Wanting an Nmos

[Tenderfoot]

I'm looking for an Nmos able to handle 24 volts at an amp or so, maybe a TO220 package and easily available. Here's the harder part I think. I want it to turn on 'slowly' I don't mean time-wise but to turn on over a wide range of gate voltage, no specific range in mind, just 'wider than most'. All the FETs I've tried in simulation turn on sharply over a half volt or so on the gate, most are used as switches I guess. I'm hoping for something more linear. Any suggestions? (yes I know it'll heat up but that's ok)

[Zero999]

All the FETs I've tried in simulation turn on sharply over a half volt or so on the gate, most are used as switches I guess. I'm hoping for something more linear. Any suggestions? (yes I know it'll heat up but that's ok)

You'll find that all MOSFETs do that. The circuit you're using is the wrong type. You need to use negative feedback to monitor the current through the MOSFET or voltage across it to gain the desired turn-on characteristic.

[temperance]

You are probably trying to solve something the wrong way around.

[Zero999]

The behaviour described by the original poster is intrinsic to all FETs, whether they're NMOS, PMOS, J-FET or P-FET. The polarity and magnitude of voltages and currents may differer but the curves are similar.

Simulation comparing the response of a small MOSFET vs to a JFET. The X-axis is the get voltage which is swept between 5V and 5V and the Y-axis is the drain current. Note the curves are the same shape.

[Tenderfoot]

which is swept between 5V and 5V

That's a very small voltage sweep!


Thanks for your above replies. I'm just asking a question, wondering if there is a mosfet to replace a Darlington pair used as an emitter follower. That gives me two diode drops in a working circuit where I would rather not loose that voltage. I know at one time there were...  vertical? fets used for audio and I had assumed they might be more 'linear', though mostly obsolete now.

[NE666]

How deep are your pockets?

https://www.digikey.com/en/products/detail/littelfuse-gsd/IXTH2N150L/5810057

[Benta]

which is swept between 5V and 5V

That's a very small voltage sweep!


Thanks for your above replies. I'm just asking a question, wondering if there is a mosfet to replace a Darlington pair used as an emitter follower. That gives me two diode drops in a working circuit where I would rather not loose that voltage. I know at one time there were...  vertical? fets used for audio and I had assumed they might be more 'linear', though mostly obsolete now.

The VFETs are legendary and were produced (and used) by Sony and Yamaha around 1980 for their audio power amps.
Unobtainium today, and not really suited for your use.

As you seen from Zero999's (green) plot, you won't gain anything from using a MOSFET, rather the opposite. Instead of 1.4 V drop you'll have >2.5 V instead.
Might be better to look for a high-gain single transistor instead of a Darlington. Another option could be a Sziklai pair.

[Tenderfoot]

How deep are your pockets?

https://www.digikey.com/en/products/detail/littelfuse-gsd/IXTH2N150L/5810057

Not that deep 'specially when delivery is added.

 Another option could be a Sziklai pair.

I believe a Sziklai pair has a saturation of around two volts?  EDit..I am wrong, don't know where I got that from.(I'd missed that a fet might have around the same voltage loss)

[Zero999]

which is swept between 5V and 5V

That's a very small voltage sweep!


Thanks for your above replies. I'm just asking a question, wondering if there is a mosfet to replace a Darlington pair used as an emitter follower. That gives me two diode drops in a working circuit where I would rather not loose that voltage. I know at one time there were...  vertical? fets used for audio and I had assumed they might be more 'linear', though mostly obsolete now.

I meant -5V to 5V. I know I messed up the text, but look at the plot.

A Darlington pair also has a fairly steep on voltage characteristic. Here's a similar plot, but with a Darlington pair, rather than FET. S1 and I1 are models for the simulator to change V1 to a current source, i.e. limit the current, when the base-emitter junctions start to conduct. I could have simply used a resistor, but this illustrates the sharp turn-on more clearly.



 Darlington Vb Ic curve.asc (1.18 kB - downloaded 22 times.)

Why do you want to replace a Darlington pair with a MOSFET?

[amyk]

It sounds like what you want isn't a MOSFET, but an amplifier.

[Benta]

Well, we're talking emitter followers here (according to post #4), so the "steepness" isn't really an issue, but an advantage for signal fidelity.

