I'm starting this thread to document my build of a 3 channel pure SIN generator with the following
specs:
==============================
3 Channels (hopefully phase locked).
0 ... 60MHz with 1Hz resolution.
4V p2p into 50 Ohm.
Microcontroller controlled.
3D sweeping capability (sweep width and resolution adjustable on each channel)
When sweeping samples analog input signal in order to hunt for optimum power out. 8bit sampling taken through digital adjustable attenuator (16bit) so that clipping is avoided and true maxima is found. This effectively increases sampling resolution to 24bits.
Ability to log scan results into PC over USB cable for analysis on PC in excel etc.
==============================
I already have a 2nd hand tektronix FG504 sig genny which is great and handy but the output, like most bench gennys, is like a wailing drunk karaoke singer when compared to a true radio quality signal required for experiments in intermodulation.
The hard work I will buy premade in the form of 3 x DDS60 modules, designed and used by amateur radio enthusiasts. The design has evolved over quite a few years and is very stable:
http://www.amqrp.org/kits/dds60/index.html
The modules will be under the control of an ATMega Microcontroller (under the screen in the photo).
I am thinking about and working on the user interface at the moment. The user can either use the small joystick switch you can see in the photo just to the right of the screen, or they can attach a rotary shaft encoder and spin a wheel to intuitively sweep through frequencies having first selected the channel and decade to adjust.
I'll post more as the build and programming progresses.
I posted in TPU section because 3 chan sin is often spoken of in theories of operation. But I would prefer to post in an "Equipment & Measurement" section. That section could include reviews of existing equipment, which would attract lots of search hits, measurement techniques and other tips on how to build and use stuff or how to improvise etc.
Yucca.
Does the TPU require AC or a sine waveform .
At higher frequencies the square wave will be rounded off a bit , a vacuum tube looks like it rounds off the edges also .
I think you are overcomplicating the task at hand , controlling the "phase" of each signal will be hard with a sine wave generator . It can be done more easily with logic lvl pulsing . Then amplified to be smooter .
Well, I'm commited now; Just purchased the 3 x DDS60 modules. Will now have to explain to my better half exactly what the hell they are lol!
Quote from: dankie on September 23, 2009, 01:55:16 PM
Does the TPU require AC or a sine waveform .
I dunno, better ask SM. As you know pure variable sin is not easy to do, variable square is easy.
What do you mean AC or SIN? pure AC is pure SIN.
Quote from: dankie on September 23, 2009, 01:55:16 PM
At higher frequencies the square wave will be rounded off a bit , a vacuum tube looks like it rounds off the edges also .
Rounded squares aren't pure freq, square waves are composed of odd integer harmonics of the fundamental. In experiments regarding intermodulation this is tantamount to cluttering noise.
It is easy to turn square into sin by using a lowpass filter set to roloff at fundamental but every time you change freq you have to alter component values. This genny will give very pure sin at any freq rock solid at 1Hz resolution. A nice bit of kit however you look at things.
Quote from: dankie on September 23, 2009, 01:55:16 PM
I think you are overcomplicating the task at hand , controlling the "phase" of each signal will be hard with a sine wave generator . It can be done more easily with logic lvl pulsing . Then amplified to be smooter .
As I said I would prefer to post in an eqpuipment section. I want a 2 chan pure sin gen anyway for stuff other than TPU research, I just thought I'd add a 3rd chan to explore Bruces ideas. The equipment will not be solely for TPU research. I'm just thinking cover lots of combinations of various RF interference phenomena in various coil configs and see if I spot any anomolies worthy of further investigation. I'm just figuring it'll please me more than watching telly because I enjoy programming micros and building things. To buy a piece of equipment with these specs would cost a few grand.
With regard to phase the chips are digital (they are a complex logic circuit with a final DAC stage) and all driven by xtal clocks, these clocks can be phase locked and so the derived frequencies can also be phase locked and adjusted with some effort in interlinking and programming.
Go for it Yucca
A great bit of kit for the bench, and future TPU exploration work.
Not sure if the feedback idea will work , as SM has mentioned you can expect to find a hole in your bench if you find the correct frequency's, might be best a put it in a flame proof box ;D
Quote from: Peterae on September 23, 2009, 02:18:32 PM
Go for it Yucca
A great bit of kit for the bench, and future TPU exploration work.
Not sure if the feedback idea will work , as SM has mentioned you can expect to find a hole in your bench if you find the correct frequency's, might be best a put it in a flame proof box ;D
Cheers Peter,
I'm putting opto isolator in feedback path. But you are right, perhaps if the "worst case" intermod combination is found in the correctly wound device then my unit may be zapped along with the rest of my bench!
p.s dont underestimate the Smith rig you are building, this afternoon I lost a 1gig memory card that was sat on the bench about 50cm away. It is now unreadable and unformattable. I was putting less than 5W into HV module.
Sorry to hear that Yucca.
I am setting up a bench in an underground disused room i have access to, to try and isolate any modern technology as anywhere else is packed full of the stuff and to try to stop radiation of the noise generated.
A good ground should be easy as well, just hope it doesnt stop the OU from showing up.
To think if Tesla powered up his giant Tesla these days he would probably knock every computer and cell phone out for miles LOL
@Yucca,
Over the past few years the 'hardware' oriented cells in my brain haven't had much stimulation.
THIS has them all buzzing again! Yes, I know. The parts aren't that hot but the combination, possibilities and the idea..... ;D
TPU or not I could find several uses for such a beast. This makes me yearn for the familiar smell of hot solder joints.
I write WindBlows/*Nix software (real coding - not 'modern' software development) for relaxation.
Please keep us informed how it goes. It sounds like you don't need help with software. If you do, send me a list of wants & needs. I'll be happy to look into it!
Quote from: BEP on September 23, 2009, 10:43:15 PM
@Yucca,
Over the past few years the 'hardware' oriented cells in my brain haven't had much stimulation.
THIS has them all buzzing again! Yes, I know. The parts aren't that hot but the combination, possibilities and the idea..... ;D
TPU or not I could find several uses for such a beast. This makes me yearn for the familiar smell of hot solder joints.
I write WindBlows/*Nix software (real coding - not 'modern' software development) for relaxation.
Please keep us informed how it goes. It sounds like you don't need help with software. If you do, send me a list of wants & needs. I'll be happy to look into it!
