Upon request from a fellow researcher, I set out to design what I hope is a useful pulsing circuit that anyone can use in their TPU research. It is by no means limited to TPU research, but it was designed with the TPU in mind.
See the document for details, but in brief the unit provides:
- 50% duty cycle square waves from ~0.4 Hz to about 13 MHz in 7 ranges.
- 3 independent outputs, OR 3 synchronized (harmonically) outputs with one control.
- Generators, drivers, and MOSFETs all in one circuit.
- Relatively cheap to build.
Comments welcome.
Enjoy,
Darren
UPDATE: 2007 AUG 04
I've completed the TP900 design with the added Pulse Width function. Pulse Width range is from 5ms to 17ns. Hope it meets everyone's expectations.
Cheers,
Darren
UPDATE: 2007 AUG 07
Revision update from TP900.1 to TP900.2 due to resistor value error on R10, R16, R22. Should be 33k. The TP600.1 design is unaffected and remains unchanged.
UPDATE: 2007 AUG 31
documents have been removed, pm me if interested
UPDATE: 2007 SEP 22
Due to lack of responses and feedback, the documents will no longer be available.
Hi Darren,
Thanx for putting this out here, well appreciated!
I'd like to ask if it is possible to adapt the design in a way that a much smaller duty cycle becomes possible. Lately a lot of ideas require a very small duty cycle. If that is possible it would make it even more flexible...
Regards
Robert
Hi Robert,
Thank you.
Yes, a more advanced design is already in the works...the TP900.
This one may have a higher frequency capability, with better phase matching on the synchronized outputs. Gate propagation delay becomes a significant factor when dealing with frequencies in the MHz range.
As a minimum, the TP900 will have a "programmable" (repeatable, accurate and matching pulse widths between outputs) pulse width output down to 5ns or so.
Regards,
Darren
Yes! Very good. This was a great idea. Keep it coming!
tHANKS Darren :)
i think we can switch in something like one shot timers to take the control over the duty cycle.
i will order the parts and get building.
Marco.
Quote from: z_p_e on July 24, 2007, 08:52:02 AM
Upon request from a fellow researcher, I set out to design what I hope is a useful pulsing circuit that anyone can use in their TPU research. It is by no means limited to TPU research, but it was designed with the TPU in mind.
See the document for details, but in brief the unit provides:
- 50% duty cycle square waves from ~0.4 Hz to about 13 MHz in 7 ranges.
- 3 independent outputs, OR 3 synchronized (harmonically) outputs with one control.
- Generators, drivers, and MOSFETs all in one circuit.
- Relatively cheap to build.
Comments welcome.
Enjoy,
Darren
Hi Darren
I like to buy one . Can you or anyone help with a prefinnished circuit?
pls.pm me your answer.
Thanx.
helmut
Quote from: -[marco]- on July 24, 2007, 12:33:59 PM
tHANKS Darren :)
i think we can switch in something like one shot timers to take the control over the duty cycle.
i will order the parts and get building.
Marco.
You're welcome Marco.
As you wanted something basic, I didn't design the one-shot into this version, but I had always intended on designing one for myself with variable pulse width included.
I would encourage you to wait for this next version rather than inserting something like a 555 one-shot...you'll be disappointed with that.
I have a very good pulse-width method in store for the TP900. I'll try to design it so that you can simply modify the existing TP600 circuit, provided you have left room for it on your board.
Regards,
Darren
okay
:)
indeed i was thinking 555 but as you said it won't do then i will look for something diffrent.
Marco.
@Darren
This is great.
I am going to wait for the TP900 and have it made locally by a small EE company.
I have a few simple questions that will help me clarify at my EE level.
1) Are the F1, F2 and F3 positive dc outputs that are pulsed on and off as per our chosen cycles per second.
1A) Is the applied pulsed voltage adjustable on the F1, F2 and F3?
2) If 1 is yes, do all three pulsed coils share the same negative?
Just as a maybe some day thing, here is a wish list;
3) If 2 is yes, can there be seperate negatives also pulsed?
