Global Free Energy Blog

The following is in reference to the High voltage experiment.

Not sure how accurate this is, but here’s the calculation for the current that 1300 volts would produce through the plastic –>

Voltage source: 1300 volts

Total thickness of plastic: ~ 3e-3 m  (3 mm)

Area of plastic: 5e-2 * 5e-2  (5 cm x 5 cm) = 2.5e-3 m^2

Plastic resistivity: ~ 1e+20 ohm-m

Total resistance = 1e+20 ohm-m * 3e-3 m / 2.5e-3 m^2 = 1.2e+20 ohms

Current: 1300 V / 1.2e+20 ohms = 1.1e-17 amps

So according to that, the current should be no higher than 1.1e-17 amps, yet for ~ a month it was a few nano amps. That means for ~ a month the current was ~ 200 million times higher.

It’s been running for several months now, and just an hour ago it was 30 pA, which is 3 million times higher. We’ll have to see if it behaves like diodes & piezos, which is stabilize at ~ 10 pA.

There might be some slight breakdown effect here in the plastic, but I’ve never seen any equations for that. The best way is probably to just measure the DC current leakage from the 1300 volt supply.

Below is a proposed high voltage experiment.

It appears there might be an unknown source of energy that flows between opposite electric charges, along the electric lines if you will. Read Unknown Energy on some correlations.

Within the diode & piezo exists an intense electric field, but the problem with these components is the resistance is extremely high, in the giga ohms. There are low resistance diodes, but the internal electric field at the junction is low in such diodes.

The Design

One obvious solution is to create an electric field by polarizing two metal plates, and use materials with low resistance, such as Graphite to produce high currents –>

The high voltage goes on the metal foil (purple), where the negative polarity goes on one metal foil, and the positive on the other metal foil. The metal foil should be completely surrounded by plastic or any non-conductive high voltage material. The above photo does not show the metal foil completely encased so as to show it. You could place a coat of silicon paste to completely seal the metal foil. These metal foil can be Aluminum foil. Inner metal foil is across the graphite plate, where two output wires come out. All metal foil can be Aluminum foil.

Experimental Results

Months ago I’ve been secretly running the above experiment, but a very cheap version due to lack of money. Hopefully someone will have more money to do a better version.

In my version, the voltage was only 1300 volts. I used a bunch of diodes and capacitors to build a voltage ladder, and ran 120 VAC (~ 170 Vpp) across the ladder. I did not have enough graphite required for this experiment for the inner solid graphite plate, so I used distilled water instead. The obvious problem with water is dealing with electrochemical reactions. Therefore, instead of metal foil, I bought a bunch of inexpensive small graphite rods from lead pencils and formed a thin solid graphite wall. The electrochemical reactions between distilled water and graphite is extremely low, and should not interfere with the results, but since there will be some residue on the parts from finger prints, etc. you will need to let everything sit & settle down for about one week. During this week most of the electrochemical reactions between the residue on the rods and water should be extinguished, or at least that’s how long it took in my experiment. So, during this week there should NO voltage on the outer metal foil. When the voltage across the output drops to lower than what you can measure, then you can apply the high voltage on the outer metal foil.

So, after waiting ~ a week, the output voltage was extremely low, and the 1300 volts was applied to the outer metal foil. The DC voltage did not change immediately, but slowly over time in began to gradually increase. It went to ~ 30 mV, and from there it began to slowly decline. That is, slowly, as in months. This peak, followed by a slow decline has been seen in diode & piezo experiments.

Since I’m using distilled water, which has high electrical resistance, I had to use a high resistance load of 10 Mohms. Yesterday the current was 0.1 mV DC, which comes to 10 pA!  There’s that magical 10 pA again, but I don’t think it has settled down yet because the day before it was ~ 5 mV (50 times higher). So far it has been oscillating up & down, but with a gradual slow decline. So over the next few days it will most likely bounce up again to a few milli volts. Hopefully … it will settle at ~ 10 pA. If  it does, then I’ll be celebrating, as that will show the same behavior as diodes & piezos. I don’t know why yet, but so far every diode & piezo I’ve tested while loaded has always settled at ~ 10 pA. Dozens of different meters have been used, and various methods. For example, a diode connected to just one thing, a low leakage 1.0 uF capacitor (discharged), both placed inside a thick metal shield for ~ 10 hours, opened metal shield and the measured voltage across the capacitor was 0.353 volts, which comes to ~ 10 pA.

