High power short pulse generator for LENR experiments

  • Hi guys,


    Reading SRI test report, I was little bit disappointed that they are using proprietary signal generator to get their 80A/100ns pulses. However, as has been stated by many replicators, probably the short and fast current pulses (i.e. with broad frequency spectrum) is essential to trigger LENR reaction.

    So I started thinking what is feasible and controllable way to produce something similar to their so-called "Q Pulses".

    Personally, I have very successful experience using high voltage gallium-nitride (GaN) FET transistors from GaN Systems for building various high efficiency power converters. Also I'm fun of STM32 microcontrollers, and there is a chip STM32F3348 with high resolution PWM timer (217 ps), exactly for precise generation of short pulses (or high frequency PWM).


    So I put together the 32F3348DISCOVERY evaluation board for microprocessor, and one floating in my lab GS66508T-EVBHB evaluation board which is half-bridge made using GS66508T transistors (650V, 30A). Also prepared some simple firmware for 100ns pulse generation (Github).

    Here is the picture:



    The control signals for high and low side switches are going from microcontroller board. The output of the half bridge was loaded to ground by matched 50ohm resistor (connected by standard RG58 coaxial cable). Input DC voltage is +300V, so at 50ohms this is 6A, which is not a lot -- but enough to start.


    Here is the picture of output voltage measured across the load (resistor) side. So the rise time is about 10ns for 300V or equivalently 10ns for 6A.




    The actual transistor (GS66508T) is capable for 30A of short pulses, however I used also GS66516T which is 60A part and more close to desired power.


    So my plan is to gradually increase the power to at least 30A using the existing half-bridge board (I should source or make some coaxial cables with lower impedances). And then switch to 60A FETs.

    As a result, I'm going to propose some ready-made hardware for LENR enthusiasts, which will be a set of evaluation and custom boards. The custom board probably will be control module using STM32F3348 microcontroller (open source and open hardware, and I will sell it from my Tindie store), with maybe some external computer software to configure power level.


    Please let me know (by liking this post) if you are interested in these efforts...

  • Please let me know (by liking this post) if you are interested in these efforts...


    asd,


    All here are interested in any LENR efforts. Most, like myself, are embarrassed to give a "like" to something someone says that they do not understand, although they do very much appreciate. Long story short, you smart guys, especially the experimenter types, will not get as many thumbs-up as we talkers, but do not let that deter you.


    That said, BEC's "Q-Pulse" has generated a lot of talk here and elsewhere. Being that it (QP) is their signature, and unique feature that distances them from the rest of the LENR+ pack, it is not surprising they keep it secret. BTW, BEC is a participant here.


    Take care, and best of luck! :)

  • These pulses require extreme care from experimenters. Have you designed UHF wideband front-ends? If not you probably are not aware of the techniques needed to make a sensitive analog circuit work reliably in the presence (same ground line, or physical proximity) of this stuff.


    The waveform shown - if corresponding to a 30A pulse - would show 2500A/us slew rate. That will inject 2.5V across any 1nH inductance - a typical lead parasitic even using good high frequency layout.


    Similarly any unshielded or poorly shielded wire carrying that pulse will inject voltage spikes into any random wire nearby, or the local ground connections.


    The two edges of the pulse are asymmetric with an HF content 10X larger on leading than falling. That is typical, and means even in a symmetric amplifying circuit you will get asymmetric rectification due to slew-rate limits. In any case most amplifiers have asymmetric slew-rate limits. Think of what this means when you are amplifying TC signals and 1mV is seen as a significant power output change?


    These effects are not so difficult to control, but only if you are familiar with UHF EMC layout precautions and techniques in challenging situations. I'm wondering how many of the experimenters using this stuff are aware of the issues here, and know the ways to diagnose and cure them?


    Regards, THH

  • I have been making 50V30A AC square waves using an adapted and very inexpensive H-Bridge circuit (as sold for control of brushless motors in models). It is driven by an astable vibrator, and seems to produce pretty damn clean square waves at anything from 1Hz to 2k Hertz. More might be possible, but not yet tested beyond 2kHz.


