Adding inductor to Full-wave rectifier for DM556

[bdunham7]

B/c of the rectified AC input.

Then you want to put it after the bridge rectifier and before all of the capacitors.  Thats going to be a pretty substantial choke, almost as big and heavy as your transformer.  The 10mH/5A version here would be an example.

https://www.mouser.com/datasheet/2/177/153_159-736947.pdf

[Analog Kid]

So, asking other respected members here (i.e., those who know a lot more than I do):

What would be wrong with simply using a large capacitor after the full-wave rectifier here, as is done in many power supplies?

Wouldn't that be sufficient to reduce ripple to an acceptable level?

And what about this business of "extending the life of the capacitor(s)": is that even a thing? What would reduce the life of the caps? would an inductor even help?

My guess®™ is that just a single capacitor (of sufficient size) should be all that's needed here.

[bdunham7]

So, asking other respected members here (i.e., those who know a lot more than I do):

What would be wrong with simply using a large capacitor after the full-wave rectifier here, as is done in many power supplies?

Wouldn't that be sufficient to reduce ripple to an acceptable level?

And what about this business of "extending the life of the capacitor(s)": is that even a thing? What would reduce the life of the caps? would an inductor even help?

My guess®™ is that just a single capacitor (of sufficient size) should be all that's needed here.

A 10mH choke before his capacitors will cut his capacitor ripple current by about half (making some basic guesstimates about the exact operating conditions, etc) and will also reduce the peak current through the rectifiers and transformer. This will also improve the power factor.  It will also reduce output ripple if compared to an otherwise identical circuit without it.  The inductor will also dissipate some power itself, so it isn't a free ride. 

A suitable 10mH 5A DC choke will probably cost a lot more than even a huge capacitor bank plus inrush limiters plus a nice lunch.  So everyone just uses capacitors these days.  If you need regulation, you just use an LDO. 

If you want to simulate various linear power supply circuits, try this:

https://groups.io/g/duncanampspsud/files/psud2_windows

Edit: this program gave me some unexpected results so I'd stick with LTSpice for now.

[Langdon]

B/c of the rectified AC input.

Then you want to put it after the bridge rectifier and before all of the capacitors.  Thats going to be a pretty substantial choke, almost as big and heavy as your transformer.  The 10mH/5A version here would be an example.

https://www.mouser.com/datasheet/2/177/153_159-736947.pdf

Actually, putting the inductor after the caps will also reduce the riple current those caps are subjected to b/c the discharge current is reduced by the inductor.

Ripple current is twofold: charge and discharge.

If the discharge current is reduced, so will the charge current (because the cap is almost charged already).

Also, no all cpapacitors care about ripple current.

Capaitors with ~0 ESR like ceramic disks don't mind immense riple current, because they don't heat up (plus they don't mind heat either).

[Langdon]

I'll just use caps then.

No inductor.

Six 680uF 63V aluminium electrolytic caps.

There is a 1 uF film cap across the bridge DC output to take some of the ripple.

[Analog Kid]

Why such an extreme amount of capacitance?
That's the kind of filtering you'd see in a high-end audio power amplifier.
You're powering a stepper driver, driving a mechanical device that by its nature generates noise and vibration.
What are you trying to do here? Attempt to make the stepper completely silent?
I mean, it's your project, so you're free to put 4080μF worth of caps in if you want to. But believe me, that would be total overkill.

[langwadt]

Why such an extreme amount of capacitance?
That's the kind of filtering you'd see in a high-end audio power amplifier.
You're powering a stepper driver, driving a mechanical device that by its nature generates noise and vibration.
What are you trying to do here? Attempt to make the stepper completely silent?
I mean, it's your project, so you're free to put 4080μF worth of caps in if you want to. But believe me, that would be total overkill.

on a system with large inertia you'd want a substantial amount of capacitance, not so much for ripple suppression but to absorb the current the steppers will force into the supply when deceleration the load without exceeding the voltage limit of the driver

[Analog Kid]

Large inertia; OK.
So essentially a big damper on the stepper's motion.
So OP: who big exactly is your stepper? And is that the reason you think you need all that capacitance?

[bdunham7]

Why such an extreme amount of capacitance?

Is it extreme?  The rule of thumb for FWB linear supplies is 1000µF per ampere of DC output.  Audio amplifiers aren't special in this regard and this rule supposedly applies to all sorts of things, bench supplies and so on.  It is a value that does a reasonable job of reducing ripple and also is enough so that typical capacitors of that size will have high enough ripple ratings to survive.  If you use less--which you can--then you have to actually check to see what the ripple rating is and then model the circuit to make sure that the rating isn't exceeded.

