My Motors got me to Tap into Radiant Energy

[JohnStone]

[QUOTE=ntc;239922]

... I did notice you probably intend to use a regulated power supply in place of batteries I did this and damaged my PSU twice and the recommendation from the supplier was to insert a very large capacitor between the PSU and the motor/s I was driving as the spikes will keep damaging it.
...

Those caps will possible not get the energy stored because of their inductance and they might make your PSU unstable.
The right measure will be to add a choke in series to + and then the big cap.

Choke: any transformer. Connect the low voltage coil as choke coil and short circuit the high voltage coil. See Utkin where he explains how the inductance increases considerably when a winding is being shorted out. It is like adding a flywheel mass to the shaft of a motor.
If you have a transormer with 2 secondaries you connect one to + and the other to - and short ciruit HV winding as well. Thus you get a current compensated choke (common mode choke).
Alterntively you might find those chokes as well at washing machines, older computer monitors, MOV and other electric equipment. Scavenge
JS

[boguslaw]

I think the energy is stored in dielectric in case of capacitor. I tend to visualize a capacitor like a resonant cavity with a lot of bouncing waves. The simplest way to prove would be to dismantle charged leyden jar and replace all metalic parts with the new one from uncharged jar.

Tha may also open a "worms can" because all electrostatics are based on the same concept....so if I'm right things would not look the same as in books. In electrostatics case the point would be to do all experiments in vacuum (obviously that is impossible without a labolatory).

The issue with protecting capacitor from overcharge from radiant energy is a big interest for me also. And here is the question : I know that the simple regulated DC supply could be made just with a zener diode and a power resistor, but the problem is the resistor placed in the power supply line limiting the output current after the capacitor. Could it be arranged in such way that the zener diode and resistor act only in case on overvoltage but not during the normal output processing ? !!! Imagine a power supply line going to capacitor and forward to load , with no resistor inline, while zener diode and power resistor in shunt to the capacitor - would that work in case of overvoltage ????

[erfinder]

I think the energy is stored in dielectric in case of capacitor. I tend to visualize a capacitor like a resonant cavity with a lot of bouncing waves. The simplest way to prove would be to dismantle charged leyden jar and replace all metalic parts with the new one from uncharged jar.

Tha may also open a "worms can" because all electrostatics are based on the same concept....so if I'm right things would not look the same as in books. In electrostatics case the point would be to do all experiments in vacuum (obviously that is impossible without a labolatory).

The issue with protecting capacitor from overcharge from radiant energy is a big interest for me also. And here is the question : I know that the simple regulated DC supply could be made just with a zener diode and a power resistor, but the problem is the resistor placed in the power supply line limiting the output current after the capacitor. Could it be arranged in such way that the zener diode and resistor act only in case on overvoltage but not during the normal output processing ? !!! Imagine a power supply line going to capacitor and forward to load , with no resistor inline, while zener diode and power resistor in shunt to the capacitor - would that work in case of overvoltage ????

If you are seeing something going back to your supply, interpret it as a good sign! You just have to make sure that you know what it is that's returning, you have to find out whether its CEMF or if its inductive kickback. Hopefully you will find that its CEMF, if it is, then you win, if its inductive kickback, reconfigure so that its CEMF. It is imperative that you get the CEMF to charge the supply and not the inductive kickback (what you guys call radiant). Once you get the CEMF returning, learn how and when it will exceed the supply voltage, then learn to control it. Once you control it, then you can contemplate methods for "mixing" CEMF and inductive kickback.

It was suggested by John Stone that you insert a choke, I don't recommend this, what I recommend is that you place two diodes between your supply and the circuit that you are using to drive the motor (one diode on the positive leg, and one on the negative leg), then place a high value capacitor after those diodes. That is all you need to do, then watch the voltage on that capacitor. You will need to adjust the timing of the device you are driving and the pulse width that its being operated at. Doing so, depending on your configuration you should see an increase activity where you are collecting inductive kickback, or the cap which is across the power supply behind those diodes will charge to a higher value than the supply. I make no promise that you will see this in your setup, I see it in my machines, and have come to the conclusion that its due to the winding configuration that I use. However I made this post because it was stated that you or another poster is seeing something go back to the power supply and cause problems. This is good! You have to tame the beast.


edit..

