AVR for 100 KVA Alternator

In its simplest form an AVR is a negative feedback loop. You have almost everything that you need except for a setpoint potentiometer and a comparator. Since you got this far I believe you have the skills to adopt the AVR circuitry from the referenced thread and make it work.

It will require some tuning to get the gain and time constant right, and you might consider putting some upper and lower voltage limits in as well. There are people on this site that are better at electronic circuit design than I am and I'm sure that they'll help you get it right.

Thank you RAMconsult for your input. I love electronics as a hobby and can do basic circuits but do not have the skills to design and build circuits from scratch. Hence my appeal to the kind experts on this forum.

Sude

sorry i cant help you. But you can concern your "AVR for 100 KVA Alternator" issue with any good specialists....

What are the reasons for building instead of buying?

What you are planning to do is very dangerous to life, limb and property......or do you have the necessary training and experience to carry out such a design?

Even secondhand ones may be available for a lot less money if you look around carefully....

Here is a new one in the USA for less than US$130:-

Genset-AVR-Generator-AVR-R449-Automatic-Voltage-Regulator-Alternator-AVR-/

Just look around like I did on the web......

I'd have thought you need to adjust the water flow manually (or otherwise) to keep the rpm within acceptable limits, to maintain the correct frequency, the AVR looking after the voltage in the usual way by varying the excitation.

It is not needed to maintain the frequency if connected to the mains, only needed for a complete private system.

I'm sure you're right, but I'm trying to think it through (connected to mains case).

Seems fine if the turbine puts out enough shaft power to run the alternator at rated output. But what if the water flow is low so turbine output down? How does the alternator know? If it gets full excitation it will demand more power than the turbine can supply. But on what basis could the excitation be adjusted?

On e.g. a plant generating from biogas (with varying gas availabilty) can measure output current, compare it with a setpoint and trim the excitation via the AVR, the current setpoint determined by gas production (typically using gasholder level). But on the hydro system described, we may not know in advance what water power is available. Unless it varies with say water level upstream of the turbine, which can be measured.

The AVR is Standard and you could buy it cheaper than trying to make it.

The AVR voltage is supposed to be variable in order to vary the DC Current that goes into the field winding so that the output voltage remains constant when the load on the Alternator varies.

The water flow must be varied by a different mechanism, called a governor, to keep the rpm constant as the load (again) varies, so that the frequency remains constant.

I hope this clarifies the concepts behing the power generation system.

Hello everyone thank you for your comments and input. Sorry for the delay in responding. Also my apologies for not properly introducing myself. I am from Sri Lanka and my name is Sudharshan Soysa - Sude - for short.

The problem is I can not buy an AVR spare from the manufacturer as I can not locate him. Also the system is quite old, installed in 1989 and it appears the spares are out of production.

The modern AVR's available have a very high excitation voltage ranging from 24 volts up. I tried adopting one of the modern AVR's but did not work as during start up, as soon as the exciter voltage goes above 12 volts the whole system act as a magnetic brake and prevents the turbine reaching working RPM. I forgot to mention that there is no mechanical governor, instead there is an electronic load bank to divert excess energy not used by the main load.

If the turbine is being prevented from reaching Working rpm, it means that the load is too high for the water flow setting! That is, provided that the output voltage of the main alternator is within the rated voltage and not higher(?).

That is why you need a Flow control mechanism to increase or decrease the flow according to the load variations.

Start by reducing the load, or disconnect the load completely and see if the AVR can regulate the Voltage output from the Alt while you regulate the flow (Manually) to keep the frequency steady at the rated value (50 Hz or 60Hz ??).

The AVR Mk5 could be applied. It would be necessary to connect the common of the excitation (S1 ?) to the output neutral. Your diagram has not reproduced clearly enough to be sure of terminal markings.

Following is what I think you have, re-drawn so I understand...

To try the AVR Mk5, you would have to add a few power components, remove the existing AVR (terminals 4,5) and connect A-A, B-B, C-C between two drawings. The UVW connections to new AVR must come from the fuses, as 1,2,3 on the existing AVR.

This is the modified Mk5....

I must check that the FET1 type and the current limit TR1 are OK for your field, but this would be the essentials.

