How I fixed my Chinese TB6560 controller (updated)

[icecreamjones]

hello, I got this board a few weeks ago and couldn't get it to run good at all. After I figured out the chinese PSU was way too wimpy (24v 4.5A rating, kept stalling out the whole drive), I still couldn't get it to go in the correct direction or run in different speeds. Was getting ready to buy something else.
I spent over a week hacking out the settings (EMC2 & Mach3) that work for me. I am posting them in case someone else is having this problem.

EMC 2 ubuntu 10.0.4 settings

Step time : 50000 ns
Step space: 1000
Dir Hold : 1000
Dir Setup : 50000
75% current
25% decay
1/2 step mode
max acc 15in ^2
max vel 1.4 in/sec

I spent probably 40 hrs testing this thing and it was completely worthless until I stumbled on to this setup. Miss steps consistently and reverse direction if you even push on it a tiny bit. Never go the right direction twice in a row. Worst of all, it wouldn't run slow. Between 20-30 ipm (5/inch ballscrew) was the only speed. That is too fast to cut on my lathe.
Now it runs perfect with full power, 300oz in nema 23 (cutting aluminum on a 1937 atlas lathe). It is just amazing how this thing went from chump to champ with a quick settings change.

This is only a 1 axis machine right now, 5/8' roton ball screw. (Will be motorizing X in a few weeks, hopefully it still runs well). I may upgrade to gecko in the future but at least I can run an actual program on my machine while I think about it.
I did use a couple of the mods from this thread, not sure if it mattered or not. I flipped the 3 optocouplers around and chipped off the surface mount current resistor.

many thanks to all those that contributed to the thread!

[Moffett8]

Thanks, I just ordered one of the Chinese boards.

Philip

hello, I got this board a few weeks ago and couldn't get it to run good at all. After I figured out the chinese PSU was way too wimpy (24v 4.5A rating, kept stalling out the whole drive), I still couldn't get it to go in the correct direction or run in different speeds. Was getting ready to buy something else.
I spent over a week hacking out the settings (EMC2 & Mach3) that work for me. I am posting them in case someone else is having this problem.

EMC 2 ubuntu 10.0.4 settings

Step time : 50000 ns
Step space: 1000
Dir Hold : 1000
Dir Setup : 50000
75% current
25% decay
1/2 step mode
max acc 15in ^2
max vel 1.4 in/sec

I spent probably 40 hrs testing this thing and it was completely worthless until I stumbled on to this setup. Miss steps consistently and reverse direction if you even push on it a tiny bit. Never go the right direction twice in a row. Worst of all, it wouldn't run slow. Between 20-30 ipm (5/inch ballscrew) was the only speed. That is too fast to cut on my lathe.
Now it runs perfect with full power, 300oz in nema 23 (cutting aluminum on a 1937 atlas lathe). It is just amazing how this thing went from chump to champ with a quick settings change.

This is only a 1 axis machine right now, 5/8' roton ball screw. (Will be motorizing X in a few weeks, hopefully it still runs well). I may upgrade to gecko in the future but at least I can run an actual program on my machine while I think about it.
I did use a couple of the mods from this thread, not sure if it mattered or not. I flipped the 3 optocouplers around and chipped off the surface mount current resistor.

many thanks to all those that contributed to the thread!

Hi all

I have bought 6 of the 4-axis TB6560 controllers. I had the same problems with 2 of these bought about 4 months ago, so I ordered more to replace these 2 and have some in stock.

The problem is that after cutting the attached g-code file the x and y axis does not return to zero. I do not hear steps being lost while cutting, and when using on some other g-code files the cards seems to work fine. Now with these letter files some of the cards loses as much as 12.27mm on x and 7.85mm on y, nothing on z.

Are there any of you willing to run an air-cut test on the attached g-code file and let me know what your findings are? Please!

I have removed tool changes and spindle starts for the test file.

One of the cards blew the y-axis after about 30mins into my test. I don't know if it worth buying these things anymore. I had a 5 axis catching fire about a year ago.

