Overunity.com Archives

Hi Everyone,

I have been around for a long time in different FE Community's. Mostly collecting information.

The biggest problem with Building and Testing Devices that I have seen, is a Reliable and Flexible Switching System. We should all work together and bring the best of us in to go from Step one to actually selling the kits to those that want them.


I propose a collaboration of users to work on a better more Reliable and Flexible Switching System!

Some initial requirements:
1: Microcontroller Controlled with a Client so as can be controlled from ones Laptop/PC.
2: Capable of fine adjustments for fine tuning of frequency's.
3: Capable of reasonable Voltages and Currents.
4: Capable of a large range of Frequencies.
5: Wide range of Switching Capabilities, Single H-Bridge, Dual H-Bridge, Single Switching mode...
6: Plenty of Circuit Protection/Isolation for the Microcontroller and also the User, fused etc...
7: Clean Slim and cost efficient for all.

I have a friend 'Selfonlypath' that is no longer online that did something similar. To give him and Dan, his friend, credit, they did a very good job but not all is still available. See attached Picture below. A few problems I had found with this circuit, supply current was not high enough from the RK-0515s at high frequency and the Fetts would not switch properly.

Some Recommendations for Microcontrollers:

1: FEZ Hydra Mainboard (Medium Speed reasonable price) (https://www.ghielectronics.com/catalog/product/328)
2: FEZ Raptor Mainboard (Fast 400MHz but expensive) (https://www.ghielectronics.com/catalog/product/499)
3: Arduino Mega 2560 (Slow 16MHz) (http://arduino.cc/en/Main/ArduinoBoardMega2560)

Personally I like the GHI Boards.

I can do the programming of the board and the Windows Client. Can someone do the Circuit Design and source cheap Chinese Manufacture? I already have a similar unit to SelfOnlyPath running and it is worth its weight in Gold. I will admit, if it can be improved on it would be great and if the entire Community can benefit and have the same access to it then it would help immensely!

I have attached some pictures of my unit so others can see the clunker I use.

All the Best

  Chris

Good day EMJunkie

I've been thinking along the same lines for a while also:

Found this about a year ago and saved it:

http://www.source-for-innovations.com/pgen.htm

Seems like it will do PLL also and can program from LCD or laptop.

take care, peace.
lost_bro

There are several considerations and they are unlikely to be met by a single solution.  For the control side considerations include:

1. Frequency capability.
2. Synchronization capability.
3. Power topologies supported.
4. Fault handling.  This is really important because it means the difference between constantly frying the power section, or just gracefully losing function.

On the power side:

5. Single ended, or half bridge.
6. Number of sections:  1, 2, 3, 4, or 6 being the most common. 1, 2, 3 for half bridges, and 1, 2, 4, and 6 for single ended.
7. Type of conduction:  continuous, discontinuous, critical conduction.
8. Trapezoidal, Quasi-resonant, or fully resonant.
9. DC or AC output
10. Voltage capability
11. Current capability


For any combinations there are a number of solutions that would work.  My recommendation is that rather than try and find a one size fits all, start with at least one class of machine, and then work on meeting your goals of low cost and flexibility within that framework.  For high power switches, Avageo has some really great isolated IGBT drivers that include lots of housekeeping functions that make it a lot easier to drive the devices reliably and survive faults.

Quote from: lost_bro on April 26, 2014, 06:02:04 PM
Good day EMJunkie

I've been thinking along the same lines for a while also:

Found this about a year ago and saved it:

http://www.source-for-innovations.com/pgen.htm

Seems like it will do PLL also and can program from LCD or laptop.

take care, peace.
lost_bro

Hey Lost_Bro,

I like it! Problem is its only good to 100KHz.

All the Best

  Chris

Quote from: MarkE on April 26, 2014, 06:40:08 PM
There are several considerations and they are unlikely to be met by a single solution.  /...

Hi MarkE,

Agreed! Its going to be hard to do a single unit that will do it all!

My unit is good from about 0.04 to 7MHz for switched DC in either H-Bridge Mode or Single Switched Mode.

I think if we think broadly and then expandable after that then we can move forward with units like the Arduino Shield concept! I think there is sufficient technology out there now that this is a very achievable task and at low cost.

All the Best

  Chris

P.S: Anyone willing to do some Circuit Re-Design and throw it out there? Even if its based on the Above design!

Fault Handling! Housekeeping! What a concept....

:'(
;D

Quote from: EMJunkie on April 26, 2014, 06:51:30 PM
Hi MarkE,

Agreed! Its going to be hard to do a single unit that will do it all!

My unit is good from about 0.04 to 7MHz for switched DC in either H-Bridge Mode or Single Switched Mode.

I think if we think broadly and then expandable after that then we can move forward with units like the Arduino Shield concept! I think there is sufficient technology out there now that this is a very achievable task and at low cost.

All the Best

  Chris

P.S: Anyone willing to do some Circuit Re-Design and throw it out there? Even if its based on the Above design!

7MHz is way beyond the regime of hard switching topologies.  You are going to have to go quasi-resonant or fully resonant in order to manage the switching losses.  Otherwise you are just going to burn your power output stage and probably the driver stage as well.

Quote from: EMJunkie on April 25, 2014, 02:28:38 AM
..... A few problems I had found with this circuit, supply current was not high enough from the RK-0515s at high frequency and the Fetts would not switch properly.

Chris

Hello All:


Please see the attached Isolated gate drive schematic:
This has a max Vcc of 35VDC and can source 9amps for fast switching... its a design that Cree is using in the SiC MOSFET demo boards.

This can be modified to suite the need. 
If you are not a fan of Avago optoisolators, then Silicon Labs has the LED emulator that uses proprietary technology to emulate the LED opto, but with a type of RF on chip Tx/Rx for really fast communication.
I'm a fan of GDTs, but the design of these is frequency dependent and having to wind a new one for each experiment can be tedious.

take care, peace
lost_bro

Quote from: MarkE on April 26, 2014, 09:08:08 PM
7MHz is way beyond the regime of hard switching topologies.

Hey MarkE,

Mine works without doing any Burn Outs! I do have issues beyond 7MHz however. Heating and so on. The 47uH Inductor does a pretty good job at keeping the RF out of the isolated 15v supply.

I guess this is why we need to think outside the box!

All the Best

  Chris

Quote from: lost_bro on April 26, 2014, 09:18:13 PM
Hello All:


Please see the attached Isolated gate drive schematic:
This has a max Vcc of 35VDC and can source 9amps for fast switching... its a design that Cree is using in the SiC MOSFET demo boards.

This can be modified to suite the need. 
If you are not a fan of Avago optoisolators, then Silicon Labs has the LED emulator that uses proprietary technology to emulate the LED opto, but with a type of RF on chip Tx/Rx for really fast communication.
I'm a fan of GDTs, but the design of these is frequency dependent and having to wind a new one for each experiment can be tedious.

take care, peace
lost_bro

Thanks Lost_Bro! We will need to do some changes to go to the next step! Electronics Gurus! Anyone keen to participate?

