• WARNING: Tube/Valve amplifiers use potentially LETHAL HIGH VOLTAGES.
    Building, troubleshooting and testing of these amplifiers should only be
    performed by someone who is thoroughly familiar with
    the safety precautions around high voltages.

DCR requirments for Chokes

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Hi Guys,
Thank you for sharing your knowledge. The question is how low a DCR is appropriate for Choke input ps ? Using the formula , if the circuit only draws 25ma so it's 350/25 = 14H the DCR will surely be high, example Hammond 157L dcr is 429 ohms

Thank you again
 
The question is how low a DCR is appropriate for Choke input ps ? Using the formula , if the circuit only draws 25ma so it's 350/25 = 14H the DCR will surely be high, example Hammond 157L dcr is 429 ohms
Maybe a bit “late to the party”, but that 429 Ω RDC is only 0.43 V/mA of series voltage drop, sumotan.

In terms of power supply droop under realworld loads, remembering that one's HV power supply L-input LC section will have good sized C followin the L, in an SET/SEP only varies 20% or so between quiescent and clipping, and for a P-P power stage (depending heavily on how close to A, AB, AB2, pure B it runs), anywhere from the same 20% to well over 100%, you're still in a pretty small range of ΔV values.
20% of 25 mA = 5 mA … ×0.43 V/mA → ±(2.2 ÷ 2 = 1.1) V​
Minor indeed.

I could be full-of-beans, but at least that's my take on it.

Just saying,
GoatGuy ✓
 
Last edited:
Hi Summer, in your CI equation where does the 350 come from?

I believe it to be a empirical rule-of-thumb which encapsulates other design criteria into a simple constant.

{goes away… calculates … scratches beard … comes back 2 hr later}

So, I was right. It is empirical, but also fairly magical. First, the '350' is the VRMS of the transformer. The rectification drop is small compared to that at high tension, so it is ignored.

BUT… that 350 = VRMS is totally coincidental … to F = 50 Hz, and rectification being full-wave bridge. Completely. Change F to 60 Hz, and while the same inductor will work, the minimal one would actually 'heuristic out' as:
L = 350 × 50 / ( mA • Hz )​
Obviously, 50 / Hz = 1 when Hz is 50. But at 60 Hz, the critical L is smaller.

In the end, we also need to remember that the use of a larger value inductor than the one specified by the heuristic will work for the intended purpose; it'll also deliver the 90% of VRMS as the filtered voltage. Only when L drops below the critical value does the L-C “choke input” filter begin to deliver (ironically) more voltage, on average. More, up to when the DC resistance of the thing starts to suck up its share.

The one big (for goat) take-away from this is … it is really pretty hard to actually formulæ-calculate what the minimum critical inductor size need be, given a particular current. Hard, because rectified sinusoids are anything but formulæ-easy integratable. Hard. I ended up with a 30,000 line spreadsheet with 20 columns. It would have been easier to use R or C or PERL, I think.

So…
L = (50 ⋅ VRMS) / (mA ⋅ Hz)
VRMS = 400
Hz = 50
mA = 40 (say)
L = (50 ⋅ 400) ÷ (40 ⋅ 50)
L = 10 Hy
That's the real heuristic. The right one.

Just saying,
GoatGuy ✓
 
Last edited:
For a true choke input filter, you need to meet the rule for critical inductance:
Critical Inductance = 350/(current in mA)

A choke input filter for an amplifier that needs 80mA DC will need an inductor of:
350/80mA = 4.375 Henry. Use a 5 Henry inductor...

Since i could not find where the 350 came from in the previous posts i wondered how that number appeared in the formula.

I have always gone by this formula that is given in the ARRL "The Radio Amateur's handbook" chapter 7 "power supplies". Under "Choke-Input Filters" ....Minimum choke inductance....
L(henries) = E(volts) / I(ma)

I have also read of more elaborate versions of this equation but essentially it boils down to this same formula.
 
Last edited:
Yes, those good old ARRL handbooks come through again. I like using simple formulas that work, that I can remember, and that I use.

Just remember, one manufacturer might tell you his choke is 7 Henrys @ 130mA DC. It may not be so (at 50Hz or at 60Hz). So a larger inductance and higher current rating gives a little margin to make the supply a true choke input supply.

Or check the part yourself (for most of us, the only test equipment we have for this is to put it in an amplifier, check the input ACV, versus the output DCV). Do not forget to account for the voltage drop due to the choke DCR. If the DCR is large enough, the DCV will be less than 0.9 x the ACVrms, even though the inductance is far less than the critical inductance.

When you are in the middle of the Pacific Ocean on a Destroyer, and the Captain wants you you to fix that broken piece of electronic gear, and there are no correct parts, you learn to figure a thing or two.

The average, unfiltered DCV, of a full wave rectified sine is 0.6366.... times the ACVrms.

It also is is 2/pi times the ACVrms.

Knowing the why and the full calculations is one thing, knowing something you can remember and use is another. Find the form you most prefer.

