Large Brushed servo setup on the cheap (WIP)

[NC Cams]

The UC3637 assimes the use of a bipolar supply ( to allow for generation of the -10v analog side) and, like said earlier, the high power is on the "high side" (+10) of tje drive as well.

We have some OLD, OLD steppers that run bipolar and they switch the +15 and -15 with respect tothe common ground which one side of the motors is always switched to. THis way they "flip" the "positives" and thus change current direction that way.

There are MANY ways to do what's being attempted. The DIY'er will typically know only so much and be forced to paraphrase and plagiarize circuitry as opposed to doing a clean sheet of paper, absolute finest/revolutionary design.

Yes a PLC or uC can be used. PROVIDING, you know how to program it and therein lies the problem - not everybody can/will and the project stalls because such services are not readily found free.

The UC3637 will get almost anybody going, fairly easily and cheaply. This metchodtakes care of current limit, PWM and pass element in simple, plug and play fashion. It has its limitations BUT it will run the motor. ESPECIALLY a "low cost, brush motor" that I seem to recall was being adapted from something else (starter motor, DC whatever, perhaps even a treadmill). Heck, the sweetness of the uc3637 circuit is that it can even be tested with a simple 9v battery to establish/change direction.

A-D has been done and relatively done so easily. The trouble with ANY A-D is resolution. 8 bit resolution gives you only 4 bit in each direction. Is that adequate resolution? I dunno. That level of steps was not adequate in R/C cars for speed control (too "steppy" as opposed to a smooth analog 'potentiometer-like" speed control). The train boys didn't mind it, however when they went A-D.

Suggestion: do it modular style. Have someone who KNOWS A-D, to the A-D conversion for +/-10.

Do the uc3637 for the PWM. Use single leg, simple current control as noted in the 3637 app note.

Ditto that for the H bridge, especially the protection via fets, schottly's or even some "active current bypassing" which is done in the Megasquirt EFI circuit to bypass solenoid flyback that the EFI injector dump like crazy. Even with active flyback, some guys had issues. PWM'ing any inductive load tends to foster those problems.

If you really want to get trick, there is an old servo motor controller that ws offered by signetics - the NE544. It works off of a 50hx pulse stretch technique that should be easy to accomplish with a PC. Via some simple op amp oscillator drives that are fed by the 544, the output can be modified to switch the motors at 3k to 20khz.

Via op amp sense of low side current, you can fold back duty cycle as a function of any current you can sense, preferrable in the low side. A variation of this ruled the RC car race circuit in the mid/late 80's. Trust me, IT WAS KISS.

The use of MODULAR segments all plugged together allows for simple, easy troubleshooting. Again, the thread is "large brushed servo setup ON THE CHEAP". The trick stuff is definitely trick but it surely won't be CHEAP and it will be COMPLEX....

Reality check time guys......

[kestreltom]

Samco et al,

Here is a link to the schematic done in Design Works Lite (you will get a screen full of weird characters, but just use "save as" to save the default file name to your desktop). If you want to open the file, zoom in, check the component values, or edit it, you can download a 30 day free version of the program I used to create it here.

I also attached a jpg file of it to this post. But it will be a little hard to read. The default image browser used by this forum downsamples the images and will not display the jpg very well. Here is direct link to a higher resolution jpg of the schematic.

Samco, I found the answers to my questions about your circuit from an earlier post of yours, so I am posting the revision of your schematic which includes a current limit.

This is for everyone to see, comment on, and use for their own pleasure - no claims to how it will work yet. I will build a prototype and test this out over the next few weeks. Hopefully, I'll have some pics and videos to post showing a successful spinning motor.

Peter, would you please take a gander at the current limit circuit and post comments and suggestions. I really appreciated your previous thoughts, and tried to implement as much of it as I was able to understand.

All of you have been overwhelming with your helpful comments and musings. Please feel free to edit and re-post this schematic with comments, this way I will know what you are talking about!

Hope this moves the ship right along...

Tom

[kreutz]

Originally Posted by kestreltom Samco et al, Here is a link to the schematic done in Design Works Lite (you will get a screen full of weird characters, but just use "save as" to save the default file name to your desktop). If you want to open the file, zoom in, check the component values, or edit it, you can download a 30 day free version of the program I used to create it here. I also attached a jpg file of it to this post. But it will be a little hard to read. Samco, I found the answers to my questions about your circuit from an earlier post of yours, so I am posting the revision of your schematic which includes a current limit. This is for everyone to see, comment on, and use for their own pleasure - no claims to how it will work yet. I will build a prototype and test this out over the next few weeks. Hopefully, I'll have some pics and videos to post showing a successful spinning motor. Peter, would you please take a gander at the current limit circuit and post comments and suggestions. I really appreciated your previous thoughts, and tried to implement as much of it as I was able to understand. All of you have been overwhelming with your helpful comments and musings. Please feel free to edit and re-post this schematic with comments, this way I will know what you are talking about! Hope this moves the ship right along... Tom

Hello;

The current limit circuit should latch the over-current condition either during the rest of the PWM cycle (cycle by Cycle peak limit) or latch an over-current flag (for error handling) and shutdown the H bridge until it is reseted.

