Design fault HY-DIV268N (TB6600)

Motors, motor drivers, power supplies, ...

Design fault HY-DIV268N (TB6600)

Postby Tony Brooks » 28 Oct 2013, 07:13

A word of caution regarding the above stepper drivers. I purchased three of these recently for use with my CNC Router that is currently in build. I haven't yet purchased a controller but I am planning to buy the MK2 as I near completion.
I also purchased 3 x Nema23 motors, 3.1Nm 8 wire versions. Before installing the motors, I wanted to check the performance so I used my signal generator to provide a clean and very precise pulse stream. I also used my digital oscilloscope to check various parameters.
Firstly I checked the available shaft torque at slow speeds just using my hand to grip a 40 tooth pulley. I was rather disappointed as the motors were too easy to stall. I was using my bench power supply set at 24VDC at 3Amps. The driver was pulling less circa 0.5A which was much less than I expected. I repeated the test on all 3 drivers and motors on both serial and parallel wiring.
One of the drivers refused to change direction so I opened the cover to investigate. One of the opto-couplers was faulty (confirmed by a swap out). While I was in there, I decided to check the circuitry for compliance with the Toshiba TB6600 data sheet. Guess what... the circuit was holding the TB6600 in 30% torque mode even while the motor was running! To confirm my suspicions, I injected a high signal on pin3 of the TB6600 to force the chip into 100% torque mode. WOW - the increase in torque was more like four fold, I could hardly stall the motors using my hand on the pulley. This was more like it! Conclusion = Happy with motors, NOT happy with DIV268N's.
I decided to replace the drivers with M542H units. I tested these in the same way and can say I am very happy with them, they pull circa 2.3A on same test then drop to 0.4A about 1/2 second after receiving last 'step' pulse. This is what I expected.
I wander how many people out there have condemned there NEMA23 motors as underpowered as a result of using duff drivers?

One more problem with the DIV268N's;-
The microstepping table on the device is also wrong!
The following table is what I compiled following my tests. If you compare this with the table that is printed on the devices you will be as shocked as I was regarding the blatent error in design.

Micro Step SW4 SW5 SW6
1/1 ON ON OFF
1/2 OFF ON ON
1/2'B' OFF ON OFF
1/4 ON OFF ON
1/8 ON OFF OFF
1/16 OFF ON ON

Bottom line - Don't assume something will work correctly just because it's new!

Hope this helps.
PS I have of course reported these issues to my supplier.

Regards,
Tony.
Tony Brooks
 
Posts: 3
Joined: 27 Oct 2013, 17:07
Location: Czech Republic
Controller: None

Re: Design fault HY-DIV268N (TB6600)

Postby Andrej » 28 Oct 2013, 08:06

Thanks for sharing this info with us. This is good to know.
Andrej
 
Posts: 2489
Joined: 10 Jan 2012, 16:09
Controller: Many controllers

Re: Design fault HY-DIV268N (TB6600)

Postby Tony Brooks » 17 Dec 2013, 20:04

For anyone interested, you can force this unit into 100% torque mode by soldering a jump lead between Vreg (pin 24) and pin 3 on the TB6600.
Tony Brooks
 
Posts: 3
Joined: 27 Oct 2013, 17:07
Location: Czech Republic
Controller: None

Re: Design fault HY-DIV268N (TB6600)

Postby Phiredog » 15 Feb 2014, 14:14

Thanks for your information. I read your post the day after I ordered a set of 4 drivers.
I opened up the unit to look at the board and was not sure which is pin 3 and 24.
Any help on the pin numbers would be great.
Thanks again
Phiredog
 
Posts: 1
Joined: 15 Feb 2014, 13:55

Re: Design fault HY-DIV268N (TB6600)

Postby Tony Brooks » 04 Aug 2014, 06:38

HI, you can see the pin numbers on the pin out diagram for the TB6600, just do a search for 'Toshiba TB6600'. You should see that pin 1 is the left most pin on the TB6600 looking from the front face of the chip, ie with printing facing you and the pins pointing down. Pin 25 is the right most pin.
Tony.
Tony Brooks
 
Posts: 3
Joined: 27 Oct 2013, 17:07
Location: Czech Republic
Controller: None

Re: Design fault HY-DIV268N (TB6600)

Postby alex_holden » 21 Sep 2014, 08:22

Hi! I'm using three of these drivers for a project I'm working on, and I've signed up to this forum because this thread is the only place I could find that is discussing their design and limitations.

Firstly, it's possible I have a different revision of the board to you. The Chinese manufacturers probably make tens or hundreds of thousands of them and I wouldn't be surprised if they change the design occasionally to fix faults or substitute components. Mine say "HY-DIV268N-5A" on the case and "www.hyu68.com HY-TB268-N6" on the PCB.

I'm using them with NEMA17 60 Oz/in 0.9 degree motors with 0.8 Ohm/1.8mH windings. For test purposes I'm powering them one at a time from a 29V/2.5A linear regulated bench power supply and controlling them with an Arduino running GRBL.

I've tested the full/half/micro stepping DIP switches on all eight settings and on mine the behaviour correctly matches the table printed on the case. I'm a little unclear about the difference between the two half-stepping modes and why you would choose one over the other (the TB6600HG data sheet wasn't particularly enlightening).

The 100%/33% current thing. On mine, they go into 100% mode whenever the pulse (step) input is high. You can test that by watching the current used by the driver and manually injecting a constant high signal at the pulse input (while the enable input is low - confusingly, these boards are active low enable). From a brief look at the circuit (there are a couple of transistors, several resistors and a couple of caps in the circuit feeding the power level select pin), I suspect it will remain at 100% current for a fraction of a second after each step, but I don't currently have access to my scope to measure it. My theory is they did this deliberately to cut down power consumption because the components are rather under-specced and they would burn up if you tried to run them at 100% current constantly for long. They did seem to heat up pretty quickly when I forced them into constant 100% mode during testing.

BTW, none of mine had any thermally conductive grease between the driver chip and the heatsink. Sorting that out ought to significantly improve their life expectancy.

The current limiter DIP switches. I'm not certain that I'm measuring this correctly and I've only run the full test on one of the three drivers, however it appears that the current limit levels are actually about half that shown in the table. I forced the driver into constant 100% mode and noted the supply current at each of the current limiter settings. Here's what I found:
Code: Select all
DIP    Claimed   Actual
111    0.2       0.05
011    0.6       0.25
101    1.2       0.5
001    1.8       0.9
110    2.5       1.4
010    3.3       1.7
100    4.2       2.0
000    5.0       2.2

I'm no stepper motor expert but that discrepancy seems pretty significant. Again, I suspect the designers did this deliberately because the drivers would quickly burn out if you actually tried to run them at 5A (BTW the datasheet for the TB6600HG says 5A is the absolute maximum limit and the normal operating limit is 4.5A).

My little motors are only rated at 2.4A/phase so it looks like I should be able to use them at the maximum current setting without much of a problem, but I don't think these drivers would perform very well with bigger motors.
alex_holden
 
Posts: 2
Joined: 21 Sep 2014, 07:24

Re: Design fault HY-DIV268N (TB6600)

Postby alex_holden » 22 Sep 2014, 06:58

After posting this, I tracked down a much longer discussion thread about these drivers over on CNCZone: http://www.cnczone.com/forums/stepper-m ... -ebay.html

In summary, I was right about there being an RC circuit that is intended to hold the drive in 100% mode for a short time after a step pulse, however the time constant is far too short to be useful. People report having tried the mod to run the board at constant 100% power and found that it quickly overheated. It may be possible to extend the pulse length (perhaps by patching in an external circuit) however these drivers are flawed in so many other ways that I'm not sure it's worth the effort vs buying a better-designed board to start with (wish I'd found that thread before I bought them!).
alex_holden
 
Posts: 2
Joined: 21 Sep 2014, 07:24

Re: Design fault HY-DIV268N (TB6600)

Postby Lordofrealms » 13 Sep 2016, 18:40

All,

Thought I would post on how I was able to turn these into very good drivers. Specifically as a replacement for the drv8825 and a4988.

First, as they ship they are crap.

Second, you accept all risk in doing this modifications if they burn your house down kill you whatever its on you.

Alright now that that is out of the way:

You probably won't be able to see the front of the chip because it will be pushed against the PCB. So lets orient ourselves with the back as the part that is thermally connected to the heatsink and the bottom as the part of the IC where the pins come out. With the front facing you and the top up pin 1 is the left most pin and pin 25 is the right most pin.

Unmount the pcb from the heatsink. Solder a jumper from pin 3 to pin 24. This forces the tb6600 into 100% torque mode. Desolder the two giant resistors (.22R 1W). They are the only component that will overheat and die at 5A and full torque. Order some 0.22R 5W Cement resistors from aliexpess, ebay, or amazon. (I got 10 for $2.50 from aliexpress, but here is a link for them on amazon since aliexpress rarely keeps the same item postings for long https://www.amazon.com/uxcell%C2%AE-Rad ... w+resistor). You will replace each of the 1W resistors with these 5W resistors. These are the current sense resistors, if you don't replace them they will burn up and you will have huge currents going to your steppers and possibly also only one coil being powered.

Once you have replaced those, lube up the thermal contact pad with thermal paste, preferably some artic silver and reattach the pcb to the heatsink. You likely won't be able to put the case back on after adding those resistors. It isn't needed, but attach the pcb with some spacers like this https://www.amazon.com/3mmx6mm-Spacer-S ... spacer+3mm (I didn't buy these so can't say if they are the right height but any nylon type washers will do).