    The nice thing about using the astable vibrator circuit (about $5 from China) is that not only can you control the frequency you can control the mark-space ratio. Sadly I didn't take any scope shots when it was on test. But I do have a photograph of the general layout of the 2 boards etc, and will post it here later. Not so much fun as building your own from scratch, but then again, I blew a few of my homebrew ones up before resorting to the China route.:(


    On the topic of inductive pick-up in nearby leads, this didn't seem to happen. Being a former radio ham, I am pretty open to the idea of co-channel interference so looking for it was part of the testing. Worth pointing out that this circuit is used to drive heater coils in a reactor, and the stainless-steel sheathed K-Type TC's actually end up smack in the middle of those coils, a position where you would expect maximum induction of 'phantom signals' to occur. Happily, nothing seen.


    The TC's are (as mentioned) sheathed in stainless steel and have fully screened and earthed leadouts to the data-logger - which probably helps a lot.. There is no direct connection between the data logger circuits which are grounded to earth, and the power supply that runs the H-bridge(s) which is not earthed but has a 'return to neutral. Since DC power input is measured before the H-Bridge circuit there is no tricky integration required either- the H-Bridge losses are considered 'trivial' and in any case are equal for both test and control parts of the reactor.


    All good fun, and keeps me out of the pub.:)In fact, I have forgotten where it is.;(


  • I'm glad this does not happen in your circuits. What control test, as a matter of course, would you do to check it is not happening? My point is that it will happen in some setups... And if not controlled there, it will easily be confused with LENR.


    THH

  • Very simple. Take a reactor and run the heater coils off straight DC. Measure the temperatures. Do the same thing with the square wave system and compare. Or- run a reactor up to any temperature you like using square wave heat, then switch it off. Check the cooling curve- if there is a kink in it at shut-down point you know you have an interference problem. Simples. In fact, switching the heat on and off and watching for anomalous bumps (or dips) in the temperature gives you a very good indication of mischeif in the machine.

  • Very simple. Take a reactor and run the heater coils off straight DC. Measure the temperatures. Do the same thing with the square wave system and compare. Or- run a reactor up to any temperature you like using square wave heat, then switch it off. Check the cooling curve- if there is a kink in it at shut-down point you know you have an interference problem. Simples. In fact, switching the heat on and off and watching for anomalous bumps (or dips) in the temperature gives you a very good indication of mischeif in the machine.


    So - I would do this.


    (1) Measure - somehow - the electrical time constant of the TC amplifier. For example by injecting a sq wave at 1Hz and measuring the rise and fall time etc.

    (2) In the actual experiment (so that leads etc are in identical positions) have a control period where the stimulus wave form is switched on and off. Observe the synchronised change in the TC output. Measure rise and dfall times. If this has the electrical time constant - it will be RFI. If it has some longer time constant it might be some stimulus-induced chnage in termperature.

    (3) For added validation do the same thing with a control experiment where no change in temperature with stimulus is expected.


    If your stimulus adds significant power to the system, so is known to change temperature anyway, the electrical (RFI) and temperature components of this can be accurately differentiated as long as the time constants are significantly different which is likely due to the thermal inertia of the system.

  • The time constant of the TC amp is tiny - they are simple linear voltage amplifiers AFAIK. But it can be measured sure nuff.


    So it should be possible to distinguish temperature chnage from stimulus with direct RFI effect of stimulus easily. This BTW is what I wanted SRI to do with the Brillouin stuff. Perhaps they will.

  • Experiments and experimenters are very welcome here.


    A number of years ago I had the opportunity to build a 1000V, 50A pulse generator using IGBTs. Pulse width was fixed at ~ 5.7 us with a rep rate from zero to 5000 pps. Proper gate drive voltage and current are key to making IGBTs turn on and off properly. Gate driver ICs are available from several manufacturers, as are pulse isolation transformers that permit stacking or bridging of IGBTs to obtain voltages higher than can be sustained by a single device. As was previously mentioned, it is important to maintain isolation between the input and output in order to avoid unwanted Ldi/dt voltages from appearing in the low voltage circuits. Separate grounds are essential, and here the isolation transformers are a help. I have used this pulse generator to heat a Celani type setup, but did not observe any excess heat.

  • The commercial “high voltage generators” of the link above are not “800 kilovolts” of course. It is induction coils for tasers. The actual voltage is only around 80 000 volts.

    The law (spark lenght/voltage) in air at normal pressure is complex, but a simple approximation is 10 kV = 1 cm.