Hi guys,
I'm thinking about picking up a Carvin HD3000 which uses the IRF chips and a SMPS. With typical older amps the caps need replaced. When I do that I usually put in 30,000uf per rail on the 60hz brute force supply. Test show that the rails are clearly stiffer.
When a SMPS is used are caps that large needed?
Thanks,
Scott
I'm thinking about picking up a Carvin HD3000 which uses the IRF chips and a SMPS. With typical older amps the caps need replaced. When I do that I usually put in 30,000uf per rail on the 60hz brute force supply. Test show that the rails are clearly stiffer.
When a SMPS is used are caps that large needed?
Thanks,
Scott
Depends slightly on the design, but usually not.
Be a little careful as excessive cap (or overly low ESR) can cause issues with the power supply control loop phase margin on some supplies, the cap size (and ratings, particularly ripple and ESR/ESL) are slightly non intuitive with switchers, especially if you are used to big iron.
73 Dan.
Be a little careful as excessive cap (or overly low ESR) can cause issues with the power supply control loop phase margin on some supplies, the cap size (and ratings, particularly ripple and ESR/ESL) are slightly non intuitive with switchers, especially if you are used to big iron.
73 Dan.
I would not add excessive (over 1000uf) capacitance to a switch mode supply. A large cap is a short until charges. Many SMPS's will see that as a dead short and will not come up.
Or an RC filter can be used if your SMPS has an external sense wire. (Won't recommend LC since modern LLC PSUs are complex enough as they are.)
Are you referring to the Capacitor/s on the Mains voltage, or on the isolated LV output of the SMPS?
The Mains capacitors will be rated at 400Vdc if only one is fitted across the mains, or two 200Vdc/250Vdc capacitors, if in series with resistor equalisers.
these are LV output capacitors.
The Mains capacitors will be rated at 400Vdc if only one is fitted across the mains, or two 200Vdc/250Vdc capacitors, if in series with resistor equalisers.
all the ATX smps I have opened up have an LC filter on each output.
Does this final LC filter affect the operation of the SMPS and it's regulation?
No, cus usually the feedback loop reading point is just after the first LC group, the second one ( the second LC filter ) cannot affect the feedback in a bad way, rather it isolates it from the load.
As for the question of this topic, one of the swiched supply's advantage is that it need low vallue capacitance for filtering purposes, just due to it's high freq. Enough capacitance should be provided as to assure a low enough voltage ripple on them, but it is not a good ideea to exagerate, too much capacitance on the outputs will be seen as it has ben mentioned, a dead short just before charging, the bigger the capacitance on the output, the more complicated and precise, the current limiting needs to be to take into account that very large shock caused by the caps charging from 0. My suggestion is that if you have enough capacitance as so at the max load current, the voltage ripple is on the order of tens of milivolts, that should be ok. Enough capacitance should be provided on the mains filtering as well, but again not too much, because the larger it is, the more stress is put on the main rectifying bridge.
As for the question of this topic, one of the swiched supply's advantage is that it need low vallue capacitance for filtering purposes, just due to it's high freq. Enough capacitance should be provided as to assure a low enough voltage ripple on them, but it is not a good ideea to exagerate, too much capacitance on the outputs will be seen as it has ben mentioned, a dead short just before charging, the bigger the capacitance on the output, the more complicated and precise, the current limiting needs to be to take into account that very large shock caused by the caps charging from 0. My suggestion is that if you have enough capacitance as so at the max load current, the voltage ripple is on the order of tens of milivolts, that should be ok. Enough capacitance should be provided on the mains filtering as well, but again not too much, because the larger it is, the more stress is put on the main rectifying bridge.
"cus" in the UK means to swear.
I am sure that is not what you meant to say.
I am sure that is not what you meant to say.
and adds stress to the Power Thermistor used as a slow charger.
You are right, i meant "because", i apologise for the mistake, sadly i cannot edit it to correct it, so again i do apologise for my mistake...
Some who are "into" text speak might type "cos" instead of because to save time.
I write in normal english. This is an english language Forum. We should not be using text speak. Partly because it will be different in other languages and will not translate reliably into english.
I write in normal english. This is an english language Forum. We should not be using text speak. Partly because it will be different in other languages and will not translate reliably into english.
I agree, thank you for the observations, i am sorry for the mistake, and it will surely not happen again.
WOW!!
Thank you for great replies.
