Interesting. Where does the positive loop occur? If the choke offers a low impedance path for the signal then to where? Usually input stages are fed further down the chain via RC filters. The choke would be connected further up the chain closer to the rectifier, what audio signal is present there? The filters would shunt the signal to ground in the power supply I would think.
The cap to signal ground provides the low impedance and the choke help to isolate each stage of the power supply . Chokes store magnetic energy and want constant current cap want constant voltage . I prefer choke loaded diodes then caps . Down fall only about .9 of the ac voltage output rather than 1.4 with a cap first stage . This let me use larger caps with tube diodes after the choke without hurting the diodes. Many good power supply tutorials on the net. Go to duncanamps.com and look at the pusd2 it free and you can build sims of your power supply .
Yes, but you didn't answer my question.
The power supply can also be viewed as a positive feedback loop . Thus the reason for isolation in the power supply so that that positive loop does not effect other parts of the circuit So yes it does effect the signal. System view which is more complex than what many like to teach . Everything maters some just not very much .
Yes inductors increase impedance with increasing frequency but there is a cap there going to ground so high frequency impedance in the choke is moot, they have been shunted to ground via the decouple cap. I think you are saying the low frequencies could be an issue if the cap isn't sufficiently large and if the LC values have not been chosen accordingly giving a high resonant frequency within the audible range. I try and set the resonance of my LC filters <7Hz. e.g. 10H and 100uF 5Hz
If this is the positive feedback loop you were talking about then you did answer my question.
Yup, got it thanks.
Last edited:
ESR is a significant part of total capacitor impedance only at supersonic frequencies, e.g. 100KHz. Actual capacitance value dominates everything else at 20Hz, and this fact argues strongly for the use of big electrolytics in high-performance amplifiers. I know of three valid arguments against:
[1] 'Lytic lifetime is shorter than film, even when both are properly selected and installed. Considering the lifetime of vacuum tubes (or any high-end audio technology these days) this argument smells like red herring.
[2] 'Lytics aren't available with adequate voltage ratings for some tube amp designs. Series hookup easily overcomes this issue.
[3] Film cap ESR can be lower. That's true, but the advantage isn't huge anymore, and it's completely negated by using a small film bypass with a huge 'lytic.
There is sometimes justification for using choke-input filters when surplus iron is pressed into service, but this is an outlying case. In my view, iron-core filter chokes have been firmly obsoleted by advances in electrolytic capacitor manufacturing and I see no reason to use them in my own designs. My opinion might be different if similar advances had occurred in choke manufacturing.
The PSU can be an unwanted feedback impedance, typically at low frequencies including subsonics. Whether positive or negative depends on the circuit details and frequency. In this respect adding a choke can make things worse, if the designer does not think about all relevant frequencies.Triodethom said:
Since ESR=Xc*DF you can lower the ESR by increasing the cap value effectively lowering Xc. But increasing the cap value will increase ripple current.
When looking at some current production high voltage electrolytic caps for SMPS I noticed the DF was quite good. With such large ripple currents you don't want a lot of internal heat being built up.
I agree, modern electrolytics are a great value for good performance.
My gear has active CCS fed shunt regs for each gain stage. The shunt device has an impedance estimated on the order of 10s of milli Ohms, the CCS on the order of 10s of Meg Ohms. The gain stages are differential with inherently high PSRR. One would think that making device changes to the CCS (and presumably changing its AC impedance somewhat) would not be audible. That has not been the case. One of the benchmarks I use to assess changes is an increase in the intelligibility of song lyrics. Either you get the words or you don't.
True You have pointed out the largest point the designer need to look at all relevant frequencies not just the main ( 60 or 50 or it second harmonic ).
This can be very effective at filtering ripple, sometimes even better than necessary so no more should be needed.
The needs of the stages looking back into the supply aren't necessarily the same, but these can be arranged separately meeting the ripple filter at a point where the supply more or less sits at DC.
Interesting comments, I didn't realize it'll turn to such debate.
Anyway like I said my amps are dead quiet using just caps but I use high values over 150 uf in the filter section. Only one of my build has very minor hum I can' get rid of with caps. I used Voltage doubler on that build, so I blame that even tho I'm not sure why it would cause it. Maybe time to try a choke, I got some here from old radios. Unless it's something to do with the general layout but I tried to do a particularly a good one so I don't know. Usually when I just throw everything in the box on test bench it works the best till I start making things tidy 🙂
Anyway like I said my amps are dead quiet using just caps but I use high values over 150 uf in the filter section. Only one of my build has very minor hum I can' get rid of with caps. I used Voltage doubler on that build, so I blame that even tho I'm not sure why it would cause it. Maybe time to try a choke, I got some here from old radios. Unless it's something to do with the general layout but I tried to do a particularly a good one so I don't know. Usually when I just throw everything in the box on test bench it works the best till I start making things tidy 🙂
your point is valid, but..
Yours design is dependent on low parasitic resistances, which will make real measurements worse than calculation
Still, my 2 stage filter gets 80mVrms ripple output, how about yours?
It will fit on half A5 footprint (2chokes, 1 twincap), but dozens of serial-paralel caps maybe not.
My biggest fear, too high ripple currents; also reverse recovery gets worse if diode takes bigger forward current (peaks).
Probably consuption can get worse too.😕 (in similar fashion like no pfc supplies)
Really low PSU ripple is rarely needed, as earlier low signal stages will usually have extra decoupling/filtering anyway. A common newbie mistake is to design a multi-stage filter to get ripple down to microvolts, but then get the grounding wrong so they get bad buzz; meanwhile the multi-stage filter has added so much resistance (or LC resonance) in the supply that they then get subsonic instability.
Historical note: the Eico HF-60 power supply just had a 5AR4 rectrifier and 40uf of capacitance for the B+ (for the output tubes). No choke. Of course push-pull and feedback can reduce hum quite a bit.
Last edited:
- Status
- Not open for further replies.