Thanks Bep, when it comes to writing a desktop computer interface (to log and maybe display scan data etc) then your expertise could be good.
Most of my time spent on this project will be software. The hardware is all pretty much plug and play modules. Of course I´ll have to fire up the soldering iron a little and also make some enclosure for it.
Stay tuned as I develop I´ll post more, probably post a short vid demoing the user interfaces main screen soon.
Quote from: Loner on September 23, 2009, 10:53:54 PM
Yucca, All I can say is "NICE!" .
(The mentioned specs are excellent! Should do the job, no matter the job.)
Cheers Loner, the tight specs are mainly thanks to the IC manufacturer and the guys who designed and build the DDS modules, awesome technology made available to hobby level experimenters!
Quote from: Yucca on September 23, 2009, 11:00:25 PM
Thanks Bep, when it comes to writing a desktop computer interface (to log and maybe display scan data etc) then your expertise could be good.
Most of my time spent on this project will be software. The hardware is all pretty much plug and play modules. Of course I´ll have to fire up the soldering iron a little and also make some enclosure for it.
Stay tuned as I develop I´ll post more, probably post a short vid demoing the user interfaces main screen soon.
If required, I can hack or create PC drivers as well :)
Hello all,
@Yucca
very nice "monster" but ... it will NOT make the coils "happy".
Otto
Quote from: otto on September 24, 2009, 12:24:06 AM
Hello all,
@Yucca
very nice "monster" but ... it will NOT make the coils "happy".
Otto
Hi Otto, why not? are you thinking square waves are needed? or maybe PWM?
If so then I could feed the outputs through digital circuitry and FETs in order to get square waves and maybe PWM from 10...90% by using comparator. My frequency in that scheme would be limited to 10...20MHz because of FET and driver speed. Of course I would still have up to 3D sweeping capability.
It will be nice for me to have a true calibrated signal source and I can make use of 1channel for lots of other things, I decided to do 3 channel because the modules are not that expensive considering the spec.
Yucca.
Quote from: BEP on September 23, 2009, 11:30:09 PM
If required, I can hack or create PC drivers as well :)
Now that might be very handy! Thanks for the offer BEP! 8)
The micro board has a FT232RL USB to Serial UART chip on it:
http://www.ftdichip.com/Documents/DataSheets/DS_FT232R_V204.pdf
Hi Yucca
I think you are going to have trouble syncing them together, this is from the data sheet and it specifys that a master clock should feed all 3 DDS chips.
Note: For those who may be operating multiple AD9851 evalu
a tion boards from one computer, a MANUAL FREQUENCY
UPDATE option exists. By eliminating the automatic issuance
of an FQ_UD, the user can load the 40-bit input registers of
multiple AD9851s without transferring that data to the internal
accumulators. When all input registers are loaded, a single
FREQUENCY UPDATE pulse can be issued to all AD9851s.
A block diagram of this technique is shown in the AD9851 data
sheet as a quadrature oscillator application. This single pulse
synchronizes all the units so that their particular phases and
frequencies take effect simultaneously.
Proper synchronization
requires that each AD9851 be clocked by the same reference
clock source and that each oscillator be in an identical state while
being programmed. RESET command ensures identical states.
When manual frequency update is selected, a new box labeled
FREQUENCY UPDATE will appear just above the frequency
sweeping menu. Clicking the box initiates a single FQ_UD pulse.
The good news is that you should be able to remove all three clock oscillators from each DDS board and connect a small board to all 3 clock pins on each DDS board as they have provided plated through the holes for pin osc chips as well as the used SM osc modules.
For true phase syncing you will need the track length the same to each DDS chip.
You will need to issue a frequency update to all DDS chips at the same time as said above.
Hi Peter,
I knew about the synching pulse supplied by micro, but the hardware clocks are a bit of a pain as you point out.
I wonder if micro coax transmission lines could be cut to length to get exact input clock phases from one shared clock?
I know it'll probably just result in chaos but before I remove the clocks and make a shared clock I will try (probably in vain) to tie the 3 seperate clocks together somehow.
I've just been writing hibernation procedure in the GUI. Across the 10V input supply I have a big electrolytic. I sample this electrolytic through a pot, tuned so its output is just above logic 1 level. As soon as the cap sags I have a few hundred ms to save the complete system state to EEPROM and shut down, I want to make a routine to write in different EEPROM positions each time so I don't wear out the first EEPROM positions.
The TPU is in the khz range .
You dont need high Mhz speeds .The 74HC series can go up to 20 mhz . Why not just go with some 555 regular square waves for such a simple task ? You are bringing a 100 pound sledgehammer for this one little nail sticking out ...
It will be alot less complicated and expensive to start from scratch and make this with cheap IC components . The most expensive IC is 2$ , most are like 50 cents . You can sim most of it with a free sim , but theres always a limit to what you can sim with free unregistered .
Bringing a PC into this is not necessary .
And everybody experimenting on a tpu needs an oldschool analog scope , any old piece of crap is way good enough , not computer scopes .
Quote from: dankie on September 24, 2009, 07:00:35 PM
The TPU is in the khz range .
You dont need high Mhz speeds .The 74HC series can go up to 20 mhz
It will be alot less complicated and expensive to start from scratch and make this with cheap IC components . The most expensive IC is 2$ , most are like 50 cents . You can sim most of it with a free sim , but theres always a limit to what you can sim with fre unregistered .
I'm not building the unit solely for TPU research. I want to experiment with RF intermodulation in small coils amongst other things. I understand this unit is not for everybody, many will see it as folly and so be it, I will still build it.
I have a full complement of 74 chips, hundreds of them, probably enough to build a small CPU. I know 74HC are fast logic, but you would only have square output. Also to get rock solid freq you would need to drive 74 from xtal clock and it would only be easy to divide that clock by 2,3,4 etc using n bit ripple counter. To design an accurate square sig source even to 500kHz capable of 1Hz resolution would be near impossible with 74.
I can guarantee to design a clean sin generator with 1Hz resolution even up to 1MHz would cost alot of time and expense and to get all harmonic content 40dB under fundamental as these DDS are capable of would not be an easy task.
Well I happen to have much experience with this , There is much much effort to be done to make a good sine wave , the wave is a stepped wave and needs post filtering .
A 1 mhz pure "sine wave" is not possible to do , you should forget that idea , a word of advice , stick with those square waves ,
A very fast clock is indeed necessary but that clock must be "variable" haha ! , good luck with finding a 100 mhz simple square wave oscillator thats not too exotic , something simple ...