All the Best

  Chris

Good idea, Why not go to a modular design, as in we could have the control board and several switching boards that will accept the signal from the control board with different switching boards for different applications, voltages - frequencies - current levels ect.

eg. a switching board would hold, the caps/regulator, mosfet drivers or switch drivers of whatever kind and all associated protections for the switching and driver solution, isolation ect.

The control board could be either Arduino or picaxe or whatever then and the user can still use the different switching boards. The switching boards could simply have a connector with the signal and power jacks so that one board plugs into the other without any wires to keep the signal traces short. Plug and play kinda. Most of us can manage to make a PCB, or we could get them made in bulk if a few of us chip in some money.

Cheers

Quote from: Farmhand on April 27, 2014, 04:46:00 AM
Good idea, Why not go to a modular design, as in we could have the control board and several switching boards that will accept the signal from the control board with different switching boards for different applications, voltages - frequencies - current levels ect.

eg. a switching board would hold, the caps/regulator, mosfet drivers or switch drivers of whatever kind and all associated protections for the switching and driver solution, isolation ect.

The control board could be either Arduino or picaxe or whatever then and the user can still use the different switching boards. The switching boards could simply have a connector with the signal and power jacks so that one board plugs into the other without any wires to keep the signal traces short. Plug and play kinda. Most of us can manage to make a PCB, or we could get them made in bulk if a few of us chip in some money.

Cheers

Hey Farmhand! Long time since we spoke!

Yes I agree! It only makes sense that the design stays in a modular form! Also it is critical to keep the signal traces short like you have pointed out! Also just as critical to keep Low Power Logic Circuitry well away from the High Voltage/Current Switching Circuitry!

The Circuit I posted is a very good circuit! Maybe if we could have someone start there and work on the component update to some better high speed smaller components?

I don't think anyone is keen to draw up the files? I know there are a few Electronics Guru's out there! Maybe they just have not yet had the need to check this thread out?

All the Best

  Chris

Might be a good idea to have voltage and current sense provisions to feedback to inputs on the micro processor so that we can use them for more control of the circuit. Over voltage or current protections ect. can be written into the code then.


Cheers

Quote from: EMJunkie on April 27, 2014, 03:02:48 AM
Hey MarkE,

Mine works without doing any Burn Outs! I do have issues beyond 7MHz however. Heating and so on. The 47uH Inductor does a pretty good job at keeping the RF out of the isolated 15v supply.

I guess this is why we need to think outside the box!

All the Best

  Chris

An IGBT with a 1us turn off time isn't going to do very well hard switching at 7MHz.

Quote from: MarkE on April 27, 2014, 07:00:16 AM
An IGBT with a 1us turn off time isn't going to do very well hard switching at 7MHz.

Hi MarkE,

I agree! Choosing the best components for the job is essential!

I have used: 5N3003
Turn-On Delay Time: 60ns
Turn-Off Delay Time: 220ns

So far this is one of my favourite N Channel MOSFET's!

All the Best

  Chris

Quote from: TinselKoala on April 26, 2014, 06:52:11 PM
Fault Handling! Housekeeping! What a concept....

:'(
;D

Nearly there TK!   ;D

You're an Electronics GURU! Feel up to posting a Circuit with Layout?

All the Best

  Chris

Quote from: EMJunkie on April 27, 2014, 08:51:19 PM
Nearly there TK!   ;D

You're an Electronics GURU! Feel up to posting a Circuit with Layout?

All the Best

  Chris

The application of the circuit really needs to have parameters placed around it.  One size fits all does not work well in power electronics.

Quote from: MarkE on April 27, 2014, 09:15:46 PM
The application of the circuit really needs to have parameters placed around it.  One size fits all does not work well in power electronics.

Hi MarkE,

Yes, I hear you and agree it will not be an easy task to achieve a "One size fits all"!

But we can get close!

The Circuit I Posted, from Selfonlypath, is that! Very Close!

All the Best

  Chris

Quote from: EMJunkie on April 27, 2014, 09:39:03 PM
Hi MarkE,

Yes, I hear you and agree it will not be an easy task to achieve a "One size fits all"!

But we can get close!

The Circuit I Posted, from Selfonlypath, is that! Very Close!

All the Best

  Chris

Chris there are a couple of things that you should look at in your circuit.

1. Critical:  The MOSFET avalanche protection is limited to the snubber network.  With a big inductive load that will either end up dissipating a lot of power or you will blow the MOSFET.  The flyback diode anode should be on the MOSFET drain, and the cathode on +DC where it is decoupled through a long inductance back to the MOSFET source.  The size of the decoupling capacitor depends on how much inductance there is back to the +DC supply.   This is particularly true given that you have a 1 Ohm gate turn-off resistor.

2. Design Improvement:  One section of the MCP1403 should be more than adequate for each MOSFET.  If you are going to drive more than 4.5A of gate current, you will have more issues with layout than driver strength.  The only reason that I would allocate by halves of that driver to one MOSFET would be if that made a big difference in the parasitic inductance in the layout.

3. Design Improvement:  You should place a 200uF capacitor in series with a 0.5 Ohm resistor across C4.  This will keep the filter L1/C4 from resonating uncontrollably when the MOSFETs are driven near 3kHz.

4. Design Improvement:  C5 should have a 47 Ohm or so resistor in series so as to limit the peak current load on the ATMEGA.  Otherwise you can kill the output pin from electromigration.

5. Design Consideration:  Z1 will only be effective if it has a very low inductance between the MOSFET gate and source.  Also, a typical zener will not act quickly enough to protect the MOSFET gate.  A TVS device rated at 16V that will trip by 20V would be a better choice.

5. Documentation Error:  The notes say D2 is for asymmetric gate drive.  It should say D3 and R4.

Quote from: lost_bro on April 26, 2014, 06:02:04 PM
Found this about a year ago and saved it:
http://www.source-for-innovations.com/pgen.htm

Gunther has put a lot of effort into this design:
http://www.source-for-innovations.com/switch.htm

I have used the circuitry myself and it works quite well.


@MarkE

Jump in here Mark, make improvements and post some EAGLE files we can use to build breakout boards.  As sharp as you are in this area it shouldn't take you more than an hour or two to get us all setup.

Quote from: MarkE on April 28, 2014, 03:59:42 PM
Chris there are a couple of things that you should look at in your circuit.

1. Critical:  The MOSFET avalanche protection is limited to the snubber network.  With a big inductive load that will either end up dissipating a lot of power or you will blow the MOSFET.  The flyback diode anode should be on the MOSFET drain, and the cathode on +DC where it is decoupled through a long inductance back to the MOSFET source.  The size of the decoupling capacitor depends on how much inductance there is back to the +DC supply.   This is particularly true given that you have a 1 Ohm gate turn-off resistor.

2. Design Improvement:  One section of the MCP1403 should be more than adequate for each MOSFET.  If you are going to drive more than 4.5A of gate current, you will have more issues with layout than driver strength.  The only reason that I would allocate by halves of that driver to one MOSFET would be if that made a big difference in the parasitic inductance in the layout.

