mosfet driver

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I'm thinking about changing to a mosFET driver stage for a BJT output stage.
I have seen this suggested by more than a few Members.

What parameters of a mosFET make them suitable as a Power Amplifier driver stage?
What range of those parameters suit drivers in a power amp?

If it makes much/any difference, I'm looking at the 50W to 150W into 8ohms category.
 
I'm thinking about changing to a mosFET driver stage for a BJT output stage.
I have seen this suggested by more than a few Members.

What parameters of a mosFET make them suitable as a Power Amplifier driver stage?
What range of those parameters suit drivers in a power amp?

If it makes much/any difference, I'm looking at the 50W to 150W into 8ohms category.

Check out the Parasound amplifiers. John Curl uses this solution in many of his desing.

Sajti
 
Not many mosfets are suitable for that job. High transconductance and and at the same time low capacitance. Just like bipolars where Cob is important for VAS, Cis is also important with mosfet. The closer to the output stage (driver, high current) the more important the transconductance.

But because there aren't many (complementary) mosfets around, you can see a trend on what mosfets are used by designers, starting back from JLH. Only the Toshiba are the new products but I haven't seen it used often (e.g. in a dual sizklai with BJT, if it is considered a driver also). I guess because it is new/rare and expensive.
 
Here is a detailed answer and table from Jay:

http://www.diyaudio.com/forums/soli...-mosfets-complementary-pairs.html#post3133236


Have also a look at the Classe amplifiers. Most of them are using lateral mosfets.
At the moment I have a fet driver, bjt OPS on my workbench. The more I am working, measuring and changing on it, the more I tend towards lateral mosfets running on there ideal bias current to eliminate bias changing of the OPS due to temperature fluctuations.
 
Thanks.

I had forgotten about the mosFET VAS Thread that I contributed to, in a very small way.
I had also forgotten that same Thread discussed Driver parameters.

It seems from Jay's work that 510/9520 as well as 710/9620 might make good driver transistors. The 510/9520 being limited to <80W into 8ohms.

Can a few of you identify the range of the parameters that make good drivers?
first impressions are that Cob of 100pF to 600pF, gm of 0.7S to 2S, Rds on of 0r5 to 3r are at least three of the necessary parameters. Does any one disagree?
 
I'm thinking about changing to a mosFET driver stage for a BJT output stage.
I have seen this suggested by more than a few Members.

What parameters of a mosFET make them suitable as a Power Amplifier driver stage?
What range of those parameters suit drivers in a power amp?

If it makes much/any difference, I'm looking at the 50W to 150W into 8ohms category.

I see an advantage and a disadvantage.

Good: a MosFet by definition has "infinite Hfe" so it makes it easier for the Vas stage to drive it.
Remember it will see the load impedance multiplied by the combined Hfe of output transistors and the drivers in the chain.
That's why on very high power amplifiers (multi output devices) often 3 stages are needed, instead of the usual 2 (output + driver) .

In fact, in very high power industrial electronics, think Metro train electric motor switching, soldering machines, etc. , IgBT (insulated gate bipolar transistor) are very common: a die integrated Mos driver and very high current/power bipolar output
Insulated-gate bipolar transistor - Wikipedia, the free encyclopedia

Bad: you'll lose extra 3 or 4 Volts across the MosFet compared to roughly 1V adding an extra bipolar and you'll have to deal with Mosfet high gate capacitance.

EDIT:
How hard can one push a sot223 with a 2W rating as a driver?

Mosfet or not, I'd feel very uneasy using a Sot223 cased transistor to drive an output power transistor.

Heat is heat and it must be dissipated.

Those 2W must certainly refer to impossible-in-practice "leads at constant 25 DegC " Lab specs.

I'm quite certain a little Sot223 dissipating 2W will burn a hole in my finger if I touch it.
 
Last edited:
I'm exploring using MOSFET drivers in a triple. So far the simulations indicate good performance with this arrangement. I'm looking at using a BJT EF stage to drive the gate capacitances, and the MOSFETs to drive the BJT output transistors.
 

Attachments

  • Mosfet Driver.png
    Mosfet Driver.png
    76.8 KB · Views: 827
I'm using MOSFET in driver stage.
The good point is: the sounding is awesome! Like when you run EF2. But light load for VAS/TIS.

The MOSFET had been using is 2SK2013/2SJ313.

But, these MOSFET very hard to find. I don't have any experience with IRF710/IRF9610. Are them is good?
 
Last edited:
.....................
Mosfet or not, I'd feel very uneasy using a Sot223 cased transistor to drive an output power transistor.

Heat is heat and it must be dissipated.

Those 2W must certainly refer to impossible-in-practice "leads at constant 25 DegC " Lab specs.

