Your best bet is probably to use a series string of super capacitors such as the Maxwell "Boostcap" range.

http://www.maxwell.com/ultracapacitors/products/medium-cell/bcap0120.asp

- A low ESR capacitor - 350F 2.5V, ~30A continuous 300A peak discharge, the size of a standard D-Cell battery, and very affordable. You will need a few in series, and a simple charge balancing circuit, but it is probably the best option. If you oversize the super capacitors to reduce the voltage drop problem (as the super capacitors discharge) you may not even need a voltage regulator, which may also save cost and space. Maxwell also do modules of super capacitors (but a bit more expensive and you don't have much flexibility in package shape). Just watch out for the high leakage current problem that super capacitors are susceptible to.

- Another option may be a super capacitor string used in conjunction with a battery chemistry such as Li Ion or Li Ion Polymer, that can handle high output currents. I came across these 1.2V NiCad batterys for compact battle robots that will do well over 20A continuous and 98A 2 second pulse time, and have a 15 minute charge time. A string of these would seem to meet your requirements without having to use super capacitors or worry about large variations in output voltage.

http://www.powerstream.com/ (under site map - "robot battle battery packs")

If you are trying to get fast recharge and high current then I think you will be stuck with either a high current battery chemistry, a super capacitor array, or both. I am not aware of any battery that can output this much current (even when supported by a super capacitor string) AND be charged up in 5 seconds.

Hopefully one of these solutions (or combinations) is helpful for your particular application. Let us know how you are going (or perhaps more on your particular application).

I agree that Super Cap solution is ideal. Battery charging and high impulse current in compact batteries is very difficult. Fuel cell was another option but will avoid in this case. I have source of current that can be used to charge Super Caps in 5 seconds so will try this option. I am going to have one problem with Super Cap is to generate a constant high current as charge decreases on capacitors bank. This is to be achieved through special circuit. I am planning fast switching circuit for this.

Fuel cells are absolutely the wrong choice for your application (they are not suitable for high pulse currents, or compact in the power range you are asking for).

If your duty cycle is 15 off, 1 on then the average current out of your power supply to keep the super capacitors charged up is only going to need to be around 5A (much more reasonable than 60A). Charging the super capacitors shouldn't pose a problem, remember, you charge up a super capacitor like you would a normal capacitor, not like a battery (in which I mean you only need to limit the charging voltage and current, the super capacitor will naturally limit its charging current, unlike a battery where careful control of the charging characteristics need to be maintained, especially when fast charging).

What is your application as it appears to just be a bench-top pulse discharge circuit?

Super Cap at 60A I feel will start warming up with 3600X0.05W = 180VI in 300ms or about 60W power loss per second. My Peak VI requirement is 60AX30V = 1800W in 300ms or about 600W power per second. 10% power loss in internal ESR or Super Caps looks high. However 15000W/kg storage capacity of the Super Cap dielectric looks great. I am not sure if Super Cap will be able to remove 60W power per second or 3600x60W = 216kW power per hour all that easily. I think I will have a serious problem here.

Hmmmm. That level of continuous power dissipation is not going to be good for those super capacitors, but it seems a bit high. Using the standard 350F 2.5V Maxwell Boostcap, I make it out to be roughly......... 14-18W peak at 60A per capacitor (when using a string of 12 capacitors in series to give 30V@60A) (using the voltage divider network formula, as the super capacitor is being used as a power supply).

I am not quite sure what the capacitors are actually rated for in regards to dissipation (the Maxwell data sheet is rather unclear). I would be interested to know the answer if someone has an idea (or at least a good rule of thumb). I am sure it will come in handy in the future.

Perhaps a small fan to cool them would be enough in this application (especially as the average power dissipated over the 15 off / 1 on duty cycle is going to be quite low). I don't think there is as big a problem as you first thought, and even with the reduced super capacitor life expectancy due to internal heating I think this is still going to be your most compact, cheapest and easiest to implement option. If you still think that internal heating is going to be a problem you could add a second string of super capacitors in parallel with the first to spread the power dissipation losses.

I would be interested in any feedback from other engineers that have actually used super capacitors in such an application, or if there is another way that is as good as the super capacitor option, other than just using a 30V 60A power supply or transformer and regulator (which is the other way to do it, although in a much larger footprint).

I have need for 18V in delivery system so I thought of using 30V power source to let it deliver power in between 38V-18V range without me having any problem. Maxwell Super Caps are actually meant for low current use for longer duration but may also deliver quick pulse power. I am only afraid if they can deliver in 15:1 charge and discharge cycle and power loss to be manageable. My final size for this need to fit in 500cc to 1000cc volume without bursting into flames.

I think a pulse energy delivery type DC-DC converter working at 1MHz frequency will also be an ideal solution to reduce burden on capacitors. Only problem will be that the DC-DC converter need to switch ON and OFF very fast with external control at 60A power rating. I am not sure if any one makes such DC-DC converters in the world. Rise time and decay time of 10us may be all right in 300ms pulse. Generally enable and disable run to several ms duration.