Quote from: dankie on September 24, 2009, 07:31:28 PM
Well I happen to have much experience with this , There is much much effort to be done to make a good sine wave , the wave is a stepped wave and needs post filtering .
A 1 mhz pure "sine wave" is not possible to do , you should forget that idea , a word of advice , stick with those square waves ,
A very fast clock is indeed necessary but that clock must be "variable" haha ! , good luck with finding a 100 mhz simple square wave oscillator thats not too exotic , something simple ...
Using DDS a "pure" sin wave is quite possible. DDS is a digital synthesizer producing a 10 bit linear approximation of a sin. So yes it contains 2^10 = 1024 discreet steps vertically but these are so fine that the output elliptical filter can pretty much take them all out. As I say all harmonic content guaranteed to be under 40dB relative to fundamental.
To test the stability of an oscillator it is best to listen to a heterodyne signal against another accurate oscillator. I have a DDS based small Yaesu VR-500 receiver 200kHz...1GHz all mode all band, and when I tune it on sideband to say 6MHz and then make an oscillator with an LC and listen to it, it is all over the place, very unstable and muddy and wide, I wave my hand near it and it zips away to another freq, I change the load impedance and it again loses freq. DDS would produce a rock solid tone with very narrowband regardless of conditions.
Don't get me wrong I'm not saying this is the perfect TPU tool. It's what it is, a 3 channel sin source with spectral purity as good as can be achieved and 3D scanning capability, I plan to keep it for life and use it in many experiments to gain a better understanding of frequency mixing in various systems.
You may well be right, square waves may be the way to go with TPU research, who's to say?
Here's a photo of the approx layout of the device on a verocard.
At the back is two 5V 1A PSUs, serial linked for 10V 1A.
Down the left is two caps, these give the unit a little time after power is unplugged, in that time it can easily save whole system state to permanent memory.
Behind the microcontroller and screen you can see 3 x DIL16 sockets, These I will plug the DDS modules into (using only one row in each socket). If I have to have the modules horizontal to accomodate a clock buscard then I will have to use 90DEG sil pins and do that.
The micro itself will have SIL headers soldered at 90deg to the board and it will also plug into the vero card.
The whole card will go into some box on standoffs with a cutout for the screen. Jogwheel on the right and output BNC connectors on the left.
Input sample conditioning circuit will be built on the front left of the verocard.
Hello all,
@Yucca
it doesnt matter if you use square waves or sine waves: A TPU will convert squares in sines. Automatically. This automatically means that "mother earth" is doing it.
A verocard!!!
Now I will tell you something.
A oscillator builded on a verocard is something like ..... its worthless!!!
I have 1 of my 3 oscillators years ago builded on such a verocard. Better NOT to use such an oscillator to pulse coils.
Yes, you get the signals, but they are not worth a cent.
And build the oscillator as simple as possible. @dankie said the same? Yes, he is right.
With such a simple oscillator maybe you can discover anomalies in your coils that will blow you from your chair and your coils will be soooo "happy" you cant imagine. Coils are "happy" when they are blown. And then your happy because then you know that you have SOMETHING.
Otto
otto
The oscillators are not built on vero they are prebuilt professionally, he is only using vero for the digital side of things, he's trying to make it so an ameature can build it without having to buy or make their own pcb's.
If you goto page 1 you can see the link for buying the DDS modules.
and yes the DDS module use a filter on the output to make the stepped DDS output pure sine.
@Yucca
i am not famillar with the atmel processor, but with the pic the EEPROM can be written to many millions of time without failure, so over use is unlikely, you could do a write verify read and if an error occurs switch to bank 2.
Again on the pic you have a built in function called brownout, which monitors the voltage rail and at a preset voltage level adjustable with some pics, you get an interrupt to save your code.
Another scheme i have used before is to save 1 byte of your parameters every second if it's different to your eeprom, this way the eeprom is gradually updated when the power is on, and then do a final check as a power down situation occurs, this way you can have little or no eeprom to write during a power down situation.
Peter
KISS
Hello all,
again and as mentioned, if you want to work on coils like an amateur without any successes you can use profi builded oscillators.
But if you are working on coils to discover "something" than you have to build oscillators with PCBs and then you may have something good.
Ever thought about capacitances in oscillators? Between every single part in an oscillator?? Capacitances in picofarads.
O yes, they are not important. But maybe something weird could "jump out" from such not important capacitances.
It was just a thought so forget it.
Otto
Hello all,
something very important for me. As I cant use PMs I have to do it in this way:
@tao
years ago you posted to discharge caps into the TPU or something like this. We were told that this is SMs secret.
Thanks for this.
Otto
Quote from: MasterPlaster on September 25, 2009, 05:41:08 AM
KISS
I agree. That's why I'm using off the shelf modules and plugging them together like lego, what could be simpler! Why reinvent the wheel when a group of radio HAMs have evolved a really nice oscillator module already over several years, you can get each module as a kit for only $30 and then you get a free sample chip. Or you can buy the modules preassembled.
@Peter, the EEPROM has 100k write cycles guaranteed, so less hardy than pic. I suppose If I just dump state to beginning blocks then it should last a lifetime if it only writes on powerdown. My caps are a whopping 1F and it has a few seconds on powerdown because the micro has a 5V reg so the 10V supply can afford to drop to 6V.
@Otto,
I am not an expert on the TPU, definitely not! You may be right, no it may not be needed for working TPU but it will help me learn more about radio like systems.
I plan to look for weird effects in lots of systems not just the TPU. I think 3 channels is overkill, I was going to do just 2, but its always good to have a spare ;D
So now you know I am only building a 3 channel radio quality oscillator. I am spending 3 months hobby budget on this thing but I hope to have it for many years.
Yucca.
Yes! This the way to get the skin effect before the conductor conducts!
Now what if you have a bias field around a conductor and then shock that? Canon balls, my friends, canon balls. Directional travelling shock waves.
Grumpy posted
Ehrenhaft.
http://www.rexresearch.com/ehrenhaf/ehrenhaf.htm (http://www.rexresearch.com/ehrenhaf/ehrenhaf.htm)
I noticed this diagram and related it to the TPU. I can't be that far off.
Quote from: otto on September 25, 2009, 06:19:04 AM
Hello all,
something very important for me. As I cant use PMs I have to do it in this way:
@tao
years ago you posted to discharge caps into the TPU or something like this. We were told that this is SMs secret.
Thanks for this.
Otto
@all (including all miserable old farts): I've just increased the GUI resolution to 100mHz. The oscillator modules have an incredible 24mHz resolution but one tenth Hz seems good enough. I had to reduce the digit widths by 2 pixels but it is still easily readable up to 1m away.
I've been in correspondence with the module designers, they assure me I should have no problems by sharing a common clock using same length runs of RG174 coax. I may also have to have measured lines to the load signal inputs. But I will try on the same bus to start with as only 1cm length between modules. If I cant assert within refclock*6 on all lines then I'll also have to use same length coax to route that signal also.
Also just picked up a bucketload of SIL pin plugs and sockets from ebay china for peanuts, shame about the 3 week wait, but no problem because I have a lot of software to write.
I had a worry today because my LCD kept crashing after many fast writes to it, I spent ages looking for a code error and now I realise that when powered by USB cable the power getting to the screen is marginal. When I power 10V through the regulator everything is rock solid.
Any ideas about menu fuctions would be welcome before I start to code, at the moment I'm thinking:
[1]
linear sweep (1,2 or 3D) (maximum ADC signal shown at end)
[2]
log10 sweep (1,2 or 3D) (maximum ADC signal shown at end)
[3]
1D graphical spectrum analyser (realtime ADC response graph plotted)
[3]
2D graphical spectrum analyser (realtime ADC response graph plotted)
Hi Yucca
Sounds amazing, most wouldnt realize or appreciate the time you will need to put into this.
The spectrum is a great idea and could be very handy indeed.
The maximun ADC peak would be good if it could log the 3 best peaks maybe in order if you are able to, i am thinking there maybe more than 1 sweet spot which could give clues to what's going on.
Peter
@GK & Otto,
Squares may be good for TPU, that's why I also want to include the following stages outputing into 3 seperate front panel BNC sockets:
variable opamp -> FET driver -> N-FET (sourcing external power)
this should give me PWM capability (10...90%) on all three channels up to maybe 10MHz. Of course they will also be perfectly phase locked and very frequency stable thanks to the temperature compensated main oscillator.
If anyone else wants to build one I will send the code and instructions to any other OU member gratis.
Yucca.
Quote from: Peterae on September 26, 2009, 05:57:32 PM
Hi Yucca
Sounds amazing, most wouldnt realize or appreciate the time you will need to put into this.
The spectrum is a great idea and could be very handy indeed.
The maximun ADC peak would be good if it could log the 3 best peaks maybe in order if you are able to, i am thinking there maybe more than 1 sweet spot which could give clues to what's going on.
Peter
lol Peter, you know more than me about time spent with MCU software, assembler language is MUCHO work and gets VERY difficult in larger projects. I have written one small system with GUI in assembler but mainly mantained other systems code, the programmers comment text is VERY needed and even then it's painful.
good idea about 3 best peaks.I would have to write some routine to record only local gradient maxima otherwise it would only record several peaks from the same overall highest curve.
Yucca.
Good work! Keep on truckin'
Quote from: Yucca on September 26, 2009, 06:02:16 PM
@GK & Otto,
Squares may be good for TPU, that's why I also want to include the following stages outputing into 3 seperate front panel BNC sockets:
variable opamp -> FET driver -> N-FET (sourcing external power)
this should give me PWM capability (10...90%) on all three channels up to maybe 10MHz. Of course they will also be perfectly phase locked and very frequency stable thanks to the temperature compensated main oscillator.
If anyone else wants to build one I will send the code and instructions to any other OU member gratis.
Yucca.
Hello all,
@Yucca
very nice work.
As least you cant damage this oscillators like Im doing it with mine because of runaways every day, ha,ha.
Otto
Otto & GK, Cheers guys, your work in TPU seems deeper than average, it'd take me weeks of reading to catch up on what you guys have done. Runaways sound exciting otto!
While I wait for bits to arrive I've designed a touch sensitive circuit, I hope to have the one metal dial be also touch sensitive so scanning can be performed with one control in one hand and no strain from pushing switch.
The circuit sets up two of the NORs as a 100kHz RC oscillator. This oscillator then pumps up the capacitance (which is touchpad to ground) the capacitance is leaked through the variable pot.
When someone touches plate the capacitance to earth increases greatly and so the resistor can't pull it down fast enough. the fourth nand buffers the output and debounces with 22uF electrolytic.
The microcontroller interprets a momentary touch between 10ms and 200ms as a valid user press.
Maybe a working TPU will hash up the operation lol, but it does work fine a meter away from a 10W spark fired aircore LC at ~6MHz.
Hi Yucca!
Hope your project went on smoothly! Can you update us on its latest progress. Thanks!
Ernest
Give it time LOL , this is quite the engineering project .
When you *build stuff* you run into trouble and have to trouble shoot and solve that problem .
As Dankie says, it does take time, I find most time on the project is thinking time away from the bench, and then programming time, probably over 30Hrs now, if I don't like the code structure I often rip it back a little and then proceed.
But here's an update:
The software is done to initial beta stage and seems bugfree. Scanning with power sampling is now complete and runs at a reasonable 1200 samples per second.
Most of my parts have arrived, see photo.
On the left you can see the 3 x DDS boards with AD9851 synthesizers and AD8008 RF amps. I can't stress enough what a superb job Analog Devices have done with these ICs; I have a very good radio receiver (scanner) that is very stable. I programmed a module to output 60MHz. Tuned the radio in on sideband reception and it was spot on. When I listen to the heterodyne signal it is really solid and clean. I can set heterodyne to 500Hz and even at 60MHz the 500Hz tone is rock solid! When you try the same with analog oscillator the heterodyne will drift all over depending on temperature, proximity of hand, phase of the moon lol etc. And when you change the output load it will disappear from the radio.
Another good thing is that the 3 seperate modules are all VERY close in frequency. I can hear a small beat between two seperate heterodyne signals. But at the moment each module has its own xtal oscillator. I plan to feed them all from one clock in the end. But I'll build it up with three seperate clocks to start with.
Also pretty pleased with the jogwheel encoder. A nice bit of kit, made in japan for CNC machine control, ballraced and strong, nice clean output signal, works well with the micro. Also I will be able to isolate its case ground and my touchsense circuit works well on it.