I am running Mach3.

Thanks in advance

Piet

[adamhixon]

Has anyone considered using the great info provided here to make their own board? That TB6560 seems to be a fine chip and does all the heavy lifting for you. The supporting hardware was just poorly implemented on these ebay rejects. The chip is available on mouser for $4.20 each. I may play around in eagle sometime. Any ideas of what an ideal schematic would include?

[doorknob]

Has anyone considered using the great info provided here to make their own board? That TB6560 seems to be a fine chip and does all the heavy lifting for you. The supporting hardware was just poorly implemented on these ebay rejects. The chip is available on mouser for $4.20 each. I may play around in eagle sometime. Any ideas of what an ideal schematic would include?

One such design is shown about halfway down the page at:

http://www.cnczone.com/forums/diy-cn...cb_mill-2.html

[JohnStone]

Hi folks,
decided to chime in after blowing :-( my "China Bang Board" as well and finding later your very helpful thread. Now waiting for the delivery of parts to replace.
Thanks a lot to all individuals sharing the fruits of your personal effort! :-)
After studying the schematic and reading all posts I understand more of the blowing of my drivers and want to forwad my notions:

1.
The 6560 requires to get the 5V first and later it will accept the motor voltage. The schemtic of this board reverses this requirement! :-( Constantly! Let me explain:
The 12V and 5 V regulator along with their capacitors perform a delayline for the 5V supply. It takes some milliseconds until the 5V is stable on the drivers.
Ignoring this the Motor supply is directed to the drivers instantly.

So accept my good advice: DO NOT do what I did - to set a switch between 24V supply and the board. It will blow soon your drivers because of this weakness mentioned above.
If you switch the mains at the input of your power supply you still do not face the power requirements but usually the drivers seem to accept this - most times. But this is still not a recommended use case!

Cause of damage:
The transistors in the output driver are aranged in an H-bridge. It is essential to GARANTEE at every time that within one pole of the H there is only one single transistor switched on at a time. This can not be done without the 5v being stable on the driver! So it is very probable that the too early voltage for the motor drive will open both transistors by itself and a short circuit will occure.
In my case it occured long enough to damge the whole driver. In other cases (switched mains) it will probably crack some parts of it so the output might still do some performance but not full spec or it will be killed later on.

An extra problem seems to be if a motor output is not conneced. My 4 channel bord was connected to X,Y,Z leaving C open. This open channel was blown first. Later Z while not being connected ..... I remember that one of you experienced the same kind of bang. This can be promoted by the heat sink not being connected to GND.

But PLEASE accept my advice. Do not connect the heat sink anywhere to GND. The currents will disturb on their trace other entities of the schematic. It is not just DC current! The best way would be to connect a lead from one screw of every driver to the corresponding ground Pin 10 or/and 15 but NOT pin 6 (this is the 5V-GND - it should not be disturbed by additional driver currents!)

I believe that this power problem is the main problem of all blown drivers. I shared my notions instantly in order to hopefully save some of your boards - wold wide :-)
Give me some days to evaluate some suggestions to face the power on issue.

2.
Reset schematic:
It is good practice to add a diode in parallel to the pullup resistor (cathode connectod to capacitor side and anode connected to supply). This is necessary in order to discharge the capacitor soon when the supply is switched off. If this is not performed you will get no reset after short interruptions of the supply. Proper reset is essential!

3.
Ther were some quesions around regarding the diods in the power input trace to the board. These are usually used in order to protect the board when reversing power and GND acceidently. Omitting these diodes will blow your board in a very very reliable way - in case of pwer reversal.

4.
Some facts reagerding my setup:
- ISEL mill with cutting area of app. 50cm x 50cm, all ball nuts, 2 Proxxon Motors 300 Watt with power supply (bought defect 200 € - only defect = Z home switch :-)
- all stepper motors 4A
- 4 channel china board (this blue one most of you have) - now blown
- PC tower case with PC living downstairs
- in the upper floor live the PC power supply, the driver board just under the roof with component side down and the 20-30V supply in the CD-drive compartment. The case provides an additional blower for this compartment - i will gladly use this support.

Adavantges of this setup:
While feeding this all in one unit through one single cord the unit has a good genuine GND condition. (I hope to omit optos completely later on) It depends on measurements - to be done.

This setup allowes to supply the on board blower with teh 12 V from the PC power supply.

4.
I feel sad because these china men convert such a good motor driver into a problem zone world wide.
Joke: What about Toshiba sponsoring them in order to increase the world wide demand for parts (spare parts) :-)

If you have additional hints for my setup please feel free to comment. It is good practice to screw an idea through different brains :-)

Regards
John

[Gall]

Just ordered a 5-axis version of this board (probably the new one) plus a digital display.

I haven't modifiet it yet (it works but not very good, now I use it to mill acrylic case for it, later I'll do the fixes).

I found that it IS possible to add optical isolation to this board! Fortunately this version has all optocouplers connected to +5V at one side and to GND at another. Using a permamemt marker I've drawn a line on the board under optocouplers (on both sides where possible) without crossing any PCB traces. Now it's obvious where exactly the sides are connected. It is possible to cut GND and +5V traces to separate input and output sides completely. I'm going to do that and install a dc-dc converter in place of 78xx regulators.

Regarding motor noise and driving problems, I think I'll remove these transistors completely without cutting traces. I was unable to measure capacitances at OSC pin without unsoldering, going to measure frequencies later.

There is one more problem with digital display board. It misses steps! I investigated its design and found that it isusing a MCS-51 compatible microcontroller (STC12C5A08AD) at 24 MHz without any hardware counters. Software polling! It is impossible to use software polling at higher step rates with short pulses. Looking for a way to replace the MCU with a CPLD or something; it were nice to find a pin compatible part but probably it's impossible.

Another problem is that the display is very dim. It is easy to solve. There are eight 2 kOhm resistors on the board - 5 near MCU and 3 on another side. They set display segment current. (There are no other 2 kOhm resistors on the board). Replace them with smaller value. I found anything around 560 Ohm would be Ok. Smaller values can lead to overcurrent.

I also got a blue manual control box for this board. It is built around the same STC MCU. It's very useful for stepper testing but quite useless in production work - no need to buy. (I bought it for testing purposes only.) However it can easily be replaced with a simple 555 pulse generator on a breadboard so a bit overpriced.

[jaclerier]

Hi guys
I read and reread this post that intend to fix this board. I followed the steps since begin and the first thing that i've done was add another 74hc14 to buff the signal steps, this realy gave me a better board, but i still have the miss steps in all the velocities so i removed the opto acoupler and take the signal directily to 74hc14 added with this modification i have a terrible noise from de stepper motor, i could hear the miss steps. i've put again the optos but i removed all the optos of dir and i've done a jumper between the pin 2 with 4 and all the four axis works very well without miss steps. i removed also the others optor i left only the step and enable optor in.
Now i have one question, how i put the diodo to protect my boad of back emf from motor and the reley?
And sorry for my english

[Gall]

Now i have one question, how i put the diodo to protect my boad of back emf from motor and the reley?

For motors: Each winding needs four diodes, two at each output (8 diodes for one motor total). At each pin, one diode to ground (cathode to motor, anode to ground) and one to +12-36V (anode to motor, cathode to power bus). Diodes to ground are on the board already (at least my board has them). Just connect each of the motor outputs to the power rail with a diode (cathode to power rail), now it's Ok.

For the relay: just one diode between the relay winding pins, also connected backwards.

[jaclerier]

For motors: Each winding needs four diodes, two at each output (8 diodes for one motor total). At each pin, one diode to ground (cathode to motor, anode to ground) and one to +12-36V (anode to motor, cathode to power bus). Diodes to ground are on the board already (at least my board has them). Just connect each of the motor outputs to the power rail with a diode (cathode to power rail), now it's Ok.

For the relay: just one diode between the relay winding pins, also connected backwards.

