Thank you. Ciss? I have looked at the curves for many of the SOT227 (pucks). The Ciss appears fairly asymptotic beyond 25 V?
What decision making applies to SOT227 versus PLUS247 for the same device?
When looking at device datasheets, what are the most important characteristics for choosing a P and N pair? Low/similar Ciss? Similar Transconductance?
The strategy for deciding upon a pair is to optimize "perfect “square law” cancellation of distortion?"
I was told that with source followers the Crss is the more important value to look at.
And this one declines with certain types at 25V considerably. Look at Pa´s Burning Amp speech 2016 and there was besides other this interesting picture....
So you must look for Ixys that go down in this way for Crss.
It is a pleasure to hear his explanations.
And this one declines with certain types at 25V considerably. Look at Pa´s Burning Amp speech 2016 and there was besides other this interesting picture....
So you must look for Ixys that go down in this way for Crss.
It is a pleasure to hear his explanations.
Attachments
Wow! What a great resource. Thank you so much. I almost missed that your screenshot was at 13 min. I thought the discussion about capacitance was over at 4m into part two. This talk answers a bunch of questions for me. I usually do not study videos. I don't have the patience. I must rethink. Thank you for alerting me to this talk. I imagine that if I was going to play with "pucks" using a typical FW PS; this P channel having asymptotic capacitance changes at a low voltage would be a candidate if I could match it up with an N channel.
http://ixapps.ixys.com/DataSheet/93001.pdf
http://ixapps.ixys.com/DataSheet/93001.pdf
Last edited:
I got itchy and ordered a few pairs of IXTH20P50P and IXTH30N60P. The capacitance and transconductance appear comparable relative to other documented push/pull pairs.
Last edited:
Nelson Pass's talk especially caught my attention when he stressed the value of saving voltage. Partly because the "pucks" would not enter into a range where the Crss was steep in the curve. It looks like Crss for most of the pucks starts to flatten at 15 V. Thus, it seems that a design may "save" 30V from the rails. A First Watt PS of +- 24V would need to limit the OS voltage to 48-30=18V AC to minimize distortion. 1.3A bias X 18V = 23W. Efficiency for Class A at 20% and, thus, .2X23=4.6W if using a puck with 2X24V rails limited to 18 V swing to stay out of Crss steep slope territory. It seems that the FW PS rail voltage (48 V potential) may be a bit low if 30V is "saved" from 48.
If a 25W Class A stereo amplifier channel amplifier idles at 250W. A single channel is idling at 125W. 25W RMS music /125W idle (bias) = 20% efficiency.
With "saving" 30 or more volts in mind, I am looking at a toroidal transformer having 4 35V secondaries. 35/.75=47V. This allows a 2X47=94V potential across a push/pull pair. To save as much as 20V from the rails. 94-40=54V AC. 54 VDC X 1.3A bias = 70W 70WX.20 efficiency = 14W. What do I need to do to target 25W with 94V rail potential while saving 40 V to avoid Crss? 25W/.2Efficiency = 125W. 54V available from the rail potential / 125W = 2.3A bias. Argh. 94VX2.3A=216W per channel. 2X216=432W for a couple of 25W Class A mono blocks to idle.
What if saving only 2X15V=30V of rail potential? 94-30=64V of available AC. 64V X 1.3A bias = 83W X .2efficiency = 16.6W. Have to increase bias to reach 25W. .2efficiency/25W target = 125W. 125W/64V= 1.96A bias. One channel idle 94V X 1.96A = 184W. 2 X 184W = 368W for a couple of 25W class A mono blocks to idle.
It appears that there is quite a price to pay if a design saves enough voltage to avoid distortion by staying away from the steep slope of the Crss curve. Further, it appears that +-47V rails might be reasonable. The rail voltage is increased, not to increase wattage, put to decrease distortion by avoiding Crss regions.
I have no idea of what I am talking about but, for better or worse, I am on the cusp of buying a 800VA toroidal transformer with 4 35V secondaries. The video made me do it.
Is it the case that 50 Volt rails have no merit in a push/pull solid state design? As I recall, DeLite got up to 100V in a single rail. Thank you again for trying to lead me in a better direction.
As I recall, E=sqrt of PR. RMS E at 8 Ohm sink for 25W = sqrt 200 = 14V RMS. 14V RMS converted to peak V =
As I recall, E=sqrt of PR. RMS E at 8 Ohm sink for 25W = sqrt 200 = 14V RMS. 14V RMS converted to peak V =
Last edited:
14V RMS converted to peak-to-peak = 40 Vpp. Saving 8 Volts of 48V rail potential for 25W RMS. If Crss suggests that there is merit to saving 40 volts, then an 80+ V rail potential makes sense (to me) +- 40 V rails. Or perhaps a cascode and low impedance buffer obviates input capacitance. Beyond me. But it seems that the most simple circuit to use "pucks" and avoid the steep part of the Crss curve might want 50 V rails.
For whatever reason, lhquam ran some of his simulations at +-40 V rails.
For whatever reason, lhquam ran some of his simulations at +-40 V rails.
I shall seek academic compensation. Apropos to your suggestion. The first article I read (this AM) was The Sweet Spot.
The final two sentences: "So here’s your chance to make simple high quality audio amplifiers better than ever. Buy yourself a cheap used distortion analyzer and go for it."
So now I am looking for Linux based oscilloscopes with analyzing software. I can't afford to read Nelson Pass publications. Don't worry, I will pay my dues and figure it out. But rather than 5 days, best to check for any progress after five years.
Last edited:
Nowadays there are many solutions to choose from. Get a good sound card, make an interface for the sound card, install the software and you are good to go.
Thank you for commenting. My half day crash course on distortion analyzers has landed on this device - PicoScope® 4262 oscilloscope / FFT analyzer. "Low-distortion signal generator." "Arbitrary waveform generator." Perhaps this contraption is self-contained. FFT spectrum analyzer for audio amplifier design.
Last edited: