Choke input Filters

Unless the diode draws current, there's no dI/dt,so the choke does nothing. When the diode DOES draw some current, the choke will at least partly block high frequencies, i.e. it will reduce by a bit the third and higher harmonics that the diode would otherwise generate.

Greetings,

I've been playing around with detector circuits which have a choke in series with a diode.

Now my analogy to this is the classic Choke Input power supply.

Langford-Smith (Radiotron Designers Handbook) says (paraphrasing)

"An input choke filter works only if the L value is greater than a critical value. Such critical value for the inductance is given by the load divided by 6*pi*f (where f is 50 or 60 Hz)"

However a knowledgeable friend of mine argues that if the inductance of the choke is above a certain value, then the diode can't see the applied voltage.

My (somewhat simplified) simulation seems to agree with him...

So what is going on, does a choke input filter stop working if the choke is too large? What am I missing?

----------

It's behaving as normal. Remember that an inductor resists change in current? Well, The larger the current the harder it is to "get started". That is, the larger inductor is better at resisting the change.

So, what does this explain your circuit? Well, When the diode is blocking no current is flowing. The moment the diode starts conduct the current is resisted by the inductor so less current would flow than normal AND it takes longer to ramp up. The larger the inductor and lower the load the longer it takes for the current to ramp up. At some point no current will flow.

Notice that conduction takes place at the same time in each graph but in each case the current and it's slow decrease as the inductor becomes larger.

With the full bridge circuit the inductor current can be constant, and which diode is on should shift as the voltage shifts.

You keep saying that the L isn't a good idea, and I more or less agree on practical grounds.

But if you're in "spare no expense" mode, or "gee this would be fun", or even "how can I make the best @#$% crystal radio in the world", I don't see why a choke filter is any worse than a cap, and it may actually do better in the context of todays lower-impedance earbuds as opposed to yesterday's high-impedance ear phones.

Sure. But as the voltage source passes through zero volts, the current continues, so the diodes have forward drop. The voltage at the left side of the inductor swings negative, partly cancelling the normal positive swings. Which is why the final output voltage is so low.

John

If the signal swing isn't much greater than the diode drop, the inductor costs you a lot of output signal.

The best setup would be a low-barrier schottky diode into a cap.

John

On 08/15/2011 01:01 AM, John Larkin wrote:

For 1 MHz and 1 V, using an MMBD301 Schottky, there's a sweet spot around 1 uH series inductance (see attached schematic). It reduces the video resistance only a very little, but it sure crispens up the impulse response.

Cheers

Phil Hobbs

Version 4 SHEET 1 1952 784 WIRE 128 64 48 64 WIRE 176 64 128 64 WIRE 288 64 240 64 WIRE 416 64 368 64 WIRE 48 128 48 64 WIRE 416 128 416 64 WIRE 416 224 416 192 WIRE 48 320 48 208 WIRE 128 416 128 64 WIRE 176 416 128 416 WIRE 288 416 240 416 WIRE 416 416 368 416 WIRE 592 416 416 416 WIRE 592 448 592 416 WIRE 416 480 416 416 WIRE 416 560 416 544 WIRE 592 560 592 528 WIRE 592 560 416 560 WIRE 416 576 416 560 FLAG 48 320 0 FLAG 416 224 0 FLAG 416 64 out FLAG 416 576 0 FLAG 416 416 out2 SYMBOL voltage 48 112 R0 WINDOW 3 -304 26 Left 0 WINDOW 123 -200 90 Left 0 WINDOW 39 0 0 Left 0 SYMATTR InstName V1 SYMATTR Value SINE(0 1 1meg) SYMATTR Value2 AC 1 SYMBOL diode 176 80 R270 WINDOW 0 32 32 VTop 0 WINDOW 3 115 104 VBottom 0 SYMATTR InstName D1 SYMATTR Value mmbd301 SYMBOL ind 272 80 R270 WINDOW 0 32 56 VTop 0 WINDOW 3 5 56 VBottom 0 SYMATTR InstName L1 SYMATTR Value {Lser} SYMBOL cap 400 128 R0 SYMATTR InstName C1 SYMATTR Value 10n SYMBOL diode 176 432 R270 WINDOW 0 32 32 VTop 0 WINDOW 3 144 129 VBottom 0 SYMATTR InstName D2 SYMATTR Value mmbd301 SYMBOL ind 272 432 R270 WINDOW 0 32 56 VTop 0 WINDOW 3 5 56 VBottom 0 SYMATTR InstName L2 SYMATTR Value {Lser} SYMBOL cap 400 480 R0 SYMATTR InstName C2 SYMATTR Value 10n SYMBOL res 576 432 R0 SYMATTR InstName R1 SYMATTR Value 1k TEXT 656 664 Left 0 !.param Lser=10n TEXT 648 760 Left 0 !.step dec param Lser 10n 100u 2 TEXT 32 736 Left 0 !.tran 20u TEXT 976 40 Left 0 !.MODEL mmbd301 d\n+IS=2.94707e-08 RS=5.99757 N=1.36601 EG=0.623476\n+XTI=4 BV=60 IBV=1e-05 CJO=2.4032e-12\n+VJ=0.4 M=0.272967 FC=0.5 TT=0\n+KF=0 AF=1

Hello again,

Well, I hooked up my Sig Gen, a bunch of components and my CRO on the bench.

