Where did you personally use or see LM317 applied to?

Thread Starter

q12x

Joined Sep 25, 2015
1,734
- This 2N3442 is a new tr that you just received ? I remember you buy some power tr in the same time I buy my 2N3773's.
I think you wanted to parallel my experiment (which was a good idea). Or is the old tr's and you cut another one?
- Its good if you have IRFZ44N mosfet, because now we have the same tr and we can parallel eachother !
 

dl324

Joined Mar 30, 2015
16,996
This 2N3442 is a new tr that you just received ?
No. I got it from HP in the late 1970's, so there's little doubt that it's genuine.
Its good if you have IRFZ44N mosfet, because now we have the same tr and we can parallel eachother !
I ended up spending all of my time testing the 60 MOSFETs. Only one was dead obviously bad. Now I need to take some measurements and check for excessive gate leakage current (caused by ESD).

I have the MOSFET version of the current sink breadboarded. I was just testing at very low power levels because I was using an AO3401.
 

Thread Starter

q12x

Joined Sep 25, 2015
1,734
the 60 MOSFETs.
What is a "60 MOSFET"? is the year? the marking on them? the total number of mosfets you have?
I have the MOSFET version of the current sink breadboarded.
give me your test circuit and tell me what you looked for and what was your thinking, step by step if possible. Obviously you are more experienced with mosfets than me. Although I have quite a number of them, I rarely use them.
I was just testing at very low power levels because I was using an AO3401.
AO3401 is a SMD 30V P-Ch MOSFET ! SOT23 package - what are you doing there?
1713289203184.png
 
Last edited:

dl324

Joined Mar 30, 2015
16,996
give me your test circuit and tell me what you looked for.
I already posted a single comparator PWM circuit in this thread. I adjust the duty cycle so the MOSFET is on for only a short time (<10%). For a P channel MOSFET, that means a duty cycle of 90% or higher. For an N channel MOSFET, the duty cycle is <10%.

I choose a suitable load resistor to get a current around 5A, but I vary the supply voltage and PWM voltages to make sure the transistor is on "hard" enough or completely off.

When I was testing P channel MOSFETS, I set the PWM supply higher than the MOSFET supply to make sure that it was completely turned off. I set the PWM supply according to what I want to test. I chose to use about 10V for the "hard" on test and use a lower voltage when I'm interested in Vgs(th).

I also use that circuit to get an estimate of Rds(on). For the P channel testing, I put a scope probe on the drain. The difference between that voltage and the MOSFET supply voltage is the voltage dropped by the drain-source resistance. Using the current in the resistor, I can calculate Rds(on). Since the duty cycle I'm using isn't necessarily low enough to prevent the junction temperature from rising, it's only approximate. To get a more accurate value, I used a 555 timer to get a short on time with an off time sufficient to prevent junction warming from affecting the reading.
EDIT: I was testing IRF9520 and IRF9540 that have relatively high ON resistances (0.48 and 0.2 ohms, respectively). That method won't work with IRFZ44 because the ON resistance is much lower. I'll have to measure the voltage drop across the MOSFET using a more sensitive voltage range on the scope.

Testing N channel MOSFETs will require a modification to the 555 circuit. It's easy to get a large duty cycle. Getting a small duty cycle requires using diodes.
 
Last edited:

Thread Starter

q12x

Joined Sep 25, 2015
1,734
I ended up spending all of my time testing the 60 MOSFETs. Only one was dead obviously bad.
If you only needed to test if a mosfet is dead or alive, you could use a simple tr tester, like my GM328A
1713330880688.png
I already posted a single comparator PWM circuit in this thread. I adjust the duty cycle so the MOSFET is on for only a short time (<10%) ..... I used a 555 timer to get a short on time....
You did? Maybe is something older in the posts, but I thought you modify or adapted something else for the opamp cct to work as a PWM. So, I imagine that you did something more simpler than that, and instead of the entire opamp circuit, you replace it with a 555 PWM+duty cycle cct, only connected to the Gate of the mosfet and then a LOAD resistor in it's Drain. I imagine....but not totally sure.
1713331302052.png
I only have some very basic knowledge on the function of a mosfet. Im not that curageous(knowledgeable or certain) like you to venture into more detailed tests. I understand what you are testing right now and how hard it is, but I cant imagine the cct you are using. I imagine only a part of it.
 
The LM317 serves a wide variety of applications including local, on card regulation. This device can also be used to make a programmable output regulator, or by connecting a fixed resistor between the adjustment and output, the LM317 can be used as a precision current regulator.
 

dl324

Joined Mar 30, 2015
16,996
If you only needed to test if a mosfet is dead or alive, you could use a simple tr tester, like my GM328A
I have one of those, but it's only good for DORA testing and identifying polarity if you didn't already know.