Do you have the original AVR? If you can identify the components then just replacing them all with ones of higher rating should work.

67model

Hello every one, apologies for the silence, and thank you for your feedback.

Thank you 67model for your advice. I was composing a comprehensive explanation of the setup as I have at present. As before I got a time out can not process message. Wonder what the time limit is to write up a post. I will attache a cropped and enlarged exciter portion of the wiring diagram. Hope it will be clear. Also I will write up the explanations of what I have done in Word and see if I can insert it to the post.

Many thanks for your time.

Sude

OK, the diagram and excitation is clear now. Whether 2HK is a hand switch Auto/manual or a relay does not matter. I attach my modified diagram - had got relay J with normally open contact - my mistake.

Is there any parallel operation of this generator with other generators or a public supply?

If no, then there is no need to use the current transformer connected to the AVR.

My next question is if you want to go the FET route for an AVR or the Uni-junction Transistor (UJT) firing circuit- Silicon Controlled Rectifier (SCR)?

SCRs do have the virtue of being robust and being the usual way for AVRs. If there is a fault in a component or construction, it is not so easy to find what is wrong.

It depends some on what components you have or can get easily. The equivalent of a UJT can be made with NPN and PNP transistors.

67model

67model, thank you for your reply. I will explain what I have done at present. I am still having problems of cannot post issues so I will brake up my reply to a few posts.

I have connected a LM317 to give 7.7 volts via a a 12 volt SMPS which activates at about 120 volts AC. The relay Common is connected to E2 of the exciter field, NC to the anode of ZP2 and NO to the out put of LM317. This setup draws about 600 mA and is sufficient to provide about 25 amps to the load at 400 volts AC.

In the original circuit relay J has no significant effect on the startup . As soon as the SCR in the AVR fires it by passes the resistor network RQ/RF. As mentioned in your post 12, I tried to replace the semiconductors in the AVR was but unable to do so as all markings were in Chinese and the components were nothing like the modern day ones. I noted that the SCR was fired by a modern looking component which I subsequently realized to be a UJT. I have come to this conclusion after seeing Lars Dasuns' circuit diagram in the thread started by Mishel. The original circuit was very similar to this.

Sude

To continue from my post 16

The answers to your question.

1) The control panel is wired up for parallel operation but the alternator is a stand alone.

2) Yes I would like to go the UJT/SCR way as that was the original concept. I think the SCR method would be more robust as that system was in operation in my set up since 1989! However if it means I have to trouble you unduly the FET way too should be ok. Either way I will follow your advice.

3) Purchasing of components is not an issue in Sri Lanka as most are freely available. In fact I have a few UJTs in my component 'larder'.

Thank you very much for your time

Sude

OK Sude,

Two points..

1. The Lars Desun circuit is designed to feed the machine field direct off the phase volts (230 rms). Since your field is about 7V for rated line volts (400), designed to be fed from a 25V rms S1-S2 supply: applying 230V could be destructive.
2. With an 100n discharge capacitor for the UJT emitter, it is recommended to have a supply of 12V to the UJT to get a good strong firing pulse. This is 1/3 of the peak voltage of a 25V rms sine wave, which does not occur till 20 degrees into the half-cycle [sine 20 degrees = 0.34]. That is assuming 25V~ is available - but the remanent rotor magnetic field at run-up will deliver a lot less.

So, regarding the J relay and Rq, these may be necessary to "kick" the auxiliary supply S1-S2 into life as soon as possible.

Some questions:-

1. Have you measured the residual Line-Line voltage off-load at rated frequency? You would need to adjust the generator to rated frequency, with excitation ON and 400V, then turn excitation off. Do excitation off/on several times and record results. The S1-S2 excitation supply voltage should be measured at the same time. This would give the available voltage S1-S2 for "wake-up".
2. Have you done an off-load (open-circuit) line voltage versus excitation current test (saturation curve) at rated speed? If you ever suspect any problem with the generator, it is useful to know what this curve should be.

I will give thought to the AVR circuit.

Regards,

67model

Thank you 67model for your quick reply. Sorry what I meant by 'similar' to Lars Desuns' circuit was the UJT firing the SCR part. I am clueless of the actual functioning of the circuit. Now I know, at least I think I know!