In my cheap version, the distilled water container is ~ 1 cm thick, and ~ 5 cm x 5 cm wide. On the outside of each plastic container (1 x 5 x 5 cm) is a large plastic plate, about 13 cm by 13 cm wide, and one or two millimeters thick. On the outside of that is the thin Aluminum foil, about 5 cm x 5 cm wide.

Usable amounts of Power

So far experimental data has shown that the current produced by diodes & piezos is high initially, but settles down to 10 pA. Once you let the component rest inside shielding for awhile, it’s capable of producing higher current again for awhile. Therefore perhaps it’s best the never allow it to reach the 1o pA level. Instead of leaving the load on all the top, load the device for awhile, remove the load, apply load, remove load, etc.

Guaranteed to work

Getting the DC current & voltage on the output is extremely easily. No tinkering around is required. Although I have not confirmed if this voltage & current is due to any ion leakage from the high voltage. I do not think so because of how extraordinarily slow the rise in voltage occurs, and because the voltage on the output foil gradually decreases over time.

So it’s guaranteed to work; i.e., produce current & voltage. The question is, what’s causing it.

For some time I forgot about UDE (Ultimate Diode Experiment), where a diode array is connected to a low leakage capacitor, allowed sufficient time to charge the capacitor, and then a mechanical switch will release the capacitors charge through an efficient LED to light the LED enough to be perceived with the unaided eye.

About a week ago I remembered UDE. Stage 4b design has now changed to UDE.

The goals of the new diode researcher:

1. Achieve personal proof that the diode produces a DC voltage. See Diode Lab. If you do not want to bother with electrometers and such, then please skip to step 2, UDE.

2. Achieve UDE for the purpose of demonstrating to academic notable scientists. This should be very simple. Place a sufficient amount of diodes in-series to achieve at least 2.3 volts DC. You can achieve high current by placing such groups in parallel such that no two diodes are in direct parallel connection. The entire diode array would be connected across a 10 nF (0.01 uF) low leakage capacitor. I recommend a Polypropylene Film Capacitors by WIMA, part number FKP2-.01/63/5. Next, connect a pin from the mechanical switch to the capacitor pins. Connect the switches other pin to an efficient LED, and connect the other efficient LED pin to the diode arrays other pin. Therefore, when the switch is on/pressed, the capacitor will discharge through the efficient LED. For the efficient LED, I would recommend part number TLCR5800. It is a 7500 mcd red LED with only 4 degrees beam width that produces 190 Lm/W. I have not tested this LED yet, but it looks great on paper. I have charged a 10 nF capacitor to 2.3 volts, and then discharged it on another efficient red LED (I don’t have the part number), which I could easily perceive with my eyes. It’s amazing the differences between LEDs. Out of ~~ a half dozen tested LEDs, I was able to see just one from a charged 10 nF capacitor. Another efficient untested LED is a yellow LED, part number OVLGY0C9B9, 20000 mcd. It is more efficient, 258 Lm/W, but it has a broader beam width of 6 degrees. Also, the yellow LED requires more voltage. Anyhow, place everything inside a thick Hammond metal shield. Drill a small hole in the Hammond shield to view the efficient LED. Mount the efficient LED up to the small hole. As far as the mechanical switch, there are various methods. Since UDE is so simple, it’s probably best to use a tilt switch. There are various types of tilt switches. There are ball types where a small ball tilts and makes contact. I no longer recommend the mercury tilt switches, as their off resistance is far too low. The off resistance should be at least 10 Tohm, preferably > 100 Tohm. Allow the diode array sufficient time to charge the capacitor. Place eye close to the hole near the efficient LED, and press the switch. Make sure the switch is on for only a fraction of a second so as not to fully discharge the capacitor. Since the LED has high resistance below ~ 1.2 volts, the capacitor should not go below ~ 1.1 V any time soon. This setup will most likely require a lot of trial and error since there are a lot of unknowns. You need to find out how many diodes in series you will need to achieve the 2.3 volts. I would recommend that you go far above 2.3 volts because diodes are so easily disturbed! You need to find out how long it takes, on average, to charge the 10 nF capacitor. It would take an *undisturbed* (10 pA) diode array 18 minutes to charge the 10 nF capacitor from 1.2 V to 2.3 V. If the diode array is slightly disturbed at say 1 pA, then it would take 180 minutes (3 hours). My highly disturbed green LED is producing ~ 0.1 pA, and would therefore take such diodes 30 hours. And perhaps the biggest unknown is can the common diode perform continuous work without becoming disturbed. This type of setup will be the ultimate demonstration to notable academic scientists, as there are no batteries inside. You could also test your unit in rural areas, perhaps even inside a cave or mine. Keep the setup far away from strong radio transmitters to prevent disturbance! Do not submit the setup to rapid temperature changes!