[mtwieg]

Thanks for your above replies. I'm just asking a question, wondering if there is a mosfet to replace a Darlington pair used as an emitter follower. That gives me two diode drops in a working circuit where I would rather not loose that voltage. I know at one time there were...  vertical? fets used for audio and I had assumed they might be more 'linear', though mostly obsolete now.

So on one hand you want a transistor whose control voltage to output current curve is "slow", but you also want something like a darlington with very low voltage drop. These two things contradict each other. You should probably give more detail on what you're actually trying to do, otherwise it sounds like you're going to get unhelpful feedback.

The VFETs are legendary and were produced (and used) by Sony and Yamaha around 1980 for their audio power amps.
Unobtainium today, and not really suited for your use.

I think you've got it mixed up; nearly all power MOSFETs are vertical, lateral MOSFETs used to be popular for audio applications (and others requiring operation in saturation) but are very rare nowadays. I think a couple companies like Exicon still make them for audiophiles though.

[Benta]

The VFETs are legendary and were produced (and used) by Sony and Yamaha around 1980 for their audio power amps.
Unobtainium today, and not really suited for your use.

I think you've got it mixed up; nearly all power MOSFETs are vertical, lateral MOSFETs used to be popular for audio applications (and others requiring operation in saturation) but are very rare nowadays. I think a couple companies like Exicon still make them for audiophiles though.

I've got nothing mixed up here, but it seems you're out of your depth.
The VFETs made by Yamaha and Sony can be characterised as almost perfect triodes or "power JFETs".
VFET is the marketing term, the scientific name is SIT or "static induction transistor".
Only two complementary sets were ever made commercially, Sony's were 2SK82 and 2SJ28, Yamaha's I don't remember.
https://en.wikipedia.org/wiki/Static_induction_transistor
Very expensive to produce, which is why they disappeared in favour of planar, trench-gate etc. technology.

[David Hess]

Shunt feedback, with a resistor from the drain to the gate, will create a gradual turn-on characteristic.  This works just as well with a bipolar transistor.

[Tenderfoot]

Time to own up, I made a silly mistake. The circuit is nothing to do with audio, nor does it need negative feedback. It's simply a fade up and fade down of some LEDs. The circuit uses a R/C and Darlington as a driver. I made it a couple of years ago and it's been working fine till it got water damaged. I thought I'd rebuild it maybe with a FET. Drew up a simple circuit in MicroCap simulator and got a strange result. Well what I'd done without thinking was use the FET as low-side switch, (NMOS is always low side switching, right) forgetting in the original I had used a high-side Darlington. This of course gave a result completely different to the way I knew the original circuit worked. I'll include a picture of what I saw, the FET result is in green, original circuit in red. Moral...  Less drinking, more thinking.

[Zero999]

Time to own up, I made a silly mistake. The circuit is nothing to do with audio, nor does it need negative feedback. It's simply a fade up and fade down of some LEDs. The circuit uses a R/C and Darlington as a driver. I made it a couple of years ago and it's been working fine till it got water damaged. I thought I'd rebuild it maybe with a FET. Drew up a simple circuit in MicroCap simulator and got a strange result. Well what I'd done without thinking was use the FET as low-side switch, (NMOS is always low side switching, right) forgetting in the original I had used a high-side Darlington. This of course gave a result completely different to the way I knew the original circuit worked. I'll include a picture of what I saw, the FET result is in green, original circuit in red. Moral...  Less drinking, more thinking.

Again, why not simply replace it with another Darlington?

Please post a schematic.

[PGPG]

All the FETs I've tried in simulation turn on sharply over a half volt or so on the gate, most are used as switches I guess.

All bipolar transistors turn on even sharper over a 0.1V on the base.
I suppose the vacuum tubes were turning on less sharply.

[Tenderfoot]

Again, why not simply replace it with another Darlington?
Please post a schematic.

Why replace the Darlington? Why not, who knows I might even learn something.

A schematic is hardly worth it. It's like I said, a R/C into a Darlington, now changed to Sziklai driving some LEDs. But ok

[exe]

How much is more linear? You can drive gate via a resistive divider... Though without a feedback (with e.g., opamp) it's prone to temperature drift and device-to-device variation.

[Zero999]

Again, why not simply replace it with another Darlington?
Please post a schematic.

Why replace the Darlington? Why not, who knows I might even learn something.

Because there is no need to. A MOSFET is an inferiour choice in that circuit.

A schematic is hardly worth it. It's like I said, a R/C into a Darlington, now changed to Sziklai driving some LEDs. But ok

The schematic is worth it. No one knew what you were on about before you posted it.

