[websrvr]

I need all your eagle files to modify your layout so I can produce some boards.

[tivoidethuong]

I need all your eagle files to modify your layout so I can produce some boards.

okies no prolem

[websrvr]

okies no prolem

Are you going to be attaching them any time soon?

[tivoidethuong]

Are you going to be attaching them any time soon?

my post inclue eagle file, you can used ulp for export libary and modyfied it

b.r

[websrvr]

You post with the zip file has your eagle board file but is missing the eagle schematic file, surely you're not suggesting I make extensive modifications to your eagle board file without the schematic?

[tivoidethuong]

resent file cadsoft

[websrvr]

resent file cadsoft

Your files have been updated.

Some of the changes:

• Modified some part footprints.
• Swapped fets for less heat.
• Removed GND traces and added ground plane for better voltage and current stability.
• Stopped traces to increase current carrying capabilities.
• Moved some parts and now all TO220 devices are bottom mounted with tab down (away from board).
• Gate resistors are now 10 ohms.
• Better performance using IRS2104S.
• Renumber some components.

[yngndrw]

I found that the TO-262 version of the IRF540 is better for mounting on the back as the tab is shorter and doesn't get in the way:
IRF540NLPBF Infineon Technologies | Discrete Semiconductor Products | DigiKey

[websrvr]

I found that the TO-262 version of the IRF540 is better for mounting on the back as the tab is shorter and doesn't get in the way:
IRF540NLPBF Infineon Technologies | Discrete Semiconductor Products | DigiKey

The TO262 version must mount with the tab against the PCB, in the updated board, they are mounted on the bottom but the tab is away from the PCB (the body contacts the PCB) and the PCB sandwiches it to the heatsink.

In this case a 3/16in aluminum plate is sufficient since the IRDs is so low that heat generation isn't an issue at 10A, the IRF540 get hot at 6A and requires a real heatsink to keep cool.


You shows a good example of dual H-Bridges using mostly SMD devices and over isolation never hurt anything except size but this design is only suitable for driving a DC motor and since your files are specific to OSHP, doing anything with it isn't recommended unless you specifically wish to tie the design and production to OSHP.



While it is nice that the quad H-Bridges have been updated for better performance and less heat and can still be updated to full SMD devices for a compact size, I'm wondering if this project is really going anywhere and if it is, when will the CPLD or FPGA will be added to the board and a usable product is available or has this project become all about H-Bridges and making a stable, high quality 6A/60V bipolar stepper driver nothing more than a fantasy?

[yngndrw]

I have them mounted on the bottom of the PCB with the tab away from the PCB like you mention. They are held against the heatsink by the PCB alone, as they are in the Geckodrive G540s.

I can't comment about the use of the IRDs as I'm not well-up enough on MOSFET selection, but Mariss warned against using some MOSFETs based on their fragility and their SOA.

I've been using my H-Bridges for steppers and they are based off of the Gecko design that Mariss released, the design files are OSH Park - You can export the RAW gerbers from Circuit Studio or fork and modify the boards as you wish.

The reason I made the H-Bridges separate is because I'm using an FPGA development board (The Digilent Nexys 3) and therefore wanted something generic that I could re-use. Personally I've just been using this to learn about CPLDs and FPGAs but there's a few other more complete designs floating around. I'm not sure it's practical to make a product out of this as you'd never be able to make it viable without a unique selling point. If you tried hard enough, you could probably save money in exchange for your own time, however.

[websrvr]

I have them mounted on the bottom of the PCB with the tab away from the PCB like you mention. They are held against the heatsink by the PCB alone, as they are in the Geckodrive G540s.

I can't comment about the use of the IRDs as I'm not well-up enough on MOSFET selection, but Mariss warned against using some MOSFETs based on their fragility and their SOA.

I've been using my H-Bridges for steppers and they are based off of the Gecko design that Mariss released, the design files are OSH Park - You can export the RAW gerbers from Circuit Studio or fork and modify the boards as you wish.