4) If 3 is yes, can there be option switch for reverse polarity per every 2nd on pulse?
5) If 4 is yes, can there be a second level per channel for frequency generator pulsing on when the coil pulse is off, and off when the coil pulse is on?
6) If 5 is yes, can there be another input/output per channel for capacitor charge/discharge control?
I am too low level EE to do this myself. But I know how I would eventually like to pulse and control components and the above would be a very good work tool. Actually for me it would be a dream come true if such a circuit was available. I could have it made for me locally.
If you ever need me to make a block diagram of the full logic let me know and I will lay it out. Like I said, I'm crap in EE but top notch in logic systems.
@wattsup.
I'm not sure you have a good understanding of how the generator works, especially at the MOSFETs, so allow me to explain a little. I am going to answer your questions 1-2, and if after you still have further questions, I'll try my best to answer them.
The Drains of the MOSFETs are not outputs really. They are better described as INPUTS. Think of the MOSFET simply as a single-pole, single-throw switch. Whatever is on its DRAIN, will be switched to ground.
Normally for pulsing coils, one end of the coil is connected to V+, and the other end of the coil is connected to the MOSFET DRAIN. So the MOSFET is switching the coil to ground. "Ground" is common to the V+ ground, so they are all the same grounds.
1) So, the MOSFET's switch the coils at the rate chosen by the controls.
1A) The pulsing voltage driving the MOSFET gate is not adjustable.
2) All 3 coils share the same negative or "common".
Again, simply think of the MOSFET as a switch, nothing more. The "DRAIN" would be the input to this switch. The other end of the switch is grounded.
Hope that explained it well enough ;)
Cheers,
Darren
PULSE WIDTH QUESTION
@All,
What range of pulse widths would folks here like to have?
Here are some examples:
1) 100ns ~ 500ns in 100ns increments
2) 50ns ~ 250ns in 50ns increments
3) 20ns ~ 100ns in 20ns increments
4) 2ns ~ 510ns in 2ns increments (deluxe $$ implementation)
Perhaps longer than 500ns? ???
Thanks,
Darren
In reality guys,
We have limiting factors to consider.
The power MOSFET takes minimum 50ns from 10 ~ 90% to turn on, and about the same to turn off. Therefore to leave some margin and also give the coils some time to energize, I would suggest that the shortest pulse width be about 500ns. Does this make sense or am I off track here?
I really doubt SM went beyond 1 MHz with his frequencies, and so I propose that the longest pulse width be 1us, which will limit the operating frequency to 1 MHz.
One RANGE position can be eliminated, and so a 6-position rotary used instead of a 12-position.
Any thoughts?
Darren
Either way I would still want to build one, but I will say since this driver could be mulit use in our kind of research, that perhaps keeping as much of the upper range as possible would be worth while still. I'm mainly saying this because you can typically buy 10hz-10mhz pulse gen on ebay for 30-50 bucks.
Thanks for designing these great units! I'll be building the 900 model soon!
~Dingus Mungus
Hello all,
as we know the highest frequency for a 15" TPU is 245kHz maybe its enough that the pulser can reach 500 - 1MHz.
@Darren
veeeery nice job done. Im waiting for the 900 design and then I will build it.
Otto
hi Daren,
considering the lightning band app.246khz they have to be as fast/short as possible.
lightning itself does not last seconds, it's there and then its gone.
with transistors and the rlc comes the dirty harmonics.
this does not happen when using crystals.
"sorry, theze are not piezo stacks, however they look like it"
that could point out to perfect frequency using quartz, so pherhaps we can do an add on for inserting crystals? (Pierce Colpitts)
However these do have responses at approximately three times this fundamental frequency, five times, seven times, etc.
Those are for example the third overtone(5000Hz), fifth overtone(35000Hz), seventh overtone(254000Hz), etc.
Marco.
Quote from: z_p_e on July 25, 2007, 09:55:19 PM
PULSE WIDTH QUESTION
@All,
What range of pulse widths would folks here like to have?
