How can any TB6560 driver design NOT violate the chips spec's?

[James Newton]

I've been reading the TB6560 driver chip data sheet.. you can find it here:
http://www.toshiba-components.com/mo...3_20080407.pdf

The interesting part to me is on page 28, where it says:

Power-on Sequence with Control Input Signals
Turn on VDD. Then, when the VDD voltage has stabilized, turn on VMA/B.
Hold the control input pins Low while turning on VDD and VMA/B.
(All the control input pins are internally pulled down.)
After VDD and VMA/B completely stabilizes at the rated voltages, the RESET and ENABLE pins can be set High. If this sequence is not properly followed, the IC may not operate correctly, or the IC and the peripheral parts may be damaged.
...
A power-off sequence should be the reverse of this sequence

So... there are a couple of thoughts that come to mind after reading that:

First, has anyone ever seen that documented in the operating instructions for a TB6560 based driver? Like all those cheap ones from China and Hong Kong that are being sold on ebay.. do ANY of those say "um... by the way, don't turn on the motor power until after the logic voltage is up, and turn off the motor power before you shut down logic power."? Or do any of them have some magical circuitry to prevent the motor power from coming on until after the logic supply has settled?

Secondly, how on earth would you even do that in a driver that depends on a separate power supply? Keeping the logic power on after motor power goes down isn't hard, the caps in the regulator circuit will do that, but the power up sequence is the trick. If you were building a complete system, and you could build two power supplies (one for logic supply and one for motor supply) and you built in a timing control that flips on the one first and then the other, then, yeah... that's possible. But if you just sell a driver, and you have the user hook up any old power supply for motor and logic without warning them... or you have an onboard +5 regulator to supply the logic power from the motor power supply... how do you set that up so it doesn't fry the chip?

I started off wondering why so many people who buy these crap TB6560 drivers end up frying the chip, and NOW, I'm wondering how any of them work at all!
http://www.cnczone.com/forums/genera...y_chinese.html
I started looking into this after reading:
http://www.cnczone.com/forums/genera...se-24.html#281 so I'm not the only one who has noticed this problem...

I'd honestly like to hear if there is a good electronic solution to that problem. I'm not an analog electronics expert, or a power systems designer by any means. Maybe there is some way of doing this power up sequence with ONE power supply that I've not heard of?

The capacitors and current limiting in a standard +5 voltage regulation circuit pretty much guarantee that it will come up to +5 volts slower than it's supply voltage turns on. I guess you could add a great huge power resistor between the motor supply and the chip and put a great huge high voltage capacitor right there so it also comes up slower but A) high wattage resistors and high voltage high value caps are expensive and B) you would loose supply power due to the resistor. That's why people didn't like the old style of linear supplies.

Another obvious idea would be a relay, but then you are controlling an inductive load and as soon as the relay breaks, you are going to get a nice healthy back EMF spike which will kill the chip anyway. (Datasheet, page 32, point 3).

I guess you could hook your motor power supply INPUT up to a relay... but then you need a separate supply for logic power again...

Bottom line: TB6560 = 2 separate power supplies and the first one has to turn on the second one after a time delay. OR... it's going to fry eventually. No?

[James Newton]

Other TB6560 documentation:
http://www.semicon.toshiba.co.jp/doc..._en_090324.pdf mostly how to (not) make it smoke.

[doorknob]

I had a brief discussion of the power sequencing topic with another cnczone poster a few months back - he was building his own driver board based on the TB6560.

I don't remember how he decided to handle it, but maybe I can find that thread.

[James Newton]

I had a brief discussion of the power sequencing topic with another cnczone poster a few months back - he was building his own driver board based on the TB6560.

I don't remember how he decided to handle it, but maybe I can find that thread.

I'd be very interested to see that... I did a search for any posts from you with TB6560 in them and I don't see any mention of that issue in those posts... could easily have missed it... Thanks for your interest.

[ToshiroTamigi]

I thought about uaing the spindle relay to do the power sequencing.
I want to connect the regulators directly to power source, tb6560 motor power switched by relay and relay coil to the 12V regulator.
In this way the motor power will be switched on after 12V regulator output has reached a reasonable value.

A 100ohm resistor in parallel should be useful to prevent power spikes on the relay contacts (the dissipations shouldn't be a problem given the limited amounth of time it will be off).

[doorknob]

I'd be very interested to see that... I did a search for any posts from you with TB6560 in them and I don't see any mention of that issue in those posts... could easily have missed it... Thanks for your interest.

The discussion about a DIY TB6560 controller starts with post #56 in the following thread:

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


I raised the issue of startup timing in post #58, and the OP mmcp42 alluded to it in at least posts #59 and #60. I'm not sure that any definitive resolution of the issue was arrived at, though.

There are some other possibly interesting quirks that I'm unsure how you would want to handle.

For example, the discussion on page 28 about start-up timing seems to say that both VDD and VMA/VMB need to reach their specified voltages before RESET goes high, but in your circuit it looks like RESET is asserted only after VDD is turned on. Are we really to believe that, as the note states, bringing RESET high (for example) in the absence of the motor voltage being present can destroy the chip?

