VFD control voltage

Hi,

I have a chinese 2kW 3 phase inverter driven spindle, bought as a pair. The VFD takes either manual speed setting or on i/p voltage.

It provides a 10V supply which it seems is intended for a 10k pot. to provide an i/p signal for speed control

I want to control it from software ( GRBL ) via S commands. In principal this works but the output of my circuit is very non linear.

I have a 3V STM ARM board programmed with GRBL, I have added a GPIO port which I'm switching on an internal timer. PMW freq is 2kHz

To protect the uP card I am going through a FOD817 opto isolator.

It's half way there but the non-linearity above 14k rpm makes it a but useless as it is. Can anyone suggest how I should improve the cct?

Sadly , the manufacturer does not provide any info on the nature of the i/p circuit, so it's a bit of guesswork know what to drive it with.

TIA.


PS I don't seem to get the graph I uploaded. Basically it is pretty linear from 0-14k ( 0-9V ) ; then it asymptotes to 10.1V. The range is good but rising too quick then flattening off.

I have not used GRBL but

my initial thought is if you can adust the PWM output to give you 0 to 8 volts for the range of spindle speed you want

then use a LM358AN op amp to amplify the DC voltage by 1.25

John

PS

Attachment 424954

if you make it so 80% of the op-amp output is across R2 the gain will be 1/0.8 = 1.25

for example R1 = 2000 ohms and R2 = 8000 ohms

gain = 1 + 2000/8000 = 1 + 0.25 = 1.25

Quote:

---

Originally Posted by john-100

I have not used GRBL but

my initial thought is if you can adust the PWM output to give you 0 to 8 volts for the range of spindle speed you want

then use a LM358AN op amp to amplify the DC voltage by 1.25

John

PS

Attachment 424954

if you make it so 80% of the op-amp output is across R2 the gain will be 1/0.8 = 1.25

for example R1 = 2000 ohms and R2 = 8000 ohms

gain = 1 + 2000/8000 = 1 + 0.25 = 1.25

---

The VFD Drive needs 0-10v

Thanks John, I do not think this has anything to do with GRBL, I just put that to give a bit of context. The software sets the duty cycle of the PWM as a linear % of the full range. I have not dragged the scope into the workshop, but I have no reason for the moment to suspect that is not working as intended.

I do not want to add extra complexity unless needed. The simplest way would be compensate the non-linearity in software.

You suggest reducing the range in order to amplify it. I don't really see the point , or that that in any way addresses the problem. I suspect my cct is poorly constructed but I don't see where. If I could understand the cause of the non-linearity I can then go about fixing it.

There is a drop across R9 going from 0V at 0 to 0.5V at 24000, not enough to explain the non-lin.

I was just thinking if the output of your circuit is linear between 0 to 8 V the amplifier would be a simple option

if the GRBL board has enough processing available the software correction is another option

John

Quote:

---

I was just thinking if the output of your circuit is linear between 0 to 8 V

---

Ok I see what you were suggesting now.

My aim is understand where I'm going wrong rather than hack it without understanding.
From the voltage drop on R9 it looks like the i/p resistance of VFD is about 20k,
I think it is the asymmetric charge/discharge resistances. 7k5 probably needs to be close to R9 to provide roughly equal current paths.

Quote:

---

Originally Posted by john-100

I was just thinking if the output of your circuit is linear between 0 to 8 V the amplifier would be a simple option

if the GRBL board has enough processing available the software correction is another option

John

---

VFD Drives can only use 0-5v or 0-10v for speed control, nothing in between is going to work for true speed control some will take a step pulse as well, most only use 0-10v and Modbus

He was suggesting amplifying the linear 0-8V part of the range to 0-10V . Then adjusting the software to produce 8V for max rpm.

That is a valid solution but I do not have room to add an extra op amp at this stage. I also want to know where I went wrong with the opto circuit to cause the non linearity. It looks like I'm on the wrong forum for electronics advice.

Hi reg

looking at your circuit again

I have two thoughts

1) is the ARM boards PWM output correct and follow a linear relationship with the spindle speed setting & pulse width ?
( may be your circuit is OK )

2) does the opto isolators photo transistor switch hard on to minimise the volt drop across the transistor ?
depending on the volt drop across the transistor the 10V supply could be too low correctly as you approach 14000 rpm



John

PS

whats the limit on the GP10 pin output current

if you can try reducing R5 the 1K resistor to 500 ohms to double the LED current

thanks for your thoughts John. I have no reason to doubt the software, it's pretty straightforward and I've checked it.

I removed the RC filter, I think the VFD must have enough internal smoothing to cope with the 2kHz chopped input ( again no doc on that ).

I realised that the non linearity is from the 7k5 being too close to the 20k input impedance of the VFD. I dropped that to 4k7 and it gets a lot straighter. I'm trying 2k2 but then I do need up the diode current as well.

If I boost diode current, anything between s20000 and s24000 gets me the max rpm.

2k7 for R8 and 330R on the diode was a lot less linear, though I was expecting that to be better.

4k7 and 1k on the diode provided just enough demand on photo transistor current to drop the max to about 10V . It's all as hacky as hell but I don't think I can do better without adding my own regulated supply voltage and I'm into diminishing return for effort now.

In fact, the mid-range oversets speed by about 500 rpm and drops back near full speed. This is probably accurate enough.