The 3 bunches of black cables are chinese wifi extender cables, these were dirt cheap and superb quality. If you bought 6m of RG174 cable and the plugs and sockets from digikey etc it would cost quite a bit. These cost â,¬5 all up delivered to europe.
Before I expand the software to 3D sweeping I will upgrade to 32kb memory. I have plans to make the 3D scanning have 3D gradient ascent to quickly hunt into maximas. To do a brute force scan (raster scan) of 1000 x 1000 x 1000 samples = 1billion samples, it would take almost 10 days to scan that at 1.2 kS/s. With gradient ascent I should be able to shorten that to hours. I will also have raster scan as well. This will be nice for exploring areas of interest when the data is transfered via USB to the PC, I will in the end write a small PC app with openGL viewer so I can see the 3D data set and peel layers back to look inside.
I will also make small RF shields with cardboard and alu tape for each module.
I've also got to work out the best conditioning circuitry for the sampling input, I probably want an optoisolator and maybe a log opamp with small knob on box front to adjust gain.
I will order a small alu chassis case to put it in next paycheck.
It'll probably be done in a few weeks. Before then I'll probably make a vid showing performance of the DDS modules.
A beautifully, extensive play of the mind and wonderful engineering!
Now when you want that device to run free at 60Mhz strap a hamster to the plate. That way you won't be around when it blows up.
Otherwise Saint Peter will be asking you "Well now son, What is that in your hand? You know we don't let that kind of technology in here. We have enough clouds as it is. And it also looks like your brought one of your own?"
--gk.
Quote from: giantkiller on October 22, 2009, 05:48:15 PM
A beautifully, extensive play of the mind and wonderful engineering!
Now when you want that device to run free at 60Mhz strap a hamster to the plate. That way you won't be around when it blows up.
Otherwise Saint Peter will be asking you "Well now son, What is that in your hand? You know we don't let that kind of technology in here. We have enough clouds as it is. And it also looks like your brought one of your own?"
--gk.
Cheers GK. I realise I'm covering similar ground you walked many moons ago. Your TPU sings vid is a classic, those sounds are real nice like an ambient track, with suprising complexity.
Lol, I don't plan to do any closing the loop (yet), all open loop with weak fuses.
I'm still wondering whether I should also use faraday cage or maybe I'd be better with a simple tinfoil hat lol?
I can't wait to start exciting some different winds with it. I'll probably play with pure sin first and then add fast comparators and FETs for squares later.
I've watched the vids lots of times and nothing screams fake, also one thing I will say is that the TPU subject is teaching me alot of new things, which is cool!
Yucca.
Quote from: BeeSee on October 22, 2009, 07:40:16 AM
Hi Yucca!
Hope your project went on smoothly! Can you update us on its latest progress. Thanks!
Ernest
Welcome to the forum Ernest, I see you registered along time ago so you probably have quite a bit of background reading done already.
To give you and all a brief idea as to what I will be trying first with this box:
I will wind a circular collector loop 6inch diam, it will have a small trimmer cap across it for tuning.
On it I will place 3 collector coils, each covering 120 degrees.
I will excite each collector with different frequencies while sampling the collectors output power and scoping its output.
I don't expect to see COP>1 on this first experiment. But it will be interesting to see the 3D power distribution and if I see clusters of power I may try concentrating on that area and try to get clusters to join up with changes to coil geometry and tuning.
It should be good watching because I plan to make youtube vids of the 3D power structure slowly spinning in 3D frequency space and gradually unpeeling to the hotspots in the centre. I wonder what shape the distribution will be? Spherical, Ellipse, Curved seperate planes, curved lines, straignt lines? hmm?
Yucca.
Million thanks for your prompt reply!
I am indeed a newbie and this is my 1st post, though I am a long registered member of OU.com.
My interest is not in the TPU (too advance for me to handle). I am interest to know more specifically how to build your 3-channel Pure Sin Generator. Hope your next video will cover the steps of the construction of this unit.
Will you all the success and Million thanks for sharing your valuable knowledge with us!
Ernest
Yucca , I hope you wont be stupid enough to offer this for free , after all that hard work .
You know theres a potential market for this ...
If somebody can only copy it , then he must buy it .
@Yucca,
Good job. This is quite the addition for any future attempts. You will be pleased. I am digging deep into my controller failure. Seems the BEMF or shockwave burned all my outputs of the uproc.
I am writing a program to toggle the output lines in a standalone mode. No usb comm. I am locking this little bugger in a cage and gonna poke it with a sharp stick!
No caveats. But I must relay this bit of advice. I put all my current efforts and resources into this device. Don't have a backup plan as in other types of test equipment to do this type TPU testing. I have shopped and I don't like the pricing behind the LF & HF multichannel signal genny with synchronization capability.
So this controller route is good. Jason did implement alot of safety features. But nothing to guard against user misconfigs, like putting a feedback BEMF at the SSR gate drive that feedsback into the controller with no firewall. And I have a firewall safety that I built 2 years ago sitting there unattached! Argh. In my zeal gained from other builds and tests I ran roughshod over this new build and cataclysmically denied myself the favor of being peaceful in my attempts. So let this be a lesson to ya! I do all the blowin' up stuff here. LOL.
I am in code land and that is very dark and addictive real estate.
--gk.
Quote from: dankie on October 23, 2009, 10:32:44 AM
Yucca , I hope you wont be stupid enough to offer this for free , after all that hard work .
You know theres a potential market for this ...
If somebody can only copy it , then he must buy it .
Dankie, I've pm'd you with more info but I plan to charge a modest amount for a preprogrammed chip with full colour laser printed build instructions and sources for all the components at the cheapest price, paypal through ebay. Existing customers will then be offered further firmware enhancements at a much reduced rate. It will make it easier for someone to build, just plug and play really. Also it will give me incentive to greatly enhance the software and produce a better product that will make a good peice of equipment with very high specs for anyones bench.
I'll buy one ;)
Quote from: giantkiller on October 23, 2009, 11:42:24 AM
@Yucca,
Good job. This is quite the addition for any future attempts. You will be pleased. I am digging deep into my controller failure. Seems the BEMF or shockwave burned all my outputs of the uproc.
I am writing a program to toggle the output lines in a standalone mode. No usb comm. I am locking this little bugger in a cage and gonna poke it with a sharp stick!
