So as I previously posted I found a Williamson 40/40 amp I cobbled together a LONG time ago, thanks again to everyone who contribute to that thread. So I gently woke the amp up, trimmed the bias to about 30 mA and hooked it to a set of Cambridge Soundworks Ensemble 3 speakers I found at a Goodwill for $40. I set it up in our bedroom, I'm driving it with an Amazon Echo Dot, and I cannot believe how nice it sounds... just so "sweet". The Ensemble 3 "subwoofer" doesn't have much reach so it's not great on hard rock but still it's a fun system to listen to folk music on.
The only downside to the system is that it won't play very loud because for reasons I can't recall I used a 24 VAC transformer so the DC rail is only about 32 V, this limits the output to about 7 W. So I'm thinking about building a true 40 W system with a 48 or 50 VAC transformer. One obstacle I'm facing is that the design is over 40 years old so most of the transistors are hard to get now, so I'm doing LTSpice simulations with parts with better availability. However no matter what I do I can't seem to get reasonable distortion results even with lots of output bias - about 90 mA static in the attached example.
Attached is my latest iteration. Some changes I've made:
#1. Changed output bias generator to Vbe multiplier for greater flexibility.
#2. Changed output transistors to FJA4313 and FJA4213 which are about the beefiest TO3P transistors I could find.
#3. Removed the output current limiting stage since the transistors look like they could withstand quite a bit of abuse.
The attached LTSpice simulation shows just under 0.5% THD at 1 kHz and 15 W output. There's a lot of odd harmonics in the FFT which made me think of crossover distortion but if you decrease the input to 50 mV pk-pk resulting in under 50 mW out, there's no visible crossover artifact. Also the distortion goes up to over 1% even though it appears to be in Class A.
I'm stumped as to the cause of the distortion, can anyone give any insight?
Thanks,
tommost
The only downside to the system is that it won't play very loud because for reasons I can't recall I used a 24 VAC transformer so the DC rail is only about 32 V, this limits the output to about 7 W. So I'm thinking about building a true 40 W system with a 48 or 50 VAC transformer. One obstacle I'm facing is that the design is over 40 years old so most of the transistors are hard to get now, so I'm doing LTSpice simulations with parts with better availability. However no matter what I do I can't seem to get reasonable distortion results even with lots of output bias - about 90 mA static in the attached example.
Attached is my latest iteration. Some changes I've made:
#1. Changed output bias generator to Vbe multiplier for greater flexibility.
#2. Changed output transistors to FJA4313 and FJA4213 which are about the beefiest TO3P transistors I could find.
#3. Removed the output current limiting stage since the transistors look like they could withstand quite a bit of abuse.
The attached LTSpice simulation shows just under 0.5% THD at 1 kHz and 15 W output. There's a lot of odd harmonics in the FFT which made me think of crossover distortion but if you decrease the input to 50 mV pk-pk resulting in under 50 mW out, there's no visible crossover artifact. Also the distortion goes up to over 1% even though it appears to be in Class A.
I'm stumped as to the cause of the distortion, can anyone give any insight?
Thanks,
tommost
Cleaning up the distortion will most likely reduce or eliminate the sound effects that you found to be unbelievably sweet. Whatever you do to clean up the distortion then, could just as easily be a backward step soundwise, so don't do anything you can't backtrack to the state in which you found the amp.
Changing the output transistors and drivers to modern, wideband output types stage may also have led to instability ( oscillation) which may be what you see as high order harmonics too. I wouldn't try to use something as fast as you now have there. Standard 2-4Mhz product should be fine and I'd be thinking MJ15003/4, TIP3055/2955 or similar in TO3P, TO247 etc. format, as you may prefer.
FWIW, I've restored and modded a lot of old amps for friends etc. and found that many widely known upgrades and mods actually reduced the listening pleasure of owners and not all were happy with lower THD, particularly when it involved significantly reduced 2nd harmonics. The huge number of JLH 10W class A fans is a testament to that preference.
Changing the output transistors and drivers to modern, wideband output types stage may also have led to instability ( oscillation) which may be what you see as high order harmonics too. I wouldn't try to use something as fast as you now have there. Standard 2-4Mhz product should be fine and I'd be thinking MJ15003/4, TIP3055/2955 or similar in TO3P, TO247 etc. format, as you may prefer.
FWIW, I've restored and modded a lot of old amps for friends etc. and found that many widely known upgrades and mods actually reduced the listening pleasure of owners and not all were happy with lower THD, particularly when it involved significantly reduced 2nd harmonics. The huge number of JLH 10W class A fans is a testament to that preference.
Last edited: