Mosfet Heating Circuits

[FuzzyTomCat]

Mosfet Heater Circuit Credit

Hi everyone,

I was looking on this forum for the credit due to Aaron on the success of the circuit because he was the person directly responsible for 555 timing circuit that is being used today that actually workes. I think somewhere Aaron even stated in the forum part of 555 circuit he got from a old "Radio Shack" manual and refined it.

I was able to take Aaron's circuit by changing the 555 timer "on" and "off" and mosfet "gate" potentiometers resistance values and construction type for easier adjustments to refine the circuit more.

The 10 ohm "load" resistor or inductor made from borosilicate glass was 100% "exclusively" my idea of style, type of materials and construction.

I have in no way so far as I know claimed any or complete intellectual property rights on this device or circuit in this Open Source project and hope to continue it's development as others are.

Best Regards,
Glen


Progression of Mosfet Heating Circuit -
Older - Schematics - Windows Live

[Aaron]

post link

I just noticed that my link to my suggested modifications post was not going to the right place

So that is fixed now, and here it is again:
Harvey's Suggested Modification

I don't know how that happened, I must have clicked on that link 5 or 6 times after posting it and worked each time, but this time it went to some other post.

Aaron, has the database been compromised? Can you compare the locked thread to the backup?

Harvey,

When I click that link, it takes me to the top of a page with multiple
posts on page 64.

It doesn't take me to a specific post. Was that from a permalink or
the actual link #?

[Harvey]

Hi Guys,

Also for those interested, this portion of the circuit, namely the inductor and the MOSFET configuration were provided by International Rectifier to public domain in 1993 for this specific HEXFET used in Glen's tests.

Poynt99 at OU reports that he has this same circuit in print as far back as 1985:

My version is even older, and is before the IRFPG50 was born:

Power MOSFET HEXFET Databook, HDB-3, 1985. It does not seem to have the unclamped diagram in any of the spec sheets, but it does reference avalanche testing using essentially the same circuit as Fig. 12a. in the "Integral Body Diode" section, App note 934B, Fig. 23.

.99

Enlarge Image

IRFPG50 DATASHEET PDF

Cheers

[Harvey]

Harvey,

When I click that link, it takes me to the top of a page with multiple
posts on page 64.

It doesn't take me to a specific post. Was that from a permalink or
the actual link #?

Hi Aaron, I used the Permalink button on post #1908, my post, but somehow it was linking to post #1904 instead. I have fixed the link since then, but it is strange that it shifted 4 posts somehow

It seems to be holding now though

[gadh]

Using Harvey's modified circuit + LEDS - results

Hi all.
1. I used Harvey's modified circuit (in short: 100k/1k/zero wire/etc…) and indeed its a lot better: I see the waveforms are more stabilized, less current draw in the 555 circuit (50mah), smaller duty cycle can be configured, and more range to test more frequencies. Thanks a lot Harvey ! Some modification I suggest from what I tested:

a. Upper 50Kohm pot.: responsible only on duty cycle, no need to be more than 10kohm (my actual value was only 2kohms when reaching the needed 3.7% duty cycle).

b. Lower 50Kohm (slightly lower…) – responsible only on frequency – when using only 50kohm (1 pot.) I only got to 3.3khz, so I added 1 more 50kohm pot and got to the needed 2.4khz freq (in about 52kohm overall resistance) so I recommend to use 2x50kohm pot. (each has 10 turns which is important to get the needed resolution).

2. Test results: COP=0.62, over 3 hours time. Exactly as I got in the former tests. I think this means my calculations are ok. Still I want to check if I'm right:

a. Ground connection of the shunt probe: when I connect it right after the shunt (between the shunt and the battery '-' pole) I do not get much negative voltage, and the range is about +-200mv. When I connect it to the 555 ground line (which of course is also connected to the battery '-' after the shunt) – I get almost equal negative voltage to positive, almost a mirror image on the X axis (I have problems uploading the waveforms from my cellphone, I'll try in the next posts) and the range is +-5volts. Which is the right connection ?

b. Measuring the energy consumed by the battery by charging with my digital charger: my calc. method Is as follows: (mah charged) x (average battery voltage during the tests) = overall watt hours used x 3600 = overall energy used in joules. Am I right ?