I recommend 10 nF since that’s easily perceived by the human eyes when charged to 2.3 volts discharged across an efficient LED. The lowest capacitance that worked was 1 nF. Anything less does not have enough stored energy.

For UDE I do not recommend the radio shack LEDs, unless you can afford to buy ~ 100 of such LEDs. The amount of diodes depends how disturbed the diodes are. I used excessive soldering on my recent radio shack LED array, that produced 199 mV for 7 LEDs. So the reference to 100 diodes is a conservative figure. Today I will buy some inexpensive diodes (non LEDs) that cost $0.031 (3.1 pennies) each that should work as well as the LEDs, perhaps better.

What is great about UDE is that it requires no electrometer, nothing but diodes, one capacitor, a Hammond chassis, and a simple switch! That is it!

After going over the data, there is good correlation between diode and any capacitor placed across it. In short, it appears the DC voltage produced by the diode is greatly effected by long term capacitance. IOW, if you place a capacitor across a diode, the DC voltage produced by that diode will slowly decay over time.

Data suggests that an undisturbed diode with high Rz might last up to 10 hours, but this appears to be an unusually long time, as most diodes may become starved of noise within minutes of being connected.

For best results, do not connect anything to the diode while it is recovering/resting. Use an electrometer to measure the DC voltage, an electrometer that has less than 10 fA input bias current. I use the INA116PA. If you build the INA116PA circuit, then you’ll need to measure the Ib before using it on a diode. Get the entire circuit ready for the diode. Before placing the diode in the input, place anywhere from 470 pF to 0.01 uF low leakage capacitor across the electrometer input. Turn it on, let it warm up a few minutes, carefully discharge the capacitor, and then time how much the electrometer charges the capacitor. The Ib is C * V / t, where C is the capacitor capacitance in farads, V is the change voltage across the capacitor in volts, and t is the time in seconds. If you are unable to obtain less than 10 fA, then try removing everything on the electrometer input and solder only the capacitor, and repeat the test. Remember, the capacitor and the electrometer should all be in mid air!!! Also, there *cannot* be any appreciable breeze because that will blow ions from the battery to the capacitor. You may struggle with this for a few days, but soon enough you will begin to understand the physics behind this in that what’s bad and what’s good for lowering Ib. Once you get that down, then you can connect the switches and redo the tests. For lowest Ib, the amount of metal surface connected to the electrometer input should be kept to a minimum. The more metal surface = more ions it will pick up. Such ions do not make any measurable difference with diodes, unless you’re measuring nanovolts, as such ions even from a bad setup will be on the order of femto amps, but you will want to have the bias current less than ~ 10 fA so that you can place a capacitor of low capacitance on the electrometer. Such a capacitor is not need when the diode is producing relatively high DC voltages, and therefore you may not want to use a capacitor since diodes don’t like capacitance. Although, for demonstrating to a notable scientists, you might want to have some capacitance to put the scientist at ease.

All of this may sound tedious and difficult, but it’s not that bad. In a few days you’ll probably have it all working well, maybe a week at most. If it takes you longer than a week, then please by all means make an anonymous post to the electronics newsgroup/Usenet at groups.google.com, or send one of the EE’s a private message. Someone will most likely help you. Or send me an anonymous email.

Once you get the electrometer Ib to less than 10 fA, then you can place up ~ 470 pF low leakage capacitor directly on the electrometer input. So if your diode Rz is 100 Gohm, and the capacitor is 470 pF, then it would take the diode 141 seconds to charge the capacitor to 95% of whatever voltage the diode is producing. See the online RC Time Constant Calculator for further calculations. This means you would leave the diode on the electrometer for ~ 141 seconds before obtaining the DC voltage, and then you would *immediately* reverse the diode input (example, if you’re using Mercury tilt switches and immediately take the DC voltage. Remember, the Mercury tilt switches that reverse the voltage should be connected before the capacitor, as this will allow you to obtain a difference in output voltage *immediately* since it will be reverse the capacitor as well across the electrometer input.