Here are some examples:
1) 100ns ~ 500ns in 100ns increments
2) 50ns ~ 250ns in 50ns increments
3) 20ns ~ 100ns in 20ns increments
4) 2ns ~ 510ns in 2ns increments (deluxe $$ implementation)
Perhaps longer than 500ns? ???
Thanks,
Darren
Hi Darren,
Personally I think option 3 is the best. I understand what you're saying about the rise time of the MOSFETS but, at some point, we might find others that are faster than than the 50 ns. It would be nice that we could just replace the mosfets and don't need to change the rest of the circuit.
I suppose you mean to want to leave an option in there to use 50% duty aswell? Like a switch that can select 50% duty or a variable (very) small duty cycle.
Hope this helps....
Robert
@z_p_e
Thanks for your explanation. This is really not what I had in mind at all. With all due respect, in my opinion pulsing negative only is a dead end. Frequencies or not.
If you apply a current to a wire and stop it before it reaches the end of the wire, you will get the Tesla effect. You cannot do this if you are pulsing the negative. There is no inrush and you are just blocking the BEMF. The BEMF is faster than the electricity flow so before the flow even stops, if the BEMF knows where the negative is, it will know where the positive is and recede back into the positive before the current actually stops. The field is faster than the current. Maybe 1000 times faster, we do not know.
This is what I am thinking is happening. When you get BEMF, the BEMF is blind and has two options. Either it can touch the negative (like a blind man using his cane to tap the ground) to then know where the positive is, into which it will recede, or, if it cannot touch the negative, it just stays put, does nothing and creates heat. This second option or action on a pair of wires is nothing, produces nothing and is analagous to the current ECD.
Now, please understand that this is my view, the view of an EE-Neophite (EEN) but because of my EENism, I see the circuits more as living circulations then as static components. This also stems from my 20 years experience in water treatment technologies working with complex systems and processes. And from what I can see even in Otto's ECD is a stuck negative, everything is trying to go through this open and closed door while you are pushing voltage forward continuously. There is no play. There has to be some play, a place for expansion and contraction for a device to live on its own. A place where energy can be conserved and not wasted.
z_p_e there are some great thinkers and doers on this board that do not all have the EE acumen but have their insights that are not masked over by such EE knowledge. Most of the time you guys are talking tech shop, we are simply lost if the shuffle so many of us EEN's will just sit back and wait it out. Because our rebuttals may seem too simplistic, so in many instances, it is better to say nothing. But, if you have to think out of the EE box, sometimes it's is an advantage to already be out of it.
So I feel what we are doing is working backwards. Most are trying to get the big flash, major output, when we have not even see one watt, grow to 2 watts, we are working trying to get 250 watts and more. If you can just get 1 extra watt and do this 250 times, there's your 250 watts. How can a whale scratch a living off of tiny krill.
In my view Otto already had a good start of an answer with his 50 turns around the finger and ONE wire going through. What if you had 50 of those in parrallel working off a simple little spark. Sparks are easy to make and very powerfull. Work out the small and you'll find the big.
This is why I asked my questions. To work out an OU unit, you will require the flexibilities given in my questions and more in order to have a fighting chance. Otherwise, with just negative pulsing, you will be wasting precious time. If you want to practise your hunting, it is best to have a good variety of traps. If you only have one trap and it is the wrong one to begin with, then you are wasting time. Same goes for fishing if you only have one lure, good luck.
I think what I may have to do to get my idea across is to make a block diagram showing the logic. I think alot of you guys are not on the right track with negative pulsing. You need the negative solid for BEMF, or you have to have a complimentary switching to send the BEMF to a capacitor when the negative is open, otherwise you're wasting juice and producing heat.
Now again, maybe I fail to see the logic of this and I would appreciate if you could explain the benefits of negative pulsing to me in laymans terms with an example of a block diagram or simplified circuit. I just don't see it. Also, I hope I did not put this in a bad way as I do respect your abilities tremendously and know your views are essential.
Thanks for all the good feedback guys, I appreciate it.
I am still digesting it however, and have been brainstorming all morning to come up with a way to make everyone as happy as possible in terms of frequency range and pulse width capability. So far it looks good ;)
Regards,
Darren
wattsup,
I must admit that I am better at understanding technical descriptions as opposed to figuring out what someone means using laymen's terms.