My recollection is the original circuit had two sections. One to sense the voltage and the other part to fire the SCR. The sense part was powered at 15 volts DC and the SCR firing part had 38 volts DC. The power for the two circuits was from a small three phase transformer with two separate three phase rectifiers. I had traced out the circuit of the original to some extent but I believe it is not that accurate. Would it be of any help to you if I posted it in the form I have?

Originally at startup the voltage would rise gradually and the first set of heaters - 2 KW - in the load bank would start firing. When the water flow is increased further the whole power system would sort of shut down momentarily and restart almost immediately with proper operation, I believe with the AVR taking over to give the necessary power regulation. The load bank handles up to 70 KW. The load is given to the factory where it is used to run mostly motors - up to 15 HP - and factory lights.

Answers to your questions.

1) At startup the residual voltage at S1-S2 is 4 volts. As the speed builds up the voltage would rise and the out put to 390 volts AC no load, phase to phase when the AVR is inoperative and the system does not give any usable power. At this state the voltage at 4-6, input to the AVR is 10 volts AC, @ 50 Hz. When 4-6 is shorted out momentarily the out put rises very rapidly to 410 volts and the load bank takes up the power generated.

2) The answer for this is no. How do I set about doing it? Would I need to use an external variable DC power supply feeding E1/E2? As mentioned in my post #16 the current draw was approximately 600 mA. However I did not do any accurate measurement and this reading may be wrong. I do not suspect the generator is at fault as at present it is operative with the fixed voltage regulator.

As it is the dry season now in Sri Lanka the turbine is shut down for lack of water and maintenance of the open channel and forebay. As such I will not be in a position to do the tests in (2) above in the near future. Once the civil work repairs are done short runs of the turbine are possible to do the tests.

My apologies for the long explanations. Thank you once again.

Best Regards/Sude

No apologies needed for length of explanation. Accurate communication of facts takes time and thought, else questions fly back.

A residual voltage of 4V at S1-S2 would not be enough to get an electronic AVR into life - hence the relay J, which probably energises at about 70% volts, opening the Rq bypass.

Here is a diagram for a Thyristor (SCR) AVR.

I have "breadboarded" the circuit in the diagram. It works as expected.
The diagrams below are oscilloscope traces of SCR current, 25 VAC voltage.
One is for V1 [TR2 base] to common at 0.78V, the other V1 = 0.99V. The negative supply, -12V, was adjusted so VR1 had enough range for V1 variation. Note the variation of SCR conduction angle.

The final diagram has voltage on C2 & 25 V AC. Note that the charging time of C2 is so short that several pulses are given by TR4 in a half cycle.

The resistors, LED lamp with ballast resistor in the SCR anode are to provide a test load. The lamp/ballast would be useful on your generator field.
With V1 = 0.78V, the LED was just flashing once/second to several. With V1 = 0.99V, the LED was lit steady.

The 1 ohm resistor was to monitor the current on the oscilloscope - you would not need it in service.

BRY39 is a "programmable" Unijunction. The 2N6028 is similar. The "firing" voltage as a proportion of supply is set by R9, R10. All resistors should be 1/4 watt, apart from R2 [1/2 W] and where marked.

If you want datasheets, I recommend www.datasheetarchive.com on the internet.

For SCR I would suggest something robust for a few amps, say ST type TYN625,
or equivalent, this is rated 25A rms 600V at 100'C case temperature.
TYN825 (800V) may be easier to get.

If you want detail explanation of circuit operation, I will give it.

Re- your question on 2), I would open the field circuit at E1 and insert a moving coil ammeter with shunt (say 10 amp range). Digital meters can be a problem on fluctuating DC current. I would first try in Manual, using the W3 rheostat to adjust field current up and down, switching in extra resistors may be needed to get below rated field current. However, a variable voltage supply would be easier for the test. Increase current in steps about 10% rated from 0, record DC field current and AC output voltage at each step. Go up to about 3 x rated field current, then go back down to zero in steps. Plot a graph. Rising and falling will be different - you have a "hysteresis loop".

Do you have a speed meter for the turbine? The saturation curve needs to be at constant rated speed or, at least, adjusted voltage to rated speed

Best wishes,

67model

Sorry, Sude, forgot to put circuit in.