You don't need to use two amplifiers and comparators.

Take a look at the data-sheet of the UC3637 in order to see how it works.


Kreutz.

[kestreltom]

Hi Kreutz,

I have a copy of the UC3637 datasheet right here on my desk. Very familiar with it.

Thanks for the feedback.

Check out the datasheet for the IR2110. There is an positive edge triggered flip-flop built in which performs the shutdown logic on both high & low sides each pwm period. This is attached to a pin labeled "SD".

You may be right about not needing 2 amp/comparators. I included one for each leg of the Hbridge partly in order to experiment with them. Hopefully I will understand how the current recirculates in the PWM + PWM control scheme. If I remember correctly, Samco said that both of the bottom fets will be on during each pwm off cycle. This is unlike any scenario I've read about yet, and I suspect that there will be regenerative current looping through the bottom 2 fets to the low side rail. IF so, one of the 2 amplifier circuits will register the current flow with a positive pulse (the comparators will tolerate/ignore a negative pulse) and will trip the SD latch in the IR2111 if the regen current is high enough.

Tom

[kreutz]

Originally Posted by kestreltom Hi Kreutz, I have a copy of the UC3637 datasheet right here on my desk. Very familiar with it. Thanks for the feedback. Check out the datasheet for the IR2110. There is an positive edge triggered flip-flop built in which performs the shutdown logic on both high & low sides each pwm period. This is attached to a pin labeled "SD". You may be right about not needing 2 amp/comparators. I included one for each leg of the Hbridge partly in order to experiment with them. Hopefully I will understand how the current recirculates in the PWM + PWM control scheme. If I remember correctly, Samco said that both of the bottom fets will be on during each pwm off cycle. This is unlike any scenario I've read about yet, and I suspect that there will be regenerative current looping through the bottom 2 fets to the low side rail. IF so, one of the 2 amplifier circuits will register the current flow with a positive pulse (the comparators will tolerate/ignore a negative pulse) and will trip the SD latch in the IR2111 if the regen current is high enough. Tom

The IR2110 indeed includes a latch gate, so it will work as a cycle by cycle peak current limiting circuit.

About using the two low side mosfets (or two high side mosfets) for current decay (it is called Slow decay), that could be used on a 2 quadrant drive but not on a 4 quadrant drive for the reasons explained on the document posted a few posts above and the links I provided. 2 quadrant drives are not used on servo applications.

[PCW_MESA]

Originally Posted by kreutz The IR2110 indeed includes a latch gate, so it will work as a cycle by cycle peak current limiting circuit. About using the two low side mosfets (or two high side mosfets) for current decay (it is called Slow decay), that could be used on a 2 quadrant drive but not on a 4 quadrant drive for the reasons explained on the document posted a few posts above and the links I provided. 2 quadrant drives are not used on servo applications.

With proper control circuitry (and 2 sense resistors) slow decay mode gives full 4 quadrant mode, and simple low side unipolar current measurement. The only difference operationally is that the braking current (from BEMF) now flows between the two lower MOSFETS and the two sense resistors.

Peter Wallace

[PCW_MESA]

Originally Posted by kestreltom Samco et al, Here is a link to the schematic done in Design Works Lite (you will get a screen full of weird characters, but just use "save as" to save the default file name to your desktop). If you want to open the file, zoom in, check the component values, or edit it, you can download a 30 day free version of the program I used to create it here. I also attached a jpg file of it to this post. But it will be a little hard to read. The default image browser used by this forum downsamples the images and will not display the jpg very well. Here is direct link to a higher resolution jpg of the schematic. Samco, I found the answers to my questions about your circuit from an earlier post of yours, so I am posting the revision of your schematic which includes a current limit. This is for everyone to see, comment on, and use for their own pleasure - no claims to how it will work yet. I will build a prototype and test this out over the next few weeks. Hopefully, I'll have some pics and videos to post showing a successful spinning motor. Peter, would you please take a gander at the current limit circuit and post comments and suggestions. I really appreciated your previous thoughts, and tried to implement as much of it as I was able to understand. All of you have been overwhelming with your helpful comments and musings. Please feel free to edit and re-post this schematic with comments, this way I will know what you are talking about! Hope this moves the ship right along... Tom