I have been experiencing this phenomena for a very long time and now take advantage of it.

Here is a demonstration from about a year ago when I first found that beast that was blowing up my power supplies.

. - YouTube



Regards

[JohnStone]

Surprise : Got Monster PCBs today fom Cornboy AND Oshpark
Starting with population just now. As suspected Oshpark delivers now double sided PCBs and those are a bit more difficult to solder because both sides suck heat from soldering iron. Later on I can compare both types of PCB.
Stay tuned.

BTW: I took an EAGLE class of 1 hour at youtube. I decided to do my next layout with EAGLE. They added an 3D export to Sketchup and a link to Element14 stock (Newark / Farnell). Now we can populate BOM with real component numbers for ordering anywhere. Additionally they have a link to LT-Spice from LT fordirect simulation from Eagle schematic.
Oshpark accepts Eagle files as source data. So we do not need any Gerber or Exellon export for this PCB manufacturer.
JS

[GChilders]

Monster boards

That's great news John. Let the testing begin.

Cheers

Garry

[ntc]

Hello Ufopolitics

Dear Ufopolitics,

Thank you for the explanations about the motor, current, coil and diodes. I have ordered a motor that should have a lower total resistance than my coil, and will work with that when it arrives. I pulled apart one of the motors I was working with and checked its total resistance, as you explained how to do; indeed it does not match my coil, as the motor resistance exceeds coil resistance.

I have also replaced my capacitors with non-polarized caps. I am referring to the caps that come after the diodes, which I use when running a motor to smooth things out. (I do not use these caps when running a CFL).

By the way, in your circuit, there is also a 1000uF/50V cap before the diodes (labeled C1); I have left that one as a polarized cap, I hope that’s right.

As for the 1n4148s after the NTE576s, I had given up on them because I burned so many of them. Originally I thought it was because I had the wrong specs, as the diodes I first got were rated .3A. Then I replaced them with 0.45A, but I still regularly burned through them when using the circuit to light the CFL. Do you think that might be symptomatic of something wrong in my circuit?

In any case, I have put them back in (and have not yet burned them). For now, I have observed that this noticeably reduces the amount of current drawn by the motor I am using. This measurement is taken between diodes and motor. When I say reduce, I mean it seems to me that fewer milliamps are drawn – as measured by the meter - in contrast to when I did not have the 4148s in place. (I would have to take them out; measure; put them back in; and measure again: before I could confirm that.)

And now I would like to share some pictures. I’m sure this will be old news to everyone here, but they are very beautiful and I hope you won’t mind my posting them. The quality leaves something to be desired, as I was using my mobile phone.

I was running a 24 VDC motor gently, at 480 Hz and duty at approx. 5% for all pictures (as scoped at output from pulser).

The first attached picture is channel 1 (CH1) on pulse output (going to drain) and CH2 on drain before the diode (with only the NTE 576 installed). One can see nicely that when output is low, drain is 38V. When output goes high, drain goes to 0V. When output goes low again, drain shoots to over 100V; quickly fall to about 2/3s that value for about 100 us; and then oscillates in decaying fashion around 38V. Then the whole thing repeats. One interesting things about the picture is the output signal. It is not absolutely sharp, either at rise or fall. However, I am not sure if it is negatively affecting what happens at the gate; though I would have to change the horizontal scale to get a clearer picture of that. From this picture, the response at the gate seems very sharp.

The second attached picture is CH1 on pulse output, and CH2 at drain after the diode (again, only the 576 installed). Here one can see that much of the voltage below 36V has been taken out. Interestingly, it takes longer for the voltage to stabilize at 38V than before, after the spike.

The last attached picture is a combination of the above two, with CH1 on drain before the diodes; and CH2 on drain attached after the diodes.

I wish you a nice weekend! All best, ntc

[Cornboy 555]

I just NEW that would happen John, can't wait for you to populate and test a board, to see if maybe we need to use lower value pull down resistor on gate to make switching sharper.