I'm not following what you mean by "pulsing the negative", and I'm also not sure why you want the bemf, or why you feel it is "faster" than electricity.
If you want the bemf, no problem...just eliminate the freewheel diode across the MOSFET. Just be sure that you DO SOMETHING with the bemf however, or it may take out the MOSFETs with HV spikes.
So if you could describe what you mean in a little more technical terms, perhaps I can see where you are coming from and hopefully explain the differences (if there are any) to what the circuit is already doing. A block diagram if you wish may be helpful.
No offence taken wattsup. I understand that you may have a different point of view in what's going on and what needs to happen, and that's ok. But I need to understand that point of view before I can comment.
What is "negative pulsing"? What is "positive pulsing"? What is the difference in your opinion?
Cheers,
Darren
in my opinion, the bemf is working against us, meaning it's a pulsed flow in the other direction as the one we want our magnetic field to go.
therefore we need to elimminate it resulting in a magnetic field which goes one way only so we can speed it up fast.
Darren, is there a way to add crystals to the circuit?
Marco.
Just a suggestion:
Instead of using the freewheel diode to direct the BEMF past the MOSFET to ground or negative or whatever - connect it so it redirects the BEMF to a counter-wound coil so the energy - as little as there may be - is used to promote the process.
You protect your MOSFET, create a faster fall-time for the coil generating the BEMF and make use of the energy.
Am I on the wrong track ZPE?
Hi ZPE
Here are the features I would like in a generic lab pulse controller:
Controller
AD9959 Quad DDS
PIC (or ARM) controller - programm in place capability.
USB control option would be great for people wanting to drive it from their computers.
Stand alone unit like -- http://www.minikits.com.au/sweep.htm
Pulse width control down to 10ns (the DDS's only output square and sine)
Output board
Opto isolated
Selectable Monostable mode for fastest pulse width for any given mosfet -- perhaps a triggered blocking oscillator design.
There are great benifits in keeping the controller and output stages separate...
I expect that the mosfet and diode selections will vary quite a bit depending on the project -- and will get blown up on occasion! I would be looking at a range of output stages including valve output stage.
The AD9959 is able to potput dar faster and more precisely timed signal then we require. But here are the specs I would anticipate using in the first instance:
30Khz - 1Mhz range
0.1Hz control
Slow sweep capability.
Fast frequency modulation on any output. - for sread band rather then off peak tuning.
Pulse width -- fastest possible by output device.
A note on mosfet rise times -- we arent driving current, we just want voltage spikes. At low current the rise time for the mosfet will be much higher than the full load figure -- so a 4V/ns device may in fact deliver 100+v/ns in the mode we are using it. This is why a triggered monostable output mode would be good -- to get the fastast pulse possible given output device and load.
cheers
mark.
Quote from: BEP on July 26, 2007, 08:22:46 PM
Just a suggestion:
Instead of using the freewheel diode to direct the BEMF past the MOSFET to ground or negative or whatever - connect it so it redirects the BEMF to a counter-wound coil so the energy - as little as there may be - is used to promote the process.
You protect your MOSFET, create a faster fall-time for the coil generating the BEMF and make use of the energy.
Am I on the wrong track ZPE?
BEP,
This is an excellent idea and certainly an option for anyone to try.
It's just a question whether it will promote the process or detract from it. Also, shunting the bemf through the original coil (as shown) may be promoting or detracting from the process as well. So one should perhaps try both ways.
At any rate, SOMETHING has to be done with the bemf. It could even be used to charge a capacitor (eventually...these spikes are likely VERY short indeed).
Regards,
Darren
Hi Mark.
Thank you for the AD DDS chip info. I'm certainly interested in these, and have been since first seeing them a few months ago.
I would definately like to develop a generator based on one of these chips, but for now, the TP600/900 will do well for many folks, and for many applications.
Perhaps I will work with Peterae in developing one of these DDS generators soon.