My biggest mistake was announcing my ambient thermal energy research. Please don’t make the same mistake!

I am so thrilled that as we speak people are conducting ambient thermal energy diode research undisclosed. The goal is to achieve a *mobile* diode testing setup that succeeds in obtaining at least 80% to 90% success rate in getting a *new* diode to produce at least 0.1 mV within one month time of purchase. Those are my minimum requirements. Each person can set his or her own limits, but just remember that a notable conventional scientist may give you only one chance.

Remember, obtaining at least 0.1 mV within one month of receiving your new diode is insufficient. You should repeat that for at least ~ a dozen times and obtain at least 80% success rate. The reason is that the interested notable scientist will want to replicate the procedure, and you want a high probability of the scientist succeeding, he or she may not have the time or patience to redo the test.

It would greatly help if you could afford to make a lot of your mobile diode testing setups to give one to each notable conventional scientist, but that is not required. Personally, I will not tell *anyone* except the notable scientists until at least one of them has succeeded in replicating my work. There is a very good reason for this.

The following could be a major breakthrough in the diode research, as it might solve the issue with placing diodes in direct parallel connection. Experimental data suggestions that 2 diodes in parallel produce half the DC voltage as a single diode, and thus half the power– P = V^2 / R. First, I’ll outline the hypothesis, and then the new proposed diode design.

The Hypothesis:
All diode models show that diodes produce a DC voltage by rectifying Johnson noise, but each model varies. Some models are more accurate that others. Science is not about absolutes. Science is not about assuming any equation is 100% perfect. The present best diode model I am aware of is where the diode depletion width varies relative to the applied *voltage* across the diode. It is a well known simple fact that the net Johnson noise across two identical resistors in parallel is less than one resistor. The net noise across two diodes in parallel is 1/sqrt(2) ~= 0.707 times less noise than one diode. The diode square law states that the resulting DC voltage from a diode, due to the rectification of an AC, is equal to the square of the AC. Therefore, the [1/sqrt(2)]^2 equals 1/2, which means two diodes in-parallel will produce half the DC voltage, and therefore half the DC power, as P = V^2 / R. Therefore, the DC voltage is relative to Rz, which is in agreement with diode experiments.

The new proposed design:
The proposed new design for diodes in direct parallel connection is to place an inductor in series with *each* diode. So if there are 10 diodes, there would be 10 inductors. Each inductor will hinder some of the AC noise *between* diodes, but it will not hinder the noise that each diode produces on itself. Every component has parallel capacitance, and therefore the noise across a diode is known as kTC noise, which is equal to sqrt(k * T / C). The inductor would decrease the parallel noise between diodes, and thus prevent some of the parallel noise cancellations. As to exactly how much noise the inductor cancels depends on the inductor.

New proposed tests:
This new diode array design that includes inductors should become Stage 5 in the research. Stage 4b is expected to be completed sometime in November to December. Stage 5 could begin in early 2010.

I know for fact that the passive component, the diode, produces a measurable DC voltage when left undisturbed inside metal shielding for a sufficient period of time. At the time of this writing, here is my recommendation –>

The circuit is very simple. Leave the gain resistor pins (pins 1 and 16) open to provide a gain of one. Do not connect anything to the guard pins because your INA116PA will not be on a PCB. Rather your INA116PA will be held in the air by the wires soldered to the pins. An air op-amp offers the lowest possible input bias current. The DC input bias current produced by my INA116PA was measured at 2.2 fA (2.2E-15 amps). Pins 3 and connect are the input pins that you will connect to the diode. Get a quantity of two “4 AAA battery holders” for the INA116PA voltage source. For my next electrometer, I will be using just two “3 AAA battery holders,” which will provide even less bias current. Less bias current is an over kill, but the only reason I’m switching to just three batteries per polarity is to minimize the size so that I can place even more diodes in my new setup. In your case, you’ll probably be testing just one diode, so it’s best to get 4 AAA batteries per polarity since I do not know if the INA116PA will work well enough on 4.5 volts. The INA116PA datahsheet claims that 4.5 volts is fine. I’ll be using new Alkaline batteries, which are over 1.5 volts each, so it will be over 4.5 volts. Also, the INA116PA will most likely work well below 4.5 volts per polarity. I get the AAA battery holders at Radio Shack. So, connect the negative voltage from your batteries to pin 8, and the positive voltage to pin 13. Connect pin 9 (Ref. pin) directly to ground. Pin 11 is the output, which you can connect to a common voltage meter. Here’s a diagram –>

The above electrometer input is left floating. I have tested the electrometer both ways, floating, and also with grounding resistors connected to a input wire. It works just perfectly fine on the diodes. You will only need to use the electrometer for a few minutes per testing period. If you’re testing multiple diodes, then it is recommended that you ground the input pins to ground in between each diode test.