My thought is, maybe in this board format, the 12a driver is a little over zealous.

Great News.

Warm Regards Cornboy.

[JohnStone]

@ntc: Thanks for your thorough measurements posted. They are very valuable. Unfortunately I lost the pic in my brain how your circuit is exactly. Will it be possibkle to post your current schematic? Driver itself would be valuable as well.
Can you confirm that those oscillations do not occure if pure resistive load at same amperage?
What is the switching time (or slope at center of edge (i.e. V/µs) at gate and drain? Those parameters might be essential for burning 1N4148 diodes.
Thanks


@Cornboy:
The pulldown at gate needs to be lower resistance at weak driver cases. only Here it is for safety only in order to prevent unsolicited ON state for FET.

The driver is NOT over zealous. If we dig into µs and ns it needs much current in order to charge the "giant" gate capacitance and override backlash fom high spikes at drain pin(drain / gate capacitance - steadily opposing). And we want to check on how many FETs we can drive in parallel for monster to get more adult later on.
This 12A driver is not more than accelerating a hammer and hit the nail. A half pound hammer can hit with up to 8 tons - for shot time of course.
Other view: If you want to fill a 100 gallon vat (parasitic gate and drain cap) every hour it would be desirable to not use the normal water tap (NE555 driver or LM339/LM324)along long waiting time but a fire hyadrant (12A driver). Fortunately we can afford this electronic "fire hydrant".

We can control / tune switch speed by gate resistor. I will measure and post these effects as well.

The driver was designed in order to squeeze as fast switching out of current technology as viable regarding effort and price. Thus I hope to get as close to radiant effects as possible. Upcoming SiC technology will be even faster. In fact whne choosing FET types we need to negotiate between RdsON, darain breakdown voltage and switching speed. We can not have oll benefits at same time.
JS

[ntc]

reply to John Stone

Will it be possibkle to post your current schematic? Driver itself would be valuable as well.
Can you confirm that those oscillations do not occure if pure resistive load at same amperage?
What is the switching time (or slope at center of edge (i.e. V/µs) at gate and drain? Those parameters might be essential for burning 1N4148 diodes.

Dear Mr Stone,

Thank you for your message. I will do my best to answer your questions.

1. My circuit. I’m afraid I have not drawn a schematic of what I am using now, because it evolved as I advanced. But it is easy to follow. I took ufopolitics’s circuit (taken from this board, ufo_555_controller_1.jpg). Then I substituted in a different voltage regulator (LM_317_VOLTAGE_REG_UPDATED.jpg , again from this board). And finally I substituted in a different pulser (the same one used by netica, from DIY Homemade Power Pulse Controller - RMCybernetics.

I use 4 mosfets, with the drain of each being wired together into one common drain attached to coil. My coil is single strand 18 AWG, 3 times 130 turns approx; total resistance 1.7 ohms; inner diameter about 52 mm, empty core. Power source is 3 in series 12V sealed lead acid batteries, each rated 3.3 Ah.

2. The oscillations. I take it you are referring to the oscillations measured on the drain when the output signal goes low (with a brushed DC motor as load). I would like to note that the oscillations were only observed on the drain before the diodes; after the diodes, the oscillations had been smoothed out.

I’m not sure I understood your question. What is a pure resistive load? Also, I’m not clear about ‘same amperage’ – is that amperage drawn from battery, or as drawn by the load (measured after diodes?)

In any case, here is what I done so far to attempt to answer your questions. I assumed that a CFL would qualify as a pure resistive load (as distinct, e.g. from a DC brushed motor.) (If this assumption is wrong – then disregard what follows, and please clarify). So I contrasted the CFL with the 24 VDC brushed motor.

CFL

I loaded a 40W CFL onto my circuit, and got it going just at the flicker point. At that point, I had 206 Hz with a 9.1% duty. Current drawn from battery was 0.34A. Current drawn by CFL (i.e. after the diodes), was 0.22A. Voltage across the CFL (i.e. after diodes) was 31V. I took some scope pictures.

