Where you got the 5-6A? The graph shows 18A.
The graph at 1A shows -600mv drift with 125° increase, that is 4.5mv/C°.
This is the temperature of the case, but the upper hard plastic in direct contact with the channel has less than 4mv/C°.
a
Attachments
There is also the lm334 as temperature sensor to adjust the bias. Both diode and lm344 are used in the Phoenician follower
This is similar principal to Bogdan's Single Ended cfa or AKSA's Nirvana, but much simpler. It measures 0.05% 17Vp 1khz and 0.08% 10khz. It is also SE character see 5W distortion harmonics. It is biased 1.27A. Max 20W. The LM334 can be replaced by two diodes in series with R8.
Attachments
Last edited:
Hi Hayk,
As promised I had a look at your two different power supply versions.
I used different input Mosfets, because your models are not available in LTSpice's library.
The thing I noticed in your images, is that one model uses 3k9 for R2 and the other one 3K8. That will have an immediate effect on loop gain and thus on THD.
I could not see what input signal you used, but I assumed 1Khz@1Volt peak.
Further I had to adjust R9 from 3k4 to 4k to get balanced currents through the IRF Mosfets as visible in the images below.
I see hardly a difference in distortion between the two versions in LTSpice, however distortion is quite a bit higher with 0.025% as your resp 0.004% and 0.006%.
The other thing I noticed is that the V/I version with two bipolars has a lower distortion and a cleaner square wave response for the same 1.4A bias current and the same +/-24Volt supply.
Hans
Attachments
Hayk,
The higher harmonics only seemed to be different because of one of the many anomalies in the .four algorithm in LTSpice.
The FFT on the other hand is way more reliable and as you see in both plots in the image below, they are absolutely identical, just as to be expected.
Hans
The higher harmonics only seemed to be different because of one of the many anomalies in the .four algorithm in LTSpice.
The FFT on the other hand is way more reliable and as you see in both plots in the image below, they are absolutely identical, just as to be expected.
Hans
Attachments
Hayk,
Because I was still puzzling why you rejected the bipolar version as a non V/I amp, I stripped both versions down to the bare level, where the lower Mosfet is replaced by a pure current source.
In your versions this current source is modulated by the input signal.
In my bipolar version the current source is modulated by the collector signal of the input transistor.
And yes, in that second case there is feedback in this collector signal involved, but for modulating the current source it makes no difference as long as it is modulated by some signal.
As a prove have a look at both circuits in the images below and their corresponding FFT's.
The current source in the left circuit is modulated by the signal through collector resistor R10 and in the circuit to right the input signal is responsible for the modulation.
In and output signals are the same, and so is the Bias.
The FFT's are absolutely identical, making it clear that the upper Mosfet acting as the voltage amplifier is 100% responsible for the end result.
Hans
Because I was still puzzling why you rejected the bipolar version as a non V/I amp, I stripped both versions down to the bare level, where the lower Mosfet is replaced by a pure current source.
In your versions this current source is modulated by the input signal.
In my bipolar version the current source is modulated by the collector signal of the input transistor.
And yes, in that second case there is feedback in this collector signal involved, but for modulating the current source it makes no difference as long as it is modulated by some signal.
As a prove have a look at both circuits in the images below and their corresponding FFT's.
The current source in the left circuit is modulated by the signal through collector resistor R10 and in the circuit to right the input signal is responsible for the modulation.
In and output signals are the same, and so is the Bias.
The FFT's are absolutely identical, making it clear that the upper Mosfet acting as the voltage amplifier is 100% responsible for the end result.
Hans
Attachments
I agree with you, your amp is superb and due to square law it surpasses many 4 transistor amplifiers proposed in Pass forum. The problem is that it has been widely publicized since J. Hiraga's criteria, that good sounding amp should have decreasing harmonics as SE. It is true if the odd harmonics are audible, than adding higher amount of even ones, as adding sugar in lemonade, you hear more agreeable sound. Since, Diy's caught the recipy, and requiring only SE character type amplifiers.
The bjt version is not square law, it can not give more than 20W.
The bjt version is not square law, it can not give more than 20W.
Last edited:
Mosfets increase in gain by current in square law, but bipolar junction transistors Ic/Vbe is square at low currents but due to early or current driven Ic/Ib at high currents it becomes on the contrary decreasing gain. I do have a 50W class A square law by 1 pair of C5200/A1943 in cascode biased 0.9A only, but it needs 2 pairs / unit. This why Mosfets are better suited for square law.
In your circuit, you see that the harmonics are not that of SE where the 2nd should by much higher than the third.
You must first resolve the power supply ripple rejection in your circuit .
You can not parallel power transistors without source/ emitter resistors, if you do apply resistors, it is no more square law.
In your circuit, you see that the harmonics are not that of SE where the 2nd should by much higher than the third.
You must first resolve the power supply ripple rejection in your circuit .
You can not parallel power transistors without source/ emitter resistors, if you do apply resistors, it is no more square law.
Last edited:
As much as I appreciate your reply, I always prefer in getting things shown before my eyes.
I took the two versions, one with the modulated current source and one with an ideal current source representing a true SE variant.
Both Amps have the same bias and supply voltage and are producing 25Watt.
In the image below, the modulated current source version at the left has even a tad less odd harmonics as the SE version, so all in all, it seems that the modulated current source V/I version that you proposed, simply behaves like a SE version.
Thank you for coming with this creative idea.
Hans
I took the two versions, one with the modulated current source and one with an ideal current source representing a true SE variant.
Both Amps have the same bias and supply voltage and are producing 25Watt.
In the image below, the modulated current source version at the left has even a tad less odd harmonics as the SE version, so all in all, it seems that the modulated current source V/I version that you proposed, simply behaves like a SE version.
Thank you for coming with this creative idea.
Hans
Attachments
Use this type of drive to avoid psu noise.
The spectrum character depends upon your commercial skills to convince your "client", see AKSA's arguments in Alpha Nirvana. The spectrum criteria was born by the Japanese after analyzing the Marantz 8b. It has the second harmonic 10db , if I do remember, higher than the third and decreases for higher harmonics. For me it is simply audio astrology. My astrology is to have no any even harmonics but only odd ones on condition that they are of opposite phase to the fundamental. Ones that tend the sine wave towards triangle instead of square.
Attachments
Last edited: