Legis
Minor Hero
Posts: 33
|
Post by Legis on Nov 17, 2010 19:52:43 GMT -5
If this has been clarified long ago in some god forsaken thread, forgive me, because I just registered to the forum.  Legis says hi to everybody! I have been reading some post regarding the parallel/series wiring of the capasitors of the XPA-1 and XPA-2 amplifiers. First of all, it is true that both amps use this wiring technique to boost caps' voltage rating. If the voltage rails were only 63V, would that mean that the maximum output to 8 ohms would be ~480-490W. Parallel series wiring doubles the voltage rating, which is why it has been necessary (and beneficial...) to implement. Second of all, it is also true that this wiring technique lowers to total capacitance to one fourth of the filter caps' total value, meaning the secondary capacitance of XPA-1 is 30,000µF and XPA-2 45,000µF. Thirdly (I think this fact has been overloooked) the fact I want to emphasize is that parallel/series wiring tehnique actually lowers the total charge of the capacitor only by half while it keeps the total amount of energy, that the capacitors can store, the same as is in all-parallel wiring of the caps.The total amount of charge (Q) of the capacitor(s) can store, is calculted: Q = C*V (capacitance in fards times the voltage rating of the capacitor in volts)
Now when we calculate the total charge of XPA-1's 12pcs 10,000µF capacitors wired in all-parallel-style and in series/parallel-style we see that the charge only drops to half when wiring the caps in parallel/series:
all-parallel-style: 12*10^-2 F x 63V = 7,56 coulombs
series/parallel-style: 3*10^-2 F x 126V = 3,780 coulombs
However the charge is always about the total energy (E) that is stored, so let's calculate it in both wiring techniques. The total energy is calculated with formula E = ½CV²
all-parallel-style: 0,5 *12*10^-2 F * 63^2 = 238,14 joules
series/parallel-style: 0,5 * 3*10^-2 F * 126^2 = 238,14 joules
(Little perspective how much power this amount energy can produce: The transformer charges the capacitor bank every 20 milliseconds in alternating current mains network of 50hz. If the whole energy of the capacitor bank is dissipated in 20 milliseconds the generated power is: P = (238,14 Joules) / (5*10-3 s) = 47628W = 47,6kW. If we let the capacitor bank's current fluctuate only by 10% between the cycles, the generated power would be 4,76kW.)
The point: This shows that the parallel/series wiring does not affect the total energy storage of the capacitor bank, nor it does not hinder the performance in any way. Actually the raised voltage rating does drop the distortion generated by the caps, so wiring capasitors in parallel/series is actually beneficial compared to all-series wiring with the same capacitors. 
References:
1) www.kpsec.freeuk.com/capacit.htm
2) hyperphysics.phy-astr.gsu.edu/hbase/electric/capeng.html
3) www.engineeringtoolbox.com/capacitors-energy-power-d_1389.html
|
|
|
|
Post by Porscheguy on Nov 17, 2010 20:23:28 GMT -5
I have absolutely no idea what you just said...... And your point is..... :-)
|
|
Legis
Minor Hero
Posts: 33
|
Post by Legis on Nov 17, 2010 20:38:59 GMT -5
I have absolutely no idea what you just said...... And your point is..... :-) Yes, my english can be hard to understand sometimes ;D. I highlighted the point at the last paragraph. |
|
|
|
Post by johndavidson on Nov 17, 2010 21:01:50 GMT -5
The problem is you are making the comparison at 2 different voltages. If we know the operating voltage of the secondary on the transformer and make the comparison with the 2 different capacitances there is a huge difference. The capacitors will charge up only to the peak voltage of the secondary. An example would be 45000uf at say 80v or 120000uf at 80v. The series capacitors would only charge up to half the transformer output voltage each.
|
|
|
|
Post by strindl on Nov 17, 2010 21:41:26 GMT -5
The XPA-2 and the XPA-1 use different voltage and temperature rated capacitors. Here are the ones in the XPA-2: 50V...there are 12 of them at 15,000 uF each...  here are the ones in the XPA-1.... 63V...12 of them at 10,000uF each.  one is wired in series/parallel combination, the other is wired in parallel |
|
|
|
Post by strindl on Nov 17, 2010 22:09:10 GMT -5
|
|
|
|
Post by sharkman on Nov 17, 2010 22:55:47 GMT -5
I have absolutely no idea what you just said...... And your point is..... :-) Yes, my english can be hard to understand sometimes ;D. I highlighted the point at the last paragraph. Legis, thank you for that info, I had forgotten what the voltage does in each type of circuit. I had wondered about this issue with some of the recent posts. |
|
Legis
Minor Hero
Posts: 33
|
Post by Legis on Nov 18, 2010 7:18:46 GMT -5
The problem is you are making the comparison at 2 different voltages. If we know the operating voltage of the secondary on the transformer and make the comparison with the 2 different capacitances there is a huge difference. The capacitors will charge up only to the peak voltage of the secondary. An example would be 45000uf at say 80v or 120000uf at 80v. The series capacitors would only charge up to half the transformer output voltage each. Yes, this is true. The maximum usefull operating voltage of xpa-1 is approx 70 Volts (the rails' maximum might still be 85v+). At this voltage the xpa-output something like 610 watts into 8 ohms. With 70 volts of operating voltage the maximum energy that the caps can store is: E = 0,5 * 3*10^-2 F * 126^2 = 73,5 joules (Btw. I had calculated the power wrong in the first post. The correct values are:
"P = (238,14 Joules) / (2*10-3 s) = 119070W = 119,1kW. If we let the capacitor bank's current fluctuate only by 10% between the cycles, the generated power would be 11,9kW"[/i])[/size] With 73,5 joules at 70V, the maximum power (of parallels/series wiring) between 20 millisecon cycle is: P = (73,5,14 Joules) / (2*10-3 s) = 36750W = 36,8kW. With max 10% fluctuation on the DC-current we get 3,68kW. What this means in imaginary max power?Let's say that the XPA-1 could output 1400watts in 2 ohms and 1700 watts in 1 ohm. 1700 watts output would mean something like 2600-2800 power consumption. Theoretically the line voltage vould be something like 52 volts at this time. At 52 volts the capacitors can store ~40,5 joules and these will translate into 20,2kW max power draw in 20 milliseconds. With 2800 watt of power draw, the DC-current would fluctuate 2,8kW/20,2kW = 13,8%. If the total capacitance would be 120,000, they could store 162,2 joules, which would translate into max draw of 81,1kW, and thus the DC-current fluctuation at 2800 draw would be 2,8kW/81,1kW = 3,45%. Let's see things at one watt. The operating voltage of the amp is 2,83 volts. this gives us the maximum energy stored in capacitors:
all-parallel-style: 0,5 *12*10^-2 F * 2,83^2 = 0,481 joules
series/parallel-style: 0,5 * 3*10^-2 F * 2,83^2 = 0,120 joulesThat amount of energy translates into power like this: all-parallel-style: (0,481 joules) / (2*10-3 s) = 240W series/parallel-style:P = (0,12 joules) / (2*10-3 s) = 60W If the draw is only a watt the DC-current fluctuates only: all parallel: 1/240 = 0,4% parallel/series: 1/60 =1,7% Conclusions and how does voltage sag affects the rms outputWith series/parallel wiring of XPA-1's caps, the DC-current in the amp fluctuates between 1,7% at 1 watt in 8 ohms and 13,8% at imaginary max load in 1 ohms (1700w output, 2800w draw). 13,8% drop in 2800w draw would drop the DC-voltage from 52 volts to 44,8, which would indicate a minumun amount of power draw into 1 ohms is approx. 2000 watts. The true RMS power draw would be something between 2800 and 2000 watts (something between 1700 - 1200 watts output at the speaker terminals) With all-parallel wiring of the xpa-1 caps, the DC-current in the amp fluctuates approx. between 0,4% at one watt at 8 ohms and 3,45% at imaginary max load of in 1 ohm (1700w output, 2800w draw). Whit this case the DC-voltage vould drop from 52V to 50V and the true RMS power draw would be something between 2800 and 2500 (something between 1700 - 1500 watts output at the speaker terminals). With music siugnal however the voltage sag does not have that great of impact. |
|
|
|
Post by johndavidson on Nov 18, 2010 10:12:30 GMT -5
The max RMS voltage at the outputs to the speakers would need to be Sqrt of P*R or 64V. The peak voltage on the rails would need to be 1.4 time VRMS or 90Vp. This would not take in to account any voltage drops across the transistor bias resistors and CE drops. This would vary based on design but you would need to add to 90Vp which would put you around 95Vp or so. This would be the lowest voltage that you could allow on the mains without having the ripple break into the output signal. So, the peak charge voltage would still need to be higher, 100Vp as an example. Now you caps would have to have enough capacity to maintain full load current draw without dropping below 95Vp. At full power the amp would need to be able to deliver 8 amps into 8 ohms with a voltage charge drop of 5 volts. It would need to do this for a time that would be defined as the instantaneous voltage of the input wave at 100Vp until the input wave returned to 95Vp. So then the problem becomes a delta V, delta T problem for a given current of 8 amps to determine the required capacitance need. Of course this will be different based on 50 or 60 hertz input power and half or full wave rectification. You would need to know the exact operating voltages of the supply and amp to do an accurate calculation.