Hi Gall tanks for your reply
take a look at the attachments marked in red an see if this right?

[jaclerier]

For motors: Each winding needs four diodes, two at each output (8 diodes for one motor total)

one more thing, i have this diode here. can i use it? will work?

[JohnStone]

As I want to prevent another blow of my drivers (thanks Mr. China - I'm not amused - definitely not!) I continue consdering more power issues. (listen Mr. China !)

Power off failure - more precisely:
When power off (especially with motors disconnected) the 5V will very probably cease long before the motor voltage in the big capacitors. Once again we see a violence of the clearly expressed requirements.
Unfortunately the motor voltage of each channel is separated by the reverse protection diode at the respective input. So if you connect motors to 3 channels they possibly survive the power off switching because the capacitors are discharged early enough. “Surprinsingly” the 4th cannel will die because it did not get rid of the motor voltage doue to the fact that there is no load at output. If you disconnect Z and switch on / off later this channel will die - “surprisingly”.

Another power consideration:
A member - Mariss Freimanis - in this forum shared some very interesting notions at differnt places. In summary he points out, that we need to consider the power coming from the motors (i.e. back EMV install condition) back to the drivers traveling through the internal or external protection diodes (the way taken does not matter) into the power capacitors. This can charge the capacitors to a dangerous overvoltage for the 6560 drivers. As there is a power separation for every driver (D15,25,35,45) the voltage can't go back to the PSU to be bufferd. It is very well understandable that the condition gets more dangerous if we applied a higher voltage initially.

Maximum ratings:
The same member stresses the fact that absolute maximum ratings are not automatically all true at same time. He ist right! So 3,5 Amps and simultaneously maximum voltage may be a serious problem. He votes for operation at reduced figures in order to have enough margin for safety reasons.

Some preliminary notions regarding a possible solution:
The intention of the separation of the voltage path (4 channels an 12V /5V) by diodes is primarly ment as a protection from reverse voltage.
Secondly they can prevent traveling noise from one channel to another. This can be true because every transient current at PWR+ will initate a corresponding current in the GND region between the drivers.
Summing up: Good idea - but - made worse by not considering other conditions above.

Now imagine the possible advantage of the following conditions:
1.
The back EMV from the motors does not occure simultaneously at each channel. If it can be shared (no diode separation) with the other capacitors it will be less serious.
Alternatively - if we want to keep the channel separation the capacitors need to be of much greater value if we exceed 24V at PSU side considerably in order to deal with the lesser safety margin.

2.
Adding voltage protection diodes of i.e. 36V would be another good idea. They possibly need to short circuit considerable energy! TBD

3.
Omitting the voltage separation by diodes would enable the big capacitors at motor voltage to supply the 12V/5V regulators. This can be a considerable help at power off condition. The 5V will be available down to 9V and possibly below. (5Volt + 2V voltage drop at each regulator) At 9V the motor voltage seems to be less dangerous.

4.
Do not operate your borad with open motor terminals. Connect a spare motor(s) or corresponding power resistors.

5.
Do not connect or switch the PSU output to the board under power. Please switch the mains and take the advantage of voltage rush in time.

6.
Ground connection from heat sink to pin 10 and 15 of each driver (PGNDA, PGNDB) or at least one of them.


Warning:
Please consider that this configuration will still not face the flaw at power on condition and might have other disadvantages. This is still an open and public thought experiment in order to evaluate some possible steps of modification.
So I ask you kindly not to modify your board just now as long you are not an expert and can conduct required measurement. (except item 4 to 6 - these are not dangerous.)

It would be easy to propose a complex setup nobody can build. But we need a real solution being as simple as possible. So give me some time for tinkering. But you are invited to share some proposals.
Regards
John

[Gall]

Hi Gall tanks for your reply
take a look at the attachments marked in red an see if this right?

Absolutely.

Please note, four diodes going to ground are probably already on the board (in the long row of diodes next to the row of connectors). You probably do not need to install them. But check the board, board versions may vary.