3. Design Improvement:  You should place a 200uF capacitor in series with a 0.5 Ohm resistor across C4.  This will keep the filter L1/C4 from resonating uncontrollably when the MOSFETs are driven near 3kHz.

4. Design Improvement:  C5 should have a 47 Ohm or so resistor in series so as to limit the peak current load on the ATMEGA.  Otherwise you can kill the output pin from electromigration.

5. Design Consideration:  Z1 will only be effective if it has a very low inductance between the MOSFET gate and source.  Also, a typical zener will not act quickly enough to protect the MOSFET gate.  A TVS device rated at 16V that will trip by 20V would be a better choice.

5. Documentation Error:  The notes say D2 is for asymmetric gate drive.  It should say D3 and R4.

Hey MarkE,

Excellent! Thank you! We are a step ahead! Anyone an Eagle expert? Anyone like to put these changes MarkE has suggested/proposed into an updated schematic?

Thanks MarkE! Excellent Post!

All the Best

  Chris

Quote from: EMJunkie on April 28, 2014, 05:00:33 PM
Hey MarkE,

Excellent! Thank you! We are a step ahead! Anyone an Eagle expert? Anyone like to put these changes MarkE has suggested/proposed into an updated schematic?

Thanks MarkE! Excellent Post!

All the Best

  Chris

Chris, you are welcome.  One thing that may be as important as the design itself is what tools you decide to use to express it in.  The last time I looked you could use Eagle for free if the design is only one page.  You might want to conduct another survey to see what tools people use.

Quote from: Dog-One on April 28, 2014, 04:35:45 PM
Gunther has put a lot of effort into this design:
http://www.source-for-innovations.com/switch.htm

I have used the circuitry myself and it works quite well.


@MarkE

Jump in here Mark, make improvements and post some EAGLE files we can use to build breakout boards.  As sharp as you are in this area it shouldn't take you more than an hour or two to get us all setup.

Doug, I use OrCAD.  Is Eagle the choice around here for low cost?

Quote from: Dog-One on April 28, 2014, 04:35:45 PM
Gunther has put a lot of effort into this design:
http://www.source-for-innovations.com/switch.htm

I have used the circuitry myself and it works quite well.


@MarkE

Jump in here Mark, make improvements and post some EAGLE files we can use to build breakout boards.  As sharp as you are in this area it shouldn't take you more than an hour or two to get us all setup.

Hey Dog-One,

Yes its a very swish unit! Very professional looking! Thanks for posting!

@All

Have seen the FEZ Hydra Board has had a price drop. This board is open source also. See: -->FEZ Hydra - Open Source Board<-- (https://www.ghielectronics.com/catalog/product/328) This is a 240Mhz ARM9 Processor (running at 200Mhz). Its .NET Gadgeteer compatible. Its got loads of extra cool features.

It does have only 4 PWM Pins. This can only allow for dual H-Bridge Mode, and single switching mode.

Ideally it would be nice to have a board with all the Microcontroller built in, but it may not be practical and may push the cost up dramatically.

Any thoughts on this?

All the Best

  Chris

Quote from: MarkE on April 28, 2014, 06:03:57 PM
Doug, I use OrCAD.  Is Eagle the choice around here for low cost?

I use DipTrace, but have had to learn EAGLE as it appears to be more compatible in the community.  Concepts are the same and there are a lot of tutorials out there to get up to speed quickly.

Quote from: EMJunkie on April 28, 2014, 06:06:50 PM
Yes its a very swish unit! Very professional looking! Thanks for posting!

Gunther likes the Propeller chip for signal generation, but I prefer the PSoC devices for a few reasons:
1. You can do hardware and software design within a single chip.
2. Cypress supports their products really well and has always offered a hand when I get stuck trying to implement a concept.
3. The Cypress design software might look a little overwhelming, but it's really slick to use and is very mature and reliable.  Oh, did I mention it's free too?

With some of the newer boards like the BeagleBone, the capabilities may go well beyond what is needed.  Having a full-blown PC on a credit card board still blows me away.  How accurately these systems can generate signals though...   Someone else will need to chime in.  I've been a little reluctant to get one to test with as I might dump far too many hours into one instead of focusing on the mission at hand.

Bottom line...  Yes, we should all have something that is easy and inexpensive to build or easy to get from someone else that has made a few of them.  The discrete component signal generators of the Stan Meyer days has long past.  On the other hand, one would be nuts to drop 400 dollars on a BK Precision pulse generator with the type of digital electronics we have these days.  A slick little controller on a board with some separate drive breakout boards is clearly the way to go.  Put everything in a handy little enclosure and we're back to focusing on our forum projects.

Quote from: MarkE on April 28, 2014, 06:03:57 PM
Doug, I use OrCAD.  Is Eagle the choice around here for low cost?

Hey Guys,

I agree, Eagle seems to be more 'Industry Standard'. I use different software also and am not confident enough to put Eagle Files out there that may have errors or bad practices that I am not yet aware of.

Eagle is Free also: -->Download Eagle<-- (https://www.cadsoftusa.com/download-eagle/) with lots of custom Library's.

Eagle does allow a full, one stop from start to finish solution. It can even get quotes for the manufacture of the completed Gerber Files. I am practising Eagle, learning its techniques so in the future I hope to be able to provide more help in this area.

All the Best

  Chris

Quote from: Dog-One on April 28, 2014, 07:09:26 PM
With some of the newer boards like the BeagleBone, the capabilities may go well beyond what is needed.  Having a full-blown PC on a credit card board still blows me away....

Hey Dog-One,

For Sure! BeagleBone... Its awesome!

In saying that I did find the documentation a little confusing when I purchased mine. Its a 200MHz Clock for the PWM's but the embedded CPU is 1GHz on the Black. I was a little disheartened that this wa snot more clearly documented at the time I purchased. All in all, this is the best board I have purchased for overall wide range of features. FEZ Hydra can also run Linux on the Embeded CPU but its CPU Intensive compared to the BeagleBone.

WOW What a price drop! BeagleBone Black - $45.00 (http://www.adafruit.com/products/1278) Thus far, this is the best option for price for our project!

Imagine VNC to your Home Power Unit to check stats and do maintenance all on a Credit Card Sized Unit! I guess we already see in in the phone technology so its not that hard to imagine...

Lurkers, Join in! we need to get the best from this thread so let us know your thoughts!

All the Best

  Chris

Quote from: EMJunkie on April 28, 2014, 07:34:04 PM
Hey Dog-One,

For Sure! BeagleBone... Its awesome!

In saying that I did find the documentation a little confusing when I purchased mine. Its a 200MHz Clock for the PWM's but the embedded CPU is 1GHz on the Black. I was a little disheartened that this wa snot more clearly documented at the time I purchased. All in all, this is the best board I have purchased for overall wide range of features. FEZ Hydra can also run Linux on the Embeded CPU but its CPU Intensive compared to the BeagleBone.

WOW What a price drop! BeagleBone Black - $45.00 (http://www.adafruit.com/products/1278) Thus far, this is the best option for price for our project!

Imagine VNC to your Home Power Unit to check stats and do maintenance all on a Credit Card Sized Unit! I guess we already see in in the phone technology so its not that hard to imagine...