Of course on the XPA1 I think the caps are all wired in parallel. If my understanding of the differential design is correct the 63V caps should do the trick. As I understand it the speakers lead are tied back to the opposing outputs of 2 amps accentually. When one amp is going positive the other is going negative by the same amount there by creating a difference of potential that would be double that on a none differential amp output.
|
|
Legis
Minor Hero
Posts: 33
|
Post by Legis on Nov 18, 2010 15:59:24 GMT -5
Of course on the XPA1 I think the caps are all wired in parallel. If my understanding of the differential design is correct the 63V caps should do the trick. As I understand it the speakers lead are tied back to the opposing outputs of 2 amps accentually. When one amp is going positive the other is going negative by the same amount there by creating a difference of potential that would be double that on a none differential amp output. Thanks for clarifying! What you mention above is very interesting, and very logical also. I actually use the same basic principle in my balanced/center tapped DIY isolation transformer. The two outer 115V taps are 180 degrees out of phase and create a total of 230V potential difference (and the two center taps are shorted and connected to the ground). With this knowledge I think there's no doubt that the total capacitance of XPA-1 is adequate to drive even 1 ohm load up to 1700-2000 watts RMS, if the transformer is up to the job.  I think that 1,2kVA tranny can output something like 3kVA for about half an hour and 6kVA for couple of minutes. Ps. what comes to the size of the transformwer in XPA-1, it is about 0,8 inch deeper and 0,8 inch wider than the transformer in Cambridge audio 840W, which is also stated to have a 1,2kVA transformer. I owned the 840W before XPA-1. Maybe Emotiva's statement of the transformer's size is little underrated, or it has bigger empty space in the middle. |
|
|
|
Post by johndavidson on Nov 19, 2010 11:40:13 GMT -5
I personally would not be comfortable running a xformer in an over load condition for the times you give but then thats me! Once the insulation on the windings starts to breakdown you are going to have a chunk of scrap metal. There is no doubt the XPA1 is Emotiva's show piece but I can hardly see the average home audio enthusiast every needing that much power. I as I have gotten older I am trying to preserve my ears. ;D As for the size differences between the 2 xformers, it could be due to many different things: the quality of the core, the specified unloaded to full load voltage regulation, the different operating voltages of the amps. |
|
|
|
Post by Porscheguy on Nov 19, 2010 12:34:06 GMT -5
I personally would not be comfortable running a xformer in an over load condition for the times you give but then thats me! Once the insulation on the windings starts to breakdown you are going to have a chunk of scrap metal. There is no doubt the XPA1 is Emotiva's show piece but I can hardly see the average home audio enthusiast every needing that much power. I as I have gotten older I am trying to preserve my ears. ;D As for the size differences between the 2 xformers, it could be due to many different things: the quality of the core, the specified unloaded to full load voltage regulation, the different operating voltages of the amps. My XPA-1's arrive today.... |
|
LCSeminole
Global Moderator 
Res firma mitescere nescit.