The reason I made the H-Bridges separate is because I'm using an FPGA development board (The Digilent Nexys 3) and therefore wanted something generic that I could re-use. Personally I've just been using this to learn about CPLDs and FPGAs but there's a few other more complete designs floating around. I'm not sure it's practical to make a product out of this as you'd never be able to make it viable without a unique selling point. If you tried hard enough, you could probably save money in exchange for your own time, however.

You've got three of the entire 6A@60V drivers in your G540 or just three of the H-Bridges?

From looking at your H-Bridge board, it's only dual and with steppers you need quads so I fail to see how you can drive a bipolar stepper motor with one board.

As nice as it sounds to have a dependable 10microstep driver, finding a complete solution that you can build and utilize is not as easy as you might think.

I have examined a couple of others and I have yet to find a solution that you can produce inexpensively for one reason or another and only being provided gerber files for a something that is outdated and could use some enhancements is also discarded.

This is about the only project that appears to be complete and relatively inexpensive but there are some issues with the original CPLD code which has never been addressed and the switch to an FPGA without updating the whole design makes it incomplete so as far as I can tell, there doesn't appear to be a usable product.

Ideally, 6A@60V / 10microstep makes for a nice driver which can be used in many application.

[yngndrw]

Sorry I meant the G250, don't know why I said G540. They are indeed a single H-Bridge so I mount two to a single heatsink. (Just using a cheap used CPU heatsink as they are a perfect fit)

I've actually found that the current CPLD code (With the couple of fixes mentioned in the other thread) actually works very well. Obviously it doesn't have the mid-band correction, but aside from that it's very smooth, mostly quiet and works well - Defiantly an upgrade from the standard "all-in-one" IC drivers and runs much cooler. I've not noticed any real heating of the drivers themselves, just the motors that I've been using.

What code issues are you referring to ? I've been running the code as given on a Spartan 6 FPGA without any issue - I just replaced the clock with a PLL/divider module from my dev board's built-in 100Mhz clock.


Speaking of Gecko Drive, they look to have just released a new FPGA-based drive - The G214. I love the DIP switches for current control but there's still no mention of the closed loop drivers which were mentioned a few years ago - I wonder what happened to those.

[websrvr]

Here's an updated board layout with SMD for all but the TO220 components and the electrolytic capacitor, trace sizes have been increased where applicable and has been tested to support 8A without any heat issues.

The driver fets have a low IRDs and are suitable for motor driver applications where minimal cooling is required, the IRF640 generates more heat.

The board has been changed to 3.5in x 3.5in providing plenty of space to reduce noise and increase cross channel rejection at high speed and also make manual/hand assembly possible.

The board was also tested with a +12V driver source instead of +15V with no discernible differences in performance but has been left intact for the purpose of posterity.

Included is the library with updated packages used in this layout so you wont have any issues making modifications as the complete project has been provided.

[websrvr]

[QUOTE=yngndrw;1983570]Sorry I meant the G250, don't know why I said G540. They are indeed a single H-Bridge so I mount two to a single heatsink. (Just using a cheap used CPU heatsink as they are a perfect fit)

I've actually found that the current CPLD code (With the couple of fixes mentioned in the other thread) actually works very well. Obviously it doesn't have the mid-band correction, but aside from that it's very smooth, mostly quiet and works well - Defiantly an upgrade from the standard "all-in-one" IC drivers and runs much cooler. I've not noticed any real heating of the drivers themselves, just the motors that I've been using.

What code issues are you referring to ? I've been running the code as given on a Spartan 6 FPGA without any issue - I just replaced the clock with a PLL/divider module from my dev board's built-in 100Mhz clock.


Speaking of Gecko Drive, they look to have just released a new FPGA-based drive - The G214. I love the DIP switches for current control but there's still no mention of the closed loop drivers which were mentioned a few years ago - I wonder what happened to those.[/QUOTEAs you mention, mid-band compensation/correction is missing and the other mentioned fixes have not been applied to the code-base.

The Spartan 6 is more powerful than required in this application however I understand the logic of utilizing resources on hand especially one that packs a punch.