Lurkers, Join in! we need to get the best from this thread so let us know your thoughts!

All the Best

  Chris

That does look like a nice kit for $45.

@All

This YouTube Video is a GEM! Its very close to what we should be looking at!

URL of what we should be aiming for: -->Dual Motor Controller Cape design for the BeagleBone<-- (https://www.youtube.com/watch?v=_bfA5D4yPq0) Instead we need Dual H-Bridge with flexible Frequency and Duty with PLL and other awesome functions! (Most H-Bridge Chips are locked to certain Frequency's or a range and are not quite flexible for our needs) MOST Importantly on top of this, we need to uncouple the channels, E.G: be able to change the Phase between channels on the H-Bridge and make them 0 Degrees to 180 degrees out of phase from each other.

This means we need FETs instead of a dedicated H-Bridge Chip.

This is awesome! An excellent start for something to aim for!

I would really like Paul Tan to join us and advise on his work if he has time!

All the Best

  Chris

P.S: Update on Links: -->Dual Motor Controller MK5 - Cape design for the BeagleBone<-- (https://www.youtube.com/watch?v=34xJIR-mD4A&feature=youtu.be) Also see Paul Tan's Blog: -->Paul Tan's Blog<-- (http://exadler.blogspot.com.au/)

P.P.S: Some more great news: -->Design Files - Paul Tan - GitHub - Dual Motor H-Bridge with dual Quadrature Encoder cape for BeagleBone Black<-- (https://github.com/Exadler/DualMotorControlCape)

Quote from: MarkE on April 27, 2014, 07:00:16 AM
An IGBT with a 1us turn off time isn't going to do very well hard switching at 7MHz.

Hello All:

Personally, for what it is worth, If I was designing something new I would try to use the cutting edge technology.

I would design the MOSFET driver stage to handle the higher required gate voltage of the NEW SiC MOSFETs.

Specifically I would incorporate the  C2M0080120D Silicon Carbide Power  Z-FET (N-Channel Enhancement Mode)  MOSFET.

PLease see attachment:

Td(on)=12ns
Td(off)=23.2ns
Trise   =13.6ns
Tfall    =18.4ns
Ciss   =950pf 

So the total Input capacitance is 5150pf/950pf= 5.421 x LESS than the H5N3003P MOSFET mentioned earlier.

This equates to less time spent in the Miller plateau when charging the discharging the gate so less heat generated and a faster overall pulse cycle is possible, not to mention the superior thermal characteristics. 

Forgot to mention that the Trr (body diode) = 40ns, pretty damn fast for an integral body diode......

And the D-S breakdown voltage is 1.2KV........


take care, peace
lost_bro

Quote from: lost_bro on April 28, 2014, 08:17:08 PM
Hello All:

Personally, for what it is worth, If I was designing something new I would try to use the cutting edge technology.

I would design the MOSFET driver stage to handle the higher required gate voltage of the NEW SiC MOSFETs.

Specifically I would incorporate the  C2M0080120D Silicon Carbide Power  Z-FET (N-Channel Enhancement Mode)  MOSFET.

PLease see attachment:

Td(on)=12ns
Td(off)=23.2ns
Trise   =13.6ns
Tfall    =18.4ns
Ciss   =950pf 

So the total Input capacitance is 5150pf/950pf= 5.421 x LESS than the H5N3003P MOSFET mentioned earlier.

This equates to less time spent in the Miller plateau when charging the discharging the gate so less heat generated and a faster overall pulse cycle is possible, not to mention the superior thermal characteristics. 

Forgot to mention that the Trr (body diode) = 40ns, pretty damn fast for an integral body diode......

And the D-S breakdown voltage is 1.2KV........


take care, peace
lost_bro

Those are very nice FETs if you are building something like a DC - AC inverter.  If the task is higher current at lower voltage, then they would not be the best choice.

Hi All,

Ok The First Poll Results:

Quote
Do you think a Reliable and Flexible Switching System is a good idea for all to have access to?
Yes                                                <-- Clear Winner 10/10 said yes.
No
Maybe
Too Dangerous for some

Everyone to date agrees this is an important issue and needs a solution.

Please vote for the Microcontroller of your choice. If you have another suggestion please let me know. Remember we are looking the best features for the best price!

All the Best

  Chris

Quote from: MarkE on April 28, 2014, 08:52:31 PM
Those are very nice FETs if you are building something like a DC - AC inverter.  If the task is higher current at lower voltage, then they would not be the best choice.

Hello MarkE

Well then I believe we have to agree on the task......
How LOW of a voltage and how HIGH of a current?

How much current *should* an O.U. device have to switch?

Usually when dealing with D.S. or S.M. devices, my biggest problem has been to switch *High* voltage , not current.

ie: EMF generated from inductor switching can give very *high* voltage spikes.......

These SiC can be paralleled very easily for higher current if need be, and have *almost*  unmatched switching speeds and one of the fastest body diodes available.... and better thermal characteristics then most lower voltage mosfets and Avalanche rated.

....they cost more and driver design is more involved than required for the garden variety MOSFET .

Maybe designing different *Driver Modules* that can be interchanged easily would be the solution.  I for one would want to continue using my SiC MOSFETs!

take care, peace
lost_bro

I may be a tad bit getting ahead of myself, but suppose we added the capability of an oscilloscope/spectrum analyzer here...

Then we would open the door to a fully self optimizing control system.  Imagine connecting this thing to whatever kind of Akula0083 inductor you have available and the system self-tunes.  All we would need are the libraries and the tuning process, code it up and let'r run.


M@

Quote from: lost_bro on April 28, 2014, 09:38:44 PM
Hello MarkE

Well then I believe we have to agree on the task......
How LOW of a voltage and how HIGH of a current?

How much current *should* an O.U. device have to switch?

Usually when dealing with D.S. or S.M. devices, my biggest problem has been to switch *High* voltage , not current.

ie: EMF generated from inductor switching can give very *high* voltage spikes.......

These SiC can be paralleled very easily for higher current if need be, and have *almost*  unmatched switching speeds and one of the fastest body diodes available.... and better thermal characteristics then most lower voltage mosfets and Avalanche rated.

....they cost more and driver design is more involved than required for the garden variety MOSFET .

Maybe designing different *Driver Modules* that can be interchanged easily would be the solution.  I for one would want to continue using my SiC MOSFETs!

take care, peace
lost_bro

I like SiC MOSFETs, they are killer for DC-AC inverter, and vector controlled motor drive designs.  But they are not what I would use for a 100V design.  You should never have to deal with spikes more than 4X Vcc.  I see a lot of people designing stuff where they put a diode around the MOSFET.  That doesn't protect the MOSFET.  When the MOSFET turns off, an inductive load will flyback, always increasing the drain to source voltage.  A properly designed clamp redirects the current typically limiting the voltage to no more than Vcc*2, but in some cases such as in some resonant designs:  4X Vcc.  A lot of the "magic" is in understanding how to do the circuit board layout to minimize stray inductance in the switching loop.

