[kreutz]

After the first set of tests, I have an small list of changes to make for next prototype revision 3.0j including some already made while I was waiting for this prototype board to arrive from the fab. Most of the changes are related to board diagnostics and serviceability, including more diagnostic LEDs for status signals and Emergency stop.

Other changes are related to more SMD integration, SMD footprints for 7 capacitors, CPLD and Attiny2313, and adding on board programming connectors.

Adding more LEDs for status signals will ease Customer Support for phone or e-mail initial troubleshooting in case of problems.

Other board changes are related to thermal management, we need a bigger board, and/or integrated fan cooling. I am still thinking about placing the power section on a separated board.

Will keep you posted.

[kreutz]

Back to the drawing board for the next prototype PCB revision

I am working on the power output now.

The power output section includes also the power supply conditioning and monitoring circuit including the 12 volt to 5 volt conversion, and optional heat-sink fan connectors.

I am replacing the current adjustment and reference voltage generation by a dual D/A converter and only one potentiometer per axis.

The logic section integrate all four translators and automatic Slow/Fast decay logic, as well as most safety circuits, including the charge pump and PC isolation and interface (I will continue to use the ADUM1400 series). The firmware is going to be totally rebuilt to make it faster, in preparation for a future 200 Khz step pulse frequency. No compatibility at all with the M-K unipolar firmware is planned.

[kreutz]

More test results.

Tested with my original KL23H286-20-8B motor in bipolar parallel configuration at peak current of 3.94 Amps /phase (2.8 Amp r.m.s. times 1.41 ) = 3.94 Apeak. We are still using the automatic torque compensation feature introduced into the Mardus-Kreutz firmware. I will post a video probably during the weekend. This time without the flywheel.

[contactirfu]

More test results.

Tested with my original KL23H286-20-8B motor in bipolar parallel configuration at peak current of 3.94 Amps /phase (2.8 Amp r.m.s. times 1.41 ) = 3.94 Apeak. We are still using the automatic torque compensation feature introduced into the Mardus-Kreutz firmware. I will post a video probably during the weekend. This time without the flywheel.

Hi Kreutz,

was the tests performed with updated firmware? the "L's" still heat up ! dont they. looking forward for the video!

RGDS
Irfan

[kreutz]

Hi Kreutz,

was the tests performed with updated firmware? the "L's" still heat up ! dont they. looking forward for the video!

RGDS
Irfan

The L6203 heats up in the low rpm and step accuracy tests, that is an expected result because of the higher motor current and chopping under those conditions. Motor heating has not been a problem, there is less motor heating than under unipolar driving.

Tested different blanking times, between 0.9 uS and 1.5 uS, we have reduced the motor noise to a minimum, but under some conditions it still shows. Since I don't like it, we will continue working on a solution to the sub-harmonic chopper operation that produces the audible noise.

Tests also showed that under a full "slow decay" or "fast decay" configuration, the performance suffers a lot, the best solution is mixed decay, as we implement it depending on the full step switching quadrant and motor direction. I am working on a mixed decay auto percentage compensation that might even be patentable, but is not tested yet.

[kreutz]

CPLD fit report

[harryn]

Hi, I have been following your thread for a while, as a "normal" 4 axis driver solution is out of my budget range, but this one seems to be within it, or close.

I am thinking about using this motor Kelling kl34H2120-42-8B, which I think fits in or close to your design range.

My router will be driving the motors rather slowly to obtain max torque, commonly 3 rpm to max 50 rpm over a total 10 turns range. One of the unique quirks of the design is that it is direct drive, so no gear reduction at all, in a sort of rack and pinion. ( There are good and bad reasons for this, but this is the current concept)

In a perfect world, I would like a driver with at least 1/64th or smaller microsteps, and would always run it in that mode. I realize this is out of your design range, but us beggers can't be choosers.

Anyway, I was thinking about buying a motor and loaning it to you for shaking down your driver board using a motor with a bit more load demand and maybe doing some low speed testing. It is ok if your driver board needs to run it in a somewhat under powered state.

Let me know what you think.

Take care

HarryN

[kreutz]

Hi, I have been following your thread for a while, as a "normal" 4 axis driver solution is out of my budget range, but this one seems to be within it, or close.

I am thinking about using this motor Kelling kl34H2120-42-8B, which I think fits in or close to your design range.

My router will be driving the motors rather slowly to obtain max torque, commonly 3 rpm to max 50 rpm over a total 10 turns range. One of the unique quirks of the design is that it is direct drive, so no gear reduction at all, in a sort of rack and pinion. ( There are good and bad reasons for this, but this is the current concept)

In a perfect world, I would like a driver with at least 1/64th or smaller microsteps, and would always run it in that mode. I realize this is out of your design range, but us beggers can't be choosers.