Short story long, the various circuits behaved exactly as theory suggests they should.

The series choke filter works just fine whether with a single diode or with full wave input, and it all works fine over a wide range of frequencies.

The only odd bit was a few squirrelly frequencies with a single diode which I put down to resonances in the choke.

So my bottom line is that the QUCs simulator must be telling lies.

Next step is to get up to speed with a better simulator. Something like LTSPICE I guess. Any recommendations?

BTW, thanks very much for the help and feed back, greatly appreciated.

......... Zim

The inductor costs you a lot of output signal _voltage_. But not necessarily output signal _power_. Go back to school, and pay particular attention to the parts of circuits 201 where they're talking about impedance matching.

The best setup depends on what you're trying to achieve, what the circuit is before, what the circuit is after, and any number of other factors. If the OP is trying to receive AM in a traditional superhet radio, probably the 'best' setup would be a synchronous detector (and, depending on his power and hardware budget that may 'best' be done after the ADC, in software). But if 'best' were always the same, then we'd only have one bit of electrical kit that would cover all of our circuits needs.

You've been designing how long, and you still don't know that "best" is a tiny word that's full of huge pitfalls?

If the inductor keeps the diode conducting long after the signal has swung negative, it costs voltage and power. Diodes that conduct all the time don't rectify. The "school" insult was unnecessary.

I was referring to the "crystal radio" in the paragraph above. I suppose you could actually build a synchronous detector into a crystal set. That would actually be interesting.

Don't be a jerk. Nobody likes jerks.

John

IF memory serves, I think I told you to use another simulator. I Already used LTspice to verify the confusing mess, because your results were very wrong from what I could see at a quick look.

Jamie

LTSpice. It's free, it works great, and you can cut & past the design files to newsgroups.

I see that you're posting from Windows, so you don't even have the "Linux" excuse (and LTSpice has specific hooks to detect that it's running under Wine, and do the right things -- which it does, with much aplomb).

If you're using a full-wave rectification setup, then that won't happen

-- only the diode to the negative voltage will be on, the other will be off.

That, sir, is something you figure out with information you learned in engineering school, if you don't learn it before.

So the statement "Diodes that conduct all the time don't rectify" doesn't signify -- because each diode is conducting roughly half the time, which is most certainly not _all_ the time.

I, too, was referring to the crystal radio comment. It's hard to get quality, high-impedance earphones. The "choke input" detector promises to have a bit of natural impedance down-conversion vs. a "capacitor input" detector which would tend to transform the impedance _up_.

So if you can get past the obvious difficulties of finding a high quality, high inductance choke, it has promise for the high performance crystal radio crowd.

If I was a trifle impatient with you for flogging a long-dead horse and I let it show, forgive me.

But stop flogging the damned horse!

That's why, gradually taking effect, nobody likes Larkin ;-)

Indeed! ...Jim Thompson

At the instant of zero supply crossing, if inductor current is still flowing, the diode summing node is negative a junction drop. That negative swing, averaged by the inductor, subtracts from the net output.

(Polarities assume diodes pointing to the right, positive desired output.)

More insults. In a situation that you apparently don't understand.

It's not impedance conversion, it's rectification loss.

Apres vous.

John

I run LTSpice under Kubuntu 10.04--works great. Mike made it wine-compatible to avoid pressure to provide a Linux version.

Cheers

Phil Hobbs

Yup. A native Linux version would have been nice, but he did such a good job with the Wine compatibility that it's almost as good.

The important thing here, is that the rectifier-inductor-capacitor filter gives an output voltage that is the RMS average of the rectified signal. The rectifier-capacitor gives an output voltage that is the PEAK of the rectified signal (at the cost of very small duty cycle of the rectifier, i.e. higher rectifier and copper losses).

Adding an inductor is a winning strategy in every sense except perhaps cost. It lowers the capacitor ESR requirements a lot.

We cannot talk of voltage 'losses' in the same paragraph as power losses: it gets EVERYONE confused. An inductor is a very good idea when integrated into the design, but hard to introduce as an afterthought. That's why the afterthoughts are getting so confused in this discussion (IMHO).

RMS average? What's that?

If the inductor conducts continuously, the output will be *less* than the signal average. Possibly much less.

That's sometimes true in power rectifiers, but we were discussing small-signal RF detectors.

The half-wave (single) diode+inductor case is messy all around.

John