The amount of gate charge that is required depends on the die size and number of devices.  If we just parallel parts then we throw more and more power away at getting the devices to switch.  All high efficiency power converters I am familiar with carefully choose the power switching and energy storage components according to the intended load.  The switch voltage and current rating should be reasonable for the task.  What might be helpful is to come up with as simple and  low cost a board as possible that just has the driver, the switch and any monitoring circuits without the isolation components.  That would plug into a board that has the isolation components.  Then you could have several different incarnations of that simple board using components that are well matched to a reasonably narrow range of specifications.  The trick is to do that in a way that people can work with these things and still keep the inductance down.

Quote from: lost_bro on April 28, 2014, 09:38:44 PM
Maybe designing different *Driver Modules* that can be interchanged easily would be the solution.  I for one would want to continue using my SiC MOSFETs!

Hey Lost_Bro,

Agree Fully! Make it as flexible as possible! Imagine the idea, Plug N' Play FET Slot for easy swap outs of FET's! I think the TO247 Package for the best range of High Speed FET's. We could use a 3 Pin Terminal Block if we had to!

@Dog-One,

I agree. Terminals for easy connection of Scope Probes for Oscilloscope and Spectrum Analyser is really important too!

@MarkE

Quote
What might be helpful is to come up with as simple and  low cost a board as possible that just has the driver, the switch and any monitoring circuits without the isolation components.  That would plug into a board that has the isolation components.

For sure this is a good idea! Even a board that has all the driver Circuits up to the isolation barrier! Then we could have a Plugin board for the switching and we could have as many as one needs of these with different FET's for different Jobs and vice versa for the Driver Circuits!

Problem is this would start getting more expensive the more boards that would need to be used. Most people are doing it tough on the money front and if this is to be a success, we need to do the best we can to keep the cost down!

All the Best

  Chris

P.S: If we have 100 people all join this movement and want a board it will make this much cheaper than 5 or 6. Common Lurkers, Join the party!

Quote from: MarkE on April 28, 2014, 10:08:18 PM
  A lot of the "magic" is in understanding how to do the circuit board layout to minimize stray inductance in the switching loop.

..... carefully choose the power switching and energy storage components according to the intended load.  The switch voltage and current rating should be reasonable for the task.  What might be helpful is to come up with as simple and  low cost a board as possible that just has the driver, the switch and any monitoring circuits without the isolation components.  That would plug into a board that has the isolation components.  Then you could have several different incarnations of that simple board using components that are well matched to a reasonably narrow range of specifications.  The trick is to do that in a way that people can work with these things and still keep the inductance down.

Well I agree, in as much as everyone will have a *different* end use for the device, we must work out a *buildable modular* type system that will accommodate a range of different MOSFETs while keeping an integrated Low stray inductance design of the *switching loop*.

What type of isolation components do you recommend?  Are we talking optos or emulators?

Do we use an integrated mosfet driver with active miller clamping?   What about a driver with integrated schmidt trigger waveform shaping?  pro, cons anyone?

What's the max. frequency we are shooting for?

I think a design for a double ended GDT driver would make a great modular board option.

take care, peace
lost_bro

Hi All,

I am going to share a Circuit that Is NOT mine Its the next step to the Circuit I posted in the first post.

Its from a friend of mine. I can not guarantee this is complete and or working.

However, this is what My Unit is based on in my original post. Its works very well, in fact I have not had a single issue with it!

Its an independently Driven H-Bridge, meaning each FET is switched individually from the Micro Controller.

All the Best

  Chris

Quote from: EMJunkie on April 27, 2014, 07:15:42 PM
Hi MarkE,

I agree! Choosing the best components for the job is essential!

I have used: 5N3003
Turn-On Delay Time: 60ns
Turn-Off Delay Time: 220ns

So far this is one of my favourite N Channel MOSFET's!

All the Best

  Chris

Have a look at this...


~A

Quote from: lost_bro on April 28, 2014, 08:17:08 PM
Hello All:

Personally, for what it is worth, If I was designing something new I would try to use the cutting edge technology.

I would design the MOSFET driver stage to handle the higher required gate voltage of the NEW SiC MOSFETs.

Specifically I would incorporate the  C2M0080120D Silicon Carbide Power  Z-FET (N-Channel Enhancement Mode)  MOSFET.

lost_bro

Could agree more.

~A

Quote from: avalon on April 29, 2014, 06:46:07 PM
Have a look at this...


~A

Hi Avalon

Yes, notice that all the High-speed applications use Surface Mount Technology.... Of course!!!

I personally find it difficult to solder the small SMD type components by hand... 

What do we do about this?

Do we source out the production to a Chinese PCB manufacturer who can populate the pcbs with SMD at the same time?

Just a thought...

take care, peace
lost_bro

Also, while we are on the subject, consider this as well...

Use it with IXDD415 and the result could be quite awesome.

~A

Quote from: lost_bro on April 29, 2014, 07:08:03 PM
Hi Avalon

I personally find it difficult to solder the small SMD type components by hand... 

What do we do about this?

Not easy indeed.
I use my 898D+ for that. It requires a bit of practice but it is doable.

~A

Quote from: avalon on April 29, 2014, 07:31:45 PM
Also, while we are on the subject, consider this as well...

Use it with IXDD415 and the result could be quite awesome.

~A

Hey Avalon,

Nice unit but the RDS on is 1.6 Ohm we need a way lower RDS on 1.6 mOhm maybe.

All the Best

  Chris

Quote from: EMJunkie on April 29, 2014, 07:37:10 PM
Hey Avalon,

Nice unit but the RDS on is 1.6 Ohm we need a way lower RDS on 1.6 mOhm maybe.

All the Best

  Chris

You are right. Besides, as always we should consider a reasonable price-reward ratio.

~A

Quote from: avalon on April 29, 2014, 07:31:45 PM
Also, while we are on the subject, consider this as well...

Use it with IXDD415 and the result could be quite awesome.

~A

Good day Avalon

That is one kick-ass integrated driver chip.......Input signal-T T L or C M O S compatible.

Each output of the IXDD415 can source and sink 15A of peak current while producing voltage rise and fall times of less than 3ns.

And The voltage range is from 8V to 30V so, it is compatible with the higher voltages needed for SiC MOSFETs.

Just need to work on the isolation part....

I have used the IXDD family of drivers and they are IMPRESSIVE.

take care, peace
lost_bro

Quote from: lost_bro on April 29, 2014, 08:31:57 PM
Good day Avalon

That is one kick-ass integrated driver chip.......Input signal-T T L or C M O S compatible.

Each output of the IXDD415 can source and sink 15A of peak current while producing voltage rise and fall times of less than 3ns.

And The voltage range is from 8V to 30V so, it is compatible with the higher voltages needed for SiC MOSFETs.

Just need to work on the isolation part....

I have used the IXDD family of drivers and they are IMPRESSIVE.

take care, peace
lost_bro

As far as I am concerned, these are the best you can get.
For quick and dirty apps I routinely use IXD_609 and very happy with the results.