67model

And the "breadboard" for circuit...

67model

Hello 67model,

Many thanks for your excellent feedback with explanations and a working circuit. I will study your circuit diagram/explanations and try to understand the working of it and get back to you with questions if any.

I will also immediately try and make the circuit and post the progress. If the programmable UJT is not available would the normal UJT 2N2646 work (a stupid question no doubt)?

Thank you for your time and effort.

Sude

Hi 67model,

Continuing from my post #23

I give below the component list of your circuit in post #21, some of which are not clear. Please give the values/correct if wrong.

R1=10k, R2=10k 1/2w, R3=3.3k, R4=8.2k, R5=??, R6=22k, R7=??, R8=10k, R9=6.2k, R10=10k, R11=16 ohms.

TR1=TR2= BC 107 or similar?

C1=220uF/16v, C2=0.1uF

Question.

1) How is the +12 and -12 volts derived for the input? Is it from an external power source?

2) How would I connect the circuit to the alternator? Do the points 4-6 (as given in the wiring diagram) or the field E1-E2 connect to the lamp/ballast?

If it is no bother I certainly will appreciate a detailed explanation of the circuit operation. There is so much to learn in this field.

I will have to post pone the saturation curve test you called for in post #18 as at present we are experiencing severe drought conditions. The weather gods tell us that a weak El-Nino weather pattern is developing over Sri Lanka.

Thank you and best regards/Sude

Hello again Sude,

Regarding your post #23 : Not a stupid question! The 2N2646 would do perfectly well. I have added an inset to the THYAVR circuit showing how it replaces BRY39 (the negative end of the 2N2546 base is B1, the positive B2). For some reason, R13 (100R) is specified as 1 watt in UJT application notes.

In reply to post #24 :

a) Your component values are correct, except I have changed R6 to 10k; R5 is 20k,R7 is 2k7 (measured -it came out of the "larder" as an ancient, all red, carbon 2k2).

All the small diodes are 1N4148 - any silicon diode of similar rating would do.

The capacitor C1 is a polyester type, a 50V DC rating type is OK. It carries a heavy pulse current (2N2646 etc rated for >2 amp peak current) - a solid foil or extended foil is recommended rather than a metalised foil with a weak end termination - capacitors designed for AC are OK (the one on the breadboard is Wima MKS4 250V~ note the construction in http://www.wima.com/EN/WIMA_MKS_4.pdf - the end of every foil is terminated in the end metal contact block [scoopage]).

BC107 or similar OK. TR3 is BC157, this a complementary pnp version of BC107 electrically - anything similar OK.

b) Reply to your questions.

1) The +12V supply can be got from the 25 VAC S1 -S2 winding using a circuit like that in the marked-up AVR (Mk5) of post #12. The C1 of 4700 mmF can be reduced to 470 microfarads. Because a sudden load disconnection on a generator can give a +50% voltage surge, I recommend the resistor and zener, LM317 can stand about 40V + 12V output = 52V - any more is sudden failure of LM317 and likely AVR components.

You do not need a -12V supply, that was just for a bench test. for your own bench test, you could use a 9V battery (I always cut the terminal ends off old 9V radio batteries - they make a free battery connector!) - use a 4k7 resistor in series with a 10k variable in place of R1.

In an AVR the negative input is from the rectified machine output - 2 resistors of 100k 1 watt in series in place of R1 would suite -280V.

2) You connect common to S2 and the top of the "25 V AC" is S1. My 25 VAC corresponds to the 25 VAC S1-S2 power winding on the generator. The anode of the SCR is in place of terminal 4 of the old AVR.

Sorry to hear about the drought, last year here was the wettest for 250 years!

But must get the hydro-set ready for the rain..

I have modified the circuit to discharge C2 to zero on each negative half cycle. The LED moves smoothly from a faint glow at V1 = 0.85V to bright at V1 = 1.1 volt, without any flickering. The new parts are R12, D2,D3. Not shown is a 10 microfarad capacitor from point V1 to common.