The electrometer and batteries are inside a metal shield. I use a large metal shield made by Hammond. I have conducted various tests with the batteries inside and outside the Hammond shield. It makes no difference.

Twist the input wires and route then threw a small pin hole in the Hammond shield, which will go to the DMM (voltage meter). The DMM is outside the Hammond shield. The entire setup, including the DMM is then placed inside a second layer shield. You can use a large microwave oven.

To turn on the DMM and electrometer, you can use either mercury tilt switches or simple mechanical contact switches. I have used waxed dental floss string to go through a pin hole in the microwave oven to the meter that will turn the switch on/off by slightly pulling the string. My new setup will use 1 Lbs. fishing line string instead. Also, I will be making my own contact switches this time, which are rather easy to make simply by placing a short thick copper wire (~ 10 to 18 gauge) that privets up and down, one end tied to the fishing line that goes through pin holes of both shields that is connected to a small weight. The weight will be enough to pull the copper wire up to make contact to another copper wire. When the weight list slightly lifted up and placed on a small ledge, the copper wire goes down and creates an open-circuit. This creates a very nice simple copper switch that will produce absolutely no measurable effects on the measurements.

For the new setup, this time I will use the aforementioned simple contact switches instead of mercury tilt switches because the mercury tilt switches will be used for the electrometer input stage. There will be four mercury tilt switches. When the entire setup is slightly tilted *forward*, two of the mercury switches will be on. When the entire setup is slightly tilted *backwards*, two of the mercury switches will be on. While the setup is tilted forward, the electrometer will be connected to a diode in *forward polarity direction*. While the setup is tilted backwards, the electrometer will be connected to a diode in *reverse polarity direction*. IOW, this will allow me to reverse the diodes connection across the electrometers input. So if the measured DC voltage is say 350 mV while tilted forward, it should be -350 mV while titled backwards. Of course you have take into account the electrometers *output* voltage offset. Note that the electrometers output offset has nothing to do with a voltage offset on the electrometers input stage. So don’t forget to subtract the output offset. Or if you wish, you can add a second stage op-amp that does the offset for you, but such pretty stuff is completely unnecessary until the final stage when you demonstrate to a notable scientist.

The next step consists of the diode. Presently I would recommend the 1N4148WS diode made by MULTICOMP. This is a 1N4148 diode that has a solid casing, not glass. The problem with glass casings is that you have to work inside a dark room since the light will shine on the diodes junction, and thus disturb the diode. I would recommend soldering 50 of such 1N4148WS diodes in-series. Unfortunately this will highly disturb the diodes– TED effect. Place the disturbed diode array inside the Hammond shield and solder/connect it to the electrometer input. Close the lid on both shields, and let it sit. Since there is no guarantee how disturbed the diode array will be, I would recommend letting it sit undisturbed for at least one month. Just to be on the safe side, do not place the setup near a wi-fi setup, although I have done extensive testing by placing my entire diode testing setup directly near a new high power wi-fi which clearly showed no change in the measured DC voltage. It is recommended that the diode array temperature not rapidly fluctuate, as this can disturb the diode state– TED effect. After a month or two, turn on the electrometer, let it be for a few minutes to stabilize, then tilt the entire setup to connect the electrometer to the diode, look through the microwave oven metal mesh grid to see the DMM reading, quickly write down the DC voltage, tilt the setup the other direction to reverse the mercury tilt switches, write down the DC voltage. There’s no need to subtract the electrometers offset, as the DC voltage is equal (Vp – Vn) / 2, where Vp is the voltage measured when tilted forward, and Vn is the voltage measured when tilted backwards. Then tilt the entire setup to neutral position to disconnect the diode array to the electrometer, and quickly turn off the DMM and electrometer. Take a measurement about once every two days.