First picture (see attached “CFL CH1 on output CH2 on drain before diode.JPG”): output and drain before diodes. I attached Channel 1 (CH1) probe to the output signal of the pulser (i.e. going to gate. Nb. In previous post I said ‘going to drain’, but that should have read, ‘going to gate’ or ‘going to drain via gate’, because the output signal of course triggers the gate). I attached CH2 probe to the drain before the 576 diode.

In the attached jpg file, we see the following. CH1 is on left y axis; and CH2 on right. Output signal goes high; drain voltage goes from 38V to zero. Then output signal goes low, and drain voltage spikes high over 200V; then oscillates around approx. 85V, dampening down over a period of about 50 usecs and staying there for a further 250usecs; then drops and oscillates around 38V, dampening down over a period of about 150 usecs; and then it stays pretty steady at 38V until next round. So as I will amplify below, there is one more set of oscillations than with the DC motor.

(By the way, these jpg files are snips of .wfm files on display in a wave reader).

Second picture (attached as “CFL CH1 on outut and CH2 on drain after diodes.JPG”): output and drain after diodes. CH1 is output signal of pulser. CH2 is drain after the diodes (576 and 4148). Output signal goes high; drain voltage drops from 38V to under 30 and trends towards 20V. Then output signal goes low, and drain voltage spikes over 200V; then oscillates around approx. 60V, dampening down over a period of about 50 usecs; then rises to about 70V; then oscillates very mildly on a downward trend to 38V.

In comparing before and after diode voltage, in general the latter is shifted down a bit, but does not have 0V and only has a first round of pronounced oscillations.

DC motor

As I was unsure about your remark concerning amperage, I got the motor going so that O.35A were being drawn from the battery. This is virtually the same as in the scenario above with the CFL. I hasten to add that the current drawn by the motor (i.e. after the diodes) was 0.08A – so this amperage is substantially different from the CFL scenario, where the current drawn by the CFL (after diodes) was 0.22A. I am not sure I could set up the motor so that both amperages are identical to the CFL. The output of the pulser was 94 Hz, with 7.5% duty. Voltage across motor, i.e. after diodes, was 70V.

Third picture (attached as “VDC motor CH1 on output CH2 drain before diodes.JPG”): output and drain before diodes. That is, this is the same setup as with the first picture, only now the circuit is powering a motor instead of a lamp. This has substantially the same profile as the pictures I posted last time. Signal at pulser output goes high; drain (before diodes) goes to 0V. Signal goes low; drain spikes to about 150V; then oscillates around and dampens down to 38V. In contrast to the CFL, there is only one set of oscillations – not two.

Fourth picture (attached as “VDC motor CH1 on output CH2 on drain after diodes.JPG”): output and drain after diodes. That is, same setup as with second picture, but with motor instead of lamp. Signal at pulser output goes high; drain (after diodes) drops to about 28V and trends down towards 20V. Signal at output goes low; drain spikes to 150V, and then smoothly slides down to 38V. In comparing before and after diode voltage, the latter does not go to zero, and after the spike falls smoothly back to 38V instead of oscillating around 38V.

In summary: presuming the CFL is an example of what you mean by pure resistive load, then:
• At drain before diodes: no, I cannot confirm that the oscillations do not occur. To the contrary, oscillations occur twice; after spike, around a high level of voltage, and then again around 38V.
• At drain after diodes: oscillations occur only once, after spike around a medium-high level of voltage; then there is a somewhat smooth arc falling to 38V.

In contrast, with the DC motor:
• At drain before diodes: after spike, oscillations around 38V and then settling.
• At drain after diodes: after spike, smooth fall to 38V.

I take it that implicit in your question is some concern that there would be oscillations for a purely resistive load. If that is correct, I would be grateful if you would explain what is your concern.

