N-Channel Mosfet Linear amp Open Circuit Cross - Conduction

[kASD]

Hi ,
Recently I came across a Friend's Mosfet N-Channel Class-AB linear amp similar to Anthony's holton amp .
When the amplifier is fed with a signal sinewave or music , With Output not connected to any speaker or load i.e open circuit operation.At increasing the input signal from zero to a finite value the amp starts heating and consumes lot of current and dissipiates the energy in the form of heat without any load connected to its output.
Are these are signs of Cross-conduction or something related to biasing problem.OR not a Problem At All.
VCC=+- 56VDC
VGS=3.9VDC
I[drain quiescent current]=75mA

With Regards
AmPmAn

 

[hunter audio]

hi

i guess

1) it may be oscillating or

2) check the output in an scope make sure the pcb design is not arising any spurious hi frquency wave form at the input

3) constantly monitor quiscent current - see if it is not running away

i have yet to try out a Holton design though every one here says they are quite stable is this the the buz circuit in the audio vedio special - then the ntc is not of the correct value

any way take care

suranjan

 

[kASD]

Commentable Thoughts

hi

i guess

1) it may be oscillating or

2) check the output in an scope make sure the pcb design is not arising any spurious hi frquency wave form at the input

3) constantly monitor quiscent current - see if it is not running away

i have yet to try out a Holton design though every one here says they are quite stable is this the the buz circuit in the audio vedio special - then the ntc is not of the correct value

any way take care

suranjan

Hi HUNTER,
I think the Feedback factor is not adequate .
The biasing is very stable indeed.Nor there are any oscillations occuring in it.
Whereas, In my own professional design amplifiers I use a Anti parallel Diode with gate resistor to over come any cross- conduction between upper and lower mosfets.

With Regards
AmPmAn

 

[subwo1]

I think I noticed on many n-channel AB amps that since the gate pull-down resistor usually supplies less current than the transistor driven pull-up, as the input frequency is increased, the amp operates more like class A. I could be missing something.

 

[sajti]

If the heat caused by cross conduction, You have no heat if You drive the amp with low frequency sine wave - say 100Hz. And if You apply high frequency square wave (10-20kHz) You will get more heat. Try this test.
If the result is more heat with high frequency try to decrease the gate resistors (I used 22ohms with good result). If it doesn't help, increase the current of the driver stage.

sajti

 

[tschrama]

Hi,

I've had done some investigations in cross-conduction in a N-ch MOSFET JLH-amp and I agree with sajti.. cross-conduction should be freq dependend. However I could not decrease the gate-snoopers below 150 Ohm, so be carefull. Adjusting the gate-snoopers might help, but you need to monitor the currents in you circuit with a osci to do that propperly.


Goodluck,
Thijs


PS
Does anyone have the schematic of the N-channel amp of aussiesamps? I was availeble for public, bus hasn't been updated and is still missing.

 

[kASD]

Commentable Thoughts

At input sinewave 50 Hz , it also shows cross conduction.
Regards
Ampman

 

[tschrama]

At input sinewave 50 Hz , it also shows cross conduction.

hmmm then something is wrong. Could you post a schematic?

 

[kASD]

Commentable Thoughts

The circuit is similar to Holton's circuit.
The R20 is creating cross-conduction to an extent, when i eliminated it the cross conduction also lowers.
In my Professional circuits our N-channel amps[which are our own creations] use Antiparallel diode with gate resistors to overcome this cross-conduction.
With Regards
Ampman

 

[tschrama]

Hi,

Thx for the schematic. I would also think R20 is the key. This resistor ensures that both output halves are driven from the same inpedance. This is important as the output mosfet's capacitances have to be charged and dis-charged at equal speed. Well that what I think, and did when building my MOSFET JLH 10W Class A, which has a similar output stage.

Adjusting this R20, while observing the current through both output halves, amp driven with a square wave will make it easy to minimize cross-conduction.

Bye,
Thijs

 

[tschrama]

BTW,

shouldn't both output resistorts (0.33 Ohm) be at the MOSFET's source.. this helps equalize the transductance of both push-pull halves and so lowers distortion. I think this error could also couse some more corss-conduction.

regards,
Thijs

 

[kASD]

Commentable Thoughts

Yeah ur rite to a point that the resitors must be at source terminals of mosfets, the R20 is key to the problem.

Whereas in our professional designs we dont use source or drain resistors when paralleling the mosfets.

In JLH Class-A amp the mosfets Exhibits cross conduction which can be avoided by using low value gate resistors say 56ohms and using anti parallel diodes to decrease Turn -OFF time .

Regards
AmPmAn

 

[tschrama]

Well, I had to keep the gate-snoopers above 150 Ohm, else oscilation became apparent on my scope. BTW The higher the voltage supply, the higher this value needed to be to keep oscilation out (i used IRF540). Maybe some sort of oscilation is causing your cross-conduction problems?

HOw does that

using anti parallel diodes to decrease Turn -OFF time .

work? Isn't that only nessecery for class B... the JLH is entirely class A, both MOSFETS should be conducting all the time, no turn-off?


Greetings,
Thijs

 

[lumanauw]

I have a similar problem with auto-oscilating class D (but my problem occurs with speakers attached). When played small volume, everything is ok. But when a heavy bass hit, it sucks current very much (far more than normal) Is this because cross conduction between upper and lower mosfet too? All mosfets are N channel.

 

[Nexus]

Hello Guys,

what do you think of the next schematic?

The bias servo needs some adaptation in order to operate accordingly.

Any Ideas?


Regards:
Nexus

 

[Nexus]

new try to send schematic....

 

[Nexus]

last try..

 

[kASD]

Commentable Thoughts

Well, I had to keep the gate-snoopers above 150 Ohm, else oscilation became apparent on my scope. BTW The higher the voltage supply, the higher this value needed to be to keep oscilation out (i used IRF540). Maybe some sort of oscilation is causing your cross-conduction problems?

HOw does that work? Isn't that only nessecery for class B... the JLH is entirely class A, both MOSFETS should be conducting all the time, no turn-off?


Greetings,
Thijs

U can only use low value gate resistors when ur driver stage current is enough, which is in our case is more than enough.

Secondly, I commented upon JLH design Biased in Class-AB mode not Class-A.
using anti-parallel diodes with gate resistors the turn off time of adjacent mosfet is decreased and the cross-conduction is eliminated.


Hi Nexus
Ur schematic is very small in size PLZ post a bigger illustrative view.


Regards
Ampman

 

[tschrama]

U can only use low value gate resistors when ur driver stage current is enough, which is in our case is more than enough.

Interesting.. how does that work? I don't understand how the bias current through the driver can effect the properties of the MOSFETs. Is 10mA enough?

greetings,
Thijs

 

[kASD]

Commentable Thoughts

Answer is simple The charging and discharging of Mosfet gate capacitance will be fast enough to overcome cross conduction, when the driver current is more.
regards
AmPmAn