3. LED usage – I added Harvey's LED circuit, and used 2 red 3mm LEDs (since they draw the same power, unlike green led vs. red one).

a. When using the 0.1R – no LED lighted at all. When removing the 0.1R (cut) - only the red one lighted. Maybe the 0.1R is not needed ?

b. The 'green' one never lighted in all waveforms I could get. Maybe the correct location of it might be between the load resistor and the battery '+' pole – to detect the reverse current ?

c. Question: across the red led (when lighted) I measured 44mv and 150ma, while the min. voltage needed is about 1.6V in a regular dc circuit. How is this possible ? and add the fact that this LED uses only 2ma max. !

[Harvey]

Hi Gad,

Thank you very much for your continued work on this

Regarding your questions for:
2.a) The correct point to place your probe reference is at the CSR leg. This should also be the central tie point for all references including the timer ground. The wire between the CSR and battery along with its inherent inductance and capacitance then simply becomes an extension of the battery itself and is excluded from the measurements. (See posts http://www.energeticforum.com/induct...html#post84686, http://www.energeticforum.com/induct...tml#post109762 & http://www.energeticforum.com/induct...tml#post109784) Please note: The Red and Green Diodes are only a tool to help dial in the preferred mode of oscillation and should be removed prior to actual measurements as they are inserted between the CSR and Timer ground connection which should be tied to the same exact point during measurements.

2.b) I have expressed my concerns regarding this method of determining battery power used. Any time you separate your voltage and current in time, you have skewed your actual energy calculations. You must only use instantaneous voltage and current together to determine the power involved. Then you may average that power over time to determine the average Joules involved. Using the average voltage from one part of the test with the average current from another part of the test is not, IMHO, a sound way to do things. So your electronic recharger does not tell you how much energy you used to recharge the batteries?

3.a) The 0.1R is based on the maximums shown in Glen's data. Your resistor and circuit operation may not be reaching that 12A spike we see recorded in his data. Try increasing the resistance of 0.1R until the 'Green' LED lights. At some higher resistance, the voltage drop across this shunt will be enough to light the LED. Remember, this is only a tool to help you find the preferred mode of oscillation settings for your resistor. The addition of this tool changes the way the circuit works, and after it is removed, further fine tuning will be required - but at least you will have a visual of the 'Drain' waveforms you are seeking so you can then reproduce them with the tool removed.

3.b) No, the correct location is as specified, between the CSR and battery (-)

3.c) 1. The fact that you are getting light with the very small averages helps to impress the importance of instantaneous events. Some of the voltage surges are high enough to produce light.

3.c) 2. The rating of the LED is typically in mW. This represents how much sustained power the device can dissipate. Therefore, as long as the average power is below the safe ratings and the peak current is below the maximum peak ratings the device will not fail. In other words, it can handle more current for short durations. That is the principle used in many PWM LED brightness controllers.

Evidently, the needed negative portion of the oscillation is not present yet in your settings, or the instantaneous voltage drop of the negative portion is lower than the required 1.6v in your 'Green' LED.

Also, I agree with you completely regarding the use of two identical LED's in this case

I've been researching a new method that may be much more accurate using a laser galvanometer. I've read that the voice coil on a hard drive can be fitted with a mirror and act as a bidirectional indicator. CandlePowerForums - View Single Post - Making a laser show scanner... just thinking ahead here to automating the PMOO adjustment process programmatically.

[gadh]

2.b) I have expressed my concerns regarding this method of determining battery power used. Any time you separate your voltage and current in time, you have skewed your actual energy calculations. You must only use instantaneous voltage and current together to determine the power involved. Then you may average that power over time to determine the average Joules involved. Using the average voltage from one part of the test with the average current from another part of the test is not, IMHO, a sound way to do things. So your electronic recharger does not tell you how much energy you used to recharge the batteries?

Hi Harvey and thanks to you for your continuous support on this too .
i had troubles before measuring the total power/energy using your suggested method and IIRC you also agreed that measuring the total energy is more reliable since interpreting the scope data might be misleading, as i also consulted with my pro colleagues on this.
My digital charger does tell me how much MAH i used to recharge the battery, so i ask you if my calc. based on this data is right ?
Also, i found that in short tests, when i do not use more than 50% of the my battery, the charging energy might be highly more than the actual used energy (discharged), so i'm looking for a way to know how to measure this right. if you have data on these PB charging energy calculations and their accuracy - please let me know.