No caveats. But I must relay this bit of advice. I put all my current efforts and resources into this device. Don't have a backup plan as in other types of test equipment to do this type TPU testing. I have shopped and I don't like the pricing behind the LF & HF multichannel signal genny with synchronization capability.
So this controller route is good. Jason did implement alot of safety features. But nothing to guard against user misconfigs, like putting a feedback BEMF at the SSR gate drive that feedsback into the controller with no firewall. And I have a firewall safety that I built 2 years ago sitting there unattached! Argh. In my zeal gained from other builds and tests I ran roughshod over this new build and cataclysmically denied myself the favor of being peaceful in my attempts. So let this be a lesson to ya! I do all the blowin' up stuff here. LOL.
I am in code land and that is very dark and addictive real estate.
--gk.
GK, I hear you! codeland is my favourite sanctuary.
Ever since I fried a 50MHz analog scope when trying to look at a cars ignition timing when I was a teenager, ever since then I am more careful.
I have thought about different schemes to guard against fast back transients should they occur. In my initial pure sin experiments I don't expect major surgeback through the oscillators, but when I go to square, well things might get rough.
Having said that, even with initial sin build I will definitely be putting isolation into the feedback path:
[1] signal in
[2] rectify
[3] smooth
[4] bleed via 1ohm 5w resistor to ground (to give a DC voltage proportional to output current)
[5] tie high side of bleed resistor with neon and avalanche diode in parallel to ground (safety path to GND on overvolt)
[6] optoisolator
[7] log opamp to give compression and so more range
[8] then into the microcontroller ADC to calculate realtime RMS power and display as necessary
I suppose it wouldnt hurt to put neons and parallel avalanche diodes to ground on each oscillator channel also, negligable capacitance is added, whether they would sink fast enough to protect my RF output amplifier chips, I don't know? Maybe you could try this after you fix your uProcessor.
Famous last words: I'll be more cautious next time?
I have pulled all chips and am using the board as a holder for just the uproc. I have tried 3 of the 4 uprocs. All the same results. With nothing but caps, diodes, + a DC2DC cnvtr the processor runs and communicates. The voltage is +5v and fine. With one uproc left that hasn't been plugged in I need to program it in another setup in case this build is breaking uprocs. Eek! I buy more, but still.
--gk.
I dont know if this is useful, but the cro i fried with my 3 channel ring had so many faults that seemed un related to the input channel. the faults were in the vertical amplifier,power supply and timebase ..separate,unrelated faults it seemed ..I had a manual . I never succeeded in fixing it. It was a nice anritsu 100 mhz
This cat we chase sees pn junctions as cotton candy..so it seems
Magamps are used in all sorts of highenergy fields including radioactive. If the acsupply to the magamp is uhf captured on a dipole antennae. Instead of just dumping the carrier so as to obtain the modulation riding the carrier we filter and store the carrier in a resonant cavity. Effecting a lowering of the frequency and a rise of amplitude. Waves sitting on top of waves and beside them. In other words taking uhf and converting it into non radiating khz. After enough power has been collected the magamp bias supply and control circuit can be powered by the ac input. The amplification occuring at a low frequency closed circuit whilt the input is open circuit. A portion of the signal used to amplify itself. Amplify. To make larger. Lots of short wavelengths summed to create a longer wavelength. The faster something changes the more energetic. Something changing 330million times a second expressed as something changing only 60 times a second
http://www.tpub.com/neets/book8/32n.htm
Quote from: Mannix on October 23, 2009, 07:16:01 PM
I dont know if this is useful, but the cro i fried with my 3 channel ring had so many faults that seemed un related to the input channel. the faults were in the vertical amplifier,power supply and timebase ..separate,unrelated faults it seemed ..I had a manual . I never succeeded in fixing it. It was a nice anritsu 100 mhz
This cat we chase sees pn junctions as cotton candy..so it seems
Hi mannix
RIP, sounds like you put some monster spikes into the CRO ground system.
were you feeding the ring with squares or sin? I'm guessing square/PWM?
Quote from: sparks on October 24, 2009, 12:24:02 AM
Magamps are used in all sorts of highenergy fields including radioactive. If the acsupply to the magamp is uhf captured on a dipole antennae. Instead of just dumping the carrier so as to obtain the modulation riding the carrier we filter and store the carrier in a resonant cavity. Effecting a lowering of the frequency and a rise of amplitude. Waves sitting on top of waves and beside them. In other words taking uhf and converting it into non radiating khz. After enough power has been collected the magamp bias supply and control circuit can be powered by the ac input. The amplification occuring at a low frequency closed circuit whilt the input is open circuit. A portion of the signal used to amplify itself. Amplify. To make larger. Lots of short wavelengths summed to create a longer wavelength. The faster something changes the more energetic. Something changing 330million times a second expressed as something changing only 60 times a second
http://www.tpub.com/neets/book8/32n.htm
Nice link about magamps, I will read that carefuly this evening.
I'm worried about spikes coming back through control coils and frying my RF amp chips and maybe also my DDS chips especially when I go to square wave mode.
Do you have any ideas for protecting semiconductor OPs against spikes? I'm thinking Transient Voltage Suppression (TVS diodes) which avalanche at say 70% of IC max and then in parallel to those put neons to take the brunt of the later current and so protect the diode junction, I'm wondering if the neon would be fast enough to take the heat. If the TVS fails it will go short so worst case is that the TVS are just pricey fuses.
QuoteRegardless of the amount of load flux or polarity of the load voltage, there is no net effect of load flux on control flux.
I see this as an isolation issue and that the mag amp can be used as a part of the feedback loop. This has been said before but I see in now a little different.
--gk.
Even tubes have a hard time with reflection. That's why we have microwave ovens nowadays. Standing waves would setup inside the tubes of the day and toast the grids. Whereas the cyclotrons create a spoked electron field that is spun. The moving charge of the electron bunch creating waves in resonant cavities of the anode as the spokes alternately align with the anode cavities. Ingenious device. And I believe it is at the heart of the matter with the tpu.