3. Switching time.

I can’t go back to find out the switching time when I burned the 4148s. However, I took one of the scope shots from above and increased the horizontal scale. So this is:

Fifth picture (attached as “CFL amplified horizontal axis.JPG”): This shows the CFL, with CH1 on pulser output going to gate, and CH2 on drain after diodes. As best I can see, the pulse signal going to the gate goes from high (approx. 10V) to low (0V) in about 30 usecs, and is a bit messy (oscillatory) on the way down. So 0.3V/usec. The drain activation – from low (20V) to peak of spike (220V) – takes less than 5usecs. So no slower than 40V/usec, and maybe even faster.

I’d be grateful if you would explain how the switching time relates to burning the 4148s.

(message continued in next post - I've hit my limit for attachments).

[ntc]

reply to John Stone - cont'd

Finally, I would like to note that all of my above measurements are on the common drain, i.e. the result of all four drains wired together. To check if I may be getting interference between them (i.e. they are discharging at different times), I scoped 2 different mosfet drains. That is, I put one scope (CH1) on the drain directly at one mosfet; and the other scope (CH2) on the drain directly at a different mosfet.

Sixth picture (attached as “two mosfets full cycle.JPG”): this shows the drain from pulse signal turning on then off. They seem very evenly matched.

Seventh picture (attached as “two mosfets spike and oscillations.JPG”): this is the same event but with an amplified focus on the spike and oscillations. Again, very evenly matched.

Which makes me think the mosfets are firing synchronously.

All best, ntc

[JohnStone]

Hi ntc,
thanks for your detailed report! You are a thorough worker and got nice results up to now.
I asked for presence oscillations and resistive load because FET stages can oscillate by themselves contributing heat to the system. Please accept that a FET is not a kind of ideal high speed relay but a complicated pet with certain bad habits. In order to check the quality of the stage it is necessary to check it step by step. In teh end we want to have fast switchin gin order to stirr radiant as much as possible.
I assume you have :

• all leads to gates same length and as short as possible
• GND from oscillator to FETs twisted with gate lines or at least close and parallel
• Supplied your generator separated i..e by a wall charger
• Connected source legs short distance, massive and to a central massive GND point being the reference for all measurements (brass or copper screw if possible).
• Connected batteries (-) to reference point referred above.

Different setup will give unreliable results and less proceeding. The hiints above and below relate to the fact that even a low frequnecy pulsing initiates vast RF effects and they are not to be neglected at our setup.

TEST:
You should disconnect for now all circuitry from drain. For every step below you shall take notes and / or scope shots.

• Disconnect all FETs at drain and gate but one single guy.
• Let the driver run and measure at gate. Check for oscillations and rise / fall time of gate voltage. The circuit of the driver suggests to have faster switch off and possibly with oscillations.
• Connect second FET and recheck like above. rise / fall time shall be somewhat slower.
• Connect all FETs and recheck like above. It might be true that the LM339 driver is not strong enough for 4 FETs as their gate contains a parasitic capacitance of about 1nF....6nF
• Connect a car head lamp as load between battery (+) anf drains and recheck like above. You should know that FETs tend to backfire to gate having an impact to rise and fall time. Possibly oscillations occure in this test only.

If you are pleased with results you can go further, knowing that your driver stage is OK. This OK state is essential in order to have reliable results later on. I want all of you to succeed but it is very difficult to find a bug in the setup if you do not check it step by step with increasing complexity. It is very essential to relay to solid circuitry in order to research the unknown.
And threfore I am very penetrating - intentionally

I am occupied just now to take same measurements like above at monster driver V5.1 and I will post results along comments. Having your lab notes available you are able to compare and find possible weak properties.
JS

[ntc]

for John Stone

Dear Mr Stone,

Thank you very much for your recommended testing procedures. I respect your careful step-by-step approach. I will do them. (I have a lot of work over the next few weeks, so it will be some time before I can complete and post the results).

Also, thank you for sharing your circuit design principles. I can see that I may need to do some reworking of the circuit. I have a question. What does it mean, ‘supplied your generator separated, i.e. by a wall charger’? At the moment, I have everything running off one power source (batteries). The coil gets the full 36V; while a regulator taps off 12V for the pulsing circuit. Is there something wrong with that setup?

Good luck with your monster driver, and I look forward to reading about your test results. Is there a schematic available that I might study?

