should be the solution..

https://audioxpress.com/article/the-capacitance-multiplier

[ElectricDruid]

That capacitance multiplier idea turns up in a few old Boss schematic, iirc. Can't name them off the top of my head, but something with a BBD in would be a good bet.

[aion]

It's already come up in the other thread, but the big problem with making the resistor smaller is that it's part of a lowpass filter, so if you go down from 100R to 10R, you need to bump up the cap value to compensate. So your 100u(?) cap becomes 1000u. And that's getting physically large and quite a bit more expensive. It makes more sense to use a slightly-larger resistor that costs cents rather than a big chunky and expensive cap to do the same job. But it's a balancing act - we want the smallest possible resistor to limit the volt drop, and the largest possible resistor to get the best filtering. So what value to choose?!? Maybe somewhere in the decade 10R to 100R?

Yeah, I guess I'm thinking about it just in terms of the extremes to avoid, independent of specific context: are there any circumstances under which a 10R resistor can become a fuse because it's too low (for example not restricting a capacitor's inrush current enough) when a slightly larger resistor would be OK in the same position?

It would assume that we would otherwise have a good reason for selecting 10R, perhaps because larger values drop more voltage than we would want in the circuit, and that the 10R resistor is otherwise of the appropriate wattage for the operating current of the circuit. But again, as with the other question, we are operating well outside the window of practicality.

That capacitance multiplier idea turns up in a few old Boss schematic, iirc. Can't name them off the top of my head, but something with a BBD in would be a good bet.

FZ-2 and BD-2 are the two that first come to mind, though I've seen it in some other more obscure ones as well.

[mozz]

If you make the resistor 10 times smaller, you don't "have" to make the cap 10 times larger, chances are you have enough ripple filtering already on your power supply. All your pedals power are all in parallel anyway.  I can't see anymore than a 25uf-47uf being useful, just taking up more space. 

[ElectricDruid]

FZ-2 and BD-2 are the two that first come to mind, though I've seen it in some other more obscure ones as well.

Thanks. I wonder which one I had in mind?!? Or maybe I'm just getting stuff confused..

If you make the resistor 10 times smaller, you don't "have" to make the cap 10 times larger, chances are you have enough ripple filtering already on your power supply. All your pedals power are all in parallel anyway.  I can't see anymore than a 25uf-47uf being useful, just taking up more space. 

I don't agree. 25uF-47uF on its own doesn't define the cutoff frequency, so I don't see how you can decide it's "useful" or not. And yes, if your power supply filtering is alrady fantastic then all this stuff we're doing inside the pedal is totally wasted. Lucky you.

[mozz]

 I don't think many power supplies these days are loaded with AC ripple. You either have enough filtering or not. Any second,third, fourth, etc pedals with resistors and capacitors are all in parallel so it all adds up. If anything i put a .1uf or .01uf across the power input. I'm not worried about any cut off frequency, if you are you need a better power supply.

[PRR]

> can the resistor be too small?

The lights   in my house dim when the well-pump starts. This is because I have 0.4 Ohms in my long skinny 500 foot feeder line. 40 Amps times 0.4 Ohms is a 16V sag, which was significant lamp-dim when it was all incandescent lamps (even cheap CFLs).

But that also means that a DEAD SHORT at my end of the feeder is "only" 600 Amps (0.6KA) of fault current. "Only" 157KW of arc-power. 

OTOH down in the suburbs of Texas they put four homes around one extra large power transformer with 30 feet of extra fat feeder to each. Fault currents were way over the old 10KA expectations, the industry moved to 22KV rated breakers        (and I have to buy them even for my 0.6KA faults). 5.3 MEGAwatts! Try stopping MegaWatts in a thing the size of a pack of Camels.

So at some point the delivery specs changed from 0.024r to 0.011r as the lowest source impedance in residential power design.

[Matthew Sanford]

Quick question, (slightly off topic) with the values of the voltage resistor; does it matter more on shared power supplies? I remember setting up a breadboard circuit and forgetting that resistor, and when plugging other things in on the same power supply, it was the only thing that worked.

