EMD Speed Sensor

[CShultz]

Has anyone here tried the new Sauer EMD speed sensor yet?  I am not a Plus+1 pro so I may be missing something, however this is supposed to be able to be CAN/Pulse/Quad configurable.. which using the service tool expansion pack they provide makes that easy enough.. but I have hooked this up via CAN and I can not seem to get a steady RPM reading using the EMD_CAN compliance block.  Sauer since then has basically told me to set the EMD to pulse and have it set to 180 pulses per revolution.  So I set the sensor to pulse pick up and set it to 180ppr, then use a standard Freq to RPM function block, and again I get jumpy readings all over the place.  Filtering the signal did not seem to help much.  IT is a pull up/pull down sensor so I have my pinconfig0 set to 3.  According to the Input Config manual I should use Pinconfig0 Set @ 3 and Pinconfig1 Set @ 0.  Looking for some help.  I have used standard Danfoss EM speed sensors with zero problem... so this new sensor which I would think is meant to be better and nicer to work with, has caused some curveballs.

[Jakob]

I have not yet tried this sensor, but is sounds a bit like the input is setup wrong.
This is your setup:
PinConfig0 = 3 = DigIn: Activated w/ Supply Voltage (Pull Down/Pull Up)
PinConfig1 = 0 = 0 - 5.26 V (mid range).

What happens if you use this setup?
PinConfig0 = (0) = DigIn: Activated w/ 5 V - Supply Voltage (No Pull Down/Pull Up)
and
PinConfig1 = 0 = 0 - 5.26 V (mid range).

regards Jakob

[CShultz]

Hello Jakob,
I have tinkered with the configuration of the input.. and it essentially shuts down any reading.  I do not recall the exact result of doing a pinconfig 0 = 0, I will re-try at some point.  However according to the EMD tech.  This is a 'push/pull output' sensor.. I take that as a pull up/pull down.  I have taken a snapshot of the page in the book and attached.

[Jakob]

A push/pull output forces the output signal to be either high or low, so if this is the case. A pull up/down resistance load should not be necessary.
But it shouldn't shut down the output either, so I don't understand why your readings fail.
Also, if the CAN signal output is jumpy, then maybe its the sensor that's the problem.

Are you getting the readout from the servicetool or a display?

Do you have an oscilloscope available to shed some light on whats going on with your voltage output?

[CShultz]

Hello Jakob,
Unfortunately I do not have an oscilloscope available.  Besides that of which comes on the service tool.
I initially had a DP200 reading my EMD readout but when it got all jumpy, I just made the program using an MC024-10 and monitored the EMD reading on the service tool with check points.  I have spoken with Sauer and they have since admitted there is some issues with using this EMD sensor with the pulse mode.  That I should take sample counts and filter them, etc.. which is bizarre to me to put a sensor on the market that isn't accurate.  They were surprised to hear I am having issues with the EMD_CAN compliance block, however the Plus+1 HelpDesk in the USA also got the same results I did so I know I am not a complete fool.

[Marbek_Elektronik]

We use the EMD Speed sensor in CAN-configuration
200ms message transmission rate to get good filtert values, better than 50ms. 200ms is maximum.

Does anybody know what "MA-Filter" means?
I find no information from Danfoss. We need it very quickly.

Our problem is the value of speed. It isn't constant. It varies over a wide range.
Momently I take an average filter. But I think the original values from EMD-sensor could be better.

[Marbek_Elektronik]

Ok, it is multi-average-filter = ma-filter :-)
But how does it works exactly?

[G30RG3]

Not Multi Average, but Moving Average filter...  And I guess It doesn't need explaining now, but in case of:
It takes the average of the x last samples coming from the hardware.   
Typically sensors take the average of all values since the last data transmission.
Some sensors also give you the option to transmit the AVG, the min, max and act value but AFAIK the EMD sensors don't do this.

On another note: I think your results will be better if you set a shorter transmission rate, and do the filtering in the microcontroller software

[Marbek_Elektronik]

Yes, of cource "moving" not "multi", just wrote by doing other things......my error.