Please contact me for further details and on-going advice.

Often someone will ask, if diodes produce a DC voltage, then why doesn’t modern science know of this. Three main reasons:

1. The newly discovered TED effect makes it almost impossible to detect the DC current and voltage produced by passive diode. Nearly two years of extensive diode measurements has consistently shown that the *slightest* disturbance to the diode almost completely eliminates the DC voltage produced by the diode. The is referred to as the TED effect.

2. The DC voltage produced by diodes requires a special setup. Experiments are showing the produced DC voltage is relative to the diodes near zero bias resistance, Rz. For example, LEDs have ultra high Rz, and thus produce high DC voltages, but such ultra high Rz (in the Giga to Tera ohm region) requires either special equipment or a charging capacitor. Since the diodes produce low current, one must wait a relatively long period of time for the *undisturbed* LED to charge the low leakage capacitor. On the other end of the spectrum, ZBDs (zero bias diodes) have low resistance, but produce ultra low DC voltages per diode in the nano volt region. One can solve this problem by placing tens of thousands of ZBDs in-series to achieve a few millivolts, but then we have the problem of measuring the DC voltage on an ultra high resistance DUT.

3. A simple unshielded diode (e.g., 1N914A) connected to a volt meter can easily produce well over 10 millivolts by rectifying RF noise such as radio stations and wi-fi routers. For this reason, an Electrical Engineer detecting a DC voltage from a diode would immediately dismiss the DC voltage as such from an unknown external noise source.

Conclusion:
My two years experience in measuring the DC voltage produce by diodes has shown that it requires a special setup and an unusual amount of patience. As expected, it appears new diodes come disturbed. Present data suggests that the amount of time required to undisturbed a diode (TED effect) is relative (not necessarily linear) to Rz. This would make LEDs require more time to recover than ZBDs. The only new diodes I’ve seen recover are the SMS7630, which are ZBDs. My SMS7630 diodes sat undisturbed for several months, followed by soldering them together, followed by sitting inside the metal shield for another two to three weeks. Then and only then did they produce their DC voltage, as it gradually increased. On the other hand, my old white LEDs that were producing a DC voltage still after two months have not recovered after lighting such LEDs.

In order for a scientist to discover the DC voltage produced by passive diodes, he or she would most likely have to use uncommon ultra low Rz ZBD’s diodes (a modern technological achievement), construct a large diode array *in-series*, allow such diodes to rest several weeks inside a metal shield, and use special low bias equipment capable of detecting down to a micro volt. Unfortunately, *every* scientist I’m aware of would never consider a few microvolts produced by such a large diode array as worth investigating. Such a claim violates conventional physics, thus severely jeopardizing the scientists career and reputation. Such a claim is too easily explained by a lot of reasons such as external noise interference. The only reason I succeed is due to my unusually stubborn and perseverance nature, and from my years of prior Spice simulations, which clearly showed that diodes must rectify Johnson noise

Preface: The following method is strict, but it is by no means required. It is a simple method, but requires tilt switches. Do not use a mercury tilt switch, as the resistance is to low. There are a lot of methods that require common parts, but you need to know what you’re doing. If you’re not an experience EE, and you want to avoid the hassle of debugging your measuring method, then consider the following method. Measuring millivolts on a 1.0uF capacitor is extremely easy *if* you know what you’re doing. If you do not want to buy tilt switches, for instance, then that’s fine, but please contact me so that I can give you ongoing advice. Otherwise, unless you are as persistent as I am, you will obtain inconclusive results.

I would invite people to perform diode experiments. I would only ask that you follow my *ongoing* advice. It is extremely easy to measure the charge on a 1.0uF capacitor! Your main obstacle will be in getting the diodes to become undisturbed!!

Here’s an outline of the procedure in my blog site. From time to time it’s updated, but I’ll post an update notice in the blog.

http://globalfreeenergy.info/2009/05/13/replication-warning/

Testing LEDs is a new part of the research. Please note that all of my LED tests were on old LEDs. It is reasonable to believe all new LEDs are lighted up at the factory to determine if they are functional. An example to demonstrate how easily disturbed LEDs are, I had white LEDs in a metal shield that were producing a DC voltage. Albeit low, but over 0.1mV. I then light up the white LEDs. After that, the white LEDs were producing less than 0.1mV. After that the white LEDs were placed in thick metal shield. After several months later, the white LEDs are still producing less than 0.1mV.

