Some of this may be residual common-mode noise and distortion, which would not appear differentially across the output. It would be nice if you had a way to view the output differentially. Does your software let you save two measurements and subtract one from the other? If so, you could measure both outputs to ground and then look at the difference, and that would give you some idea.
Yes, that's also my hope and my suspicion.
For that I would need a low-distortion balanced to unbalanced preamp. Something is always missing! 😡
Yeah, I've thought of that. You have two inputs for stereo. What would be more obvious? 😉 Only I have to learn more about ARTA and measurement.
Thank you for the advice 🙂
Looking again at your output spectrum, it appears that the noise might be repetitive. Can you expand the frequency scale to see it better? That might provide a clue about its origin(s).
I have spreaded the spectrum:
The two peaks beside k2 are looking strange.
An externally hosted image should be here but it was not working when we last tested it.
The two peaks beside k2 are looking strange.
I haven't read all the post in this thread so I don't know if anyone has mentioned
the circlotron design on the Passdiy web page. I have a little question about using
the big power Jfets. When used as a source follower will the power jfets still have a
lot less distortion than using a mosfet as a source follower?
the circlotron design on the Passdiy web page. I have a little question about using
the big power Jfets. When used as a source follower will the power jfets still have a
lot less distortion than using a mosfet as a source follower?
It looks like the worst noise is at 1 and 2x the (50 Hz?) line frequency. It also appears that, apart from the fundamental, everything is 70 dB or more below 1W. This is actually not too bad if mostly common mode, because you will see much less than that differentially across the outputs.
If you want, you can view only the common-mode noise by tying equal value resistors (e.g. 10K) in series across the output and then measuring from their midpoint to ground. If you then want to reduce this noise at the source, it looks like the best place to start is with the DC power supplies. I can offer some suggestions there if you like.
If you're asking about this particular circuit, and are comparing the IRFP240 with the SemiSouth SJEP120R100A, then I wouldn't expect to see a big difference. But I'd really want to try it to be sure.
I also think that the SJEP120R100A is a little bit better. And I plan to develop a high-end DIY kit that I want to sell.
However, in the high-end version that I want to build I also planned other transistors for the VAS.
The version with IRFP240 is to remain a open project.
You are welcome invited to join this project. Maybe you have fun to build an amp with the SJEP120R100A? The driver and the VAS is very flexible.
I hope so...
I will test it and report the results tomorrow.
That would be very nice 🙂
Hi Joe,
currently I have a terrible cold
But anyway ...
This is the measurement with two resistors (10k) in series as you've described. You can still see the 1 kHz testsignal. This is determined by the tolerances of the resistors, I guess.
As it stands I do not come around for a good instrument amplifier.
The next layout is already ready. But I'm going to torture the amp and test some other settings.
currently I have a terrible cold
But anyway ...This is the measurement with two resistors (10k) in series as you've described. You can still see the 1 kHz testsignal. This is determined by the tolerances of the resistors, I guess.
An externally hosted image should be here but it was not working when we last tested it.
As it stands I do not come around for a good instrument amplifier.
The next layout is already ready. But I'm going to torture the amp and test some other settings.
Hi moschfet,
This common mode noise has a "signature" similar to the noise you measured from one output terminal to ground, so I'd expect to see some cancellation across the output terminals. It's hard to say how much, but I think you will see at least 80 dB below 1W output, which in actual use is well below the ambient noise of the listening room and certainly in "hi-fi" territory in my opinion.
If you want to make further improvements, I think you will need a test instrument that lets you take floating or differential measurements. I use a USB-powered PC oscilloscope connected to a laptop PC running on batteries, but I know other solutions are available. Maybe the gear you already have will let you do this if you can capture L and R waveforms at the same time and then subtract one from the other.
If you don't have one already, I would also suggest getting a regulated dual DC power supply that can be set up for fixed and floating operation. This will let you temporarily substitute a "clean" power source for either the front end power supply or the output stage supply and observe any effect on the noise residual.
Get well soon. 😉
This common mode noise has a "signature" similar to the noise you measured from one output terminal to ground, so I'd expect to see some cancellation across the output terminals. It's hard to say how much, but I think you will see at least 80 dB below 1W output, which in actual use is well below the ambient noise of the listening room and certainly in "hi-fi" territory in my opinion.
If you want to make further improvements, I think you will need a test instrument that lets you take floating or differential measurements. I use a USB-powered PC oscilloscope connected to a laptop PC running on batteries, but I know other solutions are available. Maybe the gear you already have will let you do this if you can capture L and R waveforms at the same time and then subtract one from the other.
