Why does my Arduino 0-10V controller 2N2222 circuit go to maximum speed at a small PWM signal?

Question

I have built the following circuit from trial and error:

The circuit is to change the 5V PWM signal from an Arduino digital pin to a 0-10V circuit to control industrial equipment. In this case a CNC spindle controller.

I tried using an op amp to boost the PWM signal then smooth it with a filter capacitor but wasn't successful. I threw this together with a pc fan as test load and it seemed to work well. I didn't want to keep the test fan so I switched to a 75ohm 25W resistor I had on hand. It seems to work really well, I just take the voltage across the resistor and send it to the motor controller 10V input.

I get to maximum voltage across the resistor at a PWM setting of only 80 of 255. Above 80 there is no change to the voltage output. How can I get the full range from this circuit. I have heard that gates can act as capacitors so thought my PWM signal might be just charging this up too. See the proposed 10k pulldown resistor that I thought may help this.

Answer 1

Your circuit doesn't work like you think it does.

I simulated it, and got pretty much your results:

^{simulate this circuit – Schematic created using CircuitLab}

At low duty cycle, the output looks like this:

At high duty cycle, it looks like this:

The difference between the two is just a few percent in the duty cycle.

What you need would look more like this:

^{simulate this circuit}

That circuit provides a (more or less) DC output proportional to the PWM duty cycle. Since you have an NPN transistor in there, it inverts the duty cycle relationship. When the PWM from the Arduino is at a low duty cycle, the output voltage will be higher. When the duty cycle is high, the output voltage will be lower.

• R1 and R7 reduce the voltage from 12V to 10V - the maximum output of the circuit will not exceed 10V.
• R6 and C1 are a low pass filter with a cutoff frequency of about 8Hz.
• Q1 uses the PWM input to make 10V pulses at the PWM frequency. It shorts the junction of R1 and R7 to ground when the PWM input is high. This increases the voltage for the pulses, but it also inverts the pulse width.

Low duty cycle:

Middle duty cycle:

High duty cycle:

The 10V input to the CNC machine should have a fairly high impedance input. It shouldn't present much of a load to the circuit, so there should be no trouble reaching the full 10V output. If the 10V output maximum is lower than 10V, then your could reduce R1 and R7 - R7 must always be five times R1. You must also then increase C1. If you halve R1, then you must double C1.

You can increase C1 to make the output smoother, but then it will take longer for the output to respond to changes in the PWM duty cycle.

Answer 2

Don't try to smooth power PWM signals with capacitors. Your 220 µF capacitor is an equivalent of 1.45 ohm at 500 Hz. It will waste a lot of energy when driven with AC. Filtering is a good choice only for low-power signals, and 220 µF is unreasonably large in this case.

If your load can be driven by PWM, there's no need for smoothing. If your load requires proper DC, use a buck converter with a programmable reference. Generating that programmable reference with a filtered PWM is indeed an option.

Answer 3

First, ensure you don't need an isolated 10V output. Some inexpensive DIY setups use KB etc. DC motor speed controls which are designed to work with a pot, but the pot is connected to one side of the mains so you must add an isolation circuit.

Assuming that's not required here, a better approach than a transistor would be to use an op-amp. You can simply low-pass filter the output from the Arduino and amplify it with the op-amp. The output will change a bit with the 5V supply of the Arduino.

For example:

^{simulate this circuit – Schematic created using CircuitLab}

As you an see, settling time to 90% is around 1/3 second.

For a more accurate circuit, I'd use a reference such as 2.50V or 5.00V and switch that with a CMOS analog switch, but probably this is fine for CNC spindle speed. Usually a few percent is not important.