About my hybrid nofeedback AMP

ckwong99

Member
2005-05-24 6:41 am
HK
Dear All,

I had already finished this no NFB power AMP no conspicuous bad sound. However, observed from CRO the input against output signal have phase shift start from 500 Hz. Measured it will have 40 degree shift when input signal set to 250k Hz. Although, the output signal frequency no change corresponding to input frequency.

When input waveform change to square wave its output waveform change to wave shaped. Hence, the following questions want to get your help. Thanks.

1) When we use No NFB design will turn out above problem ?

2) I know this design first stage is voltage gain -> Common source as current gain so that out of phase 180 output happened. But I don't understanding what the function of resistor 47 ohm shown in attached diagram ? For my experience I can't to follow this value because the zero DC output can't be achieved until I replaced it with 500 ohm value.

3) Will the sound get better if set higher DC bias on 10k VR ?

4) At input part I added the thick line connect them , not sure it is exactly connection? Or just sticky follow the schematic showed ? Is that somewhat a tricky thing ?

Much appreciate your answer.

CK
 

Attachments

  • nofeedback_amp7.zip
    76.2 KB · Views: 380

ckwong99

Member
2005-05-24 6:41 am
HK
You means just using 1 pair of PWD MOSFET instead of 5 pairs. If I guess it right then how to see why nowadays amplify design will also use several pairs. A few study to say it can bring higher PWD and linearity but less noise. Is it correct ?

:)
Anyway, can anyone help to ask my questions ? I really want to understand it.

:smash:
Please note that my first post have typo mistake as follow :

"250k Hz" corrected to "25k Hz"
 

ckwong99

Member
2005-05-24 6:41 am
HK
To Amplifierguru :

Did you talk the 4 coupling caps between tube and MOSFET ? If so how large should be increased ? ( Current value are 0.3u per each be used )

I think the 5 pairs of PWD MOSFET are not matched. But their source resistors 0.5ohm was added.

I does not phobic NFB ..... / I can find 80% equipement using
Few articles acclaimed NO / Little bit NFB adpoted in circuit especially doing PWD AMP. It will bring accurate phase and good dumping but trade off will be less stable ... FRom my point of view, NFB just one of compensation technique to meet output ideally.

To Anatech & Amplifierguru :

Any suggestion or change can do on it ? If a bit NFB be used where I can add ?
 

anatech

Administrator
Paid Member
2004-06-06 8:31 pm
Georgetown, On
See how the amp behaves with about 6dB overall negative feedback. Adjust this figure up and down to see what effect it has.

There are many other ways to improve this amp, but you should explore these yourself. Beware of individuals making blanket statements, they normally do not actually know what they are talking about. Any artical or statement taking an extreme view should be discarded. Remember to consider the entire amplifier circuit. Not just one parameter or circuit element.

-Chris
 

ckwong99

Member
2005-05-24 6:41 am
HK
:) anatech :

I just a beginner on this .... somewhat a simple circuits can understand. However, in step to modify or design will out of my knowledge.

If it only spend a little time, can you explore the change according to your advise "6dB overall negative feedback" for me.

Much thanks for you help ! Just reply to my mail box if oversize of file after you revise.
 
Hi AmpGuru, Hi CK !

I thought that the 4 coupling caps at the bias chain would define
a low frequency roll off, but should not affect much at higher frequencies.
My first guess would have been that the gate resistors at the output MosFets in combination with the gate-drain capaciors do form a low pass. But this theory does not match with CK's observation of a stable magnitude. If we observe 40 deg phase shift at 25kHz, then I would also have expected some 20%-30% decrease of the magnitude.
But this circuit seems to show some all-pass behaviour.

CK:
Thick line in the input section:
It looks like there a simply for triode tubes working in parallel.
Probably to enable enough driving current for the output section.
Furtheron you could simply measure the signals/phase shift at several points (output of the tubes, bias chain, gates of the output fets) in the amp. This might help to find the reason.

Bye and good luck
Markus
 

ckwong99

Member
2005-05-24 6:41 am
HK
Hello ChocoHolic,

So a cool picture you looks.

