In the context of designing the PSU. Is there an advantage of using a single large reservoir capacitor of adequate size per rail vs using a multitude of sizes in series or parallel to archive the same capacitance?
As an example if a design has the need for 10,000uf per rail, would it be beneficial to have a single 10,000uf cap, or 2x5,000uf capacitors? Or is it even better to go both ways and over compensate with a larger capacitor (say 8,000uf) in conjunction with a couple of mid and smaller (say 3,000uf & 1,000uf).
Or am I just over complicating the whole thing for no reason and just go with what I can find most economically?
Thanks,
- Bill
As an example if a design has the need for 10,000uf per rail, would it be beneficial to have a single 10,000uf cap, or 2x5,000uf capacitors? Or is it even better to go both ways and over compensate with a larger capacitor (say 8,000uf) in conjunction with a couple of mid and smaller (say 3,000uf & 1,000uf).
Or am I just over complicating the whole thing for no reason and just go with what I can find most economically?
Thanks,
- Bill
Theoretically, multiple small caps in parallel would result in a lower ESR than the single large cap, improving the capacitor bank's ability to dump large amounts of current for large/fast transients.
You'd need to look at the capacitor specs and do some math to figure if this would yield a benefit at the frequencies of interest. You also need to consider the inductance of the path between the cap and the load (the output transistors).
You'd need to look at the capacitor specs and do some math to figure if this would yield a benefit at the frequencies of interest. You also need to consider the inductance of the path between the cap and the load (the output transistors).
Caps are usually paralleled to gain an advantage of lower esr Usually only needed for SMPS output filters because of their high switching frequencies and limits of loop stability. (smaller parts work better at HF).
For linear power supplies no real advantage to parallel here because it costs more for OEMs w/o much to gain. Most quality bulk caps will spec esr at 120Hz. BTW higher temp rating are indicative of further cap quality.
For linear power supplies no real advantage to parallel here because it costs more for OEMs w/o much to gain. Most quality bulk caps will spec esr at 120Hz. BTW higher temp rating are indicative of further cap quality.
for DIY costs are different
part close-outs deals
added PCB's, standoffs, size changes, wiring changes, reliability hits aren't usually caculated for total cost comparisons.
Believe me if the costs were lower you'd see it in OEM products. Their margins are so low as it is, they look at the cost everything 3 times with a magnifiying glass.
part close-outs deals
added PCB's, standoffs, size changes, wiring changes, reliability hits aren't usually caculated for total cost comparisons.
Believe me if the costs were lower you'd see it in OEM products. Their margins are so low as it is, they look at the cost everything 3 times with a magnifiying glass.
Deofol said:
For the main power supply, I think it is best to only use identical capacitors to maintain balanced current draw. For power amps, I think it is best to use two identical 105C rated capacitors per rail voltage, since this allows a small common ground bus bar. Almost every power amp with +/- rails uses 4 large capacitors screwed to a common star-ground copper bus bar. For superior high frequency response, you can solder in smaller secondary filter capacitors right next to each output transistor.
If you check the 105C capacitor reliability data you will find that large can-type screw terminal capacitors have significantly higher MTTF than small solder down caps. Ten capacitors each with a 10 year MTTF might have a combined 8 year MTTF.
this article a huge difference, back in the day (1980), and still does -
http://www.reliablecapacitors.com/pickcap.htm
178
http://www.reliablecapacitors.com/pickcap.htm
An externally hosted image should be here but it was not working when we last tested it.
178
Re: Re: Re: PSU Reservoir Capacitors (Single large vs Multiple small)
Not necessarily. You have to look at the FITS (failures per 10^9 device hours) for each specific device multiplied by the quantity. For the entire system,
MTBF = 1/ sum[ (FITS1 * # device1) + (FITS2 * #device2) + ...]
It is quite possible that multiple devices of high reliability will have a better aggregate reliability than a single device of merely average reliability. Also, the failure rate of just a couple of parts may dominate the calculation, making the quantity of other parts irrelevant.
infinia said:
Not necessarily. You have to look at the FITS (failures per 10^9 device hours) for each specific device multiplied by the quantity. For the entire system,
MTBF = 1/ sum[ (FITS1 * # device1) + (FITS2 * #device2) + ...]
It is quite possible that multiple devices of high reliability will have a better aggregate reliability than a single device of merely average reliability. Also, the failure rate of just a couple of parts may dominate the calculation, making the quantity of other parts irrelevant.
yea yer probably right. I think I was mostably hungover that day in class. IMO all the MTBF calculations assumes the designers were perfect in all design areas. Not usually the case me thinks.
Do the reliability guys get a scorecard for after the fact product lifecycle reports from the field checking their calc's for accuraccy?
Do the reliability guys get a scorecard for after the fact product lifecycle reports from the field checking their calc's for accuraccy?
infinia said:
college days...
Exactly. operating the components out of spec or not allowing for de-rating can have a big impact on reliability. The MTBF numbers are always just a "check off" item on the requirements document, so the reliability guy turns the crank on the BOM to come up with a number that exceeds the requirement and that's the last anyone thinks about it, unless you work for Boeing or NASA or a medical equipment manufacturer.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.