I want to test at a current of 5-6A (only have supplies capable of 6A), measure ON resistance, and gate leakage (because AliExpress sellers keep shipping in plastic bags).
You did? Maybe is something older in the posts, but I thought you modify or adapted something else for the opamp cct to work as a PWM. So, I imagine that you did something more simpler than that, and instead of the entire opamp circuit, you replace it with a 555 PWM+duty cycle cct, only connected to the Gate of the mosfet and then a LOAD resistor in it's Drain. I imagine....but not totally sure.
The PWM post is #229. Pull-up time will be slow, but it seems good enough at 1-2kHz.

You got the gist of the circuit. PWM signal to the gate, and a resistor load on the drain.
 

DickCappels

Joined Aug 21, 2008
10,227
At the suggestion of Ron Swanson (one of the great color computer display engineers of the 1990's, at least) I used several as modulated current sources on the dynamic convergence board on an avionics CRT display (back in the 1990's when they still used CRTs). Set the current with a resistor between the adjust pin and the output, then with a resistor from the adjust pin to ground inject a little voltage waveform into the adjust pin.
 

dl324

Joined Mar 30, 2015
16,996
How is the progress with the mosfet driver and wattage reader ?
I got distracted.

I was trying to improve my Rds(on) measurements and found that my FNIRSI Pro scope was crap because I couldn't overdrive the input to get a higher resolution display of the low voltage when the IRFZ44 was on. Then I tried a 20MHz Hantek USB scope. It was crap too.

My Tek7704A/7A26 worked as expected because it's analog so the input signal isn't "manipulated". I decided to try with a 7D20 digital scope plug-in. The one I grabbed had relay errors during selftest, so I had to get another one. I couldn't get a good display at less than the 100mV/div setting. At least I could overdrive the input without it giving me garbage.

With the 7A26, I was able to see the impact of gate capacitance charging when turning the MOSFET on (I have a 4.7k pull-up resistor on the comparator). I couldn't see that well with any of the digital scopes.

Then I got further distracted with mental exercises in trying to determine how the threshold voltages of integrated inverters change with voltage because I was contemplating making a discrete inverter using AO3400/1 to use as a gate driver. I was thinking about how I could use one of those SMD to SIP adapters to make an inverter because it has a footprint for devices on both sides. Would need to cut 2 traces on one side and rewire the S/D. I'd use a 4 pin right angle male header, instead of the leadframes shown above, to get a 4th pin for power/ground.

Lately I've been more interested in re-reading some SciFi novels than doing electronics...

I think I'll put a voltage divider on the drain, and maybe lower the source voltage, so I can get a higher resolution display of the ON voltage.
 

crutschow

Joined Mar 14, 2008
34,682
I think I'll put a voltage divider on the drain, and maybe lower the source voltage, so I can get a higher resolution display of the ON voltage.
You could also add a diode/Zener clipper at the oscilloscope input to reduce the overdrive voltage.
 

Thread Starter

q12x

Joined Sep 25, 2015
1,734
Can you lay down for me the basics of driving a mosfet, as you know it? The simplest driving method. For 1_switching, 2_amplification, 3_power driving. I have my knowledge how to do it, but I want to see your thinking and experience.
Also, In your measurements and tests, what are you looking for, specifically, until now and in the future tests.
 

dl324

Joined Mar 30, 2015
16,996
Can you lay down for me the basics of driving a mosfet, as you know it? The simplest driving method. For 1_switching, 2_amplification, 3_power driving. I have my knowledge how to do it, but I want to see your thinking and experience.
Also, In your measurements and tests, what are you looking for, specifically, until now and in the future tests.
I'll work on this when I have time.

I tested one of the IRFZ44 at around 18W using this heat sink with a 3rd piece added:
1713581600379.png
It was getting warm, but I didn't take the time to measure any temperatures.
 

dl324

Joined Mar 30, 2015
16,996
I tested one of the IRFZ44 as a current sink at 20W (8V at 2.5A) using a CPU heat sink. The temperature got up to 65C after 5 minutes and seemed to have stabilized.
IRF44-CS-20W.jpgIRF44-CS-20W-HS.jpg
I used a 25W 0.2 ohm resistor (lower right in first picture) for current sense. I didn't want to have to fiddle with one of the 8W Caddocks and a heat sink

Did another test at 25W. Ran it for 5 minutes and temperature seemed to have stabilized at 76C.

3rd test at 30W. 96C after 5 minutes.
 