A couple of things I see immediately:

1. You need some blanking (dead) time to avoid shoot-through from the top and bottom MOSFETS of each leg, probably around 300 - 500 nS

2. Be very careful with AT type power supplies, their outputs are not isolated from frame ground.

3. Surface mount or in any case low inductance sense resistors are important. Motor current cant change to quickly but the MOSFETS can change the current flow in the circuit in 10's of nS

Peter Wallace

[kestreltom]

Originally Posted by PCW_MESA A couple of things I see immediately: 1. You need some blanking (dead) time to avoid shoot-through from the top and bottom MOSFETS of each leg, probably around 300 - 500 nS 2. Be very careful with AT type power supplies, their outputs are not isolated from frame ground. 3. Surface mount or in any case low inductance sense resistors are important. Motor current cant change to quickly but the MOSFETS can change the current flow in the circuit in 10's of nS Peter Wallace

Peter,

Blanking time: from the IR2110 datasheet:
"The output drivers feature a high pulse current buffer stage designed for minimum driver cross-conduction. Propagation delays are matched to simplify use in high frequency applications."

They specify less than 20 ns deadtime, which looks too low. The IR2112 that Samco used shows 650 ns deadtime.

Do you know of a discreet component that can be inserted in series with the high side signals to the IR2110 which can be adjusted to provide the necessary propagation delay?

Computer ps: thanks for the heads up.

Sense resistors: I sourced these (2 watt/ .0087 Ohm) resistors from KOA. They look like what you specified.

Thanks!
Tom

[digits]

Hi Guys, can someone please explain how cycle-by-cycle current-limiting works on a system running in locked-anti-phase? Do you vary the duty-cycle of the fwd and reverse on-times to limit the current (i.e. modify the PWM signal to be more like 50:50 duty cycle), or do you actually end up with more dead time with all 4 FET's off?

Also, is locked-anti-phase (as described in that H-bridge secrets doc) the same as 4-quadrant drive?

Cheers.

[kreutz]

Originally Posted by digits Hi Guys, can someone please explain how cycle-by-cycle current-limiting works on a system running in locked-anti-phase? Do you vary the duty-cycle of the fwd and reverse on-times to limit the current (i.e. modify the PWM signal to be more like 50:50 duty cycle), or do you actually end up with more dead time with all 4 FET's off? Also, is locked-anti-phase (as described in that H-bridge secrets doc) the same as 4-quadrant drive? Cheers.

Locked anti-phase is one of the ways to implement a 4 Quadrant drive.

The way Cycle by Cycle limit works is as follow:

The sensed current is compared with a peak reference voltage, if the peak sensed current is higher then the output of the comparator trigger a flipflop and its output shuts-down the H bridge, the flip flop will be reseted on the next PWM cycle. That is why it is called Cycle by Cycle or pulse by pulse current limit.

di/dt depends on Power supply voltage and motor time constant, so the current increases during the PWM on period and decreases during the PWM OFF stage, if both periods are the same size (50% duty cycle), then the average current is zero.

PID action tries to increase the pulse width (speed) on the commanded movement direction, so average current tends to follow the average voltage change. The average current depends on load so under no load the current is small, but increasing the load will make the current to increase by increasing the pulse width (average voltage) in order to reach the commanded speed. When the current reaches the peak reference value the circuit will limit the torque by limiting the pulse width, thus limiting the rate of speed change (acceleration) with a given load and further torque increase.

[kreutz]

Originally Posted by PCW_MESA With proper control circuitry (and 2 sense resistors) slow decay mode gives full 4 quadrant mode, and simple low side unipolar current measurement. The only difference operationally is that the braking current (from BEMF) now flows between the two lower MOSFETS and the two sense resistors. Peter Wallace

Using slow decay on a 4 quadrant servo drive is possible but not wise. The whole purpose of a 4 quadrant drive is to reverse the movement while in forward motion by decelerating and quickly accelerating in the opposite direction. In order to do that it is necessary to decrease the current on the armature and increase it in the opposite direction as soon as possible, It is done using regenerative braking, so you don't want Slow decay, you want Fast decay.

While you can sense the decay current on the low side Mosfets (by using the two resistors technique) and use slow decay during the forward or reverse part of the motion (switching to fast decay during reversal only), it will complicate the drive circuit logic, and average current compensation (if you use a current loop) because current raise and current decay will not have the same slope.

I don't say it can't be done, it makes it more complicated than it should be.

[samco]

You guys are great!

Keep it coming.

thanks
sam