TB6600 drive from EBAY

[JohnMiller]

Here the brand new circuit diagram from my TB6600 Rattm stepper driver:

It is mostly identical with Haoyu schematic posted some time ago.

My findings for Rattm:
:-)

• Switches M1 ... M3 are correct
• Test points for Vreg (adjustmet of current) are available
• POT for driver current marked correct (silk screeen printing): CCW -> increase current (in earlyer PCBs it was marked reverese)
• Circuits with high current flow are designed quite well
• Operating voltage 45V printed on board (it should be 42V). (Others recommend 50V = absolute maximum voltage)

Some crap in control circuit needs to be corrected (if possible):
:-(

• Some blocking caps in power and control circuits missing
• Required start up sequence along ENABLE and RESET signal is not supported
• Timing circuit for TQ is very obscure and might not work
• enable circuti is active if input unconnected
• LEDcurrent violate specification for Vreg and output specification of ALERT signal
• Tuning of Vref and VFN (Voltage at RFN) do not comply to data sheet
• Mounting to heat sink without thermal conductive paste!

[frie]

Hey John,
Just curcious and also for learning, under what conditions did you drive the board: supply voltage (+amp capability) + what type (unregulated,switching, regulated), rated amps of motor, microstepping?? and how did the board die? more than one axis? what conditions+ what part blew? And also, how long did the board last under working condition? And did you check motorAmps output while driving the board?
Regards

I cannot help in building a PCB that's more in lucas's field of expertise. the flaws you mentioned i think can be worked around and for myself if this chinese p***e of c**p fails i'm buying myself a gecko drive. (or i'd better buy one straight away, but i like a challenge ).

[JohnMiller]

Hi frie,
I can not recall the conditions when my 4 channel board died. I used 30V. I was red-necked at that time and instable connections might have occured.
Those multichannel boards from China do not take care of getting 5V stable first and then switch on stepper power. Same problem at switch off. The big capacitors store power for long time while the 5V supply will go down earlyer. It is not trivial to cover these issues reliably and automatically.

The boards have a cascade of a 12V regulator followed by a 5V regulator. I would recommend to disconnect the 5V regulator from 12V input (12V regualtor is the for the blower oily and will rund muhc more cooler. and add the power by e.g. an aditional socket charger (9V ... 15V will do fine). Then you are free to decide on when to switch the 5V supply on or off. I had no chance to try this because the board still does not work after replacement of all driver ICs.

This was the reason I am interested in using TB6600. Toshiba did the job of power priority within the chips. But unfortunately most builders misunderstand this to be blessed with a free 5V supply for the control circuit.
I envision two versions: quick and dirty but complying to data sheet and another mod with more goodies.

BTW: You are right: such boards give you a feeling like hunting. And I hunt the Rattm board because I want to know what potential power and quality we get out of it. And a lot of learning possible along this hunt!

I still do not understand the reason of individual torque reduction. If I mill e.g. along the X axis only, it would be inconvenient to reduce torque of the Y and Z axis. Step jumps might occure or at least inaccuracies when the head slackens a bit in Y, Z direction because of low torque. I would accept a common TQ signal if none of the axis gets clocks any more after e.g. 2 minutes. But then we need to have an additional input.

[frie]

Forgetting is normal as long as you replace it with better material, . I always build in an emtpy atx casing which is much spacious for all extra components, thus i have a stable 5Vdc supply comming from the psu, also i came across somewhere to stop a tb6600hg card with E-stop, then shut down and start up and pull out E-stop, don't know if this will work but i think i will stick to this to be sure sounds quite reasonable. Also i'm planning on going "high" from start on the "enable" pin but with a built in very long time constant so that the enable pin will become high after the power to the board has been started. TQrobably energy conservation, board conservation and avoiding motorheat buildup (the cooler the better) but more important: To bad that board builders do not give flexibility in setting holding torque as this is better left to the customer to tune to their needs. ball over leadscrew is quite different in demand on holding torque, but maybe they will adress that in the near future (or not)
on the 4 axis i'm going to start with no mods, expect for the ones above calculating from lucas his 6msec TQ high results i'll be fine if not i'll solder out a useless resistor (in my view) in the TQ circuit and put in a capacitor making a timeconstant after the transistor of 0.1sec as charging path is faster than discharging path i think i will be ok with this mod. but hope this won't be nessecary. and still need to buy a gecko drive LOL.
All the best. Regards.

[JohnMiller]

Confused:
1. No mods
2. Stable 5V PSU from ATX

Please regard that the original Chineese design does not care for power sequence. I would not dare to operate this board with more than 12V without taking care of. The result is mostly digital: mostly overcome or destruction. No intermediate state.
You should take care of ground system. It might not be fortunate to connect ATX GND to 5V GND. Therefore an isolated (9V ....15V DC) supply for the 5V regulator (as suggested above) might avoid any doubts. This supply needes to be active FIRST and cease as very last supply at switch off.
As the big caps from stepper power conservate lot of energy voltage may last for long time. So it is recommended to add a load resistor while OFF.
And whatever you invent it needs to do well even if there is an emergency case - emergency switch off or loss/recover of mains.

What I envisioned before detcting TB6600:
1. Isolated supply for 5V regulator along manual switch at 9V side (regard well designed GND system)
2. Switch for stepper DC (SPDT switch) in order to connect a discharge resistor in off state for fast voltage decay

The sequence woukld be:
Switch ON:
- Mains
- 9V / 5V
- Stepper DC

Switch OFF:
- Stepper DC
- Wait for decay
- 9V / 5V
- Mains

[JohnMiller]

Suggestions for mods: Rattm TB6600:

Please note that:

• These mods are just now suggestions only, not tested and will be the basis for my modifications.
• As most boards are generated by copy/paste teh hints will be of general value
• As soon I get hold of the components requested I will contiunue with hands-on modification and test. The real chalenge will be to find a viable way of monting and wiring the components.
• In case of new design of a PCB some additional goodies may be added in order to complete a qualitiy design.