Posts: 20,497
|
Post by LCSeminole on Nov 19, 2010 17:51:43 GMT -5
This thread has definitely taken me back to my days in college at Florida State University. I now remember the reasons for selecting Applied Mathematics and Statistics as my dual majors and Physics as my minor. Physics just made my head spin way more. ;D
However I have found Legis and Johndavidson's discussion/work on the XPA-1 & XPA-2 quite interesting, and I actually had to read it a few times to get through with an understanding. I actually went and dug out my Kepler Physics book to refresh my memory, so thanks guys for my homework assignment! ;D
|
|
Legis
Minor Hero
Posts: 33
|
Post by Legis on Nov 20, 2010 9:39:06 GMT -5
I'm beginning once again to suspect that the capacitors are not all wired in parallel in XPA-1. The efficiency of the amp could be something like 65%. Now 600 watts output at 8 ohm speaker (we still have to calculate with 4 ohms) would mean ~930W draw power. We have to calculate with halved impedance because XPA-1 is a bridged amp and both of it's amps see a 4 ohms load. 930W equals approx 61Vrms power into 4 ohms. 61Vrms = 86,2V peak, which comes from the rectifier and the caps should be able to withstand it. So the caps cannot be in parallel. I also opened (for the science ;D) one of my XPA-1s to check out the wiring of the caps, and it seems that they are not in parallel as suspected. Check the pics out from DIYaudio: www.diyaudio.com/forums/solid-state/177541-question-regarding-differential-amplifiers.html#post2371535I think that the output in XPA-1 is limited because of the capacitance. If one would change the caps to 15,000µF like on XPA-2, I think XPA-1 would give equal output into 8 ohms as XPA-2 gives bridged into 8 ohms. Ps. I hope I'm wrong with this |
|
|
|
Post by richardyc on Nov 20, 2010 9:53:50 GMT -5
wow, those soldering job looks like it's done by a 12yr old. I am surprised Emotiva doesn't call them 'hand made' and charge you few hundred bucks more.  |
|
|
|
Post by johndavidson on Nov 20, 2010 10:18:38 GMT -5
I'm beginning once again to suspect that the capacitors are not all wired in parallel in XPA-1. The efficiency of the amp could be something like 65%. Now 600 watts output at 8 ohm speaker (we still have to calculate with 4 ohms) would mean ~930W draw power. We have to calculate with halved impedance because XPA-1 is a bridged amp and both of it's amps see a 4 ohms load. 930W equals approx 61Vrms power into 4 ohms. 61Vrms = 86,2V peak, which comes from the rectifier and the caps should be able to withstand it. So the caps cannot be in parallel. I also opened (for the science ;D) one of my XPA-1s to check out the wiring of the caps, and it seems that they are not in parallel as suspected. Check the pics out from DIYaudio: www.diyaudio.com/forums/solid-state/177541-question-regarding-differential-amplifiers.html#post2371535I think that the output in XPA-1 is limited because of the capacitance. If one would change the caps to 15,000µF like on XPA-2, I think XPA-1 would give equal output into 8 ohms as XPA-2 gives bridged into 8 ohms. Ps. I hope I'm wrong with this 6 caps are connected in parallel to provide the positive supply and 6 caps are connected in parallel to provide the negative supply if I am looking at you diagram correctly. This is exactly what I would have expected. When manufacturers list capacitance in the supply they generally total the capacitance on both the positive and negative sides of the supply. So, when Emotiva claims 120000uf to me that means 60000uf per power rail. On the XPA2 the 45000uf capacitance would also be split between the +,- supply rails or 22500uf per rail. Sorry if I implied that there was in fact 120000uf per rail.  |
|
Legis
Minor Hero
Posts: 33
|
Post by Legis on Nov 20, 2010 10:37:33 GMT -5
6 caps are connected in parallel to provide the positive supply and 6 caps are connected in parallel to provide the negative supply if I am looking at you diagram correctly. This is exactly what I would have expected. When manufacturers list capacitance in the supply they generally total the capacitance on both the positive and negative sides of the supply. So, when Emotiva claims 120000uf to me that means 60000uf per power rail. On the XPA2 the 45000uf capacitance would also be split between the +,- supply rails or 22500uf per rail. Sorry if I implied that there was in fact 120000uf per rail. So I was wrong, lol ;D. I was looking for some kind of all 12 caps in parallel with each other type of wiring . This is what happens when a layman looks at a PCB. |
|
I think that 1,2kVA tranny can output something like 3kVA for about half an hour and 6kVA for couple of minutes.