Free electrons will orbit at a given cyclotron frequency about magnetic field lines. We now have moving mass simply by providing free electrons and a magnetic field.. Where does the energy come from that allows the electron to move in a predictable drift or current. In a torroidal coil the electron drift will be determined by three forces. The polidal field the magnetic field and the gravitational field. One electron drifting 1mm a second isn't much power but billions of them is. Again where does the power come from that allows an electron to move. Does an electron moving from cathode to anode effect the cathode anode potential if the electron never reaches the anode plate? Electrons have moved but never reached the other side of the capacitor allowing the capacitor to remain charged. Unlike when a capacitor is discharged into a closed circuit there is no exchange of charged mass in the two capacitor plates. But there was an electron acceleration in the field between the two capacitor plates. Adda magnetic field perpendicular to the polidial field and the electron acceleration now takes on a curved path. Stop the polidial field and the electron movement spirals about the magnetic field lines. In a torroidal coil there happens to be a conductor field full of free electrons in geometric aligment with the electron drift about the magnetic field lines contained within the core of the torroid. One more time where does the energy come from that causes an electron to move or orbit a magnetic field line.
Then maybe sparks you can build an device and show us how you do it .
Dankie
I am asking the question about what makes electrons move about a magnetic field line or a neucleus for that matter because I DONT KNOW. I do know that whatever force or combination of forces that govern unbound electron motion is an important thing to figure out. So I"ll just keep on asking questions and researching until I know and then I'll tell everybody I possibly can as many times as I can what the fuck is going on.
Yucca
Back in the day I worked alot with unijunction triggering of scrs. They were the fastest things around with the least drop.
The ujt would have to be heatsunk to last as it is purely a resitive element but if it picksup the field reversal and triggers the scr to shunt the reflection it may work as an active transient suppression device. If it works I wouldn't waste the current on just heating though. The ujt would act as a reversed field sensor and fire the scr into conduction until the transient voltage dropped below the holding voltage needed for scr conduction.
A short video showing the stability of one of the DDS modules.
A digital radio receiver is tuned into 40MHz LSB and we hear the audio tone of it heterodyning with the DDS module also set around 40MHz.
When a Tektronix FG504 undergoes the same test it falls rather short.
http://www.youtube.com/watch?v=6nMC93-TAy0
Sorry about the crappy video/audio quality.
Nice oscillator I can see why you are worried about frying it. No way to use optical isolation for your finals?
@Yucca!
That is one sweet setup and test with good solid results!
How are you going to synch or phase adj the 3 DDS60 units?
--gk
@Sparks,
Opto isolation sounds like a good idea if I can get something fast enough. I will build this pure sin box with no protection but I will only use it to drive another box or board that then will have the finals in it, fast schmitt and FETs for squares (with a little PWM) and some beefier amp stage for sin. This box will have some isolation in it and also it power input line will have non inductive shunt for scoping total drive power. For COP calculation I will ignore the draw of the sin gennies and micro and just look at the final drive stages draw.
@GK
I will be taking all the clocks off the 3 modules and making a small daughter card with one clock and 3 same length coax runs to feed each module. One clock module should handle the fanout and be able to drive 3 DDS chips OK.
The DDS chips are 5bit programmable for phase (thru the serial bus), so 32 sector resolution. i.e. I can set each oscillators phase to these degrees:
0 ... 11.25 ... 22.5 ... etc. all the way up to 348.75 and back to 0
So I should be able to set up a reasonable 3 phase signal at 0, 123.75 & 236.25 degrees at any frequency, might be interesting to try and get stuff spinning in the double digit MHz range, coil placements could perhaps be moved slightly to compensate for the imperfect 3phase angles.
2 phase (for quadrature) will be spot on at 0 and 90 degrees.
Just ordered one of these:
http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&item=390098652853
Now I have the case dimensions I will start to plan and lay out the vero boards. It will be a tight squeeze so I'll wait for the case to arrive before I commit to soldering anything.
After thought I want to firewall the case with RF shield and have the oscillators on one side and the micro on the other. All digital and power lines will pass through feedthrough caps on the firewall.
Each RF module will also have its own shield to minimise crosstalk, these shields will be made from stiff carboard and wide aluminium tape. RG174 coax will be used to bring the final signals to the front panel.
@Yucca,
And that will equate to a full metal jacket. 8)
great work yucca ...
you have yourself a verry nice control unit ...
i was lucky enough to get many cases for for building tpu controllers inside of... they are solid alum forged ... they were originally designed for rf applacations ...
i will soon gut a unit and start building
but i want a tube unit ... 8)
aswell i have aquired a few more tube amps ... ;D
agin great work yucca ...
ist!
Quote from: innovation_station on October 27, 2009, 07:37:00 AM
great work yucca ...
you have yourself a verry nice control unit ...
i was lucky enough to get many cases for for building tpu controllers inside of... they are solid alum forged ... they were originally designed for rf applacations ...
i will soon gut a unit and start building
but i want a tube unit ... 8)
aswell i have aquired a few more tube amps ... ;D
agin great work yucca ...
ist!
Cheers ist, forged alu cases are real nice and dont come cheap, you're sitting on a few dollars just for the ALU scrap price there!
Tube controller, sounds like a very interesting build. be sure to hook us up with more info as it progresses.
I think I'll call it TRIFORCE.
I'll probably print this out, cover it with selotape and then stick it to the box.
p.s. yes I have adventured in the land of Hyrule many times lol.
QuoteOnce the goddesses had completed their tasks, they departed for the heavens, and left behind three golden triangles. In these, they put their power to govern all things; this relic became known as the Triforce.
love it yucca ...
what do you think the cost of a complete osc unit would be ..
im verry intrested in unit ..
but i ideally want a 4 freq unit with percise control right up to 1 ghz ...
for my matter manipulation device but do i really need something like that ......
ist!
great logo
Quote from: innovation_station on October 29, 2009, 10:13:52 PM
love it yucca ...
what do you think the cost of a complete osc unit would be ..
im verry intrested in unit ..
but i ideally want a 4 freq unit with percise control right up to 1 ghz ...
for my matter manipulation device but do i really need something like that ......
ist!
great logo
Hi ist,
You need one DDS module for each channel:
Quote
$29.50 for the DDS-60 Kit (without AD9851 DDS chip)
$21.00 for the AD9851 DDS chip
$80.00 for the Assembled & Tested DDS-60, including the AD9851 DDS chip
So if you can get free chip samples from Analog Devices then each module can be had for $29.50
Then you need something to program the chips frequency with, you can use a PC or you can use a standalone microcontroller like Im using. The whole micro ready built with screen and chip in socket costs $50 delivered to canada.
So depending on which route you go for the DDS module you could build one for as low as $140 all up.