~A

Quote from: avalon on April 29, 2014, 09:12:11 PM
As far as I am concerned, these are the best you can get.
For quick and dirty apps I routinely use IXD_609 and very happy with the results.

~A

Good evening Avalon

Yes, the IXD_609SI has the integrated schmidt trigger logic for waveform shaping....

I just purchased a lot of IXDD614CI which likewise have the Schmidt logic, I like the CI package because it is a
TO-220 5 pin package so I can heat sink mount along with the MOSFETs..... works like a charm.

take care, peace
lost_bro

Quote from: lost_bro on April 29, 2014, 09:28:14 PM
I like the CI package because it is a
TO-220 5 pin package so I can heat sink mount along with the MOSFETs..... works like a charm.

What a great idea! I am definitely going to get some of those.
Thanks for a super advice.

~A

@All

The TheOldScientist has done some videos on the schematics I have posted. See Below:

-->The Modular Quadratron Board from Tantratron, Part 1 <-- (https://www.youtube.com/watch?v=IiJewZ4TuwY)
-->The Modular Quadratron Board from Tantratron, Part 2 <-- (https://www.youtube.com/watch?v=osVvr5_XlFI)
-->The Modular Quadratron Board from Tantratron, Part 3. The Half Bridge Topology revisited<-- (https://www.youtube.com/watch?v=KkS9whdh0_0)
-->The Modular Quadratron Board from Tantratron, Part 4. The Demux Controller board<-- (https://www.youtube.com/watch?v=-sTdXgQLFuU)

Again, this is the circuit I have based my design on.

All the Best

  Chris

Anyone want to list some pro's and cons for Half bridge and Full Bridge switching solutions for inductive loads ?

Pro's and cons for transformer gate drives and Hi-Lo IC gate drivers ? eg. IRS2110 http://pdf1.alldatasheet.com/datasheet-pdf/view/145601/IRF/IRS2110.html

..

Quote from: Farmhand on May 03, 2014, 11:28:24 PM
Anyone want to list some pro's and cons for Half bridge and Full Bridge switching solutions for inductive loads ?

Pro's and cons for transformer gate drives and Hi-Lo IC gate drivers ? eg. IRS2110 http://pdf1.alldatasheet.com/datasheet-pdf/view/145601/IRF/IRS2110.html

..

A single transistor and flyback diode give you a backwards sawtooth current waveform:  The current rises quickly (assuming that the drive voltage is much greater than Irated/Rwinding) when the transistor is on and decays slowly recirculating through the inductor and the flyback diode.  The minimum current is set by the input voltage, the coil inductance, the switch minimum on time, and the switching period.

A half-bridge can manage zero average current in the inductor.  You need to make sure that the top and bottom transistors do not conduct at the same time or else you will have shoot through currents that can be very large and destructive.  Half-bridges are common in switching power supplies, and used in sets of three for three phase power supplies and motors.

A full-bridge can manage current in either direction in the inductor.  It is more flexible in that regard than a half-bridge.  It has the same issues as a half bridge, and current must flow through two switches increasing conduction losses.  Full bridges are the preferred form for DC motor and hybrid stepper motor drives.  The ST Micro L298 dual H bridge has been around for almost 30 years.  There are now much more advanced devices for steppers.

Transformer gate drives cannot support DC operation and fast turn-on and turn-off.  There are configurations of transformer drive that support DC using rectifiers and filters on the gate side of the transformer, but they have slow turn-off. 

ICs support DC operation for low side drivers.  Many high side driver ICs drive N channel MOSFETs using charge pumps.  These require that the low side switch turn on periodically in order to recharge the charge pump.  That can be gotten around by various means if needed using extra circuitry.

Use what works for the application and that you can easily obtain.

Thanks Mark, I might use the Data sheets and draw up a driver and Half Bridge switch circuit. I already have some IRS2110 driver chips but they are not very high current parts, maybe they can work with moderate frequencies driving IRF460's or FGA25N120-ANTD (IGBT).

If I post a schematic and test the circuit and switching wave forms maybe some folks can suggest improvements. I remember last time i set up an IRS2110 chip and switches I had some trouble with the high side drive initially.

Ta

Looking at this page http://www.powerguru.org/the-half-bridge-circuit-revealed/ and the circuit in the data sheet pdf below I made these drawings, I must confess using some kinds of IC's confuses me and this is one of them.

Can anyone see any problems with the circuit for the IRS2110 ? I know some gate resistors may be needed, if I could get operation up to about 25 or 30 kHz I would be happy. Even if I could get good operation up to a few hundred Hz I would be satisfied to begin with.

The circuit with the two batteries just signifies a split EMF. The other one with the capacitor in series with the load coil is kinda like the one from the web page above, he uses a different driver chip.

Quote from: Farmhand on May 06, 2014, 05:07:15 PM
Looking at this page http://www.powerguru.org/the-half-bridge-circuit-revealed/ and the circuit in the data sheet pdf below I made these drawings, I must confess using some kinds of IC's confuses me and this is one of them.

Can anyone see any problems with the circuit for the IRS2110 ? I know some gate resistors may be needed, if I could get operation up to about 25 or 30 kHz I would be happy. Even if I could get good operation up to a few hundred Hz I would be satisfied to begin with.

The circuit with the two batteries just signifies a split EMF. The other one with the capacitor in series with the load coil is kinda like the one from the web page above, he uses a different driver chip.

So one thing that you need to make sure of is that the circuit switches low periodically to recharge C4.  I have not looked to see if the driver chip you are using has undervoltage protection for both the high side and low side driver.  Another thing that you should consider is a two resistor and one Schottky diode network in series with each gate to suppress oscillations and optimize the waveshape.  But the BIG thing here is that you have placed your load in series with a capacitor.  The circuit you are deriving from had two stacked power supplies.  If you don't want to have stacked power supplies, and your load isn't across your output capacitor, then build a full bridge.

Thanks Mark I appreciate the tips, I think SeaMonkey already tried to school me on this and from memory I think had this IRS2110 Hi-Lo driver chip working with a capacitor voltage doubler but I failed to draw the circuit that worked for me. I'm having issues but I need to work it out myself to learn properly, I think. So I found a couple of application notes pdf's on these drivers, at first glance they might help me, if not I'm missing something, I think I'll check my solderless board for cross rail "shorts", for some reason both mosfets are half on with the circuit not even powered up. I pull em out they turn off.

I'm using the signal splitting/dead time control part of the circuit below to process the picaxe output into two 48% duty out of phase signals at 10000 kHz to test the driving circuit. I can adjust the dead time if need be to keep appropriate dead time at the switches if I can get the driver IC to work properly  ::) signals out of the CD4001 are very good.

Anyway maybe the application notes will help someone else as well.


Driver IC application notes.
http://www.irf.com/technical-info/appnotes/an-978.pdf

http://www.irf.com/technical-info/appnotes/an-1123.pdf

..