Anyway, I was thinking about buying a motor and loaning it to you for shaking down your driver board using a motor with a bit more load demand and maybe doing some low speed testing. It is ok if your driver board needs to run it in a somewhat under powered state.

Let me know what you think.

Take care

HarryN

Thanks!!,

I will really appreciate it, I will send it back as soon as I finish the tests.

There is no hurry. I still need to send the files for the new revision prototype boards to the fab as soon as I finish doing changes, they take a couple of weeks to come back from China, then I need to populate the board and reflow before continuing with the tests...so, it is going to take some time yet.

Regards,

Kreutz

[harryn]

Hi

I don't yet have the motor - need to buy one for this testing. Any chance you can add smaller micro stepping to the driver ?

BTW - I have used a USA based board supplier that is extremely easy (and fast) to work with. www.4pcb.com I had a board set made with very thick copper and gold plate for a power project I was working on.

They also have a number of partner firms that do board stuffing.

The price I paid was possibly a bit higher than the China version, but the support and "know for sure when it is coming" was great. I also try to support USA suppliers when I can due to the extreme trade deficit we have here.

[kreutz]

Hi

I don't yet order the motor - need to buy one for this testing. Any chance you can add smaller micro stepping to the driver ?

BTW - I have used a USA based board supplier that is extremely easy (and fast) to work with. www.4pcb.com I had a board set made with very thick copper and gold plate for a power project I was working on.

They also have a number of partner firms that do board stuffing.

The price I paid was possibly a bit higher than the China version, but the support and "know for sure when it is coming" was great. I also try to support USA suppliers when I can due to the extreme trade deficit we have here.

Hello;

Thanks for the link, I used to send my prototype PCBs to Advanced Circuits, the quality is excellent as is the turn around time. Recently I moved to the fab in China because of the big difference in prices, the quality is excellent, so far. Paradoxically my friend lost her job due to her company relocating to China or India.

It is good to support our suppliers, but most of them have already factories overseas too, I have been receiving component samples from Malaysia and Hong Kong.

I can't justify spending five times the amount I already pay for the prototypes, since I don't do it for profit.

Back to the motor drives; Smaller micro-stepping modes need changes, not only on the firmware, but probably on the hardware too. The problem is the current regulation loop. Simple control loops, like the one most of the manufacturers use, use blanking to avoid switching noise from interfering the PWM. Introducing a blanking time implies a minimum PWM duty cycle, below which, there is no current regulation. That is why I decided offering microstep modes to 1/16 only.

There is another detail, most microstep positions don't have a detent torque associated, so it is easier to lose position when not on a full step detent. You can add a gearbox to the motor and get both, torque and position, using 1/16 microstep or less, since you don't need high rpms anyways.

I think your motor is a better candidate for our unipolar drive. You can make it yourself. By using our automatic torque compensation (on the Mardus-Kreutz firmware), you won't lose torque compared to the bipolar configuration.

[kreutz]

Some of the Changes done on Rev 3.0j PCB

-Repair SOT23 footprint on my PCB library for the 2N3904 transistor.
-Place a GND test point on the Right side of the board.
-Place a PC_GND test point near the Emergency Stop Test point.
-Place a separate LED for Diagnostics/Programming Status.
-Recalculate Reference filters/Adjustment circuit to limit the Reference voltages to 0.51 volt peak.
-Place LED for Emergency Stop.
-Change the location of the +5V regulator for better thermal management
-Add Charge Pump Enable/Disable option
-Replaced PC_Voltage input connector by USB-B Connector.
-Place LED for PC_Voltage Status indication.
-Add two 1200 uF High ripple current @ 100 Khz rated capacitors to the Vmotor inputs.
-Add pull-up resistors to all PC inputs/outputs, no only the ones redirected to the auxiliary "External Interface" connector (included on the previous revision).
-Add the CPLD and Micro-controller's "In System Programming" connectors to the board.
-Replaced "through hole" Current Sensing resistors by SMD Resistors.

----------------------------------------------------------------

Possible new changes (that will increase the cost of the board)

Add USB based setup and eliminate all jumpers and potentiometers on the board.
(this option is being considered but I don't want to add more complexity and cost)


.

[cephjedi]

USB? I was under the impression that Mach software couldn't take advantage of the USB. I heard PCMCIA parallel adapters have worked for some people, but USB->parallel adapters won't work. Please tell me that's old information, because I would LOVE to hook mach up via USB!

But alas, if this board won't work with Mach, I won't be able to take advantage of it.

Otherwise, I'm still fascinated by the evolution of the board. Carry on Dr. Kreutz!

-Jim

[cephjedi]

Heh heh, I may have answered my own question.