Thanks.
But what ist the frequence of hardware sampling?
If there are 64 last data values, what ist the time of it?

Ok, I will use 50ms and compare it with 200ms for data transmission time.

[G30RG3]

Hardware sampling info: Hey Danfoss, you'll need to answer this one!
In the past, with the KPP sensors, it was easy enough to see the way the signal was generated, since we could count the actual dents or flanges on the speedring.
But I have no idea how the EMD sensor works internally.

[Nilla]

Hi guys!

To get the fastest support I would recommend you to get in contact with PLUS+1 Helpdesk on plus+1helpdesk@danfoss.com.

I'm not sure where you found the info on MA-filter but I don't think that will solve Marbek's initial issue; "Our problem is the value of speed. It isn't constant. It varies over a wide range."

As I understand Marbek, although the motor that the sensor is used for keeps a constant speed, the sensor doesn't. There are no details on the accuracy or tolerance of the speed range, only that it is 0-2500 rpm. This is how the sensor works.

As the EMD sensor is managed outside our business unit, and the question is hardware related, I will try to send your feedback to their engineering department.

Best regards
Nilla
PLUS+1 Helpdesk team

[Marbek_Elektronik]

Thanks Nilla.
Our customer asked helpdesk and they only send a short datasheet.
So, I asked here and I hoped to get information.

The sensor is mounted on an oil pump and the output of sensor is not constant.
We don't know how constant the oil pump is running. May be not constant over 360 degree.
But also the sensor seems to be not constant in output even if there would be a constant input speed.

If engineers give us documents of Speed over angle and other datas, this would be very interesting for us.

Now, I will test  rate 50ms and 200ms and will test ma-filter 64.

If I had to programm this sensor, I would suggest a modus of the CAN transmitter to send on the same position, not on constant time intervall.
At higher speed it should send every 360 degrees 1 message for example.
At lower speed every 180 degrees, but on same positions.
The message should be the speed and then the pulses and time since last meassurement.
Ok, I am dreaming.......

[G30RG3]

Hi Nilla, Marbek ...

https://assets.danfoss.com/documents/DOC152886482077/DOC152886482077.pdf  Or google "L1017819 pdf"
See page 8 for info and 14 on instructions to set the parameter.

BR
G30RG3

[G30RG3]

If I had to program this sensor, I would suggest a modus of the CAN transmitter to send on the same position, not on constant time interval.
At higher speed it should send every 360 degrees 1 message for example.
At lower speed every 180 degrees, but on same positions.
The message should be the speed and then the pulses and time since last measurement.
Ok, I am dreaming.......

Wouldn't like this behavior myself, because that would mean very long intervals when the motor is close to standing still, or even no updates at all when the motor stops...  OTOH at high speed you would get lots of data on the CAN-bus. The actual interval period of the MA filter should be able to filter out variations in angular velocity, but the techinfo isn't clear about the base clock on the MA filter.  I expect it to be maximum 64 pulses of the encoder in its highest resolution (180 pulses per round, so about 1/3 of a round) but it could be anything.

What bothers me more:  It takes about 2 rounds of the motor/sensor before there is output after a power up. So to use the sensor as a position encoder means you need to get an initialisation setting after every power up..

[Marbek_Elektronik]

We use the CAN mode:

We use MA-Filter Parameter 64, because we get the best signal.

And we use Transmission Rate of 10ms.
If you use longer Transmission Rate, 200ms, the signal is not better! It's the same signal as you get if you choose 10ms and look only every 200ms to it. There is no average beause of waiting 200ms.

But, the sensor is not very good: Signal is not smothed.
For example, it is running with constant speed of 200 rpm,
the signal is : (in 0.1 rpm)
2000
2018
1990
2029
2036
2003
2026
2030
2008
2026

There are big tolerances! So it is very difficult to regulate the speed of an oil motor.

Why does speed measurement is so unexactly?

jpg: blue is the current of valve, orange is the rpm