[James Newton]

I thought about uaing the spindle relay to do the power sequencing.

Right, I'd thought of that as well... it does mean you can't jog or move the axis until the spindle is running...

I want to connect the regulators directly to power source, tb6560 motor power switched by relay and relay coil to the 12V regulator.
In this way the motor power will be switched on after 12V regulator output has reached a reasonable value.

Not sure where you are getting 12V from... is that the logic supply voltage for your driver? I guess the board itself regulates that down to 5V?

Keep in mind, if you just switch the OUTPUT of the motor power supply, you will kill the drivers via back EMF when you power down. You MUST switch the INPUT to the power supply, and have a separate logic power supply.

A 100ohm resistor in parallel should be useful to prevent power spikes on the relay contacts (the dissipations shouldn't be a problem given the limited amounth of time it will be off).

Well... I'm not an analog / power engineer, but I'm pretty sure that will not prevent the back EMF kick.

[James Newton]

The discussion about a DIY TB6560 controller starts with post #56 in the following thread:

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


I raised the issue of startup timing in post #58, and the OP mmcp42 alluded to it in at least posts #59 and #60. I'm not sure that any definitive resolution of the issue was arrived at, though.

Thanks for finding that doorknob... As you say, it appears to have not been resolved and later in the thread he moves to other drivers.

[ToshiroTamigi]

The relay will not be managed by spindle output. It's coil will be connected directly to the 12v regulator onboard.

The emf will be handled by the diodes at tb's outputs(togheter with the tb6560's internals). I only need to revise the topology of the supply diodes to better cope with them.
I'll post a schematic as soon as i'll found a couple of free minutes.

[James Newton]

Mariss Freimanis posted this on another thread but I wanted to quote it here because A) it's brilliant (as usual) and B) it points out exactly what (I think) happened with the TB6560:

I would like to offer some general opinions regarding monolithic motor driver integrated circuits. A monolithic motor driver IC integrates low-level control circuitry with a power stage able to drive a motor in a single, neat and ready to use package. That is the promise anyway; unfortunately the execution usually falls short for one or more of the following reasons:

1) Custom (application-specific) ICs are very expensive to design. Usually a big client approaches an IC manufacturer and says "I have motor application for a printer and I will need 10 million ICs designed for controlling this motor to these exact application specifications."

The client signs a contract that allows the IC manufacturer to sell the resulting design after the client's needs are met in exchange for monetary considerations.

The IC manufacturer then free to sell the IC in the marketplace as a generic device. Unfortunately, the device only meets one customer's requirements exactly, the one who commissioned the device in the first place. Everyone else afterward who buys it has to deal with the idiosyncrasies built into the device as specified by the original client.

2) An IC manufacturer learns from their client base and attempts to build a generic motor IC on their own which should satisfy everyone. The manufacturer may be a top-rate IC designer but that doesn't mean they are top-rated motor drive designers.

Allegro comes to mind here; they have nicely executed ICs that make very fundamental design mistakes in understanding what's required to make a motor happy. An example is their recirculating, non-recirculating mode switching depending on what angle you are on the sine-cosine reference. It is a naive misjudgement of practical drive requirements.

I don't mean to single Allegro out, there are design criticisms for all the other monolithic ICs as well.

3) Integrated circuits are exquisitely sensitive to temperature gradients. They depend on a constant temperate across an IC silicon chip. Manufacturers constantly combat thermal gradients across an an IC to to guarantee it will work properly. To an IC designer, a motor drive is an IC from hell. There is the motor section that generates enormous heat and that heat interferes with proper operation of the control section of the IC. Motor drive ICs are not reliable devices; it's why they smoke easily.

4) Manufacturer's marketing departments over-spec the devices they sell. A good example is the L297 / L298 chipset. It was touted as a 2.5 A rated IC even though no sensible engineer would ever use it with motors over 1 Amp.

I have seen the L297 / L298 chip-set, the various Allegro offerings and the currently popular Toshiba IC come and go. In my opinion they are all doomed to failure for the reasons mentioned. Don't worry, there will be another new one next year and it will have the same fate.

Mariss

The TB6560 is a perfect example of #1 above (not to mention #3 and #4 as well). This thing was designed to be used in a /system/ where the power supply sequence was carefully controlled. It obviously had a logic power supply, and a microcontroller that would fire up, hold the chip in reset, THEN enable the motor power supply and complete the power up sequence with the TB6560 control lines. Easy when you are designing a complete system, quite difficult when you are designing a separate stepper driver.

[pminmo]

My experience with the 6560 failing is the quality of VMA and VMB at the chip leg itself. However, if a design is just thrown together not taking into consideration proper power supply timing for power up and power down, that will be an issue as well.

As to the Mariss post, some points are valid, some are Gecko sales oriented. I'm sure STmicro engineers would take issue with him....

In general though, monolythic stepper ICs are geared to consumer products, specifically printers. CNC usage is not why they are designed, but some of the chips work fine for hobby machines.

[James Newton]

Hey Phil, thanks for taking the time to reply. Could I ask for a tad more information? What do you mean by "quality"? Things like kick back? wire coming loose? Spikes? Over-voltage?