Jog wheel is optional, the micro board already has buttons to control the thing with. Also you need to find case, wires and sockets etc, but you probably have that stuff already.
As youre a long standing member I could email you the software to program your micro board with. Just plug into a PC USB port and upload it with one mouse click.
I don´t know about going up to 4 channel GHz with variable frequency and phase, that´d be a serious bit of kit :o
thank you yucca!
for your responce..
the reason im intrested in upto 1ghz is because i have 2 skids of thease and they are stable to 1 ghz and i think they are quite a few watt rf amps ...
here is a picture
ist!
btw in my limited teck knowalage .. i beleave thease rf amps or TRUNK AMPS... amp both ways send and return ...
as i have said i have not played them yet or hardly even looked at em i know a few teckies that know thease things well so when it comes time .. i will ask advice
Quote from: innovation_station on October 30, 2009, 08:26:13 AM
thank you yucca!
for your responce..
the reason im intrested in upto 1ghz is because i have 2 skids of thease and they are stable to 1 ghz and i think they are quite a few watt rf amps ...
here is a picture
ist!
btw in my limited teck knowalage .. i beleave thease rf amps or TRUNK AMPS... amp both ways send and return ...
as i have said i have not played them yet or hardly even looked at em i know a few teckies that know thease things well so when it comes time .. i will ask advice
Hi Ist, After googling a bit it seems Trunk amps are either one way or two way. They are high spec kit with bandwidths typically from like 5MHz to approaching 1GHz.
In the higher frequencies you will lose signal unless everything is tricked out with coax runs that are carefully impedance matched.
I've found that above say 10MHz you have to be real careful with wiring capacitance and inductance, every piece of wire becomes an LC component.
A pure sine wave is the easiest thing to keep intact, but if you want to propogate pulses then its a real art to keep that pulse together and not have it collapse into a load of unphased sin harmonics.
P.S.
I've got through a few parts of the youtube vid about the guy who teleported to mars, it's fantastic listening! Thanks for the heads up.
Just finished writing a double buffered graphics API in C++ for the NOKIA 3310 LCD screen:
http://www.youtube.com/watch?v=dHV8wELuOa8
I am moving to this screen instead of the NOKIA 6610 screen because:
its cheaper.
it seems easier on the eye than the 6610 screen.
it will make case layout and installation easier because the screen board has 4 nice mounting holes on it.
@Yucca
Awesome!
Cheers stprue,
I've finished the main freq screen and also decided on the layout for the main menu.
FREQU TUNING:
Gives the main frequency screen where user is able to adjust any decade of any channel.
PHASE TUNING:
Gives the phase screen where user is able to adjust phase of any channel.
LIN SWEEPING:
User is able to set start and end points of a linear sweep and also specify the number of milliseconds that the sweep should take. The unit will output a trigger signal from a fourth BNC connector so a scope can be hooked up and used as a frequency response monitor for the coil under test.
LOG SWEEPING:
As above only the sweep is logarithmic, sweep step is proportional to frequency.
HARMONIC LOC:
User can set multiplication factors for f2 and f3. For example if f2 is set to 2.000 and f3 is set to 3.000 we will get the first and second harmonic. When harmonic lock is active the main freq screen will show a small padlock in bottom right. Then if the user sets say f1 to 1MHz then f2 and f3 will automaticly go to 2MHz and 3MHz. Of corse any multiplication factors can be entered.
At the moment it will be 1D sweeping, In the future I will expand to 3D sweeping by putting a second level into the sweep menus and maybe combining LIN and LOG into one item to give me another menu position (any ideas for another menu function are welcome)
.
;D
BRO I LOVE IT!
YOUR ROCK AT WHAT YOU DO SIR!
IST STANDS BACK IN AWE ....
YOU MAKE THAT WORK BRO ... I HAVE ABOUT HUMMMMM 1000 USES FOR SUCH A UNIT ... ;)
I JUST CANT BUILD SOMETHING LIKE THAT ... SO ILL BUY SOME FROM YOU IN THE FUTURE .. 8)
PEACE IST!
Quote from: innovation_station on November 08, 2009, 09:39:17 AM
;D
BRO I LOVE IT!
YOUR ROCK AT WHAT YOU DO SIR!
IST STANDS BACK IN AWE ....
YOU MAKE THAT WORK BRO ... I HAVE ABOUT HUMMMMM 1000 USES FOR SUCH A UNIT ... ;)
I JUST CANT BUILD SOMETHING LIKE THAT ... SO ILL BUY SOME FROM YOU IN THE FUTURE .. 8)
PEACE IST!
OK IST, I might make some up for request in the future and sell for cost and postage to long term OU members, you'd have to box it up and provide 12V DC supply, small gell cell would be ideal for low noise and isolation.
Yep, you could hook it up to any coil setup that you can imagine, maybe put your coil power out into a speaker, let the unit alone for a few hours while it explores millions of frequency combos. If you come back and the speaker cone is blown out then you know you've hit something good lol.
I DO FIND THAT FUNNY YOU MENTION THAT ...
ill explain why ...
i had this amp for car audio ... it was given to me ... it was blowen 1 channel ..
this is the conclusion i draw ...
i know the guys that were in the car when it blew i know the song that was playing ...
and now i say i know why the fet blew to particles.. lol ;)
the volume the location of the car at the time of the blow up of the amp the song
infact i will go so far to say it is probally engenieered this way .... it is a consumer world still as we know it ...
this song might of had a few 808 samples in it ;)
ist!
@Yucca,
nice project you have going, I sure like the blue light on the screen. I hope once you finish it and start to hook it up to coils, that you optically couple to the drive MOSFETS in order to prevent any potential back EMF voltage spikes from ruining such a beautiful board, because there will be spikes for sure. (just listen to Otto crying about all the transistors he's fried...lol)
EM
808 and 303 are nice instruments!
Quote from: EMdevices on November 08, 2009, 01:14:37 PM
@Yucca,
nice project you have going, I sure like the blue light on the screen. I hope once you finish it and start to hook it up to coils, that you optically couple to the drive MOSFETS in order to prevent any potential back EMF voltage spikes from ruining such a beautiful board, because there will be spikes for sure. (just listen to Otto crying about all the transistors he's fried...lol)
EM
Hi EM, thanks, your idea to isolate outputs is good and something I hope to do.
can you (or anyone else) recommend any isolation device that will cover 10kHz to 60MHz? Of course, the cheaper the better.