You mentioned using the MOSFET on the low side. This is why it gave a sharp turn on. It had nothing to do with the type of transistor. The difference is in your schematic, you have the transistor configured as a follower, which has a voltage gain of just below 1, which buffers the capacitor voltage to the output, minus the threshold voltage. Moving the transistor to the low side, converts it into a common source/emitter amplifier with a high voltage gain, hence the large suddent change in voltage across the LEDs for a small change in voltage across the capacitor.

You could swap the Sziklai for a low threshold voltage MOSFET and it would give a similar response, but it will still have a greater voltage drop, due to the threshold voltage being slightly higher than the Sziklai pair.

If you really want to learn something, then redesign the whole circuit to use a microcontroller and dim the LEDs with PWM, using a MOSFET.

[Tenderfoot]

If you really want to learn something, then redesign the whole circuit to use a microcontroller and dim the LEDs with PWM, using a MOSFET.

That would teach me nothing, I was programming in assembly decades ago, and using PICs shortly after. Using a microcontroller and code to replace one cap and one resistor is, I think, pointless. You of course may disagree.
Regarding your comment about using a low side transistor, I agree, as you would see if you read back.

Thanks for everyone's comments and suggestions (and I do mean everyone). I'm guessing this post has come to it's natural conclusion at this point?

[David Hess]

All the FETs I've tried in simulation turn on sharply over a half volt or so on the gate, most are used as switches I guess.

All bipolar transistors turn on even sharper over a 0.1V on the base.
I suppose the vacuum tubes were turning on less sharply.

The turn-on characteristic for MOSFETs and bipolar transistors is defined by their transconductance, which is the change in output current for a change in input voltage at the gate or base.  Bipolar transistors have higher transconductance than MOSFETs, so they switch more abruptly.

[MariuszD]

Using just the MOSFET will be difficult, but a two-transistor amplifier will allow you to achieve the desired characteristics like low dropout voltage and better linearity.

My circuit uses PMOS and NPN transistors, and it can be easily converted to NMOS and PNP.

This circuit used to control LED strips has one drawback: current only flows thru the diodes above 18V. This is an area for further improvements.

[mawyatt]

The turn-on characteristic for MOSFETs and bipolar transistors is defined by their transconductance, which is the change in output current for a change in input voltage at the gate or base.  Bipolar transistors have higher transconductance than MOSFETs, so they switch more abruptly.

Theoretically a MOSFET can have a higher transconductance than a bipolar at a specific current since the transconductance is proportional to W/L and a bipolar transconductance is fixed and not dependent on size or any device parameters other than Ic/Vt, where Vt is the thermal voltage kT/q and thus the bipolar device is not dependent on anything but collector current, temperature and 2 physical constants. On the other hand the MOSFET gm is directly size (W/L) dependent and can be scaled for a specific transconductance, of course for a normal size die area the bipolar has the higher gm.

Another interesting aspect of MOSFETs is the Threshold Voltage can be controlled so they conduct well below the bipolar 0.6~0.7 volts and can be created as Depletion or Enhancement mode devices.

Interestingly way back ~2000 when we were working with IBM SiGe bipolar devices a discussion was to create a Depletion Mode Bipolar SiGe Device!! The device physics indicated that a specific Ge profile grading within the SiGe bipolar base could create a sufficient electric field to cause significant collector current to flow without any base connection even at low collector potentials and somewhat independent of collector potential. Altho this idea gained some traction we couldn't get funding to continue and the Depletion Mode Bipolar Transistor was abandoned.

Best

[Zero999]

If you really want to learn something, then redesign the whole circuit to use a microcontroller and dim the LEDs with PWM, using a MOSFET.

That would teach me nothing, I was programming in assembly decades ago, and using PICs shortly after.

Are you sure? Have you done it recently? Things have moved on a bit since then. How about programming in C?

Using a microcontroller and code to replace one cap and one resistor is, I think, pointless. You of course may disagree.

That wasn't the idea. There are other reasons why it might be a good idea. The main one is PWM means the transistor can run cooler so a smaller device can be used i.e. TO-92, rather than TO-220.

Regarding your comment about using a low side transistor, I agree, as you would see if you read back.

Thanks for everyone's comments and suggestions (and I do mean everyone). I'm guessing this post has come to it's natural conclusion at this point?

For future reference, please post the schematic in the first place. A lot of time has been wasted guessing.