This AVR, of course, works with an inductive field load. This makes a big difference to the current in the SCR. I substituted the 245V~ winding of a 60W mains transformer, with a parallel diode, for one of the 120R resistors in the test load.

Its inductance was measured as 29 Henries by feeding it from 30.1 V AC (the 25V AC winding) via an 8k2 resistor, getting 18.0 volts across the resistor and 20.06 volts across the transfo. So its reactance was 8k2 x 20.06/18.0 = 9k25 = 2 x pi x 50 Hz x L. So L = 9250/(2 x 3.142 x 50) = 29.4 Henries. Measuring the phase angle of its current, passed through the 1 ohm resistor, gave a value of about 1200 ohms for its effective resistance. So its L-R time constant L/R was 29/1200, about 24 milliseconds, about 1.2 cycles of 50Hz mains.

This shows the current with the LED indicator dim...

As you can see, the inductance resists a change in current and the current does not go.to zero as it did with a resistive load.

Here is the current with the LED bright...

What you see is the current jumping up to about half its peak value when the SCR is turned on. This is because it is taking over the inductor current which was circulating round the diode in parallel with the inductor while the SCR was off. When the current has passed its peak, because the voltage is falling, the diode begins to take over the inductor current, until the SCR is carrying zero current and turns off - until fired on the next cycle. Note the SCR is off for about 12ms of the 20 ms cycle, so the inductor current is circulating through the diode and decaying for 12ms.

If you work out the exponential decay of current in an L-R closed circuit with L/R = 24ms for 12ms (half the "time constant") you will find 12 ms is about 0.6 of initial current - so a change on the oscilloscope trace from 3.8 divisions at peak to 1.9 divisions at minimum i.e. 0.5 is close to expectation. Note 3.8 x 0.05 V/division = 0.19V; that is 0.19 amps with a 1 ohm resistor.

That is enough for one day.

Regards,

67model

Many thanks 67model for your post#25 giving the redesign of the THYAVR to use the 2N2646. A minor clarification B1 of the UJT, I presume is the out to the SCR via R11?

Pressure at work does not give me much time at present to experiment. However I have assembled the circuit on a breadboard but have not fired it yet. As it is quite late now I will recheck the connections and test it tomorrow. Will post the results as soon as I can.

Best Regards/sude

You are correct, Sude, R11 goes to the SCR gate, as on the main circuit diagram.

I will write a circuit operation description and look forward to reading your test results.

Does your turbine have a speed indicator?

Regards,

67model

Hello 67model,

In reply to your post #27: Assembled the circuit on the breadboard and fired it - BANG -worked as you predicted. I have made minor adjustments as given below for the bench test.

1) Substituted ballast resistors on SCR circuit with 12v/5W auto lamp.

2) 25V AC was provided with 230/9 volt transformer.

3) -12V DC to the input of TR2 was via variac and 230/16 volt transformer with one diode, (anode) feeding R1 and other end (+) connected to common.

4) Substituted R4 with 27K trimmer pot to achieve V1= 0.4 to 6.45 volts with VR1 for 0 to 230 volt variation on variac.

5) When variac out put is zero V1= 6.45 volts and bulb glows at maximum. As the voltage is increased V1 steadily drops and 0.4 volts and bulb is a faint glow with no flickering.

6) Conclusion - your circuit works beautifully as designed by you.

Question time:

1) In your post #23 you mention, negative input is machine output with two 100k/1W resistors in series in place of R1. Shouldn't a diode be present in the circuit? Does the negative half cycle provide the necessary (-) input?

2) Wouldn't it be good on a safety point of view to put an isolation transformer to (1) above?

3) Or couldn't I feed the "sense" voltage with a small step down transformer with appropriate changes to the input components for the lower voltage?

4) I presume the out put voltage of the alternator is set by VR1.

5) The way I understand the working of the circuit in a nut shell is; when V1 is > 1 volt the SCR is turned hard on and as the line voltage reaches the set voltage - 220V AC - the SCR is fired less and less to maintain the out put at the desired value. A brief explanation would be very welcome.

6) I have access to a scope. How would I connect it to see the firing pulses?

7) On the design for Michele the FET failed when the welding transformer was used. The reason I requested an AVR design with a SCR was primarily to over come this issue. Would a similar failure occur in this design as well? I personally think this design will stand up to on/off loads caused by a welder.

