Hello every one,

I have a Grundfos upm3 pump with pwm control function. according to the data sheet I need to make a pulse according to the screenshot below.
For the pulse I generate a pwm signal of 400 Hz with a duty cycle of 30%. According to the documentation of the pump I need to add a dc offset of 1 volt.

So i created the circuit (see screenshot). I would like to know if I am doing it wrong or if there is a better way to do this.

 

I tried also a comparator circuit. Like below


I get this on my scope:


How do I get rid of the down slope?

 

You've got it wrong. The 1V spec is not an offset. The UPM3 manual shows the PWM input model as having an opto-isolator with constant-current control.
It needs 4-24VDC for a logic 1, and under 1V for a logic 0. Current is around 6-10mA, freq. 100Hz-4kHz. so you can just connect it to a 5V digital output on an MCU.
The pump's PWM feedback output is also opto-isolated, just need a pullup resistor to a MCU digital input if you want to readback pump speed.

 

You've got it wrong. The 1V spec is not an offset. The UPM3 manual shows the PWM input model as having an opto-isolator with constant-current control.
It needs 4-24VDC for a logic 1, and under 1V for a logic 0. Current is around 6-10mA, freq. 100Hz-4kHz. so you can just connect it to a 5V digital output on an MCU.
The pump's PWM feedback output is also opto-isolated, just need a pullup resistor to a MCU digital input if you want to readback pump speed.View attachment 262842

Thank you for the information. I have overlooked at that page of te manual.

For my reference, it is posible to connect a 5v pwm signal with a resistor of approximately 500 ohms in parallel of the ref and pwm in connection, to get a constant current?
Or is that done by the circuitry of the pump?

Thank in advance.

 

The pump's PWM input already has constant-current circuitry for the opto-coupler, so no resistor is required. I think around 100 ohms max. would give some short-circuit protection yet still meet the 4V minimum. The diagram shows around 6mA typ. current but their spec is <10mA.
The pump's PWM output is raw opto-transistor with 470R resistor and zener (protected), so a pullup to say 5V is needed to the digital input.
These PWM communication signals are in a standard- "VDMA Einheitsblatt 24224 "Wet runner circulating pumps - specification of PWM control signals" but I have not read it.

In your 2nd circuit, it is best to not use an op-amp as a comparator, because the op-amp input gets saturated and takes a long time to recover. The LM358 is slow and does not like high +ve voltage input. I think this caused the slow ramp or down slope you measured.

I've had problems with larger Grundfos UPM pumps, their firmware is not the best.
The firmware is designed such that you install the pump and program it once. If you change the configuration sometimes the pump will go crazy i.e. constant-pressure mode low P setpoint but is running full speed!

The fix is to reset settings to factory defaults, then program the configuration again, and do not make drastic changes to the modes.

 

Hi all, just found this thread while doing some googling on the subject. How difficult would it be to make a temperature controlled PWM signal for these pumps?

To give a little back ground. When fitted in boilers, these pumps are almost always controlled by PWM signal to give a specific temperature difference across the pipework. ie: if the boiler has 50°c on flow and 45°c on return, it may slow the pump down to increase the difference between the two.

They use these pumps in domestic systems, external from the boilers in some cases, where they are left on a fixed speed. It would be ideal to be able to make a small circuit, that could have 2 temperature probes, one on flow, and one on return, and be able to set it to target say a 5°c temperature difference. Where if it was 6 or 7° of a difference pump would run full speed, and if it was 2 or 3°c of a difference pump would slow down a little, all with the aim of keeping it as steady as it can around 5.

Yet while every pump has this input pretty much these days, I've yet to see a differential controller available for this purpose.

 

I think it's best done as an Arduino project. Why? Because changing the control algorithm - is it proportional/linear with temperature difference or just a high/low switch, do you want to take in account the burner is on, setting any hysteresis etc. is easily done changing the software. Changing hardware is soldering, moving parts etc. less flexible.
I do see many low cost pipe-mount thermistors or DS1820B temperature sensors out there. With ESP8266 or ESP32 you can WiFi enable it as well, have it generate a webpage or graphs etc.
I find better control with Grudfos is Constant Pressure (for hydronic heating systems) as that throttles the pump according to heat demand, with how many zone valves are open/closed.