What caps for V3
I am struggling to come to a conclusion on what type and value of the PSU caps I need to use per rail per channel of Turbo v3. I tried the NP suggested 10000uF, 63V caps in CCCCRCCCC for each rail in the V2 variant and I still got hum related to ripple. I am saying that this is ripple related hum because it disappears at lower bias levels. The Toroid I have used for V2 is a 1000VA, 24V at secondaries.
Either my caps are fake or are not upto the quality expected. What is the spec I should look for in terms of ESR and Ripple Current? Is there a minimum value I need to look for? Also any suggestions on the brand and model of the caps which are affordable for a DIYer? I have seen some fancy caps being sold for over $150 per pair, cannot afford that.
I may be asking questions which are asked over a thousand times on this forum, but I am looking for help from a F5T perspective. Also I am not an expert in caps and power supplies. Please bear with my ignorance.
Thanks.
I am struggling to come to a conclusion on what type and value of the PSU caps I need to use per rail per channel of Turbo v3. I tried the NP suggested 10000uF, 63V caps in CCCCRCCCC for each rail in the V2 variant and I still got hum related to ripple. I am saying that this is ripple related hum because it disappears at lower bias levels. The Toroid I have used for V2 is a 1000VA, 24V at secondaries.
Either my caps are fake or are not upto the quality expected. What is the spec I should look for in terms of ESR and Ripple Current? Is there a minimum value I need to look for? Also any suggestions on the brand and model of the caps which are affordable for a DIYer? I have seen some fancy caps being sold for over $150 per pair, cannot afford that.
I may be asking questions which are asked over a thousand times on this forum, but I am looking for help from a F5T perspective. Also I am not an expert in caps and power supplies. Please bear with my ignorance.
Thanks.
With that much capacitance I suspect that you might have fake or very low quality caps.
You don't need to pay exorbitant amounts for caps, but it is nice to know that you are getting genuine parts. Panasonic, Nichicon, Rubycon, Cornell-Dublier all make caps in that value and although not inexpensive, nowhere close to $75 each.
Something else occurred to me - how are your grounds set up? My first F5 was hummmm prone until i adjusted the location of the PSU and speaker ground connections...
You don't need to pay exorbitant amounts for caps, but it is nice to know that you are getting genuine parts. Panasonic, Nichicon, Rubycon, Cornell-Dublier all make caps in that value and although not inexpensive, nowhere close to $75 each.
Something else occurred to me - how are your grounds set up? My first F5 was hummmm prone until i adjusted the location of the PSU and speaker ground connections...
bigE those numbers are correct. exept for the last one. the rail voltage is to low to reach 144W at 8ohm. same for number 2and as buzzbord said. transformer rating of 3times dissipation is a good way to go. 2 times is minimum. i like 3 times. that gives you more headroom and lower temp in the transformer. but when a transformer reach a certain size, then bigger is not allways better
I can't get 144 Wrms(288 peak) out of a 50 Volt rail? Why not? 50 volts through 8 ohms is just over 6 amps.
6*6*8 = 288..... What am i missing?
buzz,
Thanks for the ESP site reference, I'm going to check it out.
I get the lowest ESR, and the highest ripple current of the size and value I find. I use Panasonic from Mouser:
ECE-T1JA273FA Panasonic Electronic Components | Mouser
At $20.48 each, fairly cheap when you figure 27,000 uF at 63 Volts, 6.59 amps ripple current, and ESR: 0.017 ohms.
That would be $7.59 per 10,000 uF.
Rush
Thanks. Looked at the caps. Seem pretty good. Digikey is cheaper by $5 each. Ordered 16 of them.
Cheers.
In some cases, the mere fact that the F5 away from other devices may result in loss of humm.
The power grid has several harmonics from various devices used by itself and neighbors, it also influences the humm, and is difficult to alleviate.
I advise the use of a filter Electrical network 30A of Lambda (TDK is now), this filter does wonders throughout the audio system.
I use Lambda MXB-1230-30A 30 where I made a box with outlets
Thanks, I will check out the Lambda stuff. I can hear my Toroid making different noises when my water pressure pump keeps going on and off....
BigE.
you will not get 50VAC on the output with 50VDC rails
but you are not far off
How do you make the adjustment?
If you want the schematic say, I send by PM.
Attachments
for 144W of ClassA power you will need a bias of ~3.1A
the PSU will need to maintain ~ +-53Vdc while delivering that 3.1A of bias.
The ripple on the supply rails will be considerable.
I would expect a 0-38Vac, 0-38Vac 900VA (12Aac) transformer to maintain that +-53Vdc, with acceptable ripple, if you have +-50mF to +-100mF for a monoblock assembly.
Another unhelpful post to augment my previous one?
the PSU will need to maintain ~ +-53Vdc while delivering that 3.1A of bias.
The ripple on the supply rails will be considerable.
I would expect a 0-38Vac, 0-38Vac 900VA (12Aac) transformer to maintain that +-53Vdc, with acceptable ripple, if you have +-50mF to +-100mF for a monoblock assembly.
Another unhelpful post to augment my previous one?
6times to 10times is often stated for a ClassA power Amplifier.
Another unhelpful post?
AndrewT,
It would be helpful to know how you obtained your numbers?
In the F5 article, Pass uses:
Peak watts = (I^2*R) where I = 2*bias and R=8 ohms.
He then divides this by two for rms power. When using a 3 amp bias, that is (6 * 6 * 8 )/2 =144 Wrms. This can be achieved by a rail voltage of V = I*R = 48 volts.
This is into a purely resistive load. We know speakers are not like that. The real output power will vary with power factor - a measure of the phase difference between V and I. But, pass ignores that in his calculatuion of power for the F5, as do I.
Also, the notion of 6 to 10 times is 6 steady state and 10 times peak. Which makes me think the 10 applies to class a/b operation, not the. Steady state class A.
It would be helpful to know how you obtained your numbers?
In the F5 article, Pass uses:
Peak watts = (I^2*R) where I = 2*bias and R=8 ohms.
He then divides this by two for rms power. When using a 3 amp bias, that is (6 * 6 * 8 )/2 =144 Wrms. This can be achieved by a rail voltage of V = I*R = 48 volts.
This is into a purely resistive load. We know speakers are not like that. The real output power will vary with power factor - a measure of the phase difference between V and I. But, pass ignores that in his calculatuion of power for the F5, as do I.
Also, the notion of 6 to 10 times is 6 steady state and 10 times peak. Which makes me think the 10 applies to class a/b operation, not the. Steady state class A.
I agree. This is a lot of real estate being spent on toroid. This thread cannot be just that for one individual's clarity. One can use the PM route.
So far, I've been the only person that has been trying to formulate things, obviously incorrectly. AndrewT responded valuably to my first effort, from which I made formulas... but AudioSan says they're off by a factor of two.
From the problems in formulating this, either people are not capable of doing it, or they just won't tell anyone how. Are you really content with someone just posting numbers without knowing where they come from? Maybe you can DIY without knowing how to size a transformer and blindly follow others, but I can't.
From the problems in formulating this, either people are not capable of doing it, or they just won't tell anyone how. Are you really content with someone just posting numbers without knowing where they come from? Maybe you can DIY without knowing how to size a transformer and blindly follow others, but I can't.