Quote from: Farmhand on May 16, 2014, 06:21:10 AM
Thanks Mark I appreciate the tips, I think SeaMonkey already tried to school me on this and from memory I think had this IRS2110 Hi-Lo driver chip working with a capacitor voltage doubler but I failed to draw the circuit that worked for me. I'm having issues but I need to work it out myself to learn properly, I think. So I found a couple of application notes pdf's on these drivers, at first glance they might help me, if not I'm missing something, I think I'll check my solderless board for cross rail "shorts", for some reason both mosfets are half on with the circuit not even powered up. I pull em out they turn off.

I'm using the signal splitting/dead time control part of the circuit below to process the picaxe output into two 48% duty out of phase signals at 10000 kHz to test the driving circuit. I can adjust the dead time if need be to keep appropriate dead time at the switches if I can get the driver IC to work properly  ::) signals out of the CD4001 are very good.

Anyway maybe the application notes will help someone else as well.


Driver IC application notes.
http://www.irf.com/technical-info/appnotes/an-978.pdf

http://www.irf.com/technical-info/appnotes/an-1123.pdf

..

Those are excellent application notes that cover the subject of the high side bootstrap power supply very well.  10MHz is very fast for power MOSFETs.  Is that your timing resolution, or your cycle time?  A good frequency range to work with MOSFETs without too much worry is between 50kHz and 300kHz.  Slower and the output filter energy starts to get big.  Faster and you have to start paying a lot more attention to layout issues which are much worse using a solderless breadboard than on a circuit board.

OK Now I see the problem, I'm sure you'll find it easy to believe that I had the driver chip upside down hahaha, no harm done it works exactly as it should I think in bootstrap mode, it can be used with a floating supply as well.

The drive circuit posted above with the split supply is basically from the data sheet and works ok but I'll check it to be sure.

I used a 1 uF bootstrap capacitor, and the test setup is 12-24 volts supply.

No not 10 mHz, 10 kilohertz from the picaxe to the signal splitter and 5 kHz to each mosfet 48% duty for the test to see if i have the circuit right.  :) scope capture attached.

Some more reading and I can set to work.

P.S. IRF840's were being driven while the shot was taken.

..

Quote from: Farmhand on May 16, 2014, 08:04:42 AM
OK Now I see the problem, I'm sure you'll find it easy to believe that I had the driver chip upside down hahaha, no harm done it works exactly as it should I think in bootstrap mode, it can be used with a floating supply as well.

The drive circuit posted above with the split supply is basically from the data sheet and works ok but I'll check it to be sure.

I used a 1 uF bootstrap capacitor, and the test setup is 12-24 volts supply.

No not 10 mHz, 10 kilohertz from the picaxe to the signal splitter and 5 kHz to each mosfet 48% duty for the test to see if i have the circuit right.  :) scope capture attached.

Some more reading and I can set to work.

P.S. IRF840's were being driven while the shot was taken.

..

5kHz sounds much more reasonable.  The waveforms show the dead time that you intend.  It sounds like you are off to the races.

Seems to work ok, I did put 10 Ohm resistors between the driver outputs and the gates, at lower frequencies like under 400 Hz the hi gate drive drops a fraction I think. It seems to switch transformers quite well, I tested a ferrite core transformer up to 200 kHz and also a steel core at 400 Hz. The 10 Ohm resistors seemed to take out a lot of the noise I could find at 200 kHz

I've noticed before that the gate driver chips I used for low side switches (TC4420, 6 amp parts) can get quite warm to hot when I try to use them at over about 900 kHz, and previously I have used two low side mosfets out of phase to switch the same coil so as to share the load and better handle the higher frequency I wanted.

So I'm thinking that it could maybe be beneficial to make a four phase signal to switch two half H bridges one after the other to share the load of switching the same coil, all the previous components in the signal train seem to be able to handle higher frequencies no problem. Effectively either double the reliable switching frequency or halve the work on the switches and driver chips.

Keeping dead time minimal while avoiding shoot through current seems to be the challenge, I'm sure that can be better adjusted with a different signal arrangement though. If I power the signal components from 15 volts rather than 5 volts they should work quicker. Only the micro is limited to 5 volts. Unless I'm reading the data sheet wrong it looks like a CD4049 chip which I have, if it is powered by 15v can translate a 5 v signal into a 15v signal, I think the data sheet is telling me it will do the opposite so I'll have to try it and see.  :)

..

Hi All,

What Microcontroller do you think is best for this project? Price wise and Feature wise?

BeagleBone (Black Series)   -     3 (75%)
FEZ Hydra                          -     0 (0%)
Arduino Mega                     -     1 (25%)

All the Best

  Chris

@All - I have been in conversation with my friend Albert. We have been talking about pricing on the setup he has.

Can anyone make any recommendations on a cheap Chinese Manufacturer that can also place the components?

All the Best

  Chris

Hi EM, is that a setup you posted previously in the thread ? I was thinking about the poll and maybe the low votes was because in essence all the micro's do basically the same stuff so it's difficult to say. I'll buy and use whatever I can get code for or write code for.  :)

..

Quote from: Farmhand on May 20, 2014, 04:03:13 AM
Hi EM, is that a setup you posted previously in the thread ? I was thinking about the poll and maybe the low votes was because in essence all the micro's do basically the same stuff so it's difficult to say. I'll buy and use whatever I can get code for or write code for.  :)

Hi Farmhand,

Yes but the V5 Version. I may have posted V1 or something.

Not sure I understand you fully, if I do understand, then I would have to disagree.

Microcontrollers are awesome. Micro's are the heart of any system, or better, the brain. At all costs we need to protect our Microcontrollers. The whole point of this thread is to achieve a result and that result is to use a Microcontroller that can control a Highly Flexible Highly Reliable Switching Unit.

A unit that can do most anything we want it to do, this can be controlled by the Microcontroller...

So a Micro controls a top of the line Switching system, this switching system can switch all sorts of Frequencies, it can do all sorts of Duty Cycles, It can be an Isolated H-Bridge, Isolated Half H-Bridge, Single Isolated Switch and so on. Multiple Frequencies on each Isolated Single Switch...

So This Switching System is only a part of the equation, not the full System and that's why I wanted to know what people thought of the Microcontrollers, remembering some are much better for this sort of thing than others. For example Raspberry Pi is not one I would use, Arduino Duemilanove/Arduino Uno is last on my list, many Micros are not really satisfactory for a project like this. Of course this is my opinion.

There is no way a Microcontroller can switch any sort of Load, maybe at most 1/2 a watt, with most Microcontrollers. The Tantratron is rated to 1600VA.

The Micro might shut down the entire system in the event of an overheat for example...

All the Best

  Chris

Hi Chris,


I'm the designer of the circuits you mentioned in the start of this thread.
The design is well proven so far in many hard switching cases, up to 500kHz. Soft switching could go to few MHz.
Of course to every design there are improvements possible, but the point is that the design has proven to provide good results, and is flexible and modular.


Attached is a picture of the latest board we use.
I'm using several such boards (self-made), which are called now "IPC" (Isolated Power Channel) and comes in 4 or 8 drivers, so IPC-quadra and IPC-octa.
Many people are suffering by trying to build themselves H-bridges, but have either no galvanic isolation, is not modular, or does not work reliable.
The intend I had, together with my friend and partner Selfonlypath, is to create the above for our private research.
However we understand the community could benefit from this as well.