I guess one of the advantages of SMPS is smaller filtering caps.
I think I'll give it a try...
Thank you for great replies.
I guess one of the advantages of SMPS is smaller filtering caps.
I think I'll give it a try...
You can upset an SMPS with too much load capacitance.
The SMPS on power up is trying to supply a short.
Some SMPS think this is a fault condition and wont start up properly and go into protect mode.
On one of my PIC based SMPS designs the PIC allows a short for a couple of hundred milliseconds then if it still there it assume a fault condition.
If the output has large capacitors the PIC will assume a fault, shut down and put on the error LED.
The SMPS on power up is trying to supply a short.
Some SMPS think this is a fault condition and wont start up properly and go into protect mode.
On one of my PIC based SMPS designs the PIC allows a short for a couple of hundred milliseconds then if it still there it assume a fault condition.
If the output has large capacitors the PIC will assume a fault, shut down and put on the error LED.
Building an Switch Mode Power Supply is not an easy task, you need special tools like ESR meter for measuring the ESR on the filter caps, L-Meter for measuring inductance, and more importantly you need an oscilloscope with wich you cand o just about anything. You also need experience in working with SMPS's, in understanding them, understanding how they work, experience in designing the board layout wich is very important. So yes the SMPS unit has great advantages over analogue ones, they are usually more efficient ( much more on the high performance ones ), they are less bulky ( smaller in size and weight ), they require less capacitance and much smaller transformers, wich can make them cheaper in some cases, but you pay the price in complexity wich is much greater in a switched supply, also it could be argued that SMPS's could some times be more noisy than the analogue versions ( by "analogue" ones i mean the power supply unit basest on the line transformer, i hope it is clear for anyone ).
And i remind you how simple ( at least math wise ) to design and built is one based on a line transformer, you just need the transformer, a rectifier bridge, and 2 large filter caps, and that is it. The choice is yours.
And i remind you how simple ( at least math wise ) to design and built is one based on a line transformer, you just need the transformer, a rectifier bridge, and 2 large filter caps, and that is it. The choice is yours.
I disagree that a SMPS is usually more efficient.
Most well designed 50Hz transformer >100VA has an efficiency of about 95% or better at full load. Limited to about 65% of VA rating for capacitive load (rectifier + large cap bank) we end up at about 93% efficiency. Losses in rectifier is about another % or so.
Efficiency increases with reduced load, and quickly as losses are predominantly I^2*R , and I reduces with load.
You be hard pressed to get to 92-95% efficiency for a switched power supply and maintain that at a reduced load, as much of the losses is either linear with load (switching) or squared (current ie load) , even if you can reach it for the upper ranges. A very well designed resonant topology can reach and go beyond 95%efficiency for a limited power output.
Also, power loss in SMPS are predominantly focused on small volyme semiconductors, adding bulky heatsinks and critical overload control to prevent catastrophic failures.
In linear supplies themost of the power loss is spread over the whole transformer which in itself is heatsinking it away.
Swithing power supplies are selected on other merits than efficiency, like costs (in massproduction) , weight, W/volume , coolness...
Most well designed 50Hz transformer >100VA has an efficiency of about 95% or better at full load. Limited to about 65% of VA rating for capacitive load (rectifier + large cap bank) we end up at about 93% efficiency. Losses in rectifier is about another % or so.
Efficiency increases with reduced load, and quickly as losses are predominantly I^2*R , and I reduces with load.
You be hard pressed to get to 92-95% efficiency for a switched power supply and maintain that at a reduced load, as much of the losses is either linear with load (switching) or squared (current ie load) , even if you can reach it for the upper ranges. A very well designed resonant topology can reach and go beyond 95%efficiency for a limited power output.
Also, power loss in SMPS are predominantly focused on small volyme semiconductors, adding bulky heatsinks and critical overload control to prevent catastrophic failures.
In linear supplies themost of the power loss is spread over the whole transformer which in itself is heatsinking it away.
Swithing power supplies are selected on other merits than efficiency, like costs (in massproduction) , weight, W/volume , coolness...
95% or better for a 50hz power transformer at full load seems to me a bit optimistic, but still in some cases a linear supply comprised of a well built power transformer, rectifier and filter caps, may have better efficiency, so i should have been more specific. The great advantage of switched supply comes with voltage regulation at very good efficiency, and in this aspect SMPS's are far superior than linear ones, but they are not easy to design.
- Status
- Not open for further replies.