Originally posted by Golgoth Stef, Bruno's arguments do seem to make sense. To actually hear the difference you could try testing again while deliberately creating mains disturbances eg by having somebody action a power drill somewhere in the house. Do keep us updated
I can try to connect a large drilling machine directly on the same power supply plug!
I prefer to have no problem.
OA51 said:
chrisb03 said:
wytco0 said:
tiki said:Hi,
I know, it's not my competence, in my humble opinion it would be better to avoid mixing the themes between the threads. I would kindly ask you to discuss these supply problems in the appropriate thread. It will be easier to find any special information later, not only for me.
Thank you very much indeed!
For minimum capacitor requirements see Bruno's post. I for myself do not see any disadvantage for too big rail capacitors, except overkill and inrush current without a softstart circuit.
Best regards, Timo
Golgoth said:
Just moving this topic to appropiate thread.
The snubber caps are base on existing designs. Link to one is already mentioned in this thread by ericpeters.
http://www.zero-distortion.com/techno/powersupply/powersi_05.htm
Chris
chrisb03 said:
one of the problems is probably the capacitor impedance vs frequency curve
only some batteries give good results ( a thread talked about alkaline )
optima batteries are renowned ( high current peak , very low ESR )
www.optimabatteries.com
alain
5 channel amp-supply
Okay,
I have read this stuff. I have 5 UcD400's. If I am correct I need the following (minimum) supply (my main speakers are rated 8 Ohm):
-> 5*400/2=1000VA toroid, or should I take the 210W/8Ohm?-> 5*210/2=525VA toroid)
-> 0.5*0.1/8=6250uF, 10 times for 5 channels (or 6 10,000uF caps)
Correct?
Frans
Okay,
I have read this stuff. I have 5 UcD400's. If I am correct I need the following (minimum) supply (my main speakers are rated 8 Ohm):
-> 5*400/2=1000VA toroid, or should I take the 210W/8Ohm?-> 5*210/2=525VA toroid)
-> 0.5*0.1/8=6250uF, 10 times for 5 channels (or 6 10,000uF caps)
Correct?
Frans
Re: 5 channel amp-supply
Cap sizing is correct, but transformer should be sized as follows:
- Power factor (W/VA) for this type of supply can be as low as 0.5, in other words you need 1/0.5=2 VA/W
- Watts needed: 5*210=1050
=> VAs needed = 1050W * 2VA/W = 2100VA
Note these computations leave zero margin for impedance dips: if you want to allow for impedance dips at half the nominal i.e. 4 ohms you should double the above values, both caps and toroid (BTW there is no point in oversizing further than a factor 2 as the UcD400 will never draw more than 450W anyway, thanks to its OV and OC detections)
FransDHT said:
Cap sizing is correct, but transformer should be sized as follows:
- Power factor (W/VA) for this type of supply can be as low as 0.5, in other words you need 1/0.5=2 VA/W
- Watts needed: 5*210=1050
=> VAs needed = 1050W * 2VA/W = 2100VA
Note these computations leave zero margin for impedance dips: if you want to allow for impedance dips at half the nominal i.e. 4 ohms you should double the above values, both caps and toroid (BTW there is no point in oversizing further than a factor 2 as the UcD400 will never draw more than 450W anyway, thanks to its OV and OC detections)
Hi,
Michel, apart from my possibly insufficient calculation (e.g. missing transformer impedance) I cannot fully agree to your statement about the higher recharging current. This current is, in my opinion, primarily dependent on the discharging one. The discharging current takes a tangible amount of charge from the caps while the approx. 10ms discharging time, independent of the cap value. So the recharging current should be determined by: the recharging time, the voltage drop (which is indirectly proportional to the capacitance), the capacitors combined impedance (ESR, ESL) and the source (transformer) impedance. The frequency dependent influence of the capacitance on capacitors impedance is constant due to the more or less constant recharging voltage rise. In fact, if the source voltage follows a sine function, the recharging current will vary over recharging time, so there is a dependance of the current waveform on the capacitance, but this is not valid for the averaged recharge current.
Again, That's only my opinion.
Regards,
Timo
Michel, apart from my possibly insufficient calculation (e.g. missing transformer impedance) I cannot fully agree to your statement about the higher recharging current. This current is, in my opinion, primarily dependent on the discharging one. The discharging current takes a tangible amount of charge from the caps while the approx. 10ms discharging time, independent of the cap value. So the recharging current should be determined by: the recharging time, the voltage drop (which is indirectly proportional to the capacitance), the capacitors combined impedance (ESR, ESL) and the source (transformer) impedance. The frequency dependent influence of the capacitance on capacitors impedance is constant due to the more or less constant recharging voltage rise. In fact, if the source voltage follows a sine function, the recharging current will vary over recharging time, so there is a dependance of the current waveform on the capacitance, but this is not valid for the averaged recharge current.