But AI seems to agree with you that this is way too much:

[Analog Kid]

FWB = full-wave bridge

[bdunham7]

FWB = full-wave bridge

Yes.  ?

[Analog Kid]

I find it annoying when people use obscure acronyms without spelling them out. GPS, ADC, PWM? No problemo. But I had no idea what "FWB" meant until I looked it up. Friends with benefits?
I don't want to have to refer to a glossary just to read posts here.

[bdunham7]

Friends with benefits?

Sir, this is EEVBlog! 

I also try to avoid jargon and shorthand in the initial mention, but we're in a thread on power supplies with full wave bridge rectifiers so I thought I was OK.

[Langdon]

The original rectifier board had like one humonguous 10000µF capacitor.

4080µF is nothin'

[Analog Kid]

How big is the stepper?

[Analog Kid]

Why such an extreme amount of capacitance?

Is it extreme?  The rule of thumb for FWB linear supplies is 1000µF per ampere of DC output.

Interesting; never heard that one.

Thinking out loud here, it seems to me that to properly size a power-supply filter capacitor to remove ripple, one would need a formula that takes the following variables:

• output current
• frequency
• desired level of ripple as a fraction of input voltage

Did I leave anything out? load impedance? voltage?
What would be that formula? Something having to do with the capacitor's time constant, I'd guess.

[bdunham7]

Thinking out loud here, it seems to me that to properly size a power-supply filter capacitor to remove ripple, one would need a formula that takes the following variables:

• output current
• frequency
• desired level of ripple as a fraction of input voltage

Output current is taken into account, obviously since the rule would have the capacitance scale linearly with maximum design current.
Frequency is assumed to be 120Hz/half-wave rectified humpy-wave.  AFAIK the rule doesn't change in 50Hz-land since it isn't particularly precise to start with.
Voltage takes care of itself.

Desired ripple control, ESR and power factor are all intertwined and the rule sort of balances those to a reasonable result.  Too much capacitance with low ESR may yield a very poor power factor and high RMS-to-average ratios.  In a regulated supply or an amplifier using an unregulated supply (and thus depending on PSRR not smooth supply voltage) you just need the minimum voltage to not drop down below the point where the load functions properly.

You can simply model it in LTSpice to see if you come to the same conclusion.

[Analog Kid]

So no formula? Not even as a starting point?

[Xena E]

That AI quote is BS*

The rule of thumb would be 1μF per mA of current.

Doesn't apply in the case of a choke input filter.

If the intended target is the fundamental ripple, a tuning capacitor can be added across the choke, personally I'd just go with lots of capacitance and screw the power factor rather than buy a big choke...

BS as an abbreviation of Bovine scat, (not an acronym, an abbreviation 😉).

X

[bdunham7]

So no formula? Not even as a starting point?

Linear power supplies are fairly challenging to model numerically, partly because there are so many variables that all matter enough to worry about.  Start with the transformer--you have to know its effective output source impedance and reactance.  The rectifiers have a varying V_f and shutoff characteristics. 

However, if you want a method of estimating needed capacitor size, I'd use a first-order hold-up approximation to start  At 60 Hz, you get an incoming peak every 8.3ms.  Assume the capacitor charges to that peak and then determine what minimum the voltage must stay above as the capacitor discharges (the hold-up threshold).  Figure out how large the capacitor must be to supply the load current for 8.3ms without dropping more than the difference between your peak and the holdup threshold.

A 1F capacitor will drop 1V every second as it delivers 1A.  So C = 1A*1s/1V = 1F.  Sub in your values, say you're willing to accept 5V of ripple and the load current is 500mA.  C = 0.5*.0083/5 = .00083F or 830µF.  Now factor in some second order corrections such as the capacitor not actually charging to the peak (due to resistance) and the hold-up time actually being less than 8.3ms because the rising wave will catch up some time before the peak.  Both of these factors reduce the needed size, perhaps to about, hmmm, about 500µF or so? 

[Langdon]

The rectifier is already soldered and made.

Its gonna be 4080uF whether its necessary or not.

Its already done.

In retrospect, 4080uF was probably not necessary, but who cares?

I got those caps free from the old electronics classroom in my school.

They were never used, but are old.

Unfortunately, they are only 85c, not 105c.


The original rectifier used a single 10000uF 40V cap with a max surge voltage of 50VDC for a 38.5V circuit!

Only 1.5V below the max!

[Analog Kid]

I'm still curious, if you don't mind answering: how big is the stepper you're driving?

[Langdon]

It's a 4.6A synchronous/stepping motor powered by a DM556 (maybe just a DM542 pretending to be a DM556)

The rectifier works great (I tested it today)!

[Analog Kid]

OK, that's pretty beefy, so you're probably doing the right thing here.