I wonder if there would be an interest if I would start manufacturing and selling these boards to anyone who is interested in this stuff?


Note that the board has an Arduino Mega landing. I know the Mega is not the fastest and coolest on teh market, but it is sort of a standard, which is also important for modulatory.
The good thing about the ATmega is that the pulses are predictable with clock-precision, which is a must. This is not the case where boards are used with an O/S (like Rasperry Pi).


In fact we are checking if our future platform could be FPGA based. In this sense, I'm closely following the development of a new board by Jack from Gagdetfactory - see http://forum.gadgetfactory.net/index.php?/topic/1876-the-next-generation-papilio-help-me-shape-it/


In anycase some additional analog circuitry is required as front-end for the measured feedback signal conditioning, and threshold control.


I think there could be value if we share and/or sell these.


What do you think?


-Dan

I think that there is value in a well documented and well behaved driver for experimenters.  Finding the right combination of price and performance might be a little tricky because I think that most experimenters do not realize all the things that can go wrong with power drivers and therefore why it is worth some money to buy a well engineered one.

Quote from: dancombine on May 28, 2014, 11:54:13 AM
Hi Chris,


I'm the designer of the circuits you mentioned in the start of this thread.
The design is well proven so far in many hard switching cases, up to 500kHz. Soft switching could go to few MHz.
Of course to every design there are improvements possible, but the point is that the design has proven to provide good results, and is flexible and modular.


Attached is a picture of the latest board we use.
I'm using several such boards (self-made), which are called now "IPC" (Isolated Power Channel) and comes in 4 or 8 drivers, so IPC-quadra and IPC-octa.
Many people are suffering by trying to build themselves H-bridges, but have either no galvanic isolation, is not modular, or does not work reliable.
The intend I had, together with my friend and partner Selfonlypath, is to create the above for our private research.
However we understand the community could benefit from this as well.


I wonder if there would be an interest if I would start manufacturing and selling these boards to anyone who is interested in this stuff?


Note that the board has an Arduino Mega landing. I know the Mega is not the fastest and coolest on teh market, but it is sort of a standard, which is also important for modulatory.
The good thing about the ATmega is that the pulses are predictable with clock-precision, which is a must. This is not the case where boards are used with an O/S (like Rasperry Pi).


In fact we are checking if our future platform could be FPGA based. In this sense, I'm closely following the development of a new board by Jack from Gagdetfactory - see http://forum.gadgetfactory.net/index.php?/topic/1876-the-next-generation-papilio-help-me-shape-it/


In anycase some additional analog circuitry is required as front-end for the measured feedback signal conditioning, and threshold control.


I think there could be value if we share and/or sell these.


What do you think?


-Dan

Welcome Dan!

I think this thread has proved about the value of a Reliable and Flexible Switching System! People that have an idea about the area of research we all are investigating already see value here.

Isn't it funny how thing go around in circles and come back to the start sometimes! Three or so years ago I said to Albert how beneficial a Board like yours could be to people.

Please feel free to post your product range when you're ready. I am sure people will buy.

All the Best

  Chris

Quote from: MarkE on May 28, 2014, 05:42:03 PM
I think that there is value in a well documented and well behaved driver for experimenters.  Finding the right combination of price and performance might be a little tricky because I think that most experimenters do not realize all the things that can go wrong with power drivers and therefore why it is worth some money to buy a well engineered one.

Hi MarkE,

I fully agree. Its critical to have something that will suit most all purposes.

Have we come to any hard decisions in the thread other than "Reliable and Flexible Switching System", not really. We don't have an improved design, we don't have any Circuits Designed, really we are no further than the first thread that I posted a few months back.

I wonder why some groups of people can get things done and other groups of people can not?

Procrastination or fear of being ridiculed, I don't know. Isn't it a shame how so many people can join for the good cause but not really do anything...

Apologies for my attitude today, I have seen some rather disheartening things occur in the last few weeks and it saddens me. Rant now over...

All the Best

  Chris

Chris I think that the big issue is setting out constraints.  I would think that something capable of 100W peak hard switching is probably reasonable:

Say 32VDC input max, 4A current, half bridge or full bridge.

Half bridge solves the question of what to do about flyback protection.

Then the next issue is over current fault protection.  Some sort of cycle by cycle and burp burp for severe faults is really critical as part of the driver.

Then the next issue is what measurements if any feed back to the controller.

I think that once specified the design work is not demanding.  The best performance can be had from surface mount parts, but that depends on whether you want to supply finished assemblies or kits or instructions only.

Chris, MarkE,


The design I have was improved over the years and is very stable. I made several prototypes and versions, and works now very well.
here's a picture of v5 of the board: http://users.skynet.be/fc372083/public/boards/2014-05-29%2009.32.37.jpg

Since then I made some minor adjustments and the board is now called IPC-quadra v6.
The changes are mostly some small layout changes (not related to the core of the isolated power switching), and make it universally usuable to interface with any control board (Arduino, Beaglebone, etc.)
The specs are:
- 1kV galvanically isolated for control signals
- 1kV galvanically isolated between all 4 power channels (as such user is flexible on how to connect those together, either H-bridge, push-pull, etc...)
- control signals 3 - 5V compatible
- power section: max 600V & 4A (10A with re-enforced copper traces by soldering) (so 6kVA reachable, and tested in vitro as such)
- operating frequency: DC to 1MHz (depending on choosen power transistor)




I'm willing to make a batch of 25 such boards, if there would be enough people interested.


Regards,
-Dan

Dan, maybe the thing for you to do is publish specifications and then do an IndieGoGo or Kickstarter campaign to cover your internal cost for 25 - 100 boards and something for your time.  If the campaign fully funds then you can offer to sell the boards for what you think is a reasonable price. 

Hi MarkE,


yes I've been thinking about a Kickstarter project.
The practical drawback for you guys is that this will take time before boards can be available.


If there is a small community of 25 interested, I can start a small batch quickly, and later move to a larger Kickstarted based volume.


-Dan

Dan, you are being generous by taking such a risk on people's say so.  In your shoes I would collect enough deposits first to at least cover the fixed costs.  Essentially you could set-up KickStarter to do that.

Ok Guys,

Please let us know who is interested! See attached File. No PLL, No Microcontroller, this is bare board with Components Soldered and ready for you to plug in your gear.

Price coming tomorrow when confirmed.

All the Best

  Chris

P.S: Poll Results:

How Much would you be happy to pay for a high quality Switching Unit with advanced PLL?

$2000                 -  0
$1000                 -  0
$500                  -  2
$300                  -  0

It looks like you have done a lot of diligent work.  Good luck.

Hi All,

NOTE: This is not my Board and not my design. I am ONLY providing an OPTION for you to look at. You are responsible for making your own decisions and coming to your own hypothesis. I can say, there is nothing this good on the market that I have found!

IPC-quandra Price: €276.00

Includes: PCB, all components soldered & tested, including power IGBT and diodes
Optional: demux (2 IC & BNC connector)

I can provide contact details for purchasing if you are interested.

All the Best

  Chris