If you don't have one already, I would also suggest getting a regulated dual DC power supply that can be set up for fixed and floating operation. This will let you temporarily substitute a "clean" power source for either the front end power supply or the output stage supply and observe any effect on the noise residual.
Get well soon. 😉
By the way, I just realized that 80 dB below 1W is about the same as the 300 uV spec that Pass Labs used to quote for their X250 Supersymmetry power amplifier. So this seems like a good benchmark unless you need the amp to sound quiet when driving horns or other very high efficiency loudspeakers.
Just one more (I hope!) comment on this. There is one important thing you may be able to check using the tools you already have. Take a look at the ripple waveform on both sets of power supplies. If you see any ringing or HF bursts, you will want to add some damping to the affected supply. Something as simple as 0.1 uF (+/- 10x) across the secondary may be enough, but there are other techniques and they are well described elsewhere. If the problem exists and you address it, you will probably see improvement in the noise spectrum.
Thank you for the many hints. I first have a look on a chapter of common-mode distortion.
I've looked at the power supplies with an oscilloscope, but I could not see anything striking.
In the next order comes a INA114 🙂
I have made this little test:
When I play sinus of 17kHz(I just can hear it if it is loud enough) with a level at full output. You can hear noise , but it is no louder than other amps.
I have a LM3886 and the noise is comparable.
I've looked at the power supplies with an oscilloscope, but I could not see anything striking.
In the next order comes a INA114 🙂
I have made this little test:
When I play sinus of 17kHz(I just can hear it if it is loud enough) with a level at full output. You can hear noise , but it is no louder than other amps.
I have a LM3886 and the noise is comparable.
In recent days I've tested a few transistors. At the input stage, I have now Paired BJTs.
With the paired transistors, the switch on is also much quieter (without the softswitch).
I still have a few IRFP4310Z from another project. There were the fat Mosfets not working particularly well.
But in the circlotron they working very well. With the Circlotron they have complex resistances brought to glow 😉
I'm still waiting for my order with the INA114 to finally show fine FFT plots.
Some changes I also did by some values. I will show you later the complete circuit.
With the paired transistors, the switch on is also much quieter (without the softswitch).
I still have a few IRFP4310Z from another project. There were the fat Mosfets not working particularly well.
But in the circlotron they working very well. With the Circlotron they have complex resistances brought to glow 😉
I'm still waiting for my order with the INA114 to finally show fine FFT plots.
Some changes I also did by some values. I will show you later the complete circuit.
The INA114 has arrived 🙂
And I've also ordered some of 2SC1845/2SA992.
These are very low-noise transistors and I want to use them in the first differential amplifier. Unfortunately, I see relatively few amps with these transistors.The most likely use 2SC2240/2SA970.
Does anything against these transistors?
Does anyone have experience?
And I've also ordered some of 2SC1845/2SA992.
These are very low-noise transistors and I want to use them in the first differential amplifier. Unfortunately, I see relatively few amps with these transistors.The most likely use 2SC2240/2SA970.
Does anything against these transistors?
Does anyone have experience?
Nice work, I have done a hybrid circlotron in multisim, a time ago, I have made time to go als for a version who have allfet.
Change just mosfet for the semisouth one, distortion dropped in circlotron to 0.5 procent from 1.6 procent with the 2sk1058
goodluck, very nice design you have.
Change just mosfet for the semisouth one, distortion dropped in circlotron to 0.5 procent from 1.6 procent with the 2sk1058
goodluck, very nice design you have.
Attachments
Thanks 🙂
I have made some several tests and I think the IRFP240 is one of the best choice for linear driving. The SIC-fet will not be much better, just a little bit.
Bob Cordell and Nelson Pass are also fans of the IRFP240 and they know what they doing.
The low price of this device invites to match them.
I have made some several tests and I think the IRFP240 is one of the best choice for linear driving. The SIC-fet will not be much better, just a little bit.
Bob Cordell and Nelson Pass are also fans of the IRFP240 and they know what they doing.
The low price of this device invites to match them.
That,s right, I will test it let you now later.
mosfet amps who use complementary mosfets don,t sound wel except if heavenly class A because of the differends between them, the fault excists in the high regionen of the amp, where the most important things happen with sound reproduction, matching is the most underastimate thing in audio wonderland.
I go try simulate the Vfet IRF240, has little much capacitance tough.
regards
mosfet amps who use complementary mosfets don,t sound wel except if heavenly class A because of the differends between them, the fault excists in the high regionen of the amp, where the most important things happen with sound reproduction, matching is the most underastimate thing in audio wonderland.
I go try simulate the Vfet IRF240, has little much capacitance tough.
regards
- Status
- Not open for further replies.