[ElectricDruid]

I don't think many power supplies these days are loaded with AC ripple. You either have enough filtering or not.

Yep! We agree about that!

Any second,third, fourth, etc pedals with resistors and capacitors are all in parallel so it all adds up.

You'd have to draw that for me because I'm not seeing how that works from where I'm standing. The effect of the filters doesn't add up. Each filter is for one pedal alone, and any effect it has is for that pedal alone. Even if they did add up, the losses in passive filters are terrible when you start stacking them up, so it wouldn't work even then.

If anything i put a .1uf or .01uf across the power input. I'm not worried about any cut off frequency, if you are you need a better power supply.

I'm trying to design stuff that works for as many people as possible, not only people who have nice quiet power supplies. If I can do that while not causing any problems for people who *do* have nice quiet power supplies, why would I *not* want to make that improvement? Seems like a no brainer to me.

[Rob Strand]

Based only on the formulas, they would be less susceptible to everything since they barely resist the current flow. Op-amps aren't greedy like LEDs, so they don't need their current limited externally. A low CLR would not cause the current draw to increase over what the op-amp or other active devices are trying to draw.

But, I have also heard anecdotes of people having issues with ~10R PSU resistors burning up even in relatively small 9V op-amp circuits with a typical current draw, say 3 duals. It was speculated to be due to the inrush current as the filter capacitor is charged during power-on.

This would assume an otherwise correctly-wired circuit... obviously there are many other issues if we add reverse polarity or shorts to the mix.

Does the math check out on this? Could it be possible that a combination of a resistor that is too small with a capacitor that is 'too good' (meaning low ESR and high value) may cause issues?

It's true small resistors can fry, especially with larger caps.

It you short out a transformer for a short time the current greatly exceeds the rated current yet the transformer doesn't instantly self destruct.    If we hold the short too long then damage will be done to the windings.   This is because the windings are overheating.  The outside of the transformer won't get as hot as the winding.

A resistor's rating is sort of like that.  The power rating is continuous however its pulse power rating is somewhat higher.

A capacitive load effectively shorts the supply at power up until the cap charges through the resistor.   A larger cap will increase the length of time it takes to charge.

The resistor sets the turn on current to V / R.   It also sets how long the cap takes to charge.

Clearly if the series resistance is larger than R = V^2/P = 9^2/0.5 = 162 ohm for a 0.5W resistor and 9V supply, it doesn't matter what size the cap is since the resistor can handle being shorted across the rails indefinitely.

A small resistor makes the peak current high but it decreases the duration.    If the resistance is smaller than 162 ohm the power will exceed the continuous power rate of the resistor.

Here's where things get more complicated.   Exactly how large can the current be?  To some degree we expect we can handle larger currents or powers provided they aren't too long.

There's no simple way to derive what a resistor does in this situation.  You need pulse power data for the resistor from the manufacturer

Here's an example for a 1/2 W resistor,



The red dots show the pulse power for a 9V rail with a resistor value shown on the right column.  The cap values are 100uF and 1000uF.  There's two pulse ratings: a single pulse and a repeated pulse.  If we were to turn the unit off and on at say 0.5sec  then for tp/ti = 100 we expect the multiple pulse case to apply.  This point is marked in orange.  For the 1000uF cap we might expect 30 ohms to be a safe choice.  For 100uF it's not so clear to apply the single or multiple pulse rating.   We can play it safe an use 10 ohms or we could push our luck and use 3 ohm.   The thing is these are minimum resistances so we shouldn't be pushing our luck.

Be aware different brands and models of resistors will have different curves to the above, even if they are 1/2 W resistors.

The values are at least in the ball-park of where people see issues.    Also note the load of the circuit adds to the power dissipation in the resistor.  If a pedal is pulling so much  current as to approach the continuous rating then that will decrease the allowed pulse power at turn on.    This case is more complicated.

[merlinb]

I don't think many power supplies these days are loaded with AC ripple. You either have enough filtering or not.

Agreed. The filter is mainly there to decouple pedals from each other to prevent unwanted feedback, not for ripple reduction.

[Rob Strand]

Agreed. The filter is mainly there to decouple pedals from each other to prevent unwanted feedback, not for ripple reduction

The earliest Boss pedals which incorporated a transistor active filter where ACA powered devices ie. ACA unregulated 9V (12V no load).    (I'm not including transistor + zener regulators.)

OD-2 1985 ACA
DS-2 1987 ACA

Not all ACA powered devices of the era had the active filter.   Early PSA powered devices didn't have the filter.  Early on PSA powered devices were higher current pedals, where the ACA power circuit (resistor + diode) wasn't suitable.   Regulated power adaptors weren't common back then so it was a special requirement necessary for higher power pedals.

The later PSA pedals used the filter on circuits which had discrete opamps and on higher gain circuits.  No filter on low gain and unity gain circuits.

That makes the original motive for the active filter circuit ripple based.

[imJonWain]

I often wondered if part of original idea of it was BOM/cost reduction.  You've already got to buy tons of transistors for all the pedals you make driving the cost per transistor super low.  So you add qty 1 per pedal so that instead of needing an expensive single 47u or 100u+ cap per pedal you can use another 10u that are already used in a bunch of places to the same effect.  In sufficient volume it probably saves a good amount of money and it's one less different part to keep around.  I assume similar for the simple discrete opamps designs.

[Rob Strand]

I often wondered if part of original idea of it was BOM/cost reduction.  You've already got to buy tons of transistors for all the pedals you make driving the cost per transistor super low.  So you add qty 1 per pedal so that instead of needing an expensive single 47u or 100u+ cap per pedal you can use another 10u that are already used in a bunch of places to the same effect.  In sufficient volume it probably saves a good amount of money and it's one less different part to keep around.  I assume similar for the simple discrete opamps designs.

Pretty much all Boss pedals use 100uF with or without the filter.

The resistor + diode in the ACA (unregulated) forms a decent filter with the filter cap.  When that resistor + diode is removed the extra filtering (from the active filter) helps.

You also have this problem for the ACA pedals, which bypasses the resistor + diode,
https://www.diystompboxes.com/smfforum/index.php?topic=131773.msg1281490#msg1281490

[merlinb]

That makes the original motive for the active filter circuit ripple based.

Yes. How does that answer the OP's question in 2024?

[R.G.]

It's horses for courses. You can't design an effective filter for just what you want to filter out unless you already know what it is you want to filter.

Well, of course gross overfiltering works too if you're not losing signal doing it. A 50/60/100/120 Hz filter generally does a good job on higher frequencies in the audio range, too.

I love to speculate what was going through the original circuit designer's mind as much as the next guy, but this is tempered by the fact that I realize that I can never really know what they were thinking. Er... if they were thinking. :-)

[Rob Strand]

Yes. How does that answer the OP's question in 2024?

It doesn't.  No one else answered the question either.  The thread was derailed into filtering early on.  I'm just giving evidence for the motivation Boss's for incorporating the active filter.

My first post answers the OP's question.  Pulse power ratings are something I've researched in the past for professional products.  There's a reason why resistors blow up at power up and there's a reason why manufacturer's provide pulse power data.

I love to speculate what was going through the original circuit designer's mind as much as the next guy, but this is tempered by the fact that I realize that I can never really know what they were thinking. Er... if they were thinking. :-)

You can get an idea by looking at each case where the active filter appears and when it doesn't appear.  Basically finding a pattern (or lack of a pattern) from many data samples.   Speculating cold might give you three engineering possibilities but speculating with hard data can narrow things down.   For the Boss pedals there is a pattern: sensitive pedals have the filter, pedals which put noise on the power rails do not.

[bluelagoon]

For the Boss pedals there is a pattern: sensitive pedals have the filter, pedals which put noise on the power rails do not.

Scuse my ignorance Rob, but which type fx pedals would be more inclined to putting noise onto the power rails, as opposed those that don't, or as you state are more sensitive ? Thanks.