If you parallel those caps, you'll increase the total pain inflicted on the capacitors.
When you parallel caps or otherwise increase PSU first cap capacitance, you'll get yourself closer for the PT secondary 'seeing' a short. This will increase the ripple current the first cap has to deal with, not decrease it.
Of course for better filtering, and countering the above mentioned increased pain, just increase the first resistor after rectification and before first cap.
When you parallel caps or otherwise increase PSU first cap capacitance, you'll get yourself closer for the PT secondary 'seeing' a short. This will increase the ripple current the first cap has to deal with, not decrease it.
Of course for better filtering, and countering the above mentioned increased pain, just increase the first resistor after rectification and before first cap.
Certainly not.
Of course it depends a little on the conditions, but if you double the number of "first" caps, ripple current will be anywhere in between a little less (if using very little capacitance) to roughly half (if using decent capacitance).
Under no conditions will ripple current across each cap go up if you add more caps.
Of course, the most reasonable approach would be to split the total capacitance over several caps, keeping the total capacitance the same.
Hello,
I don't use a capacitor input ( first part after the rectifier) ever, and I always try to avoid adding series resistances in the filter chain to the Finals, like the plague. ALL my Ls are 20 Ohms or less in DCR.
The reason for the careful bypassing is because all caps are the worse part in audio, are JUNK, are narrow band. None are capable to play back linearly, full range musically, "all" the music, in a SE DC amp.
Jeff Medwin
Strike my previous comment. You're of course completely right; my brain was not with me when I wrote that.
The individual current load per capacitor cannot increase, but the total load does - this can be bad for the PT or the rectifier.
Member
Joined 2009
Paid Member
Yes, BUT....
Look at the circuit in post 175 above, the first filter cap is rather isolated from the rest of the amplifier. Between this first filter cap and the rest of the amplifier is a i) pair of paralleled hollow-state diodes, ii) a choke and iii) at least one more filter cap.
EDIT: I'm probably going to add another filter cap between the hollow-state rectifier and choke. A quick simulation shows that a 20uF cap here (well within the ratings for the rectifier) will be very effective in further reducing h.t. ripple and does not adversely impact the transient response of the power supply.
It seems to me that the first filter cap is as much IN the circuit as the mains substation down the street so why worry too much ? What worried me was this first filter cap has to be a lot more robust than your average tube amplifier power supply - it is directly after some low impedance SS diodes. But folk are saying the Atoms are up to the task which is good to know since I have several of them with no other use.
Last edited:
I would skip the ATOM cap and go with a good low ESR cap designed for line power supplies.
The ATOM caps have rather high ESR in relative terms compared to modern caps designed for switcher power supplies.
I can't even find a datasheet listing ESR or dissipation for the ATOM series, no max ripple current, no nothing.
Look at Nichicon VX series for comparison, at least you can get some useful information in the datasheet, like maximum ripple current, dissipation factor, etc.
I like the KHM series better (105C), lower ESR and dissipation factor, but not as good a selection for low capacitance values.
The ATOM caps have rather high ESR in relative terms compared to modern caps designed for switcher power supplies.
I can't even find a datasheet listing ESR or dissipation for the ATOM series, no max ripple current, no nothing.
Look at Nichicon VX series for comparison, at least you can get some useful information in the datasheet, like maximum ripple current, dissipation factor, etc.
I like the KHM series better (105C), lower ESR and dissipation factor, but not as good a selection for low capacitance values.
Hello,
You are not understanding what I am talking about. A single 20 to 50 uF supply cap in a SE DC amp B+ filter can NOT play full range music.
The addition is only four or five small uF value film caps, which effect differing frequencies, such that the amp plays the speaker linearly. In almost all cases, I am adding no more than 1 uF TOTAL of selected ( carefully , by ear ) values, to get the amp to play music.
How much of a stress can it be, with the tube rectifier seeing a choke input filter, going from 20 uF to 21 uF, or 50 uF to 51 uF ?
That which you bring-up is a non issue, and not what I was discussing at all.
Jeff Medwin
Time for you to download and study wide band WIMA DC LINK capacitors !! Industrially priced also.
Jeff Medwin
I don't bother with PP caps for the first filter, only for later stages close to the amp stage.
Low ESR caps designed specifically for line power supplies work better in my opinion. Or at least are at better price-point for the function.
For example the 100uF 400V EKXJ451ELL101ML50S cost $3.77 q1, compared to the 80uF 700V (not available in 400V) MKP1848680704Y5 at $32.98.
Low ESR caps designed specifically for line power supplies work better in my opinion. Or at least are at better price-point for the function.
For example the 100uF 400V EKXJ451ELL101ML50S cost $3.77 q1, compared to the 80uF 700V (not available in 400V) MKP1848680704Y5 at $32.98.
Last edited:
I was not responding to you, but to Bigun in post #176.
I find parallelling caps a very bad idea in almost all cases. Only exception I make is before the regulator; I put one film cap in the raw supply, capacitance not critical, something like 220nF or more sometimes. It's role is to suppress oscillation. I tend to put it close to the secondary, often parallel the first cap. I also put a small 1600V film cap right across the secondary, before rectification, for the same purpose.
After the regulator, one big film cap. Usually 10µF 600V MBGO type, or sometimes 20µF / 400V. This is not needed (it's for oscillation suppression purposes), and not in the signal path, since my amps draw the exact same current in all normal (non clipping) conditions.
But in the actual audio circuit - I never parallel caps, never. I get full audio range and more, above and below hearing.
Look into properly driving your tubes' grids, for starters.
Member
Joined 2009
Paid Member
I find myself in agreement with this. If I were buying parts from scratch this is what I'd do. Since I have the Sprague's to use, I may as well do so.
Maybe you just found the secret sauce for Jeff's infamous full frequency sound... 😉😉😛😀:
Decca were first; but they used resonance prone sheet-metal bodied cartridges to achieve ffss 😀
On another note, here are a few tests on the source impedance of some supplies by Bjorn Kolbrek.
..and it's still a choke input supply with a small capacitor up front
Have used this style with HT, bias supplies, heaters and solid state amps with consistent results.
..and it's still a choke input supply with a small capacitor up front
Have used this style with HT, bias supplies, heaters and solid state amps with consistent results.
Member
Joined 2009
Paid Member
Power transformer ringing ? you'll see an RC snubber in my design schematic 🙂
But the other issue is the resonance you get from the h.t. choke and filter caps. The usual recipe is a large L and C to push this frequency low as possible but the approach I'm taking does not use a large L or particularly large C so care is needed.
But the other issue is the resonance you get from the h.t. choke and filter caps. The usual recipe is a large L and C to push this frequency low as possible but the approach I'm taking does not use a large L or particularly large C so care is needed.
With regards to having a high series impedance that has a low component of series resistance, what you'll get is lower AC support from the previous capacitor, or higher AC isolation. The flow between them will largely be DC needed to replenish consumption so the later capacitor should be sized for supporting the load. It could also be bypassed.. at least to some frequency using a small decoupling impedance which might be sized to swamp parasitics (for example). Thus leaving one to two capacitor positions significantly involved in the signal path.
This is where you'll find DCR useful in the early stages. The required damping being whatever is needed to do the job.
Member
Joined 2009
Paid Member
