My build is presented here (post #180).
That is an integrated amplifier, honestly preamplifier gave me more headaches than amp.
If you got all parts, build is very well explained in the guide by Brian, DC is stable, sound is great.
For VAS I have used E graded transistors.
My V2 is done old-school way, transformer and regulator.
I had no intention to compare it with commercial or other DIY amps just because it is a gift for my son who recently left the nest and started adult live. This amp and vintage ISOPHON speakers are to give him a nice sound in his new living space.
My home amp is also on Exicon transistors, shown here.
Just one comment on JFET, as there no grade in P channel you may get quite different IDSS with it.
For EU builders, I have bought LSK/LSJ from www.ib-fluck.de (no connection, just to mention as these are not easy to source).
That is an integrated amplifier, honestly preamplifier gave me more headaches than amp.
If you got all parts, build is very well explained in the guide by Brian, DC is stable, sound is great.
For VAS I have used E graded transistors.
My V2 is done old-school way, transformer and regulator.
I had no intention to compare it with commercial or other DIY amps just because it is a gift for my son who recently left the nest and started adult live. This amp and vintage ISOPHON speakers are to give him a nice sound in his new living space.
My home amp is also on Exicon transistors, shown here.
Just one comment on JFET, as there no grade in P channel you may get quite different IDSS with it.
For EU builders, I have bought LSK/LSJ from www.ib-fluck.de (no connection, just to mention as these are not easy to source).
No they are not in the 10N 10P but every other they say integrated body diode. If you have in stock just physically test them for body diode. I have Hitachi and Renesas and they do have body diodes.
Here is from profusion website. The ECX10N20 page.I thought both Exicon devices have integral protection diodes. The boards have a provision for an external diode (D96 & D97). I choose to populate these as a "belt & braces" approach to safety. There's no harm in using them if you're not sure.
Maybe some else can chime in? The Exicon databases aren't explicit about it (that I can see).
It is clear that it has got Integral protection diode:
The schematic shows the gain is only x6.
If I want higher gain I just increase R7.
Is that correct?
If I want higher gain I just increase R7.
Is that correct?
I am aware that higher gain will increase THD and reduce bandwidth.
But not everyone wants to use a preamp.
But not everyone wants to use a preamp.
I concur, if it is covering the audio spectrum who care about 40 kHz, maybe your cat or dog will even be pleased with your amp. Since your amp now rolls of at a lower frequency, the higher THD may not increase that significantly it may even decrease.
A gain of just 3dB is twice the previous level and may be more than you need. I know one does not want to turn the volume higher than about eleven oçlock then it seems that the amp is not powerful enough. 🤔
A gain of just 3dB is twice the previous level and may be more than you need. I know one does not want to turn the volume higher than about eleven oçlock then it seems that the amp is not powerful enough. 🤔
You can increase R7 from 500 to 1K, to gain increase gain to 20.8dB (11x). C3 and C4 can be dropped from 15p to 4.7p to maintain the same stability margin.
Thermal stability for the DC operating points should not be affected. Sims show that DC offset drifts less than 5mV and Bias less than 5mA from cold to hot.
If you do this, A 2Vrms source should be able to drive the amp to clipping (input sensitivity of 1.63Vrms for 40W).
Increasing gain this will increase distortion as power levels rise. Here are the results from a few quick sims at 1kHz:
1kHz at 1W into 8R
R7 500 = 0.0011%
R7 1000 = 0.0018%
1kHz at 25W into 8R
R7 500 = 0.0051%
R7 1000 = 0.0089%
1kHz at 40W into 8R
R7 500 = 0.0696%
R7 1000 = 0.1213%
In my opinion, this is simply the price to be paid for performance from a simple design with less open loop gain versus more complex designs. Open loop gain is 51dB at 1kHz. I designed for the most NFB I could run without causing stability issues. With the as-built 15.5dB gain, loop gain analysis shows a ULGF of 1.25MHz with 23dB and 64 degrees of margin. Loop gain at 20kHz is 37dB.
But this is all just my opinion. There's no hard rules here. Built it however you like. I understand the low gain may be a concern to some prospective builders. I can update the build guide for options on increasing gain with an explanation of the tradeoffs.
Also, please be aware this is all based on a quick simulation. There may be other impacts to doing this that are revealed yet.
Thermal stability for the DC operating points should not be affected. Sims show that DC offset drifts less than 5mV and Bias less than 5mA from cold to hot.
If you do this, A 2Vrms source should be able to drive the amp to clipping (input sensitivity of 1.63Vrms for 40W).
Increasing gain this will increase distortion as power levels rise. Here are the results from a few quick sims at 1kHz:
1kHz at 1W into 8R
R7 500 = 0.0011%
R7 1000 = 0.0018%
1kHz at 25W into 8R
R7 500 = 0.0051%
R7 1000 = 0.0089%
1kHz at 40W into 8R
R7 500 = 0.0696%
R7 1000 = 0.1213%
In my opinion, this is simply the price to be paid for performance from a simple design with less open loop gain versus more complex designs. Open loop gain is 51dB at 1kHz. I designed for the most NFB I could run without causing stability issues. With the as-built 15.5dB gain, loop gain analysis shows a ULGF of 1.25MHz with 23dB and 64 degrees of margin. Loop gain at 20kHz is 37dB.
But this is all just my opinion. There's no hard rules here. Built it however you like. I understand the low gain may be a concern to some prospective builders. I can update the build guide for options on increasing gain with an explanation of the tradeoffs.
Also, please be aware this is all based on a quick simulation. There may be other impacts to doing this that are revealed yet.
At 1 Watt 0.0018% vs. 0.0011% is no big deal.
Looks very good.
Thanks.
Looks very good.
Thanks.
Yep - 1W is not the issue. The transients where power levels will spike are the "issue". But this largely depends on how high your nominal listening level is. 1W nominal with a 14dB crest factor reaches 5W at the peaks. Which coincidently is just below the 5.75W Class A region with 600mA of bias 🙂.
This is probably where many will operate with efficient speakers... assuming you care about your hearing.
This is probably where many will operate with efficient speakers... assuming you care about your hearing.
Sensitivity is normally measured/related to noise floor not output power. For instance sensitivity is 100 uV for 10dB S/S+N level in some defined bandwidth, say 20kHz. More accurately call it input level for rated output @ 1 kHz.
Last edited:
I've seen it described like this, but I see your point. It may cause confusion on what it's describing. I'll tweak the language before updating the guide.
Yep. ±30V V1 PSU and ±36V V2 PSU
One other note. I ran those sims with KSC3503D / KSA1381E since these are readily vs the 2SC3503E / 2SA1381E @benpe and me built with.
I also briefly simed TTC004B / TTA004B. These perform close to KSC3503D / KSA1381E at 15.6dB of gain. But as gain increases, the gap widens significantly in favor of KSC3503D / KSA1381E.
One other note. I ran those sims with KSC3503D / KSA1381E since these are readily vs the 2SC3503E / 2SA1381E @benpe and me built with.
I also briefly simed TTC004B / TTA004B. These perform close to KSC3503D / KSA1381E at 15.6dB of gain. But as gain increases, the gap widens significantly in favor of KSC3503D / KSA1381E.
- Home
- Amplifiers
- Solid State
- Ultra Amplifier Build Thread