I am going to print an 80mm fan bracket to attach to the heatsink for additional cooling. It probably isn't needed, but better safe than sorry with long prints/cnc projects.

For those unsure how to hook this up to ramps, enable goes to enable+, step going to pul+, dir goes to dir+, and jumper all three -s together and run to a gnd on the ramps. Note: don't use the gnd next to the vcc on the driver socket of ramps as it is the driver power gnd not the signal ground and will give you issues. If you are going to use a gnd from the ramps driver socket use the one almost directly across from the enable pin.

Let me know if you have any questions thanks.

Let me know if you have any questions or if you still have trouble after this.
Lordofrealms
 
Posts: 3
Joined: 13 Sep 2016, 18:23

Re: Design fault HY-DIV268N (TB6600)

Postby Lordofrealms » 15 Sep 2016, 01:34

All,

Two additional things.

1st:
Here is a case I made with a 40mm fan mount. It will go on even after replacing the resistors.

http://www.thingiverse.com/thing:1771631

2nd:
Once I get the 5W resistors in, I plan on trying 2 of them in parallel to give a current sense resistance of .11Ohm. This should double the current setting the drivers apply. I then plan to attempt running them at 7A (3.5A jumper setting). I am still having some issues getting the speed I want from my steppers and my steppers are not at their thermal limit yet (around 50C max temperature). With the 40mm fan (a very cheap low cfm one) the drivers stay below 50C at 5A and the heatsink, resistors, and diodes stay around 30C.

7A should not affect the long term life of the driver even though it is above the specs. (Again do the modification at your own risk). I will post again after I have made the modification and everyone know the results (it will probably be 2 weeks or so for the resistors to come in).
Lordofrealms
 
Posts: 3
Joined: 13 Sep 2016, 18:23

Re: Design fault HY-DIV268N (TB6600)

Postby Lordofrealms » 29 Sep 2016, 06:05

All,

I finally got my 5w 0.22Ohm resistors in. Unfortunately, logic and data sheets apparently do not apply to this board. I ended up wiring the resistors in SERIES not PARALLEL as I earlier suggested. Parallel decreased my performance and series greatly increased my performance both with no heat related issues or buildup. With the two resistors in series, I am not able to run the drivers with no active cooling. Apparenty, all of my heat buildup was coming from the current sense resistors.

I am not sure what the true current output either rms or peak is as I don't have a modern oscillascope. If there is a good way to measure the current with just a multimeter I would be happy to contribute.

My best guess is that I am now pushing somewhere around 7A rms per coil based on my prior drv8825 amps and axis speed and my current axis speed. This driver has taken me from stable at 35mm/s with the drv8825s to stable at 120+ mm/s.

Also, in case anyone is running marlin on a mega, I was running 1/16 microstepping with 640 steps per mm. Above 75mm/s the mega on the arduino could not process all of the steps and would lock up. This may be because I am using auto bed leveling so there is some matrix calculations I think it does for every move, but none the less, reducing microstepping to 1/8 and 320 steps per mm immediately solved that issue and I am now stable at 120+.

If anyone has any ideas on verifying my exact current per coil or if anyone has any questions, feel free to reply or pm me.
Lordofrealms
 
Posts: 3
Joined: 13 Sep 2016, 18:23

Re: Design fault HY-DIV268N (TB6600)

Postby bobdxcool » 11 May 2017, 22:17

Tony, I want to drive a NEMA34 (5.5 amp motor, 3.6mH, 46kgcm) stepper motor using the TB6600 driver (http://www.ebay.com/itm/4A-TB6600-Stepp ... 2005148156).

So to increase the current limit, I can solder 0.47 ohms in parallel with the current sense resistors (R620)?
Also, as it has been stated in this forum that these drivers are designed to provide only 30% torque. But as per one of the posts, connecting a jumper between Vreg (pin 24) and pin 3 on the TB6600 would provide 100% torque always, right ?

Just as an alternative I was also considering ordering another stepper driver from leadshine (http://www.omc-stepperonline.com/leadsh ... -p-77.html). Do you have any idea about this driver ? If so, would this be a good one without needing any modifications to drive at 100% torque ?
bobdxcool
 
Posts: 2
Joined: 11 May 2017, 22:15
Controller: Other

Re: Design fault HY-DIV268N (TB6600)

Postby Deltaflyer » 17 May 2017, 10:21

hello bobdxcool,
The TB6600HG chip inside the motordriver can deliver MAXIMUM (peak) 5A. Permanent load is 4.5A. This information is taken from the data sheet of Toshiba's TB6600HG.

Greeting Stefan
Deltaflyer
 
Posts: 19
Joined: 05 Jun 2015, 06:35
Controller: Mk3


Return to Electronics



Who is online

Users browsing this forum: No registered users and 2 guests