Saturday, October 21, 2017

Those crazy MOSFET designs I throw out - how accurate can they be?

I was looking at a design program and wondering if it would be useful. Here is the simple explanation of what it would do for me. First the data entry screen and what I have to tell it. Notice it doesn't even ask for a transistor type.
I entered all the data and 'designed' the amp.The only transistor specific data they ask for  is Vth and gfs which are datasheet items.

Notice the data sheet gives a range. 0.8 to 2.8 volts is a rather large range for such a critical parameter to swing. So they ask for the 'worst case' or highest value. Anyhoo, you enter that little data and click continue and it 'designs' the circuit as shown on the right. So what are we doing here?
Examine the graph and notice the lines go flat as Vds increases. When I first 'discovered' the MOSFET (in the mid 70's?) this was the first characteristic that caught my attention. Being in instruments and controls I was always in need of a fixed current with varying loads. If you take your MOSFET and tie the gate to the drain, place them in series with an ammeter and power supply and adjust for 1ma current the voltage you have set the power supply for will be Vth. In the early days of the JFET they would make a few 100 thousand or million and then test the Idss. One batch might have 3 or 4 ranges of Idss and they would be different components. Example J111, J112, J113 or J174, J175, J176, J177. The sorted components could be expensive because of the testing or you could get untested parts and sort them yourself. If  you looked at my earlier post on MOSFET amps you may have noticed I like self biased circuits with feedback. It compensates for some component variation. Anyhoo, let's look at some biasing and driving test.
I set Vcc at 3 volts and the bias is 1.4VDC with a 1 millivolt signal on it. 1000uv in and 700uv out is not much of and amp.
I upped the bias to 1.5volts and now get 3.5mv. Better but still not great.
With 1.6volt bias I get close to 18mv which is not bad. Notice I'm still using 3volts Vcc.
With 1.7volts my signal drops to 200uv. 1000uv in and 200uv out is not good. So it can be seen that the Vth is a critical value and the datasheet states a 2 volt range is acceptable? Notice I'm still using 3volts for Vcc. My best performer was with 1.6volt bias. Let's test the circuit with different Vcc.
Vcc 1.5volt just won't work for me.
Been here before 3volt Vcc is the best so far. What effect will raising Vcc have?
Do you see any difference with 6 volts?
How about 12volts?
The MOSFET is rated 100volts so we try that. Do you see any difference? Remember the constant current? This circuit has the same current as the one with 3volts. The swing in gate voltage causes the same current variation and the drain resistor drops the same signal. The only difference is drain voltage and power consumption. The current is about 2ma in all the circuits.


No comments:

Post a Comment