First of all the coupling caps was no help on HF phase shift due to I had proveded on Saturday. You are right !!

:smash:
"But this theory does not match with CK's observation of a stable magnitude. If we observe 40 deg phase shift at 25kHz, then I would also have expected some 20%-30% decrease of the magnitude."

I need to take note and measure again for above assumption.

Beside, I agree with your thinking. All-pass is the designer expected. Are you sure that the phase shift is the killer of a bad wave-shaped ouput, when input signal is square wave. More or less feedback is a kind of compenstation. Does it a major problem point to affect it ?

What do you think on the R51 - 47ohm resistor ? I increase it to 450 ohm can have stable zero DC output. As well as better the sound I felt.

Anyway thanks to you concerning and discussion. I will try to probe the circuit output somewhere again.
 
Hi CK!

I am wondering about different R45 and R46.
I cannot exactly read the names in the schematic, but I guess
R45 is one of the gate drive resistors of of the upper MosFets and
R46 one of the lower sider gate drive resistors.
From my perception the large capacitor of 10nF between gate and
drain should easily dominate vs. the nonlinear (and different N-chanel vs. P-chanel) gate charges of the Fets.
In order to improve the phase shift and rectangular response, you could try two steps:
1. Decrease R45 in to 220Ohms and check if the amp is still stable,
I would guess yes.
2. Decrease the gate-drain capacitors to 6.8nF

R51 is influencing the bias of the output stage and your step to increase its value should help to reduce the crossover distorsions, but will generate more heat, due to the increased idle current.
As long as you are not facing any thermal runaway, it should be fine.
On the other hand I WOULD expect thermal runaway, because
there is no thermal compensation and no degeneration resitors.
Are you using exotic MosFets, which offer a negative temp. coefficient already at low currents?

Bye
Markus
 

ckwong99

Member
2005-05-24 6:41 am
HK
Dear Folk,

All you guessed are right. I have seen a few circuits like this, they would like to use different value of R45 and R46. Perhaps all the N/P Channel MOSFET with a certain kind of behaviour need to match.

PWD FET is Hitachi J56 , K176 model. I have bought it many years ago.

I will follow your instructions. See how the best of sound will blowing out.

Thanks
CK
 

ckwong99

Member
2005-05-24 6:41 am
HK
Hi ChocoHolic,

Just complete some meausrement FYR.

Constant Input : AC 0.42rms / 1.2Vp-p

1) 14.8vrms / 44vp-p / @1k input

2) 14.14vrms / 41.5vp-p / @10k input

3) 13.1vrms / 38.5vp-p / @20k input

4) 9.16vrms / 28vp-p / @50k input

5) 4.62vrms / 17vp-p / @100k input

About phase shift 40 degree at 20kHz. I found such problem also exist on the 4 output copuling caps. Therefore the problem is in the voltage gain area.
 

ckwong99

Member
2005-05-24 6:41 am
HK
You give me best idea. Once, I get a result will report to you.

It have already replied from Tube Column .......

> 40 degree shift when input signal set to 25k Hz. ... square wave its output waveform change to wave shaped.

Of course. The tubes are a fairly high impedance. The MOSFET inputs are big capacitors. It is going to have a high frequency droop, and detectable phase shift before obvious roll-off. While some folks will disagree, I think 40 degrees at 25KHz is "nothing". (Put a mike right against a tweeter and measure 20KHz phase.....)

THanks to all your valuable inputs.

CK
 

ckwong99

Member
2005-05-24 6:41 am
HK
Status report :

Still have 45 degree shift on tube output when it disconnected to the middle stage. Even the 4 small value CAPs had been dismounted.

A new question discovered...

When I attempt to do a maximum power output test. Seems the the clipping single was introduced into the speaker out. When peak voltage was raised in 53VAC at 500 Hz. There is also same as the value 53VDC to the Drain of the output stage.

At that time very bad and 'zzz' sound to be heard.

Should it be increase the DC supply at output stage ? I find this FET output similar design they'd like to use 75-90vDC supply.

Will it bring more benefit when using higher DC ?