Output filter Caps

Yes and no. Yes the "rewards" are a higher reserve peak current that gives better low frequency extension and control, frequency dynamic range etc as stated by Sean above. But...

By increasing the supply capacitance, you are forcing the transformer to push out more current through its secondary windings. The xfmr VA rating is determined by the DC power on the load side, temperature rise of the secondary, and the secondary voltage. By doubling the filter caps you are increasing the xfmr secondary current draw (what goes out of a cap must come in). Since the xfmr VA is fixed and the filter caps provide a pure capacitance load, the xfmr does not produce more power with the increased current flow. Instead, it is dissipated as heat. Heat which can damage the xfmr and the circuitry nearby (diodes, regulators, etc). The xfmr is now producing more reactive power and less "working" power. The xfmr is an integral part of the PS and, as far as I know, cannot remain a constant with respect to filter cap sizes.

And, for God's sake, be careful about monkeying around with filter caps. They have enough energy stored to kill you if you short them with yourselfdom - even months after the power is switched off.

Back in the mid-1980's, I had a Hafler DH-220 amp which I wanted to modify myself. I went to a local computer parts supply dealer, and got a pair of high quality computer-grade power supply capacitors that were twice the capacity of the stock filter caps in the Hafler. This mod made a really major improvement in not only the deep bass response of the amp, but also made some improvements to the mid-range and upper frequencies. I was able to get the replacement caps at wholesale price -- had these caps been part of the stock Hafler amp, it would have increased the retail price of the amp by a factor of 2.

Thanks fellas for your responses. If I do decide to double up the caps is there anything out of the ordinary in the wiring? It is obvious how the current caps are wired. Would I just need to solder extra wire from the positive and negative leads and attach them to the leads on the new caps? Also, I plan on standing in a bucket of water with my entire body wrapped in tinfoil as to protect myself from both fire ants and electricity. (Kiddin)

Lincoln: In response to your email about replacing the stock 22,000MF caps with 27,000MF caps...

From what I remember, the DH500 is rated at 250wpc/8ohms. Which means that it is probably a stable voltage source down to 2 ohms (250wpc/8ohms 500wpc/4 ohms 1000wpc/2ohms). Therefore a 1000 VA transformer is MOST LIKELY it's xfmr size, assuming it is sized for a minimum impedance of 2 ohms. The current output at 1000 wpc is 32 amps and the peak to peak voltaqe would be 63V. The RMS current is calculated at 22 amps and the peak current @ 1000 watt is 1.4 RMS or 31 amps. Assuming 1,000MF for each amp RMS, that's where the 22,000MF is derived for you amp (which is what you stated in your thread).

You can use a pair of up to 31,000MF (up to the peak current) caps without taxing the xfmr. If you really want to go all out, I would use two 15,000MF caps is parallel. Reduces the output impedance by one half and makes for better performance.

Gs5556: I mean no disrespect, but your figures and assumptions are way out in left field on this one. The amp already has two 22,000 uF caps in it from the factory for a total of 44,000 uF's. Why would anyone want to reduce total capacitance by a very measurable amount ( 44,000 uF's down to appr 30,000 uF's ) ? Surely, Hafler & crew did not produce a product that was severely underdesigned to the point of smoking the transformer under normal use conditions.

Besides that, the Hafler DH-500 does not "double down" or act like a voltage source as impedance is halved. As far as i can recall, it is not even remotely close to doing so. This tells me that the amplifier is of a current limited design and can use all of the "reserve" or "buffer" in the power supply that it can get.

I think that Linc wanted to install more ADDITIONAL capacitance to what was already there. As such, he can simply drop the caps into the circuit and wire them in parallel. As mentioned, the major draw on the transformer will come when first powering up the unit. After that, it is simply "topping off" or "refreshing" the charge that originally took place upon firing it up.

Having said all of that, i do agree that using multiple smaller caps ( two 15,000 uF's ) would be better than using one 30,000 uF cap. This could be broken down further into using a combo of multiple sized caps ( 470, 1000, 2200, 4700, etc... ) to achieve the same appr value. Only problem with doing so is that the wiring can become a mess. That is, unless one fabricates some type of sub-board with a wiring harness / adapter to compensate for such.

For the record, i have six preamps* that came with 1000 ( one thousand ) uF's of capacitance from the factory. They now have 42,000+ uF's of capacitance in them using the aforementioned "staggered value" approach. While the stock transformer would give out if it had to deal with in-rush current that would be required to refill this quantity of capacitance if asked to do so on a regular basis, i avoid this by never turning the preamp off. In fact, these preamps do not even come with power switches from the factory.

I also have another preamp that came from the factory with 44,000+ uF's of capacitance in it. While this unit has a very "beefy" ( for a preamp ) transformer, it is WAY below what one would find in an amp the size / power rating of a Hafler DH-500. As such, i would not hesitate to increase the amount of filter capacitance / power supply reserve. Nor would i fear problems from doing so as long as the precautions that i mentioned earlier were attended to. Sean
>

* If you are wondering why i have so many identical preamps, I like this preamp and it is no longer available. As such, i have purchased several of them and performed similar modifications to each. Some of these have different factory options though, such as different gain levels in the phono stage, factory upgraded / redesigned output stages, etc...