All best, ntc

[prochiro]

Hello ntc
Yes there is. Both need separate power supply and the power to your boards circuit, if only 12 volts, ends up as at 10 or less to shoot to fets. This should be 15 to 18 volts so as to give fets a full 12 volts for hard slamming. Good work so far. There are a lot of details but you are chopping them up, one by one.
Dana

[JohnStone]

Hi ntc,
The circuit to study is in thisV5.1 PDF. I took Cornboy publicly with me and posted the sequence of building and testing. Later on I will add my posts to PDF in order to issue a final version. But before I will do thorough measurements in order to verifiy all features requested. Up to now I found the micrel driver to be amazingly fast. This night I got 5ns rise time at free run and +1ns with one gate connected. This suggests that switching losses will be neglecteble even at high current. and we will be able to drive more than 2 FETs at a time. That is good news!!!!

Regarding separate feeding the driver / oscillator.
GND:
Any circuit is not restricted to DC or low frequency but has all capabilties of RF. Usuall this is of no concern for blinking LEDS and similar actions. But our setups produces a vast amount of RF distubance and therefore we need to obey some precautions in order to get them below severe disturbance.
BTW: We never would be allowed to sell those circuits commercially because of regulations because they are wild RF transmitters.
One major intrusion of RF and artifacts rom high current is a bad GND system. Therefore I recommnd to have ONE central MASSIVE GND reference where you connect source legs of FETs, GND of motor GND of battery and GND of driver. ONE single point where the leads arrive in star shape.
In audio applications you get humming if you do not obey those rules. Apart that you have the guarantee that you have no displacement currents traveling any unkonwn trace on GND level causing voltage differnces.

You might consult this site additionally.

Supply:
similar but not so desasterous your power supply (+) behaves. It is bad habit if you share any PSU line of your oscillator / driver with high current traces. Therefore I suggest to use a simple cheap wall charger (or separated battery as separated PSU for oscillator, driver freeing it from any other part of pulluted areas of power circuit.

I suggested 15 Volt minimum in order to get 12V regulator performing well. But 18V or even 24V or 30V will be OK. Please check voltage regulators for heat at higher voltage. One issue to be measured will be if we get any advantage by increasing 12V to i.e. 15V at gate driver. I doubt it but I do not know just now.

As we investigate the unkown we need to eagerly build our setups in order to not generate multiple dirty artifacts thet cover our sweetspot we search for. Ic we can control the previously unkonwn we can check on how we can simplify i.e the supply circuits. The rule I to change only one parameter at a time and study its effects. But if you have a dirty setup you never will be able to study or control one singel parameter because you will have a feedback from power circuit and you will not suceed. Imagine the feedback of a speaker to the microphone... teh setup does well within certain limits und will runaway above a certain feedback rate. But at ouer setups we will not listen to feddbacks - we stand there blindly and disturbed.

If you rework your circuit you may post a pic in order to enable me to comment. And feel free to ask. There are no silly questions but answers. The major goal isfor you to learn in order to enable you performing well by yourself.
My concern ist to give all of you a solid basis in order to succeed.

EDIT: I just measured my Micrel driver with scope and found oscillations. I replaced the GND lead of the probe by a very short wire and most of them were gone. Please note that such a wire of some inches causees disturbance. Therefore keep all wires as short as possible!!!! The best smooting of signals is to prevent disturbing them.
JS

[JohnStone]

Today I retracted to my lab with a special brand of wine

Cheers Cornboy! Excuse please, I had to rotate the bottle about 120 degree regarding origin in order to align it to my glass!
and spend nice time with monster driver while taking thorough measurements.
Current and future results will be found at lab note LN006. Up to now I took all essential scope shot along explanations but FET not being loaded at drain pin. My procedure is to go step by step in order to not mix artifacts making me blind for causes.
Adding load will be somewhat more complex because load (especially load spikes) backfire to gates opposing driver actions. Another concern will be to prove protection circuit to act at strains as expected.
I hope you can understand my explanations - else ask!
Cheers!
JohnStone