If you wonder what an application of the 0.6366... is, the old VOM with a d arsonval meter movement and a rectifier to measure ACV. It is calibrated to measure the rms volts of Sine Waves. Any other wave shape, and it responds to, and reads out, in Average volts, not rms volts.
 

PRR

Member
Joined 2003
Paid Member
Resistance in the *first* (before-cap) choke has no large effect on final DC value, but damps the turn-on surge.

In this sim, V7 is 400V peak, so 283V RMS. With choke input we expect 254V DC. Choke DCR is varied from 1 Ohm to 600 Ohms. The final DCV value hardly changes. However the (unlikely!) 1r DCR peaks to 455V for a dozen mS; the 600r DCR peaks to 335V. While the peak is too quick to blow-up an electrolytic cap, even brief 455V on a 300V cap would worry me, 335V for 30mS is no worry at all.
 

Attachments

  • sumotan-32.gif
    sumotan-32.gif
    19.2 KB · Views: 205
The "critical" inductance value is based on what is refereed to as the boundary between Continuous and Dis-Continuous for the inductor ripple current... The handbook equations are not always applicable.... If your going to use a handbook equation, you should understand where these numbers are derived from... In many cases the author will add in their own "fudge factor"... You ideally do not want to be at the critical inductance value..since that is a boundary region...but rather well into the CONTINUOUS region for some design margin... It all falls back to the basic equation E = di/dt * L
 
PRR,

The transient voltage at the beginning also depends on the DCR of the secondary (or DCR of the half primary in the case of a full wave secondary w/center tap).

I am not as sure of this, but I believe the DCR of the primary also is a factor, as it translates 'upward' due to the turns ratio.

cerrem,

As I said in some recent threads, I do not always trust manufacturer's choke ratings, either for current or inductance. That is why I recommended using more than critical inductance. Know your vendor.

Also, what happens if later you decide to reduce the current of the output tubes, change tube types, change output transformers that have lower current rating.

Changes have implications, especially if there was little or no margin there to begin with.

There is good engineering, so so engineering, and the original Tacoma narrows bridge.
 
As already pointed out, critical inductance is the bare minimum needed under optimum conditions. It's value depends on both applied voltage and current draw, so formulas with only a current term are incorrect, or only correct for a particular voltage.

A simple but correct number is usually stated as: critical inductance in Henrys equals the equivalent resistance of the load in K Ohms. Just be several times this under all load conditions and you're golden.

All good fortune,
Chris
 
Only add resistors if you need to because of tube rectifier ratings.

Back in the tube era, you could not get large enough electrolytics to get the ripple low enough. That forced designers to use at least one choke and sometimes two. These days, you can get big electros, so you can just use a big electro and not have a choke at all. A tube rectifier will not cope with that, but you can use silicon diodes which will.

BUT, tbe current in a simple scheme of solid state diodes and a big electrolytic will have very high peak currents, and with the carrier storage in the diodes, may generate noise which gets into the signal path of the amp, and even other devices in you house.

A tube rectifier and a choke-based ripple filter is a very quiet system noise-wise. And it's good at blocking noise from other sources coming down your mains wiring. There are easy ways to fix this in a solid-state + cap power supply though.

Because a choke lets you avoid high peak currents, there is less volt drop in transformer windings. So regulation is better, and there's less thermal stress on the transformer, so it can be a little bit smaller.

All tube rectifiers are rated for the maximum size first filter cap they can tolerate, and the minimum plate-to-plate resistance (the transformer secondary resistance) they can tolerate. Go outside these ratings and the tube will arc over internally, with spectacular (and possibly expensive) results.

If, in a given application, you will go outside the tube ratings for min resistance and/or max filter capacitance, you can fix that by adding resistance. Only add resistance for that reason. You don't need it for anything else. Resistance worsens regulation and makes negligible effect on ripple suppression when a choke is used. (Different story of there is no choke)

Your friend who was taking about resistance giving more stable tube operation was probably referring to these factors without fully understanding or explaining them.
 
Last edited:
… you could not get large enough electrolytics to get the ripple low {requiring} at least one choke and sometimes two. {Now} we can get big electrolytics, and not have chokes at all. Tube rectifiers don't like that, but you can use silicon diodes which will



• noise
• stability
• peak power I² power loss
• needing silicon for big cans
• resistors to mitigate higher V


Yes. One thing often lost in these discussions is that the main goals of designing a power supply, like at the top of the list are:

• № 1 - quiet, ripple-free DC output
• № 2 - stiff variable-load regulation
• № 3 - reasonable line-variation independence
• № 4 - piped in A/C noise suppression
• № 5 - very high reliability
• № 6 - inexpensive implementation
• № 7 - possible slow-turn-on HV ramp​
It is remarkable how readily some of these are achieved with pretty simple CRC, CLC, LCRC, CLCLC and other chain-of-low-pass filter implementations. № 1, easily. Nix on № 2, though, and № 3. But excellent on № 4, № 5 and № 7. Not necessarily achiving № 6. This would be for valve rectified CLC, with small C₁, modest L₁, and rather large C₂ (reflecting Keit's position on “electrolytics, big, are pretty cheap these days” paraphrased).

Thing is … there is “another way”, that is equally inexpensive, yet optimizes all 7 criteria. Requires some sand … but so silicon rectifiers be!

Using symbol S as 'series regulator', the optimum topology is basically:
C₁LC₂SC₃​
The first (not very large) C₁ accumulates a bit more than the minimum full-wave-bridge peak-to-peak energy to turn 100% ripple into perhaps only 5% to 10%. Being small, it doesn't impose a harsh peak-charge-flow current spike on the transformer (or the rectifiers). You could use valve rectifiers, or silicon. Makes little difference …

Then L, again, modest value, because the C₁ is doing the “heavy lifting” of averaging out energy storage and release. L is there as an outstanding high harmonic buzz-of-rectification filter. A few henries is all that is needed. Keeps costs down.

Then C₂, the “medium size” (equal to C₃) reservoir. Gets ripple down to 1% or so. Sets up stable-enough DC for the series regulator.

Then S … a MOSFET-type high-voltage series “source follower” style regulator. Its GATE is driven by a Zener-clamped stack; having Zeners-in-series is cheap and effective at setting a fairly reliable (day-to-day, season-to-season) regulation reference voltage, but not necessarily a particularly precise one (since that “precision” doesn't matter).

Finally another C, C₃ … which is the reservoir that totally squashes whatever series impedance the S regulator has, and brings it down to mΩ. Same value as C₂.
________________________________________

I bring this up (again and again) because it is so easy to implement, so robust, so reliable, and delivers such remarkable ripple-and-noise free output. With all-sand components, nearly forever. Until the capacitors dry out and quit.

Finally — and this is an excellent exercise for those of us who like to compute such things — one can even achieve № 7 slow-turn-on HV ramp as well. When the Zener-clamped voltage reference stack is fed with a very high value resistor linked to another electrolytic capacitor, its charge-up rate can be remarkably slow. I use a series of dirt-cheap 4.7 MΩ resistors (5 of 'em) and a 4.7 µF, 450 V electrolytic (or two 10 µF's in series, each 350 V). The ramp up rate takes nearly 30 seconds to come up to the Zener clamp.

Anyway… that's the main idea.
I admit … its off-topic a bit … but I really believe the 7 points are THE point.

Just saying,
GoatGuy ✓
 
Yes. One thing often lost in these discussions is that the main goals of designing a power supply, like at the top of the list are:
• № 1 - quiet, ripple-free DC output
• № 2 - stiff variable-load regulation
• № 3 - reasonable line-variation independence
• № 4 - piped in A/C noise suppression
• № 5 - very high reliability
• № 6 - inexpensive implementation
• № 7 - possible slow-turn-on HV ramp​

Anyway… that's the main idea.
I admit … its off-topic a bit … but I really believe the 7 points are THE point.

Just saying,
GoatGuy ✓
here you should go:
https://www.diyaudio.com/forums/solid-state/
bye bye
 

Nice “poke-in-the-eye”, fellow DIYer.

A bit heavy handed: one could easily use all-vacuum devices to achieve the same end. Cathode-follower with error-amplifier, glow-discharge voltage reference bottles, direct-heated valve rectification. To achieve exactly the same end… of points.

Perhaps you missed the main point of all, “№ 0 - design the amplifier around the desired topology, the power supply to best support it.

This, being the "Tubes" forum, lionizes valve circuits for most every purpose. Yet, in the end, № 0 remains key.

If you for example, or anyone else for that matter, wish to implement your power supply using hand-made selenium rectification stacks in conjunction with hydrogen-mercury vapor rectification, by all means. Go for it.

If you desire points № 1, 2, 3, 4 and 7, AND to use ALL vacuum components and series regulation by cathode-follower + error amplifier. You'll definitely achieve № 0

You'll get everything except economy ... and with entirely the same commentary as I posted above in general.

Anyway…
Nice poke.
I got it.

Just saying,
GoatGuy ✓
 
Last edited:
claudiomas,

Huh?

The original post was about chokes and tube rectifiers. Yes, the thread went off tracks several times and came back to the original tracks several times.

You posted a link that took us to "Solid State" 'Talk all about solid state amplification'.

I personally do not have a problem with using solid state devices in tube amps, but many do have a problem with that. But as such, we both agree to use this tube forum for tube amps. The title of the forum is not: "Tubes Only". We can all judge what is appropriate, even if I do not always do that myself, forgive me.

I think we can all agree that sometimes we will disagree.

I always say that if all you have in your toolbox is a hammer, you will respond to all problems as if they are nails.

And GoatGuy, you beat me to the point.
 
Last edited:
And GoatGuy, you beat me to the point.

LOL… You framed it substantially better than me. The "Tubes" forum isn't exclusively about tubes-for-everything, but rather it is amplifier-by-valve centric and less pedantic in censoring the supporting circuitry blocks.

After all … we regularly read posts of all sorts of biasing and anode-loading ideas using LEDs, Zeners, little semi-custom current regulator boards, and so forth. All very solid in their state of affairs. Even valve-MOSFET cascodes!

Thank you for the support, friend. Warmly welcomed. GoatGuy ✓
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.