Best Regards,
Gad

[Harvey]

mAh vs Watts vs Joules

Hi Gad,

One Watt applied for one second equals one Joule.

If your recharger has a very accurate output voltage that does not fluctuate during recharging, then you can use this value in your calculations.

Recharge Voltage x Recharge Current = Recharge Power

Recharge Power x Time = Energy.

Where Power is in Watts and Time is in Seconds

Example:

Recharge Voltage = 13.8 V
Recharge Current = 800 milliamps
Time = 2 Hours = 7200 seconds

Recharge Power = 13.8 V x 0.8 A = 11.04 Watts
Energy = 11.04 x 7200 = 79,488 Joules or 79.5 kJ rounded to a single decimal place.

Notice here that I used the recharge voltage, NOT the average test voltage - this is important as it keeps the power calculation in the right time frame.

[gadh]

test #10 - report

Hi all.

test #10 results and questions:
1. COP = 0.54, again failure to see gain.

2. Green LED lighted ! i found that my led was broken so i replaced them both to 2x5mm leds, real green and red this time, and they both lighted when the parallel resistor (i use 500R pot.) was > 1 ohm (the more resistance - more light).

3. waveforms - better this time (see attached images). although the freq. was only 90khz. i'de better check the COP when using a higher freq, but i did not find such in this test.

4. the test lasted this time about 5 hours. but i was limited by the 555 battery (9v 200mah) capacity - i used 180mah in 5h = 36ma (10ma when no load connected). its a big improvment in the 555 circuit power consumption (Thanks to Harvey) but its still limited (ses next paragraph - related)

4. when i connected the 555 circuit to the 12v battery instead of my regular 9v one (as Harvey showed in previous posts), i could not reach any resonance in all pot. range ! the freq. remained always on the original 2.4khz.
is there any way to change that ?

Gad

[Harvey]

Hi all.

test #10 results and questions:
1. COP = 0.54, again failure to see gain.

2. Green LED lighted ! i found that my led was broken so i replaced them both to 2x5mm leds, real green and red this time, and they both lighted when the parallel resistor (i use 500R pot.) was > 1 ohm (the more resistance - more light).

3. waveforms - better this time (see attached images). although the freq. was only 90khz. i'de better check the COP when using a higher freq, but i did not find such in this test.

4. the test lasted this time about 5 hours. but i was limited by the 555 battery (9v 200mah) capacity - i used 180mah in 5h = 36ma (10ma when no load connected). its a big improvment in the 555 circuit power consumption (Thanks to Harvey) but its still limited (ses next paragraph - related)

4. when i connected the 555 circuit to the 12v battery instead of my regular 9v one (as Harvey showed in previous posts), i could not reach any resonance in all pot. range ! the freq. remained always on the original 2.4khz.
is there any way to change that ?

Gad

Hi Gad,

1.

2. You will want to adjust the 500R you have there, to the minimum value that still allows the Green LED to light dimly. Then you will want to adjust your Gate pot to get the 'preferred mode of Oscillation' for the brightest green and dimmest Red possible. From looking at Glen's 5 hour videos, this is not a stable setting, it 'dolphins' up and down over time and you have to wait a bit for it cycle. But your goal is more Green which corresponds to more negative current. For those that have digital scopes, or analog scopes with math functions, this corresponds to the 'mean' value on the CSR being as close to zero or negative as possible.

3.

4. a

4. b It sounds as though your connection between the first adjustment pot and the 555 power connection could be disconnected or misplaced. Another thing that occurs with a 12V supply is that the IRFPG50 now has a much better gate voltage and so the switching is typically crisper. This is also where the timer gets a more stable retriggering. Glen noticed that things improved after he charged up the timer battery, but he also noticed that the power batteries had a residual charge after recharging that needed to be 'burned off' by about 15 or 20 minutes of operation before he could get the preferred mode of oscillation dialed in. Up until that point, the circuit would 'float' around in aperiodic mode somewhat randomly and the 'mean' value on Channel 2 would not come down where he liked to see it. The entire system seems to be quite interdependent and small changes may take minutes to take full effect.

