• WARNING: Tube/Valve amplifiers use potentially LETHAL HIGH VOLTAGES.
    Building, troubleshooting and testing of these amplifiers should only be
    performed by someone who is thoroughly familiar with
    the safety precautions around high voltages.

The Hybrid Class A amplifier

Status
Not open for further replies.
Jump to Latest
#61
sajti said:
If You use the MOSFET as source follower, the most important capacitance is the Cgd. That must be as low as possible.
The biggest problem is not the frequency response, but this capacitance is nonlinear, and can results distortion.
Click to expand...

sajti, Thanks for your quick response. I agree with you that Cdg is
non-linear. Of course the lower Cdg is the better. But if No-NFB
amplifier case , the impact to the sonic is much lower than NFB amp.
My FET has Cgd 100PF given by the datasheet.
 
#62
revintage said:

This is how I will build my Schaded hybrid:
Click to expand...

Hi, revintage. Thanks for you quick response to show your hybrid amp
schematic. It is very interesting for me. I quickly analyzed it and made
a simple diagram to analyze NFB topology as shown in the attached
photo.

You applied DG feed back for MOS FET. I assume it is an inverting op-amp.
Therefore feed back resistors consist of Rf and rp of 6H30. The total
gain is given by following formula(assuming IT ratio is 1:1).

G=Mu x Rf / rp

In this formula, rp is not linear. It depends on the signal level as makes
"Tube SONIC". Anyway, you are using simulator so that you will get
distortion vs signal level relationship. I guess it is "SOFT DISTORTION"
like tube amps.

I am now designing similar concept hybrid amplifier with "TUBE + Class-D"
instead of "TUBE + MOS FET".
 

Attachments

  • fettheory.gif
    fettheory.gif
    11.8 KB · Views: 724
#63
Hi ja2dhc,

You have totally misunderstood the schematic! None of your conclusions are right! And that is a good thing 😉 !

Please download LTSpice(www.linear.com) and the Schade-paper from 1938(www.petemillett.com). This could add valuable knowledge. I can then supply you with the sim file to play around with. I can also sim your circuit for you.

If you had checked the schematic properly you should have seen that the IT is 4,5:1.

About the feedback over the MOSFET it gives you a gain of 3 and also makes the characteristics being triode- instead of pentode-like. A FFT-analyzis will show this. Input impedance is also very high and more dependant on capacitances and the inductances of the IT and capacitances of the MOSFET.

So the 6H30 works into a HIGH impedance in the midband(1M). At 20Hz it works against the primary inductance and will be ca 13k. At 20k it is working against capacitances in the IT and the MOSFET and will probably be in the ballpark of 3-500kohm.
 
#64
revintage said:
If you had checked the schematic properly you should have seen that the IT is 4,5:1.
Click to expand...

My point is that your amp's characteristics is SOFT DISTORTION like
tube amp, not precise analysis. In my previous post, I indicated
"assuming IT ration 1:1". If 4.5:1 is used the formula is simply changed
as follows;

G=Mu x Rf / rp /4.5

The above includes Mu and rp which are the signal level dependent
factors, like tube amp so Tube sonic is obtained. This is my point.

I am using TINA simulator instead of LTspice. I want to see your
simulation result of SOFT DISTORTION.
 
#65
Hey ja2dhc,

Still you didn´t get it 😉. Still your formula isn't valid here! There is NO more, what you call "soft distortion"(What is that?) at hand than in the one you built! My input stage works exactly like yours even if yours see no inductance and lower capacitances.

Think you have to read Schade first: When feedbacked this way the characteristics go from pentodelike to triodelike for the MOSFET.

I also have Tina so I can make a file for you to make the concept easier to understand.

Again: Input impedance at the transformer primary is, theoretically, high at over 1Meg in the midband, ca half of that at 20kHz due to MOSFET and transformer capacitances and ca 13k at 20Hz from the 100H primary inductance.

Talking about SOFT DISTORTION, please let me know the figures of your input stage loaded by the 4,7k resistor having to deliver the required voltage 😉.

Will try to find a 6688/E180F spice model to compare.
 
#67
Hey,

Didn´t find a E180F/6688 model so I will sim your amp with a D3a instead. Or do you have one? If it works in Tina it will work in LTSpice.

Your amps outputstage might probably have somewhat lower THD than mine due to the high inherent feedback of the sourcefollower.
 
#68
revintage said:
Your amps outputstage might probably have somewhat lower THD than mine due to the high inherent feedback of the sourcefollower.
Click to expand...

Thanks so much. I've not yet implemented sim for my hybrid amp
due to lack of models of MOS FET and 6688. As I posted before in the
#53 post, actual measured characteristics are comfortable for me.
There is no necessity to do sim now. As you see from the photo in #53,
it is a typical soft distortion. THD is gradually increasing in accordance with
the output level. This characteristics are given by 6688, not by MOS FET
which is strongly NFB'ed by itself and no distortion. In net, my amp's
sonic is 6688 sound. This is the keypoint of hybrid amp.
 
#72
Hey W.......,
I speak out of over 40 years of experience with tube-gear, NOT from "seeing" computer models (even if they are a great help when designing)! You might like distorting triodestages (even if it is mostly 2nd order), I dont!

If you happen to like triodes with "too" hard resistive loading, why on earth are you so keen on the great "constant voltage source"(you had another name for it) we discussed in another thread😉 ?
 
#73
revintage said:
Hey W.......,
I speak out of over 40 years of experience with tube-gear, NOT from "seeing" computer models (even if they are a great help when designing)! You might like distorting triodestages (even if it is mostly 2nd order), I dont!

If you happen to like triodes with "too" hard resistive loading, why on earth are you so keen on the great "constant voltage source"(you had another name for it) we discussed in another thread😉 ?
Click to expand...

So, something is controversial in your logic about what I like, what I don't? 😉

You suggested to load a triode on very low resistance. Or am I missing something deeper? 🙂
 
#74
You suggested to load a triode on very low resistance. Or am I missing something deeper?
Click to expand...

Apparently you are missing the whole picture here😉 !

ja2dhc uses 4,7 anode resistor. I said that resistance was to low so I suggested 12k as that is close to the most you could use if you wanted to stay with 10mA and get some filtering, with 3k between filter caps.

Personally I prefer iron and I would definitely like to see a choke or a "CVS" in ja2dhc ś anode-circuit.
 
#76
revintage said:


As I see it, there is extremely high distortion for a triode driver. This is due to 4,7k/270k loading the triode. You will get enough bandwith AND low distortion with a higher value anoderesistor.
Click to expand...

I would try to divide the resistor into 2 pcs of 2.2kohm, and make some bootstrapping from the amplifier output. If the MOSFET has the gain about 0.95 that makes the anode resistor up to 50kohms...
The bootstrap capacitor should be 47-100uF.

Sajti
 
#77
Hey Sajti,

It would work, but then you leave the NON-NFB concept😉 .

Maybe better to arrange higher B+ together with higher value anode-resistor. 12k is possible with present 270V. I would go for 400V.

A CVS, CCS or choke/transformer would also help.
 
#80
revintage said:
Let´s be pragmatic about this: As you feed a part of the signal back to an earlier stage, what else could it be 😉 ?

Another thing, shouldn´t we look at a sourcefollower as highly feedbacked😉?
Click to expand...

Theoretically it's positive feedback.

The sourcefollower has 100% local feedback.

Sajti
 
Status
Not open for further replies.
Top Bottom