Design of Resonant Input Choke PSU

[Yugo]

Design of Resonant Input Choke PSU

I have intention to design resonant choke input PSU for A class amp or a pre amp,but I approached some difficulties regarding the way and formulas for calculation of damping resistor and solving the resonances of the supply.

---LC
Let's assume dual mono PSU,tube or SS rectified LC .
---LC
Mains f=50Hz.

Lcrit.=Rload/940 or ~Vout/Imin (Volt-mA-Henry).
For example 20H.

But this is for a non resonant input choke supply.I want my PSU to be of a very low DC impedance with big iron cores and according to this calculation

http://www.qsl.net/i0jx/supply.html

my first choke would be designed for 2.2H and with very,very low DC res.,say 10 Ohm.
Is this valid calculation?

Next I calculate for 100Hz resonance:

f=1/2p sqrt(LC) and determine the resonance cap of 2-3 times the B+ voltage.Say 1uF of high quality dielectric (mica,PP).

Now the problem begins.Because the Rdc of choke is very low the paralel resonance of the trap will not be well damped and we have to use a resistor in series with the resonant cap(or in series with the choke but then my goal fails ????) to stop the circulating AC currents and to damp the trap.
Any ideas how to calculate this resistor?

Morgan Jones sugests 10nF in series with 10k and all that in paralel with the choke but I suppose this is not for the resonant choke input supply.
I also found on diyparadise site resonant choke without R


http://www.diyparadise.com/tunedchoke.html

and with R and in the latter it doesn't match my calculation!!!.


http://diyparadise.com/mar06/gg.html


Is it in real world PSU possibe to achieve 0,5 Q factor with real end reasonable pasive elements?

The Q=R *sqrtC/L for paralel trap and needs increasing the R of the choke for good damping which is in contrary with the Q factor of the next series LC - Q=(1/R)sqrt L/C which needs decreasing the R of the choke for good damping.Also this will be additionally damped with the Rload.
The resonance of the first cap with the first choke(which is series resonant trap and form double pole filter with 40/decade or 12dB/octave roll off) will be easily calculated again but as per Hagerman calculator the result for the damped operation is 400 mili Farads???!!!!!!!
Where is the mistake?

http://www.hagtech.com/theory.html#choke

After that I must try to found the next resonance of the Pi filter between first cap,second choke and second cap and at the end between second choke and second cap separately.

Shall I add the two last caps as in parallel and also add two chokes as in paralell (all together the two branches) and then simply make calculation for the Pi filter,or simply calculate for one branch??

Where shall the resonances of all resonating circuits be? Between 1Hz and 10 Hz for good phase behavior,or they can be around 20 Hz?
And are all those rules applicable also for the SS input resonant choke PSU dealing with chokes of 10 mH and 20kuF?

Quite difficult isn't it?

Your comments and opinions will be highly appreciated!!!

Regards,
Yugovitz

[Tweeker]

Being that the pre operates in class A, it is quite possible to achieve a non resonant supply of low q with reasonable parts as we arent worried about super low resistance and regulation. A Q less than .7 will be fine and wont ring. Also with a class A amp you wont be ringing said bell so often, its transients that ring it.

Use a high resistance choke, a pair of ~220nF across the choke with their other ends tied together and to ground. For the first section use a large electrolytic cap, follow it by an RC section with a high quality polypropylene capacitor (motor runs are great here, no need to pay the audio tax.) A bleeder resistor can aid in meeting any minimum choke current demands. What voltage and current do you need?

Resonant choke supply part values often have to be trimmed by experiment. Most of the people running resonant supplies are doing it due to transient regulation issues (a choke likes to try and maintain the same current through it) with single side band linear amplifiers. Also at thousands of volts huge effective H values are needed to keep the AC currents down in the inductor. At 50hz, AC peak choke curent= VACRMSin/(1156*L). For a 10H choke at 3,600VAC this would be 312ma.

Unfortunately PSUDII does not do resonant supplies, however, the stepped load function will pick up ringing.

[Johan Potgieter]

The 10nF-10K pair is used to suppress high voltage transient peaks as the current through the choke (choke input filter) goes to zero during the rectifier off-phase. Also, you probably use full-wave rectification, in which case the frequency is 100 Hz (120 Hz for USA).

But rectified 100Hz go with a lot of harmonics (it is not a pure sine wave), so one cannot get the h.t. line entirely clean with simple resonant filtering, though that helps. Then, a further problem is the exact inductance. There is large varying d.c. through the choke even with a class A load, and the inductance is inversely varying with it. In the end, nett gain in ripple suppression would have to be a matter of testing.