With the exception of the problem stated in 4b, it sounds like you are on the right track with the other things. Part of the low 'COP' value could be related to excessive losses in the MOSFET. The 'On' resistance of this device is around 2 Ohms. So it will heat up. This is especially true if we are keeping the resistor 'on' too long. It only needs to be on long enough for the inductor to charge - that's it. So as soon as the charge current starts to flatten out, there is little value in keeping the MOSFET on any longer in that cycle. In other words, the charge current will begin to increase on a curve after the MOSFET is turned on. When that curve begins to level out, you are draining the batteries for very little advantage. It is different for every inductor, but for those we are working with, it is typically less than 15µs on time. And on the flip side of this, you do not want the charge time to be cut short. If you turn the MOSFET off too early (which is exactly what happens sometimes during the aperiodic retriggering) the inductor does not have sufficient time to fully charge and this gives you a wimpy BEMF amplitude. The sweet spot is when you get almost a full charge in the inductor (~90+%) And turn off the MOSFET with a clean sharp pulse. This gives a good amplitude BEMF on the Drain pin and that is what drives the negative current, (Green LED) and produces the greatest impact on the load resistor heating.

[gadh]

Thanks Harvey for your suggestions.

about the waveforms (4b) - could you tell me exactly what was not so good according to my attached images ? the upper section reprsents the load probe (500-600v) , which i think goes up and down "cleanly" and exactly what you've said. doesn't it ?

[Harvey]

Hi Gad,

The upper trace is your Drain pin on the IRFPG50 correct?

You can see in that trace when the MOSFET is turned on. The supply line at the Drain pin is at battery voltage just prior to the turn on sequence, and then it drops to near ground during the on time. So the supply is about 1/5th of a division. If this is 24V, then your BEMF spike is about 14/5th's of a Division, or about 336V. This would mean that the scope was set somewhere around 120V / Div. Your knob shows 2V/Div and the uncalibrated lamp is off, so it would seem that you have a separate attenuation (about 100x) between the circuit and the scope. If this were truly 200V/Div, then it would be showing your supply line to be nearly 40V. Of course at the settings shown with that intensity, your trace width alone is almost 12V thick so it's not real precise.

What I find curious, is all the ringing that is going on in the lower trace just as the MOSFET turns on. Both of your probes are tied to the exact same physical reference point correct? If the lower trace is across the CSR, then it represents the instantaneous current flowing through the MOSFET. It seems unlikely that this current would be ringing like that unless the MOSFET was ringing like that as well and if it is, it should be evident in the top trace. It may be there, but the trace resolution is such that we cannot see it clearly. Either way, it is not a clean turn on and it could mean that the gate itself is ringing like this.

That short upward slope in after the ringing is your inductive-resistor charging, and it looks as if we are turning it off too soon. See if you can lengthen the on-time a bit more to get that to a point that it is a little wider and more horizontal prior to the turn off point.

The MOSFET seems to be turning off rather cleanly and that is good. All of the negative ringing that is going on after the turn off completes is the BEMF collapse taking over and driving the circuit. Any positive current seen at that point would be capacitive pass through in the MOSFET, while the negative current is passing through the Body Diode. It is possible that we don't really have positive current there and what is really happening is that we have driven the Reference point down instead, thus making it appear that a positive current is flowing. This can happen when the battery inductance sloshes back and forth after the MOSFET cuts off the current flow. In other words, zero volts is not staying at zero

So, widen your on-time a little bit and see if that increases the BEMF spike.
And, check your gate signal to see if it is ringing when it goes high - if it is then we may have to find out why and try to correct it. That may be adversely affecting the retriggering.

[gadh]

Hi Gad,

The upper trace is your Drain pin on the IRFPG50 correct?

You can see in that trace when the MOSFET is turned on. The supply line at the Drain pin is at battery voltage just prior to the turn on sequence, and then it drops to near ground during the on time. So the supply is about 1/5th of a division. If this is 24V, then your BEMF spike is about 14/5th's of a Division, or about 336V. This would mean that the scope was set somewhere around 120V / Div. Your knob shows 2V/Div and the uncalibrated lamp is off, so it would seem that you have a separate attenuation (about 100x) between the circuit and the scope. If this were truly 200V/Div, then it would be showing your supply line to be nearly 40V. Of course at the settings shown with that intensity, your trace width alone is almost 12V thick so it's not real precise.