Cheers,
Darren
Quote from: -[marco]- on July 26, 2007, 01:55:47 PM
in my opinion, the bemf is working against us, meaning it's a pulsed flow in the other direction as the one we want our magnetic field to go.
therefore we need to elimminate it resulting in a magnetic field which goes one way only so we can speed it up fast.
Darren, is there a way to add crystals to the circuit?
Marco.
Marco,
A crystal could be used, but not sure why?
Using a crystal eliminates, or at least severely limits frequency adjustment unless dividers are use. With dividers comes resolution limits too.
Crystals provide stability, but are inherently difficult to "adjust". One solution could be to use a PLL with a crystal reference osc., but again the frequency range and adjustment is limited. Also, this is becoming quite a different animal than the original intent of the circuit. I would sooner go the DDS route than use crystals and PLL's.
So, in the end a DDS unit might be the best. However, don't give up hope for the TP900. It will have a good frequency range and PW adjustment.
Regards,
Darren
PS. I hope to post the TP900 circuit very soon.
the bemf can be extracted by using diodes and then fed/redirect into the same bifilar coil only the other winding, so it cancels itself out ,i have already tried that and you can also capture the bemf in capacitors etc..
Marco.
Hi Darren
I see you are doing a Stirling job over here. ;)
The DDS would be very nice but inherently the cost of building would climb and take some time to get the software up and running, until we have a clearer vision of the exact function of the control we require I think it would be best to keep design time and build costs down to a minimum.
The TP600 should be able to keep most experiments on track and using CMOS it's difficult to get much cheaper.
I am a bit of a module fan myself and tent to build modules that plug together to create the required circuit.
Your above circuit, I would build each oscillator as one small board that plugs into the sync board which plugs into the fet driver and fet output board, that way if the oscillator or the sync board needs to be modified and redesigned for other purposes we just need to rebuild a small board instead of the whole board.
Also using photo etching to make the PCB's would mean you only need to cad the oscillator once and build 3 times as a module.
This method has been very handy with my own board as I can unplug a pic processor and plug a 555 timer based oscillator directly in place, and now thanks to your great simple oscillator circuit I can plug one of those in for a complete set of frequency ranges when I get round to cad-ing it up to a PCB.
Peter
@Darren
good work, keep it up.
I believe when SM refered to a pure frequency, he meant low jitter. That means a quartz crystal reference somewhere. Whether PLL, dividers, DDS doesn't probably matter so much as long as jitter is low.
This rules out R/C oscillators of any kind, should jitter be of extreme importance.
Low jitter also means short leads, lots of low-inductance by-pass capacitors, etc.
GHz layout, not wires all over the table.
@All,
take a look at my latest ideas at:
Fastet, narrowest pulser possible
http://www.overunity.com/index.php/topic,2582.msg41929.html#msg41929
Rat Race version 2
http://www.overunity.com/index.php/topic,2582.msg41932.html#msg41932
Rat Race version 1
http://www.overunity.com/index.php/topic,2582.msg38506.html#msg38506
Regards, Earl
the transistor setup will generate the unwanted harmonics all over the place, that is why the tubes work much better and crystals do not generate these unwanted harmonics they are stable the only thing which alters their frequency is, temperature.
when using transistors we may also need to use pre amp like tools to reach proper feedback like Steven said ,tubes are much diffrent.
M.
To answer a PM request I offer the following:
@Earl,
Crystal oscillators are not exempt from frequency jitter at all. In fact there are many schemes designed around dealing soley with this issue.
RC oscillators may exhibit frequency jitter as well, but not likely worse than its crystal counterpart. Crystals excel mainly in stability over temperature mainly due to error reduction through frequency division.
At any rate, the TP600/900 is meant mainly as an experimental tool, and should not be worse in performance than the tube oscillators that SM started off using with his testing.
Darren
well there seems to be a diffrence in the harmonic content, it is written in an verry intresting old book i am currently reading (1966) i will examine it somewhat further.
Marco.
"I am using 6BQ7-A tubes for the input and phase inverter because they are VHF amplifier triodes designed to operate in Color TV at very high frequencies and so you can imagine how crystal clear my high frequencies are in my stereo amplifier."
