Blog Post Actions
Group Actions
Engagement
  • Author Author: jc2048
  • Date Created: Date Created
  • Views 1291 views
  • Likes 13 likes
  • Comments 12 comments
Related
Recommended

Experimenting with MOSFETS: Threshold Voltage

jc2048

 

Previous blogs:

 

Experimenting with MOSFETS: Total Gate Charge

Experimenting with MOSFETs: Transfer Characteristic

 

Introduction

 

More experimenting with MOSFETs. In the last blog I looked at the transfer characteristics but it wasn't very easy to see from the graph I plotted where the threshold was. The threshold is measured at a drain current of 250uA and that is down in the noise for the current probe I was using. So this is a different approach to finding the threshold voltage.

 

Circuit

 

Here's the circuit I ended up with.

 

 

The current through the MOSFET is turned into a voltage by the source resistor. That gets compared to a voltage derived from a voltage reference and the op amp drives the gate in such a way as to make them match. Once I've adjusted it for 250uA flowing in the drain, the gate voltage (relative to the source) is the Vgs threshold voltage. [In case it's not obvious, I'm copying the constant-current load circuit that has been the subject of several blogs on the site here.]

 

Here's the circuit built on a small piece of pcb material.

 

 

The op amp I used was an OP177. It's a precision op amp and maybe a bit 'over the top' for a simple experiment like this but it was close to hand. Although all the specs are for +-15V operation, it works fine for this on a single supply of 12V. The 47nF decoupling capacitor is underneath the package in case you're wondering why you can't see it. The gate resistor was simply being used to span the gap between the op amp output and the gate's pad on the board - it's not necessary for the operation. I used a 15 ohm resistor, but only because I had plenty (it's an odd value that I don't seem to design with much). The voltage reference I used was an LT1009. It's far too good for this but it was lying around on the bench so it got thrown into the pot. I didn't bother with any additional compensation for the op amp.

 

First Test

 

I set the current by adjusting the preset potentiometer whilst  looking at the drain current with a bench multimeter. Once I'd set it to 250uA, I got these readings for an IRL730 part - a threshold voltage of a fraction over 3V.

 


That's right in the middle of the range the manufacturer allows themselves in the datasheet

 

 

One thing I noticed a bit too late on in this is that their spec is measured for the case where the drain voltage is the same as the gate voltage. How much difference that makes I don't know. It might be simply to keep the dissipation down as the threshold voltage is temperature dependent.

 

More Tests

 

Having got this working, I thought I'd have a go at measuring a group of parts and see what the spread on them was. I don't have many IRL730 devices, but I did find an old tube with 17 STP55NF06LSTP55NF06L parts in it, so I tried those. The datasheet I've got gives a minimum of 1V and a typical of 1.7V (they are billed as being 'logic level' parts); it doesn't give a maximum figure.

 

This chart shows the Vgs for each of the 17 parts sorted by ascending voltage. Ten of the parts cluster around 1.8, the rest have values up towards 2V. Not quite as good as their typical value suggests but the spread is only 0.2V. Doesn't look like there's a very high chance of getting a 1V part.

 

 

 

 

 

If you found this interesting and would like to see more blogs I've written, a list can be found here: jc2048 Blog Index

  • Finally, this is a BSS138 device in an SOT23 package.

     

     

    Here they are again reduced to a dV per degree C value:

     

    BSS138 -2.05mV/C
    BS170 -3.25mV/C
    STP55NF06L -6.06mV/C
    IRF730 -6.67mV/C

  • Here are two more. The STP55NF06L is a 'logic level' power device in a TO220 package, the BS170 is a 'small signal' device in a TO92 package. By coincidence, they had similar threshold voltages, so I put them both on the same graph so we could see how they compare. As you can see, the power device varies much more than the small-signal part.

     

  • in reply to jc2048

      wrote:

     

    ... I'm having fun with these blogs. ...

    +1

  • When I saw in datasheets that the threshold voltage was normally measured with the drain voltage the same as the gate, I was curious about how much effect varying the Vds voltage would have on it. Simple way to find out was to try it. Here's a graph of threshold voltage versus Vds for an IRF730 part.

     

     

    As you can see, the threshold voltage dropped as the voltage between the drain and the source increased, but the effect was very small: it only fell by just over 3mV as the Vds went from 1V to 10V.

     

    I did that test quite quickly, after allowing the whole circuit to warm up for several minutes, and stepped down the Vds values as well as up with a comparison between the resulting Vgs(th) values to make sure I wasn't just measuring the effect of the temperature rise from the increased dissipation at the 10V end. But I wasn't very confident in the result - I really needed a better idea of the way the threshold voltage changed with temperature.

     

    The next day I did a simple test where I measured the Vgs(th) for Vds=10V as the whole circuit warmed up. Here's the curve I got over a period of 15 minutes. The MOSFET dissipation is 2.5mW (10V x 250uA), so not too much in the way of self heating. That gave a change of 6mV before it settles.

     

     

     

    Something worried me about that curve. It seemed to go in the wrong direction - the curves in the datasheet imply that the Vgs(th) temperature coefficient is negative - so what was I looking at? It's more likely that it's the way the constant-current circuit behaves - the meter I was using to show the drain current could only show it to 3 digits.

     

    That wasn't very helpful so, finally, I've rigged up an ad-hoc temperature test. It isn't very sophisticated - the TO220 package is sitting on the top surface of a ceramic power resistor and I'm measuring the tab temperature with a thermocouple - but it's good enough to give me an indication of how the threshold varies with temperature. This one was done with Vds being the same as Vgs.

     

     

     

     

     

     

     

    To return to the first plot, the 3mV change I saw there is nothing compared to how the threshold voltage moves with temperature and it's highly likely that some proportion of it is down to self-heating anyway. In a real circuit, if the threshold voltage was important to you in some way [it's not really if the device is used as a switch or inside a feedback loop], you are going to be battling the change with temperature and the change with Vds looks to be quite minor.

     

    I'm having fun with these blogs. I know all the information should be available in books or on the web, but it is interesting doing the experiments for myself.

  • Hi Jon,

     

    I decided to make a little more permanent test jig. I ended up using a TLE2142 OP amp since I didn't have an OP177. I also adjusted the 4k7 and the 2k7 in series with the pot so that i could use a 1K pot. I have found that the 250 uA remains quite stable so a small 10 turn trimmer was used. I originally thought that the pot would need adjusting with each different MODFET but of course this just depends on the stability of the series 3k9 Source resistor. Any how it has been a lot of fun and I still have quite a few MOSFETs to go.

     

      

    CLICK IN THE BOX TO SEE THE PICTURE

     

    I use alligator clips with wires and pins to attach to the board.

     

    John