Voltage limiter with isolation and filter

[hapetter]

Im going to measure frequency from a hydro generator.

task:
The original signal is from the potential
transformer (pt) connected to the terminals of the
generator. This signal is first fed to an isolation
transformer to isolate the digital side. Then the
isolated signal goes through a filter and a voltage
limiter. The filter removes the high frequency harmonics
from the signal. The voltage limiter plays a very
important role. The governor not only measures the
frequency during the normal generation operation but
also has to measure the frequency during the start-up
and shut-down operations. During the start-up, the
output voltage is produced by the residual magnetism
from previous excitation and is very low. The ratio of
the isolation transformer should be small. But after the
start-up and the excitation of the generator, the
output voltage of the generator goes up to the normal
value and the output voltage of the pt is 220V (or
1lOV). So the voltage limiter is necessary in this case.

The rest of the measurement is taken care of a microcontroller.

I need some help to design the isolation transformer and the voltage limiter with filter. What electronics can I use for this task? maybe there is some electronics that contains both? Shoulb be for print card production..

Thanks!

Best Regards
Hans Pettersson

 

[ScottyUK]

Hello hans,

I have to ask... why re-invent the wheel? Proximitor type pickups have been used for years to determine generator speed, and for a synchronous machine, knowing speed implicitly tells you the generator frequency. A proximitor usually gives a single pulse per revolution or possibly a pulse train if it is using a toothed wheel as the source. In all likelihood your machine will already have some form of sensor either for the overspeed protection or the vibration monitoring. I'd seriously look at tapping one of the existing signals.


----------------------------------

If we learn from our mistakes I'm getting a great education!

 

[hapetter]

Pickup is ok for speed detection, but not for speed control.
The requirements are 0.005% resolution at nominal speed (50/60Hz)

 

[MacGyverS2000]

Pshew... that's a 2.5MHz sample rate, minimum, at 60Hz. Are you sure you have the poles on the motor to check that?

Dan - Owner

http://www.Hi-TecDesigns.com

 

[hapetter]

From the AC signal: I will measure the time T, from one zoro-level crossing to next. I will use a microcontroller to detect this and using 2 16bit timers and a clock frequency of several MHZ to acchive the resolution.

 

[MacGyverS2000]

To make it accurate to 0.005%, your voltage needs to have less than 15mV of ripple... that doesn't sound too fun to design around, either, but then again I never did enjoy my power engineering courses.

Dan - Owner

http://www.Hi-TecDesigns.com

 

[hapetter]

The PT signal is first fed into a isolation transformer and then trough a filter and voltage limiter. This is what I need help with.. the selection of isolation transformer and voltage limiter with low pass filter. Let me attach the hardware diagram. The components right to the voltage limiter is replaced with a microcontroller

 

[davidbeach]

As ScottyUK asked, why reinvent the wheel? A prox detector on a toothed wheel could give you hundreds of pulses per revolution, an encoder on the shaft somewhere could give you thousands of pulses per revolution, but you want to use a noisy signal that gives you six pulses per revolution?

 

[ScottyUK]

Hans,

Pickups are definitely used for speed control on big utility class turbines where overall control loop performance holds the speed setpoint somewhat better than 1 rpm in 3000. That's 0.033% shaft speed control accuracy using a control loop where the final control element is usually a hydraulic servo valve controlling a fluid flow.

I don't think your 0.005% resolution is unattainable using a proximitor with a toothed wheel - the method used by the QSS card from Emerson's old WDPF system gives a theoretical 0.006% accuracy although over the full operating temperature range they will guarantee [±]0.5Hz (0.016%). That's from a standard control system card built to a design which is at least 20 years old.


----------------------------------

If we learn from our mistakes I'm getting a great education!

 

[mgtrp]

From my uni days, I recall a couple of students doing a project that used phasors to calculate the frequency of a VT signal much better than zero-crossing methods. No idea of the hardware that they used though.

 

[Skogsgurra]

Hello hapetter.

My friends here seem to ignore the complications involved in mounting a toothed wheel and a proximity detector (or inductive pick up coil) on the machine.

Measuring voltage frequency after some filtering is a perfectly good way of doing it. You do not even need much filtering. A passive two pole RC filter works well. And if you expose the microcontroller input to the whole voltage, you get very sharp and precise "zero" croosings ("zero" meaning input threshold of the input).

Damage to the chip if doing so? None. All micros that I know about have substrate diodes that can take care of around ten mA. Just use the current limiting properites of the passive RCRC filter. A good design will keep current in the substrate diodes below about 1 mA.

Have a look at

http://www.gke.org/presentationer/files/Utveckling av frekvensrelae.pdf

pages 15 - 19.

Good luck! BTW, why do you need such a high accuracy?

Gunnar Englund

http://www.gke.org

--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...

 

[ScottyUK]

Skogs,

Your friends know that most of these machines already have such a toothed wheel because it's used by control and protection systems which are installed by the manufacturer. But I agree, if there isn't such a wheel then it does complicate things somewhat. Hans should be awake soon - we'll see what he says.


----------------------------------

If we learn from our mistakes I'm getting a great education!

 

[hapetter]

Good morning folks

This is hydro, not gas or steem turbines. The nominal speed is low < 750rmp. and as low as 100rpm sometimes. And the pickup signal has more noise from what I have seen. Many tooth wheels on existing units have down to 4 teeth. Never seen as many as 100 in hydro.

The common way in hydro is to use pickups to detect speed up to nominal speed, since you dont have PT signal at low speed. Then for synchronizing and speed control we use PT signal. Then pickups are backup if PT fails

Anyway. I like to make this frequency mesurment board. I need help to design and select components for voltage limit and filter circuit.

Anyone?

 

[hapetter]

And. The high accurancy is a requirement in hydro. Given in IEC and IEE standards.