I'm quite certain a little Sot223 dissipating 2W will burn a hole in my finger if I touch it.
I am not planning to run a 2W sot223 mosFET device @ 2W
I currently use 15W to 30W BJT devices @ between 200mW and 1.5W with added heatsink.
A 2W device cannot work at 1.5W even with massive PCB pad cooling.
But could it survive 100mW, or 200mW, or 300mW?
That's the question:
How hard can one push a sot223 with a 2W rating as a driver?
 
I'm exploring using MOSFET drivers in a triple. So far the simulations indicate good performance with this arrangement. I'm looking at using a BJT EF stage to drive the gate capacitances, and the MOSFETs to drive the BJT output transistors.
you are running the pre & drivers hot.
2W through the latFET drivers and what through the pre-drivers?
What is the voltage across the 220r (R45) emitter resistor?
 
A 2W device cannot work at 1.5W even with massive PCB pad cooling.
But could it survive 100mW, or 200mW, or 300mW?
That's the question:

SOT223 is slightly better in specification compared to 4-pin DIP package (such as IRFD110).

In general SOT223 is 2W device with more than 1.25W capability (according to specs, without heatsinking).
IRFD110 is 1.3W device with 1W capability (according to specs, without heatsinking).

In DOGC-H (Boraomega) LTP, each IRFD110 is biased with 10mA giving around 0.5W with acceptable temperature here in "hot" country. So I cannot see why the SOT223 cannot withstand the same power.

Look at how extra heatsink is used in DOGC-Mk3 (I think from Boraomega himself, the pic).
 

Attachments

  • Bora_DOGCMk3.png
    Bora_DOGCMk3.png
    456.6 KB · Views: 459
Last edited:
you are running the pre & drivers hot.
2W through the latFET drivers and what through the pre-drivers?
What is the voltage across the 220r (R45) emitter resistor?

Yes I am running them 'hot'. The MOSFETs are running 25mA and the pre-EF is running about 8mA. At your target output power range the dissipation would be significantly lower due to the reduced rails. I have built the OPS in its SlewMaster form and am looking to rework it into the MOSFET driven variant.
 
Cordell's models exhibit a lower threshold voltage in simulation than the physical devices do in real life. This means some values need to be adjusted between the simulation and real life, such as the pre-driver Re needing to be larger in value.

I don't have much practical experience with SMD so I can't offer you a valid opinion with respect to what is acceptable in terms of dissipation. If you know the approximate dissipation you can estimate its temperature rise and decide if that will be acceptable or not.
 
It's the cooling effect of the PCB pads that I can't estimate.
Easy for discrete heatsinks to find Tc and then de-rate for temperature.

They tell us that 2W assumes X sq inches of pad area.
Does that mean pad on both sides, or only the one solder attach side?
do PTH make a significant difference?
does it mean no cuts or holes within the area?
does it mean no other pads for the other pins in the area?
does it mean equally around the drain pad, or can it be offset slightly, or offset to completely to one side.

There seem to more questions in my head than I have seen answers for.
 
Cordell's models exhibit a lower threshold voltage in simulation than the physical devices do in real life. This means some values need to be adjusted between the simulation and real life, such as the pre-driver Re needing to be larger in value.
What is the operational Vgs for your latFETs at the {25mA+output base current} that you are running them at?
 
Last edited:
It's the cooling effect of the PCB pads that I can't estimate.
Easy for discrete heatsinks to find Tc and then de-rate for temperature.

They tell us that 2W assumes X sq inches of pad area.
Does that mean pad on both sides, or only the one solder attach side?
do PTH make a significant difference?
does it mean no cuts or holes within the area?
does it mean no other pads for the other pins in the area?
does it mean equally around the drain pad, or can it be offset slightly, or offset to completely to one side.

There seem to more questions in my head than I have seen answers for.

Hi,

the pad is necessary only on the soldering side.
The logical solution to use tetragonal area, with no cuts, or holes on it.
You can use SMD heatsink attached with soldering.

Sajti
 
It's the cooling effect of the PCB pads that I can't estimate.
Easy for discrete heatsinks to find Tc and then de-rate for temperature.

They tell us that 2W assumes X sq inches of pad area.
Does that mean pad on both sides, or only the one solder attach side?
do PTH make a significant difference?
does it mean no cuts or holes within the area?
does it mean no other pads for the other pins in the area?
does it mean equally around the drain pad, or can it be offset slightly, or offset to completely to one side.

There seem to more questions in my head than I have seen answers for.

Andrew,

Here is an application note that may be useful >HERE<.

From what I can gather from a few quick searches, the more generous the thermal pad is the better for obvious reasons. It is also considered good practise to place a thermal pad on both sides and connect them with an array of PTH vias. I suppose there would be nothing to stop you from also gluing a small metal tab to the top of the package to aid in cooling.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.