How important is transformer output?

[DreadPirate]

I'm needing a ss amp with a great deal of dynamic power for some real power hungry CS3.5 speakers. These are notoriously low impedance. When I'm assessing amplifiers from the power side, how critical is the transformer size or is capacitance the key? If for example, I am looking at units with 600VA vs 2000VA and both have 120,000uF capacitance, is it a given that the 2000VA will be far superior, given more or less equivalent wattage rating and technology? Is it that simple or are there other things to consider?

[AndrewT]

Hi,
what is the nominal impedance of the speakers?
What is the minimum impedance of the speakers?
What is the recommended power to drive your speakers?

[DreadPirate]

My question is poorly posed, I think. The speakers in question are 88dB/W/m, nominal impedance of 4 ohms with minimum recommended power of 40W (though I think 400W @ 4 ohm is more realistic for normal use), but my desire is to understand the relative importance of capacitance and transformer output in the ability of an amp to drive speakers. Using a water analogy, the capacitance is a storage tank, the speaker load someone opening and closing a valve, the transformer some kind of pressure throttling device. If I'm only lightly opening and closing the outlet valve, the amount stored in the tank is more than sufficient to keep up with my demand and the throttling valve throughput adequate to replenish the tank. Once I impose more serious demand, at some point, the amount stored becomes taxed and the ability of the throttling device's throughput begins to have an impact on meeting the demand.

I'm using an Aragon 8008BB with 2200VA transformers and 120,000uF capacitance to adequately drive these speakers. If I were to use an 800VA transformer in the same amp instead, with same capacitance, and given a very demainding speaker load (there must be some bottom limit to what is practial or in use generally), would the system be taxed or not? Is the 2200VA severe overkill when 120,000uF is present, even, for let's say, the most demanding speaker loads, a Magnepan, perhaps?

[AndrewT]

If I were designing an amp to drive these 400W into 4ohm speakers, then I would specify a 600 to 750VA transformer for each monobloc and use ~+-40mF per channel.

[tinitus]

I would take an amp module designed for say 300watt/8ohm, with say 4 output pairs each channel
Lower the supply voltage to +/- 50-55Vdc
750VA rating Andrew suggest should be fine for each mono
2 pairs of 33.000uf each channel should be ok
Use heavy heatsinks

With good amp modules and done properly I expect it will be better than Aragon

btw, isnt Aragon true balanced amps? They may not like low impedance

I believe that amps should be designet for at least half the impedance load of the actual speaker impedance

[Michael Chua]

Originally posted by DreadPirate
real power hungry CS3.5 speakers. These are notoriously low impedance

The CS3.5 is actually not that difficult a load. It is still above 4 ohms. A notoriously low impedance speaker would be the Watt-Puppy 8 where impedance hovers around 2 ohms.

For a normal amplifier, a 800VA transformer is more than enough for 400W @ 4 ohms. 2200VA is an overkill. For high performance amps, 2200VA with 120,000uF per channel is not unrealistic.

Regards
Mike

[cbdb]

You need both a large transformer and lots of capacitance to keep the rail voltage from having excessive ripple (dropping below the no load voltage). The capacitance is the current source when the transformer is off (not sourcing current) which is typicaly 80% of the time at high power output. When the transformer is on it has to be large enough to "fill" the capacitors in the short time its on. At full power into a low impedance there is a max ripple voltage before it effects the sound. There is a certain combination of transformer/ capacitance size that will meet this criteria and if either one is to small you will have excesive ripple. After this point increasing one without the other will not be benificial. (No matter how much larger the tranny is, the cap voltage will still drop x amount when it is the current source).

So going from 2200VA tranny to 800VA, you willl no longer need 120,000uf. Its a waste of caps.

[h_a]

keep the rail voltage from having excessive ripple (dropping below the no load voltage)

Using a cap-only psu you will always get ripple, no matter how much caps you add. If you're really that concerned, go for an actively regulated psu. Sure, not easy to do, but very effective. Feel free to mail me if you're interested in pcbs

Anyway, if you want a high-current capable amp, I would think more in terms of an amp that is designed for that (large number of output transistors for example, high power drivers if BJT...). No use for a large cap bank and toroid if the output stage is not up for it.

Have fun, Hannes

PS: power needs are usually also largely overestimated...just use your multimeter to measure voltage across speaker terminals when playing music. You could be surprised

[AndrewT]

Originally posted by h_a
power needs are usually also largely overestimated...just use your multimeter to measure voltage across speaker terminals when playing music. You could be surprised

Yes and no.

Average power needs are often overestimated.
I posted data on speaker terminal voltages when testing a new satellites + bass speaker system. The voltages were tiny and surprisingly the bass speaker voltages were much less than the satellites drive signal.

Peak transient power/voltage/current needs are often underestimated.

Allow at least +20dB above your average level and some say that +30dB is required for very dynamic sound sources.

If one is ticking along @ 200mW into 92dB/W/m 8ohm speakers (about 80dB at the listening position) and you need +20dB that is a peak transient of 100dB then the power requirement jumps to 20W and that is far below what real sounds in real life sound like.

Using the +30dB for transient overhead requires 200W into 92dB speakers to generate that 110dB.

Is it Gedlee that suggests we should be aiming for 130dB SPL for transient peaks?

Heatsinks do not need to be designed for transients!!!!

[h_a]

Interesting, Andrew.

It would be, however, interesting to know where these figures come from. Especially +30dB headroom will be completely useless if one likes to listen to contemporary rock or pop, where the loudness war kills the dynamic range.

Also not everyone has neighbours that tolerate 110dB nor a listening room with a noise floor below 30dB...

Last I'm wondering what all these 5W amp owners think about that headroom Or the Aleph-owners. Anyway, the headroom can be easily examined by shoving your music onto a PC and having a look at the waveform.

Have fun, Hannes