Again, That's only my opinion.
Regards,
Timo
chrisb03 said:
what i mean is that for the same value say 6800uF , large can capacitor is often better than smaller one
in your case , i would try 1x6800uF at first , and then 4x6800uF to define the impact on the sound
For example , if your speakers have already very damped bass , 4x6800uF will degrade the presentation
I think there is no rule for power supply caps , except a minimal and maximal value , but for normal listening levels at home , reasonable values often give better results
alain
Found Bruno's post in Ucd180 thread, post 294
"Yes you can overdo on caps, at least when the power supply is hard-rectified (normal supply without chokes). The current through the rectifiers becomes more and more spiky and the magnetic fields coming off the hookup wires and even mains cords can potentially start degrading the performance of the rest of the kit. Even if you can't hear the buzz directly, you do hear increased "veiling" of the sound stage. This is one of the reasons "audiophile" mains cables sometimes have an audible effect. I don't like it when mains cables start to make a difference.
The 20000uF per rail that I have on my "esoteric amps" is really as far as I dare take it. I'm planning on a PFC supply to address this."
"Yes you can overdo on caps, at least when the power supply is hard-rectified (normal supply without chokes). The current through the rectifiers becomes more and more spiky and the magnetic fields coming off the hookup wires and even mains cords can potentially start degrading the performance of the rest of the kit. Even if you can't hear the buzz directly, you do hear increased "veiling" of the sound stage. This is one of the reasons "audiophile" mains cables sometimes have an audible effect. I don't like it when mains cables start to make a difference.
The 20000uF per rail that I have on my "esoteric amps" is really as far as I dare take it. I'm planning on a PFC supply to address this."
chrisb03 said:
Good digging Chris, it confirms what I suspected: increased peak cap recharging current (i.e. rectifier current), which follows from increased rail capacitance, can be harmful. So until Bruno comes up with his PFC supply (PFC = Power Factor Correction) we should not exceed 20000µF per rail.
So the definitive supply capacitor sizing rule for a two-rail class D amp is:
(Number_of_channels * 50000µF / R) < Caps_per_rail < 20000µF
where R is the speaker impedance in ohms
Oops the forum doesn't like "mu" symbols obviously, I meant:
The definitive supply capacitor sizing rule for a two-rail class D amp is:
(Number_of_channels * 50000uF / R) < Caps_per_rail < 20000uF
where R is the speaker impedance in ohms
Which means, BTW, that a single power supply of this kind should not power more than 20000/50000*R = 0.4*R channels! That's a maximum of 0.4*8 = 3.2 channels -> 3 channels per supply for 8 ohm speakers, and 0.4*4 = 1.6 channels -> 1 channel per supply for 4 ohm speakers
The definitive supply capacitor sizing rule for a two-rail class D amp is:
(Number_of_channels * 50000uF / R) < Caps_per_rail < 20000uF
where R is the speaker impedance in ohms
Which means, BTW, that a single power supply of this kind should not power more than 20000/50000*R = 0.4*R channels! That's a maximum of 0.4*8 = 3.2 channels -> 3 channels per supply for 8 ohm speakers, and 0.4*4 = 1.6 channels -> 1 channel per supply for 4 ohm speakers
Golgoth said:
(very strange, my "mu" symbols display correctly now
Hi Eric,
No, it just means they would be even happier if they had a PFC supply, or if they shielded their modules from the spiky rectifier current disturbances (faraday cage around each module).
Note that, as Bruno hinted at in his post quoted by Chris, the rectifier current spikes could be smoothed out by inserting suitable chokes upstream of the capacitors. This would double up as a nice mains filter BTW.
Correct me someone if I am wrong, I find that a 2mH choke inserted between the rectifier and the hot end of the 50000uF rail capacitor would yield a low pas filter with a cutoff frequency 1/(2*pi*SQRT(L*C) of 16Hz, i.e. well below the 100 or 120Hz full wave rectifier current frequency.
Mmmm... it does look like a nice and simple way to achieve both passive PFC and mains filtering with oversized storage capacitors, but we would have to make sure that we don't run into nasty overvoltages when consumption changes suddenly, eg when a silence follows a loud thump. In other words we would need to make sure there is enough damping resistance in the RLC circuit: this may be a case where excessively low capacitor ESR could harm
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.