I indeed use an x100 probe, and its calibrated and accurate. my 2 probes are in the exact points where theu should be: 1st - on the drain leg of the MOSFET (GND - to 555 gnd), 2nd - at the shunt (GND-right after the shunt)

I'll check what you've suggested. Assuming by "on time" you meant duty cycle, and i'll send images as before and also of the gate.

Thanks alot
Gad

[gadh]

Hi Harvey.

1. I attached 2 series of images:

a. 600v - shows the regular waveforms , "not clean" - with the small wave after the larger one + gate pot. waveforms as well.
b. 400v - this is the only mode i could get a "clean" waveforms at the load probe. However, you can see that the avgerage voltage on the shunt is clearly positive so this mode is useless for testing. how can i get also an avg. negative voltage ?
Also, this "clean" mode was very hard to maintain. the gate pot. was not enough sensitive so i could not hold this mode for more than a few seconds even when i connected a 10ohms resistor in parallel to it, in order to increase its sensitivity.

2. Q: when i connect the LEDs circuit, where should i connect the GND ?(the cable that connects the 555 GND to the battery '-' pole) - after the LEDs (CSR-LEDS-GND-battery) or before them (CSR-GND-LEDS-battery) ?
please see the images i attached: when using "after" (i think thats what your diagram showed) - i get different waveforms than the ususal, and only when using "before" - i get similar waveforms to the known ones. of course when i get similar i can rely on these settings when i disconnect the LEDs as you suggested, but when i get diff. waveforms i could not rely on them.

3. Test #11 - COP 0.66 - better than before but still no gain. this time i tweaked a lot the various pot. and i think i got the best "green" led luminosity so far, relative to the "red" one. i think the green was almost lighted as the red or equal - so how this did not affect the COP to be > 1 ?

Thanks
Gad

[Harvey]

Hi Gad,

Good work on the Green LED


Your third image is closer to what we are looking for. The current (lower trace) tends to level off and right then we have that steep vertical drop indicative of a clean switching of the MOSFET to "OFF". At this point all of the energy stored in the magnetic field of the inductor becomes BEMF on the drain pin. The faster that turns off, the more voltage you get at the BEMF pin.

In the first image, you can see that we do not get that clean steep turn off, and this results in some of the field energy being allowed to flow through the MOSFET in a linear fashion where it becomes lost. Observable in the amplitude being reduced in the BEMF spike.

We can see in image 2 that the inductor does not have sufficient time to fully charge, and therefore the magnetic field does not contain as much energy as it could and this also leads to a reduced BEMF spike.

It would be helpful if you could include your time base settings for the shots also so we can see what the frequencies are that are involved.

The probe should be connected directly across the CSR and both grounds for the probes should be connected to the same exact point. This is absolutely necessary to avoid allowing current to flow through the scope ground wires. We know that the LED's will elevate the reference by their Vf, but this in not important to the CSR accuracy and the 1.6V differential in the top trace is unreadable as the trace itself is ~7.5 times thicker than that differential. Therefore, even with the LED's in place, the current reading is still very accurate and the Drain reading margin of error exceeds the elevated value.

However, it should be noted that the inclusion of the LED tool adds resistance to the path and therefore limits the maximum current that can flow in the Inductive Resistor. This is why it is only a tool to be used for dialing in the settings and it should be bypassed (with a short thick wire) or removed during the caloric tests. The waveform timing should not change much when the tool is bypassed. But the waveform amplitudes should increase in both traces.

==================================================

I have a test for you that may need to be done to boost our confidence in your electronic method of battery energy calculation.

1. Place an automotive headlamp* on your 12V battery with a current meter in series with it.
2. Use your scope or another meter to measure the voltage across the lamp.
3. Take readings every 6 minutes of voltage and current
4. After 30 minutes stop the test and calculate the energy dissipated using the readings taken.
5. Install your charger to replace the energy in the battery and calculate that value after it is fully replaced.

How close does the replacement value match the dissipated value?
If they are nearly identical, then we can have good confidence in this method. If they are not, then we need to add that margin of error to the 'COP' calculations.

BTW, how long are the wires between your batteries and the load resistor and what gauge are they?

Cheers,

Harvey



* I suggest this because it is a high current device and will accelerate the test.