A:
Logic supply (5V) added and Vreg disconnected.

B:
- The 100nF (C88) is not close enough. An additional cap shall be added (*C101)
- Reset circuit added (values need to be tested). The diode perfomrs teh discharge of *C1 inorder to get a defined start condition if power returns.

C:
- Electrolitic cap shall be replaced by a low ESR type (usually used at switched PSUs)
- Added 100nF ceramic cap for buffernig higher frequencies

D:
Reverse ENABLE signal to be inactive if input unconnected.

E:
Control of ALERT LED by TB6600 now driven by part of U5 (74HC14).
This measure resolves the problem of input pins 1 and 13 to be unconnected as well.

F:
Debouncing for UFN (value needs to be tested). Might resolve uncontrolled singing of motors and some anomalies.

G:
Rework of RFN for compatibility to data sheet and lower loss at RFN. For ease of modification the existing RFN are not replaced but doubled.
See section "I" as well.

H:
To be negotiated:
- Resistor required between output 74HC14 and base of transistor! Or use of a transistor with integrated base resistor (e.g. PDTC114T)
- Value of delay shall be modified. But any increase will increase the delay of turn-off the torque reduction.
- Other designs use an IC 74HC123 instead in order to get a superiour result. Unfortunately this solution cannot be added by hand to Rattm PCB

I:
Rework of Vref control for compatibility to data sheet
See section "G" as well.
Due to the normal tolerances of components it is essential to adjust the current along measuring Vrfn at test points.

J:
Buffer cap added locally to U5 power.
Please not that U5 pin 2 drives the transistor Q2 very!!! hard (section H) generating noise at power pins.

[_ID_]

Regarding point I:

In datasheet is written 0.11Ω ≤ RNF ≤ 0.5Ω, 0.3V ≤ Vref ≤ 1.95V

You made a voltage divider with 22k and 2k2 resistor and 10k trimmer. If you turn the trimmer fully CCW you will get 1.784V and fully CW you will get 0.322V.
You added 3 resistors in parallel to existing 3. So you get 0.11R (point G). This is lower limit but still ok regarding the Datasheet.

Considering this the settings for the 100% current would be between 5,40A and 0.97A.

I might be completely wrong here- I'm just thinking loud

[_ID_]

Woltage divider made of 16k resistor 5k trimmer and 1k5 resistor would make current between 1A and 4.38A, leaving 6 parallel resistors of 0.68 Ohm at NFA/NFB.



R_NFA=R_NFB=0.11 Ohm
CCW state: R1=16k, R2=6k5 =>V_REF=R2/(R1+R2)x5V=1.44V => I_OUT=V_REF/3/R_NFA=4.38A at motor windings
CW state: R1=21k, R2=1k5 => V_REF=R2/(R1+R2)x5V=0.33V =>I_OUT=V_REF/3/R_NFB=1.01A at motor windings

[JohnMiller]

Yes ID! You are right!
But we need to account for tolerances of resistors to be 5%.
We need to guarntee in worst case that we can adjust 4.5V minimum. Of course there is the other extreme where we overshoot like pointed out by you.
Either we use 0.5% resistors or we decide to start with POT in the middle range and do fine adjustemnts along measuring Vref with a meter.

I will recalculate all values later on. I want first to show what measures are necessary in order to get the Rattm board in the range of specification. But I am not confident that it is worth to do it. I just get familiar with Eagle CAD in order to do my own layout and share it as well. Lots of tricky measures need to be negotiated.

BTW: What simulation SW do you use?

[lucas]

That are a lot of changes, there won't be much left of the original board.

The 30% torque reduction will indeed be too low for some machines, depends on leadscrew pitch. Better is to do like the THB6064: change the Vref setting and then you can choose the amount of reduction you wish.
This unit also has the plugin connectors with open sides, these come loose easily and blow the driver if it's the motor one.
The TB6600 datasheet isn't clear on the Vreg output, the pin description says "decoupling capacitor" but they use it also for the alert and mode outputs. Some current can be drawn from it but how much is a mystery, I asked Toshiba without succes.
This internal 5V regulator is the advantage of the TB6600 vs the THB6064, if you don't want to use it the you're better off with the THB.
It has more features: more microstepping settings, decay settings, torque reduction build in...

PS: sorry for the late reply.

[JohnMiller]

HI Lucas,
thanks for your hint. After I studied the datasheet I am convinced. And yes they have resolved at THB6064 the mandatory sequence of power switch ON / OFF as well. That is good news.
I studied the public information provided along your kits and I feel now that any own effort is obsolete. Now I decided to spend same money I have spent before for Chineese drivers again - but into something working fine.

At your kits you have a heat sink. How warm does it get on long term and heavy duty. Is it recommended to add a blower to the heat sink?
What gauge and what length do you recommend for motor connection? Is it necessary to mount the drivers in very proximity to motors as to travel with the mill movements.
Rgds John

[lucas]

Good decision...

The chip is quite hot at currents above 2.5Amp and some other components also heat up depending on the supply voltage.
Best solution is a small breeze from a fan to keep everything cool, same applies to TB6600.
Cables can be several meters long if a decent gauge is used, I normally use shielded 0.75mm².