The only *new* diodes I’ve tested were the SMS7630, but even they were sitting in the lab for several months before being soldered together. After that, the soldered SMS7630 arrays were sitting inside a metal shield for at least three weeks. I do *NOT* recommend using SMS7630 diodes because according to the recent discovery the produced DC voltage (and power) has been relative to the diodes near zero bias resistance, Rz. Presently, I recommend IR (infrared) photodiodes and LEDs. IR photodiodes/LEDs tend to have lower Rz than visible spectrum photodiodes/LEDs, so the DC voltage they produce is less than a typical visible spectrum LED, but present data suggests that the rate at which diodes recover is relative to the reciprocal of Rz. The exact recover rate equation is unknown, but it appears to be non-linear. Therefore, it is believed that IR photodiodes/LEDs will produce less DC voltage than visible spectrum LEDs, but they should recover at a faster rate.

The only new LED experiments so far have been performed by one person. Unfortunately, this person completely ignored my recommendations time after time. This person has/had four LED experiments going on at once. The good news is that this person saw *all four* of his separate LEDs simultaneously charge their capacitors at a rate of ~ 1mV per day. The 1mV is very low compared to the record, which is 353mV (0.353 volts). This person still needs to improve his measuring technique. For example, out of numerous measurements there was one incident where one of his capacitor-only tests slightly charged the capacitor. This makes his error rate relatively low, which adds some credibility of all four of his LED’s charging the capacitor, but there’s no reason for such accidents to occur.  So it appears that it’s possible to get new LEDs to produce, but it appears *new* diodes are disturbed. Disturbed diodes still produce DC current/voltage, but at a much lower level.

Please contact me if you’re interested. Together we can determine the best diode for your tests, and I’ll provide ongoing advice so that you can obtain an undisturbed diode. One thing I know for certain, highly shielded undisturbed diodes produce DC current across a load, and such diodes can produce at least 0.35 volts and 23pA. If you have the perseverance, then I can help you obtain your proof by witnessing a diode charge a low leakage 1.0uF capacitor to well over 0.1 volt in a highly shielded setup.

Very Important tips on how to replicate the LED experiments.

The diode is not as an Alkaline battery. The DC voltage produced by diodes is believed to be from purely *random* ambient thermal noise that always exists in all matter; e.g., at room temperatures there’s over 1 billion joules per m^3 in solids. Furthermore, the diode is exceptionally sensitive to external sources, and is easily disturbed. Just touching the diode is enough to change the diode state from producing pico amps to femto amps. Diodes made of a transparent casing such as the 1N914 and LED’s are also exceptionally sensitive to light. For example, shining light on my white LEDs changed them from producing a measurable DC voltage to a non-measurable DC voltage.

There’s a good chance your diode is disturbed, unless your diode has been sitting inside sufficient metal shielding for a long time. The amount of time required for a diode to recover is unknown, but data is beginning to suggest that diodes with higher near zero bias resistance (e.g., LED’s) require more time to recover. I noticed that it took ~ two to three weeks for my SMS7630 diodes (excepti0nally low zero bias resistance, ~ 5500 ohms) to recover. It took about three weeks for the 1N914 diodes to recover, but IMO they were only *slightly* disturbed. In over two months of resting, my white LED’s  are still highly disturbed; e.g., they are producing less than 0.1mV DC.

This research requires a lot of perseverance! Unless you’re a senior EE with low signal measuring experience, don’t expect to slap together an LED onto a low leakage capacitor and get quick conclusive results. You need to take common precautions. My blogs contains all necessary notes to succeed. When time permits, I’ll compile a single page.