[kreutz]

USB? I was under the impression that Mach software couldn't take advantage of the USB. I heard PCMCIA parallel adapters have worked for some people, but USB->parallel adapters won't work. Please tell me that's old information, because I would LOVE to hook mach up via USB!

But alas, if this board won't work with Mach, I won't be able to take advantage of it.

Otherwise, I'm still fascinated by the evolution of the board. Carry on Dr. Kreutz!

-Jim

The USB connector is only an easy way to get +5 volt and Ground from the PC Power supply for the non isolated side of the PC interface by using a standard USB cable (with the added advantage of a 100 ma maximum load current protection included on the PC side). The USB communication is not being used so far. Mach3 works with the parallel port interface already included on the board, so there is no need for a Break Out board (B.O.B.).

There is a copper trace, on the board, you cut it if your PC has a 3.3 volt Parallel port, it is easy to re-solder the bridge if you change the PC later. We might install a jumper or switch for the same functionality in future revisions, if there is demand for it (5 volt /3.3 volt).

[kreutz]

New Rev 3.0k

Divided the board into 2 smaller boards. The first one has the four translators, charge pump, and all the PC interface (including B.O.B. functionality and signal isolation). The other board has the output stages and low voltage power supply regulators. The safety functionality is on the power board.

Main advantages:

-Logic board does not have to be a 2 Oz Cu board (more expensive).
-There is more space on the power board for thermal management.
-Changes to either of them, in future revisions, will not affect the other one.

Added interface cables and connectors will increase the cost but will help with translator's In System Programming (by making unnecessary extra ISP connectors on the board).

[contactirfu]

New Rev 3.0k

Divided the board into 2 smaller boards. The first one has all the PC interface (including B.O.B. functionality and signal isolation), and translators. The other board has the output stages and low voltage power supply regulators. The safety functionality is on the power board.

Main advantages:

-Logic board does not have to be a 2 Oz Cu board (more expensive).
-There is more space on the power board for thermal management.
-Changes to either of them, in future revisions, will not affect the other one.

Added interface cables and connectors will increase the cost but will help with translator's In System Programming (by making unnecessary extra ISP connectors on the board).

Looks very well promising Kreutz, that way we can also have an option of goin in for more powerful drivers when they are available! OR you can also provide options for switching to higher or lower power versions just by changing the power board. Something like junking a inline 4 to a V6 in my future jeep! GOOD LUCK!

[kreutz]

New Rev 3.0k Update, Revival of the "Kit"

The power board has been divided into 2 smaller boards with two axes on each one. This change will revive the DIY assembly for the "through hole" components on the power boards, the SMD components on those boards will be "pre soldered".

The logic board is all SMD and will be completely assembled, programmed and tested before shipping.

We will also offer a couple of DIY, double axes, unipolar output boards for up to 8 Amps r.m.s. per phase and 100 volt, They will be more sophisticated than the Mardus-Kreutz boards (more safety features), and will work using the same logic boards without modifications, the SMD chips will be pre-soldered on those boards too.

[kreutz]

Zeroing on L6203 chip heating

In order to control L6203 heating at very low speed, and during Idle, it is necessary to find another solution to the current decay.

I went to the Internet search engines and looked around for published solutions, most current circuits employ either Slow decay or Fast decay all the time, slow decay permits lower dissipation by using the inherent synchronous rectification capability, but can't be employed on the second and Fourth quadrant of the sinus waveform due to current waveform distortion (due to current control limitations), Fast decay permits a better current control but has a lot of torque ripple and package heating is high. There is an alternative synchronous rectified Fast decay, which is less demanding on power dissipation due to the synchronous rectification effect, but requires extra complexity (reverse current control).

A combined Fast/Slow decay while solving the current loop control problem, and having better efficiency, does not completely solve out heating problems in the very low speed range including idle. Part of the problem is already solved using 50% automatic current reduction after 1 second inactivity.

So, a good thermal management solution has become an "Slowing down" issue on our project, due to the low cost constrain on the design and PCB cost to area proportionality. Testing new approaches, while interesting, take more time and stresses a little more my friend's depleted financial resources.

I will make a minimum 20 boards run in the form of semi-kits (part pre-assembled (SMD), part DIY "through-hole" components soldering) after the current Rev3.0k prototype is tested. Heat-sinks will not be included, I will also provide heat-sink diagrams with dimensions.

.

[kreutz]

More Changes

Added Charge Pump Activity LED in order to help troubleshooting.

-------------------------------

To our friends in Europe:

The first kit run is not RoHS Compliant. So, they can't be imported into the EU (due to EU regulations). On those kits, all the components and PCBs are in RoHS compliance, what is not compliant is the solder paste I am using now. Compliant kits will be sold under a different Revision Level.

Kreutz

[contactirfu]

Think its gonna be a great kit kreutz......the most documented and trouble freee I feel.

Thumbs up to U!