PULSE WIDTH CONTROL QUESTION:
What would you guys prefer for the fine pulse width adjustment:
a) a 6 position rotary for fixed stepping (good for repeatability and matching) OR
b) a potentiometer to vary the PW continuously (good for fine adjustment, but will have to rely on control position for approximate repeatability and matching)
?
In either case, there will be a rotary for selecting the Pulse Width RANGE, similar to the Frequency RANGE control. So, two controls for Pulse Width; one is for RANGE, the other for fine adjustment in that range.
Let me know your thoughts, thanks.
Darren
Darren, ho abou both and a switch between to select either one of them?
then we can also easy measure the value of the variable one so when we find something we can write it down.
it has to be as percise as possible when it comes to tunning.
Marco.
Quote from: -[marco]- on July 28, 2007, 03:46:49 AM
Darren, ho abou both and a switch between to select either one of them?
then we can also easy measure the value of the variable one so when we find something we can write it down.
it has to be as percise as possible when it comes to tunning.
Marco.
Marco,
Having both can certainly be done, but I'm not sure how practical that is. Please keep in mind that the pulse width will not be that critical. However the frequencies and matching between them in "sync mode",
will be critical. As such, I am surprised no one has asked for a "fine" frequency adjust control yet.
Because the pulse width is not that ciritical, I think it best the control be a continuously-variable one, i.e. a potentiometer. Then simply noting the control's position on the panel will allow you to repeat the setting and roughly match it between generators.
For the most part, the
Pulse Width has nothing to do with tuning. That's done by altering the frequency. Once the frequency is set, the pulse width can be adjusted for a minimum effective setting. Once set, the Pulse Width control can be left alone, while the frequency control can be adjusted for further tuning.
Cheers,
Darren
@z_p_e
Thanks again for your trying to understand my laymens terms. By pulsing negative or positive, I simply mean the opening and closing of negative or positive. ON and OFF.
I think it will be much simpler for me to contract a local small EE company that has recently made for me some special DC to DC converters.
I think also that we need new thread of Glossary of OU Terms where Post #1 can hold the growing list and the thread can be used for consensus building amongst members. This way we can have one place to refer to for terms to use on the board. I would not have time to be the Master of Cerimonies so if anyone has the guts, be my guest. Maybe open the first 4 posts and name them A - F, G - L, M - R, S - Z.
Question: What is a peep or peeps? lol
Quote from: wattsup on July 28, 2007, 01:12:11 PM
@z_p_e
Question: What is a peep or peeps? lol
;D Peeps = People
thaelin
OK folks,
The TP900 is finally done! :)
I estimate that I've put in at least 60 hours in designing the pulse circuitry alone, and I've been at this 16 hours already just today. My eyes are goin' nutty already :o
I'm quite happy with what I came up with, and perhaps even a little proud of my acomplishment. Its not an HP instrument, but HP never made one of these, that's for sure!
I managed to retain the frequency range of the TP600, and implemented a precision pulse generator circuit to boot. I hope a 5ms to 17ns pulse width range will please those that are interested.
The Pulse Generator circuit turned out to be a big project all on its own. It was a challenge, but I enjoyed it.
See the first post, on the first page here for the TP900.1.pdf document. I hope you can appreciate the effort that went into the design....enjoy.
http://www.overunity.com/index.php/topic,2831.msg41474.html#msg41474
Regards,
Darren
Hi Darren,
Thank you verry much for the circuit :)
as you might have noticed i also jumped into the tubes.
i decided to take the advized route first tubes and then solid state.
yesterday i noticed the importance of perfect sine waves and the question arizes if pulses can be used instead.
there is still a lot we need to find out.
Marco.
My apologies guys.
Sometimes there is an error or two in the first release (especially with no checker).
I guess I must have done a copy and paste with a resistor then forgot to change its value. R10, R16, and R22 have been corrected to 33k.
I have updated the document to TP900.2.pdf available again here:
http://www.overunity.com/index.php/topic,2831.msg41474.html#msg41474
Regards,
Darren