For now, please consider the following outline –>

1. Initially I was against applying anything to the LED, but now I highly recommend coating the LED with a good layer of liquid (at room temperature) rubber or some other drying liquid so long as it has ultra high electrical resistance. Places such as Harbor Freight sell rubber that is in liquid form at room temperature. What ever you do, do *NOT* heat up the LED since that will disturb it. Perhaps someone could test the resistance of liquid white-out that is used for whiting out writing errors on paper. The idea is to prevent an appreciable amount of light from entering the LED.
2. Presently I would recommend an IR photodiode or LED. It will not produce as much DC voltage, but present data suggests they will recover in a shorter period of time. If you have the means to conduct numerous diode experiments simultaneously, then also try a green LED and the 1N914 diodes. For undisturbed diodes, expect ~ 10pA * Rz. Rz for 1N914 diodes is ~ 9Mohms, which comes to 0.09mV. If you please 10 1N914 diodes in-series, then expect 0.9mV, but remember that this is for undisturbed diodes. As far as LED’s, an LED is not an LED. The near zero bias resistance (Rz) greatly varies.  A lot of LED’s have relatively low Rz, and are expected to produce less than 0.1mV. For example, my green LED produced up to 353mV (0.353 volts), yet the most voltage my white LED’s produced was far less than a millivolt.
3. Get a high impedance voltage meter. I highly recommend the AM-240, which I bought new for $40. This DMM is amazing with input resistance well over 10 gigaohms.
4. Do *NOT* touch the LED with your fingers. Try to wear rubber gloves. ***GENTLY* connect the LED to a low leakage capacitor. Since the AM-240 has ultra high input resistance, I recommend a 1.0uF low leakage capacitor. If you must solder the capacitor to the LED, then do a quick spot solder job. Don’t worry about a professional solder. Do *not* use solder flux.
5. Connect your entire setup ready. Presently, I use small Mercury switches such that when the entire setup is slightly tilted it will connect the LED to the low leakage capacitor. Also, when tilted even further yet, another set of Mercury switches will connect the LED & capacitor to the voltage meter. This offers no noticeable disturbance to the diodes.
6. Short the low leakage capacitor. There are various ways. You can use very short clip leads. You can use small male/female pins where you simply slide the male pin inside the female pin. Let the diode ***REST*** inside a good metal electrical shielding for a long time, at least three weeks. The diode *MUST* rest while connected to nothing! That means no capacitors, no loads. The rest period depends on the diode and how disturbed it is. The longer the better. I use two layers of metal shielding– a thick Aluminum shield made by Hammond (~ $20), placed that inside a microwave oven. Someone suggested that microwave ovens contain radioactive materials. I’ve tested diodes inside a lot of metal shields without a microwave oven, and it makes no difference.
7. After ~ one day, unshort the low leakage capacitor. Be very careful not to get your fingers or body near the capacitor. If you used clip leads, then use long tip fliers to unclip them. You can put some soft tape, such as duck tape, around both tips of the fliers to help with gripping, and possibly reduce friction. Use the long tip fliers to unclip the clip leads. Let the diode remain unshorted and connected to nothing so that it can continue to rest.
8. After ~ one month, slightly tilt the setup such that the diode can begin charging the low leakage capacitor. Let the diode charge the capacitor for 10 hours.
9. After 10 hours of allowing the diode to charge the capacitor, turn on the AM-240 voltage meter. Short the AM-240 input for at least a few minutes. If possible, use different type of meter leads such that the AM-240 leads are shorted on the AM-240 side. Unshort AM-240, then *quickly* tilt the entire setup a bit more such that the second line of Mercury switches connects the AM-240 to the LED/capacitor. The AM-240 will quickly show the DC voltage. Within say 5 seconds you should have your reading. Then tilt the entire setup back down again such that it disconnects the AM-240, but the LED should still be connected to the capacitor.
10. After another 14 hours (a total of 1 day) take another voltage measurement– turn on AM-240, short AM-240, unshort, quickly tilt the setup, take measurement.
11. Continue to take measurements every 24 hours until the measurements are no longer changing at a relatively fast rate, in which case take one measurement every two days or longer. Long term measurements disturb the diode.

Please read the above steps very carefully! For conclusive results, it’s best to conduct the above experiment without the LED to test your measuring abilities. This would be a capacitor only experiment to see if your measurements are slightly charging the capacitor. Such experiments should not charge the capacitor.

Present data suggests that either the daily measurements eventually disturbs the diode, or the capacitor is sufficiently starving the diode of noise. Either way, it’s likely that the LED/capacitor will reach a peak voltage, and from there it will begin to slowly decay over the weeks. Simply disconnecting the LED and letting  it rest is sufficient to recover the LED.

As far as how long to allow the diode to rest is unknown since it’s unknown how disturbed your diode is. Working with LED’s is a new part of this research, and I have never used a *brand new* LED. All of my LED’s were old, and resting inside a container. My SMS763o diodes were new, but I they too were sitting unused for several months. My 1N914 diodes were ~~ 5 years old.