[Miles Prower]

Originally posted by Yugo
Design of Resonant Input Choke PSU

I have intention to design resonant choke input PSU for A class amp or a pre amp,but I approached some difficulties regarding the way and formulas for calculation of damping resistor and solving the resonances of the supply.

I would advise to simply fuggeddaboudit. This is a very "faddy" kind of thing. Of course, parallel resonant traps are used quite frequently in RF circuits. However, in those cases, it's needed to trap one frequency, or a narrow band of frequencies. Even though your ripple frequency is 100Hz, it's not a pure sinusoidal 100Hz signal. There are harmonics as well. Trapping out the 100Hz fundamental with a resonant trap will let through all the nasty, off resonant, harmonics moreso than a ripple filter of conventional design.

[EC8010]

Exactly. A resonant filter can be tuned to suppress the 100Hz fundamental (I did it once) but it then lets through increased levels of the harmonics compared to a standard LC filter. Nowadays, I'm so much more concerned about getting rid of those higher harmonics that I choose chokes partly for low shunt capacitance.

[Yugo]

Thank you very much for your explanation and suggestions.
The idea was to design resonant choke input because it can be made with a smaller core for the first L and than for the second chokes to wound them in several distributed sections thus decreasing the capacitance which can result in a couple thousands pF or maybe much less.
So in this way (I think this idea is not so bad),we can deal with all those nasty high order harmonics.
Although you say that the first choke will pass the nasty frequencies,I don't see reason for such worryies because at the end of the filter the last LC section will suppress the all unwanted harmonics.
Nevertheless,I admit I am not so 100% sure about this.
Have anyone made examination with spectral analyzer regarding this issues?
As per MJ 10nF-10K and 220 nF pair of caps will serve for suppresing turn on -off peaks from the diodes and their reaction with the chokes.This is major problem with SS diodes and L input,less with vacuum rectifiers and second chokes.
The PSU will be for Class A amp consuming around 100mA with B+ of app.300-400 V.The problem which arouse was the calculation of the deQ-ing first choke/resistor in series with the resonant cap.
Which formulae shall I use for this calculation?
1)Rdamp=L/(C*R)
2)Q=R*sqrt(C/L)
3)Rdamp=sqrt(L/C)


Best regards and Happy Easter!

Yugovitz

[Tweeker]

The pair of 220nF caps are an alternative to the 10k 10nF traditional approach. The 10k 10nF method degrades high frequency filtering.

[Yugo]

The 10k 10nF method degrades high frequency filtering.

....which means it is better to use as a final solution?

Regards,
Yugovitz

[Tweeker]

I designed this before I knew what your requirements were. This is the lowest DCR critically damped supply having mV ripple I could come up with without resorting to gigantic computer grade capacitors. The type specified are inverter grade electrolytics, with an oil motor run as the final cap. One could increase the value of the last resistor for phono duty if desired, though ripple is already <2mV. 150R would knock it down about a decade.

Hammond rates its chokes fairly conservatively, it may be possible to run at 100ma without the choke saturating on ac current peaks. Peak AC current in the choke should be around 26ma with 50hz power.

See diagram for preferred input choke snubbing method.

[Johan Potgieter]

Originally posted by Yugo
The 10k 10nF method degrades high frequency filtering.

No. The impedance of a 50uF capacitor at 200 Hz is only 16 ohm, decreasing as f rises. In series with the 10K that gives >60 dB reduction, and there are other capacitors further along the line. There is no trace of extra h.t. supply deterioration from these components on a scope/spec.analyzer.

The danger of input choke turn-off transients also exists with tube rectifiers (they turn off as quickly, but usually have higher peak inverse values than s. diodes). Depending on the Q and internal C of the choke, spikes can be several KV with high current amps. The 10 nF then also needs to be several KV, typically ceramic (capacitance drift or tolerance not important). The same precaution would be necessary with the caps over the diodes. The diodes are somewhat protected by the inductance of the power transformer windings, but that may cause ringing itself. It is pretty difficult to quantise this, as even lead inductance may enter into the picture with very short transients. It is also difficult to measure, as a scope probe with a few pF of shunt capacitance could influence the spike (and the probe could blow!).

After all that, I would not worry too much about choke input filter hum etc. reduction. I never had difficulty with ripple coming from that. As said, at the input or pre-amp stages where ripple would be most troublesome, there would have been several further decouplig RC pairs also in the h.t. line.

Regards.