Greg Ball SKA / contruction & application notes

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Hi all there, possibly interested in these SKA amps from amplifierguru Greg Ball:

possibly some of you are ventilating the idea of hooking up such a SKA (GB300D version) with an ESL, but are afraid it might get unstable due to the heavy reactive loading of the amp.

As I am soon getting a pair of ESL's myself (www.cacacity.de), I had the same fear (due to bad experience in the past with BJT power amps). So I got in touch with the manuf and asked for the equivalent load cct. I was told that the amp "sees" a load similar to 2R2 in series with 7.5uF. After asembling such a load with a battery of foil caps (8.1uF) and power resistors, I made some tests and the results are here to be seen:
The pics show the current, measured over the 2R2 resistor. Input to scope was via a BNC cable, 1:1, at 5V/cm.

pics:>

The rather very low load impedance reduces the overall lavailable loop gain of the SKA 300D output stage (common source topology), thus stabilizing the whole arrangement :) .

In contrast, many "usual" BJT amps on the market rely on an EF output stage (common collector topology). It is known that EF love to oscillate at the presence of acapcitive loads. That's how oscillators are built! :smash:

For those interested in SMPS applications in audio power amps: it works fine! I switched two 320W, 48VDC (6.7A max) together for a bipolar supply. Due to the stray fields at 100 kHz, I would strogly reccommend to have the "power house" placed in a separate shielded cabinet. Using common practice EMC techniques, you get a clean and rugged regulated power supply. I feed 3 GB300Ds with it, using plenty of reservoir caps (10mF on each rail, close to the amp module) and, by former experience, have added a 3.3uF MKP 630V propylene cap at each side of the rail, in closest proximity of the PCB, to improve dV/dt performance. 100kHz ripple is below 0.5mVpp, 100Hz ripple is 200mVpp at full load and it sags 350mVDC at full load. Those interested in cct details (schematics etc) should send me an email.


so long,

stoeffle.++++
 

Attachments

  • test-ska&esl kopie.jpg
    test-ska&esl kopie.jpg
    69.1 KB · Views: 601
Hi Stoeffle,
nice clean outputs at 2kHz.
What sort of voltage produces the 8App output across the combination?

When you quote 350mV droop on full load is that into one GB300 or all three amps?

Driving 3channels from one dual polarity supply is a little unusual.
Do you see any downside? Other than needing a total of +-30mF decoupling (a lot less than usual for rectifier PSU).

What sort of power is available from GB300 when using loaded Vrails of +-47.65Vdc?

At these lower supply voltages, what minimum (reactive) impedance is safe for these amps?
 
SKA load test

Eva said:
This is a non-sense test. The reactance of 8.1uF at 2Khz is just 10 ohms, so any amplifier will laugh at that.

Try 20Khz for a more sensible test.


Well Eva,
will give it a try at higher frequencies, to calm your spirits.
At least, if the amp had trouble with this load, it would show this with ringing, don't you agree? At least other amps I tried, did ring.

stoeffle.++++
 
AndrewT said:
Hi Stoeffle,
nice clean outputs at 2kHz.
What sort of voltage produces the 8App output across the combination?

When you quote 350mV droop on full load is that into one GB300 or all three amps?

Driving 3channels from one dual polarity supply is a little unusual.
Do you see any downside? Other than needing a total of +-30mF decoupling (a lot less than usual for rectifier PSU).

What sort of power is available from GB300 when using loaded Vrails of +-47.65Vdc?

At these lower supply voltages, what minimum (reactive) impedance is safe for these amps?

Hi Andrew,

Vout at SKA 300D terminals was 80Vpp.
The load has an angle of 77.4deg.

Yes, it may be a bit unusual to drive several amp modules from a single supply. This mainly due to possible xtalk effects from one module to the others, via the supply rails.
I decouple each pair of rail reservoir caps (10mF on the medium to treble unit, 30mF on the subwoofer module) with power schottky diodes, so one amp does not "see" the other ones directly.
On the other hand, Gregs topoloy has a magnificent PSRR performance.
But, as the supply is a regulated one, capable of feeding almost 7 Apms continuous duty into the load, at a 100kHz rate, I did not see a problem there. I experience some little xtalk more often not via the supply rail lines but due to other coupling mechanisms inside a common cabinet. My subwoofer module unit is mounted at the back of the subwoofer box itself, linked with 3x 2.5sqmm (=AWG12) power cables to the common DC power source.

As the GB300G has 4 FETs at each polarity leg, and Rds on is claimed to be 1/2 Ohm (RFIP9240), we get 1/8 Ohm for the trannies.
So, with a 4 Ohm load assumed, at 47.5VDC supply, assuming a .1 Ohm cabling on each leg, we get a figure of 10.98 Ap. Thus, some 241W on the 4 Ohm load.

Note: the SMPS is adjustable +/- 10% Vout. So, if required, the DC supply may be crancked up to some 54V. This may sag lets say .5V, so with 53.5V at the output trannies and same load, you would get 12.4Ap => 306W.

The 48VDC supply already is some kind of overkill for my ELS, as the manuf states 40Vp max. At this level, the membranes of the ELS already may touch the stators and sound pressure level, I was assured, was torturing loud.
Basic efficiency of ESW15023 is 86dB at 2.83Vrms at 1m, so extrapolating, at 28Vrms we sould get 20dB more = 106dB...

Greg Ball states that he does not reccommend to feed loads with less than 4 Ohm. So, If you have to drive such kind of loads, I would suggest to test your amp to see what happens, also in regards to thermal stability. The main reason I assume is, the load impedance influences the overall gain loop, by reducing the gain in the output stage. This may impair the THD & etc specs, I guess.

so long,

stoeffle.++++
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.