Last edited:

Thread Starter

q12x

Joined Sep 25, 2015
1,734
- I believe you drive the mosfet directly, not pulsed, exactly as we drive the BJT-NPN's. Thats what I can guess from your pictures. This was my thought as well. Although, the correct rule is to drive it pulsed. Very interesting experiment, nevertheless.
Probably a bad idea:
- I thought of a way of getting the wattage from a pulsed mosfet. Theoretically, let's say Im driving a mosfet until is reaching 50*C with a certain size heatsink, right? And I measure the *C until I get exactly 50, for 10min or more, until is stabilizing, and not increasing. The stabilization occurs when the heatsink is totally and completly hot, uniformly ! Thats why it takes some time until it does.
Now, I take out the mosfet, and insert a NPN ... relatively the same wattage as the mosfet ... here im not very sure about the size,wattage,I or V of the NPN compared to the mosfet, but if we dumb it down only to W to be in the same ballpark, I say it will be ok. Ill dumb it down further, only to the package size, because in the very end the disipation surface size of the tr matters to be the same. So these are my rules to compare and find an "equivalent" for this specific test. So, the next step is simple. Run this NPN up to 50*C, write down everything, lets say specifically the Wattage is the only thing that interest us, and then, by analogy, we will say that the mosfet was running at the same wattage as the NPN, based on the temperature dissipation reading !
My test might be wrong, because in reality the NPN might run at 15W@50*C while the mosfet might run at 44W@50*C. I am aware of this, but...it's all I can think of. And I had to spill it out as stupid as it sounds, in the hope it will give you further ideas, inspire you for a better way to test the pulsed mosfet. So, be inspired and come with a better solution !

- On another side, based on your cct that you collected, it appears that this is the correct way, or at least the way "they" did it to test a mosfet. Direct power mosfet driving, and not pulsed! So maybe you are on the right path? Heck, you are an american, you can ask these questions directly from the source, the factory they are making mosfets ! Say to them, to show you the cct for testing comercial mosfets. If it will be similar with this one, we are on the wave. But 50% I believe you are doing the right thing and pulsing is just a way of minimizing heat even further. But classic is direct, no pulsing. Is what I think.
Also, in every datasheet they are specifying the Wattage for direct (and for pulsing) Wattage !
your cct:
1713763301989.png
- If I remember correctly, in every datasheet they are specifying the Wattage for pulsed, AND at a fv of 1kHz. So in theory, you can collect the voltage in 1 pulse, probing that mosfet Source-Drain, (exactly as I did with the NPN on its C-E). Eh? Even -I- can do this ! Hmmm. Interesitng.
- I got another idea! I think this next cct might do the pulse job correctly (and not only for mosfets but for BJT-NPN as well). As I see it, Q4 instead of being linked directly to Vcc, it is linked to this 1kH pulse, and it's role is to dim down the voltage( pulse intensity) thus mosfet will lower its conductivity. Practically, we are lowering the voltage Gate of the mosfet, but in this case is a pulsed voltage. Interesting, right?
1713763301989 copy.jpg
 
Last edited:

dl324

Joined Mar 30, 2015
16,996
I believe you drive the mosfet directly, not pulsed, exactly as we drive the BJT-NPN's. Thats what I can guess from your pictures. This was my thought as well. Although, the correct rule is to drive it pulsed. Very interesting experiment, nevertheless.
You seem to be preoccupied with pulsed vs continuous. If something didn't vary, there'd be no point in having a MOSFET or BJT in the circuit. The current sink circuit drives the MOSFET or BJT "continuously", but the voltage applied to the gate or base depends on the pot setting.

You can use BJTs and MOSFETs in similar manners. One of the main differences between them is that BJTs are controlled with current and MOSFETs are controlled with voltage.

I tend to use BJTs because that's what I first learned. MOSFETs weren't generally available, cost more, and couldn't handle as much power. Now the main obstacle for me is cost. I can't buy a P channel MOSFET in a through hole for a dime, so I've stuck with BJTs.
I thought of a way of getting the wattage from a pulsed mosfet. Theoretically, let's say Im driving a mosfet until is reaching 50*C with a certain size heatsink, right? And I measure the *C until I get exactly 50, for 10min or more, until is stabilizing, and not increasing. The stabilization occurs when the heatsink is totally and completly hot, uniformly ! Thats why it takes some time until it does.
Operating a device to get a fixed temperature is only useful in cases where you're trying to regulate temperature. Usually, we're more interested in a circuit that does what we want. If power dissipation raises heat to an unacceptable level, we can use a bigger heat sink instead of using a device that can dissipate more power or tolerate more heat.
On another side, based on your cct that you collected, it appears that this is the correct way, or at least the way "they" did it to test a mosfet. Direct power mosfet driving, and not pulsed!
Some parameters are measured under pulsed conditions. That doesn't mean that the device needs to be operated that way.

If you want to vary the on time for a device, you can use a PWM circuit.

I used a PWM for checking current capability in the MOSFETs because I didn't want to have to bother with heat sinking. On resistance is measured with a pulse circuit with a duty cycle and frequency that minimizes junction heating because it affects on resistance.
 
Last edited:
Top