[websrvr]

Sometimes it's worthwhile to pay a few extra dollars to get something that's a step above the rest without getting into a commercial price-tag, finding a suitable 10microstep driver is one of those things, unfortunately, there doesn't seem to be a complete solution, this thread is a prime example, and just like many others, it's incomplete for one reason or another so rushing out and producing boards isn't advised.

We've seen some updated H-Bridges suitable for the project but the rest of it seems to have fallen by the wayside, the CPLD portion is removed from the design, the original design doesn't provide any eagle files and no FPGA alternative has been created and aside from your Spartan 6, finding a stable, reliable and basically bug-free project without any performance issues is what prevents people from looking at the driver as a viable solution and hoping for a day that some one posts one.

Performance is an issue with the CPLD code not being updated and while it is suggestive of potential fixes and solutions, not applying it to the code makes it useless to most people.

The H-Bridge design I posted is definitely capable of providing 10A without excessive heat or requiring a massive heat sink for cooling, using the TK40A06N1's, sub-ing out the IRF640 for another one that would generate less heat or removing it entirely and adding a 60V to +12V buck and driving the FET's with the IRS2104S's as recommended makes more sense but again, no complete eagle files exist so making the changes more time consuming and annoying.

Of course if I had black and white images of the original board layout and a full schematic in eagle format it would be fairly simple to get a layout done suitable to work from.

I've had several inquiries about this project and it seems I'm not the only one who believes this project has value but disappointed in that no useable solution is available to build after more than five years.

You could argue everything is provided and it most likely is, the fact that no files can be use to generate a board immediately without doing some major work is why it's useless and the code not being updated with all the suggested fixes is why it's also useless to the majority of users.

Basically if eagle files are available, I can do a lot myself but also have someone who's fluent and skilled who can rework the design quickly but neither one of us has the time or interest to draw anything from scratch and build 50 iterations for testing and development to get it to a point it can be built and used.

Drawing packages for production places like OSHP are OK for small runs during initial project development stages but a quick project killer for production usage since you can't easily alter the design and no one else accepts the files.

I have outstanding production capabilities, what I don't have is a product to produce that is known to work without any major issues so like everyone else interested in such a driver, my projects sit on the back burner waiting for such a project to become available.

[tivoidethuong]

websrvr, i advise proasic3 fpga

[websrvr]

websrvr, i advise proasic3 fpga

Do you have eagle files and code that use this FPGA?


If not, advising it isn't helpful unless you making the schematic in eagle, building a couple of samples, tweaking your code to perfection and testing the hell out of them in a short period of time.

[tivoidethuong]

now i use epm240, i will replace proasic3 soon



b.r

[websrvr]

now i use epm240, i will replace proasic3 soon



b.r

The EPM240T100 (C5N or I5N) I'm told is an inexpensive and suitable FPGA for this kind of project, do you have eagle files for this?

A set of eagle files that will produce a board and code that works are all pretty much what people are looking for yet you are talking about future MCU changes and while it might be great, you don't have a prototype and working code or eagle files for the proposed board so it's not helpful.

A working solution which includes the eagle files to produce the board is what makes a project of interest to the general hobby community who is looking for a solution now, not in 2 years time.

Turning a working solution into a 3-axis or 4-axis board doesn't require a lot of work but if you don't have files to start from working solution it's a waste of time for any immediate results.

[websrvr]

As much as I hate to say it, this project is just not suitable for any real use purpose since there does not appear to be a complete set of eagle files in which the board can be produced from.

While it's nice to hear that updates and upgrades are being considered and changes in the FPGA/PLC to enhance functionality are being considered, without files to produce a working board, this project is considered a complete waste of time.

To suggest that another project is better suited without providing a link to the project is arrogant and insulting to say the least, many people are looking for a 10-microstep stepper driver and I have yet to see any project that yields a working product, even this one cannot be confirmed because there is no way to produce a working sample due to the lack of eagle files so I must join the ranks of the disappointed and continue my search for the mythical device.