Replace IRF by Lateral Mosfet

[lurker]

Is possible to change the fosgate output IRF9540 / IRF540 with 2SK1058 / 2SJ162?

If yes, how to do? I have the schematics.
p.s. excuse my poor english

 

[davidsrsb]

Laterals have very different characteristics, with much higher rds and much lower gm. You would not normally expect to see more than a few volts Vgs on a vertical fet in a linear audio application.

 

[Upupa Epops]

Lateral mosfets have also quite different biasing voltage, so direct replacing isn't possible....

 

[richie00boy]

And the pinout is different, requiring some innovative soldering and rewiring.

You would also have to disable the amp temperature compensation.

I really wouldn't bother.

 

[AndrewT]

Hi Richie,
do you need to disconnect the bias temp comp?

What are the disadvantages of leaving the comp connected?

 

[AndrewT]

Hi David,

You would not normally expect to see more than a few volts Vgs on a vertical fet in a linear audio application

Have you got this the right way round?
The lateral FET has a low voltage bias requirement and the vert has a high voltage requirement.

 

[ilimzn]

do you need to disconnect the bias temp comp?
What are the disadvantages of leaving the comp connected?

Yes, and if you don't the bias current will drop well into crossover distortion when the outputs heat up. This is assuming the existing compa can actually generate a bias voltage for the gates that is low enough (~2V instead of ~6V).

Have you got this the right way round?
The lateral FET has a low voltage bias requirement and the vert has a high voltage requirement.

Low BIAS voltage requirement, but not low voltage requirement. The threshold voltage is low (~1V) but the gm is also low so more Vgs is needed compared to verticals, at the same output current. The lower threshold voltage advantage is lost very soon as output current increases. Unless the driving section has separate higher voltage power rails, as a concequence of using laterals, this exact higher Vgs presents an increased rail losss, and a net decrease of maximum power, compared to verticals.

 

[davidsrsb]

Some VFETs have low threshold too ("logic level")

The IRF540 is fully turned on at Vgs=5V, the lateral requires Vgs=10V just to get above 10A

 

[AndrewT]

Hi,
thanks gentlemen for another angle on Gm. It helps me get the difference between LFET and VFET into my mind.

Biasing the push pull LFET output stage with a BJT driver pair.
Should the bias temp compensator be attached to the driver heatsink? to keep the bias voltage to the gates substantially constant.

David,
most LFETs have an Id max = 7A. Only >=10A when driven in one shot short term peak. So you should never see Vgs=10Vpk when on any music signal.
After allowing for temperature derating your current limit is even lower and if it is a short circuit current then the SOAR limit brings permissible Id much lower still.

 

[richie00boy]

Yes you could include thermal compensation on the drivers, but as their temperature remains reasonably constant and not that hot, in practice it can be skimped on.

I think davidsrsb was just making a point about the big differences in transconductance between lateral and vertical devices with his comment, rather than stating a true 'rule' that is physically practicable.

 

[ilimzn]

Hi,
thanks gentlemen for another angle on Gm. It helps me get the difference between LFET and VFET into my mind.

Actually, perhaps the proper way of referring to these devices is LMOS and VMOS. Reason being, there is (or rather was) such a thing as a VFET - a power JFET. Most small signal JFETs are lateral, i.e. LFET.

Biasing the push pull LFET output stage with a BJT driver pair.
Should the bias temp compensator be attached to the driver heatsink? to keep the bias voltage to the gates substantially constant.

This is usually not necessary because the driver pairs are degenerated by their emitter resistor(s), so their effective tempco is lowered, unless we are talking big currents and small resistors.

most LFETs have an Id max = 7A. Only >=10A when driven in one shot short term peak. So you should never see Vgs=10Vpk when on any music signal.

Actually, you might be surprised. First of all you need to check what is a 'short term peak'. For instance, on many parts this is defined so that substantially higher than IDmax(DC) can happen on a normal audio signal half-wave (keep in mind that to a large degree the complementary devices share the load). This is important because even at midlevel loudness, amps DO clip. By your reasoning, one does not need gate protection zeners 😉
There are a couple of other examples where you might get unexpectedly high Vgs, such as clipping (even without a load) with stages that have higher rails for the driver stage, oscilation (god forbid), shorts, very reactive loads. As ritchie00boy said, the example was illustrative, not necesairly real, but it is good practise not to let these thing to chance.

After allowing for temperature derating your current limit is even lower and if it is a short circuit current then the SOAR limit brings permissible Id much lower still. [/B]

True, but consider what happens when you short circuit the output. The NFB fails (since it's input is shorted) and any signal slams the gates all the way up or down. Depending on how you drive your outputs, you may get a full rail voltage as Vgs. MOSFETS in general are FAR less tolerant of Vgs limits being broken, than od Id limits being broken - the likelyhood of survival is a lot higher in the second case.

 

[AndrewT]

Hi Illimzn,
thanks again.
I agree that zener protection to prevent excessive Vgs is necessary.
Weird voltages can accidentally appear on the gates. Clipping is an accident that should not happen during music reproduction but probably inevitable and should be protected against.

I have changed the Zeners on my Sugdens from 12V to 7V5 to help control excessive output current. Additional zeners inserted directly from gate to source pins. Retained the standard 12V at the driver emitters.

Vgs = rail voltage is why the VAS transistor must also be protected if the output protection triggers. (assuming VAS > driver> Vmos).

 

[AndrewT]

Hi Ritchieboy,
yes, David was just comparing

the big differences in transconductance between lateral and vertical devices

But he took the lower current devices (Lmos) outside their operating envelope while keeping the higher current (Vmos) inside the envelope. That makes the comparison invalid. If he were to multiply the Lmos device quantities up to closer match the current ability of the Vmos devices then the high current drive voltages of the Lmos would be much closer to the Vmos.
Which brings us back to the original Question is it possible to replace Vmos with Lmos, probably not worth making the change due to;-
1. the need to check and probably change the design to suit Lmos
2. the considerable cost of multiple driver replacements for what level of performance improvement, if any.