Your design is very similar to the original. I now believe I got messed up with the original circuit as I could not locate the "sense" part of it. I am very grateful to you for your time and effort put to design and prototype an AVR for my application. Now that I have a working circuit I will commence making a circuit board to put it all together.

Thank you and best regards/Sude

A sensible adaptation by you to use available components and test gear for trial voltages, Sude!

In reply to your questions 1 & 2. You are correct that diodes are necessary and that the AVR (Mk5) circuit got the generator output voltage reference from the negative half cycles.

Also, although a connection direct to generator neutral was made for simplicity on the AVR(Mk5) your existing AVR diagram looks as if isolating transformers were used inside the AVR and you probably still have these transfo.

Also, since connection is from all phases UVW to terminals 1,2,3 on the AVR it looks as if you have 3 phase rather than 1 phase sensing for the AVR - the Delta voltage being transformed to a lower voltage.

So, the following circuit is suggested to get the feedback voltage to the base of TR1.

I am sorry the diagram is not perfect - I was experimenting with the EAGLE printed circuit board software to draw the circuit diagram. R4, VR1 and TR1 are on the existing THYAVR circuit and R1A, R1B replace resistor R1.

In brief, if the secondary voltage of the two transfos is 40V rms corresponding to 400V rms line-line [10:1 ratio] then you will get almost 95% of their peak as a DC value from common to cathode of D1, D2, D3. That is 40 x 1.4142 x 0.95 = 53.7 volts.

Since the voltage across R4 and VR1 is 12 volts less 0.8 volts [for 12V DC supply] with the SCR just conducting you have 11.2 volts across 8k2 + 0.5k = 8.7k [VR1 at mid point]. This must be compensated by the negative voltage from the rectifier (ignoring the 0.8 volts which should be added to the 53.7 volts in exact theory) so,

(R1A + R1B)/8k7 = 53.7/11.2 = 4.8. Hence R1A + R1B = 4.8 x 8.7 = 41.7 k

Use say 20k and 22k = 42k in this example.

The voltage rating of C3 should be at least twice the normal voltage of 53.7 in this example.

You are correct on 4), 5), 6).

To see firing pulses I suggest connecting a scope from common to R11 - R14 junction.

Set triggering so it just works on a +1 volt peak input (the scope might have a calibrate/test output of about this level) in the Y triggered (not free running) mode.

The trigger pulses are very short (roughly RC = C2 X R11 in length - 0.1 x 10^-6 x 16 = 1.6 microseconds). You will need a fast timebase speed and the trace may be very faint unless it is a storage oscilloscope - it will only light the screen for 1.6 microseconds in every 20 milliseconds. You may need to look in the dark with brilliance at maximum. Also the scope would need to have a bandwidth of about 10 MHz if the apparent voltage is not to be less than true - the UJT fires at about 6V and a volt is lost across the UJT.

Regarding 7) I do not expect failures if you have a good robust SCR - they operate off with no current or on with very little voltage and switch quickly. Operation with both high current and high voltage together is a lot of watts. Multiply watts by time and you have energy - it is the energy that kills a device by melting the silicon. The only caution is not to economise on the firing pulse - I suggest raising the 12 volt supply to say 15V.

One thing that occurs to me is to connect another diode from D2, D3 junction to common, anode to D2-D3. This avoids high reverse voltage on D2 D3. You need generous voltage ratings because spikes on tripping off a short circuit could be thousands of volts on the generator main windings.

I will post a functional description later - you have the essence of it.

Regards,

67model

On account of "burning midnight oil" I missed out some things I meant to write.

1. On the circuit for the 3 phase, full wave, voltage sensing a) the diodes can be 1N4006/7 - BYW100 came out because it went with the symbol b) the transformers have a thermal fuse in their primary windings which you do not have or need.
2. The transformer secondaries need correct polarity. If you get it wrong, the 3 phase voltages will not make a symmetrical "" with equal voltages on every side [U-V, U-W, W-V all equal on secondary]. Just reverse the connections on one transfo secondary if that happens. If you want "right first time" quality: do a polarity test as done with generator protection CTs and VTs. "Flash" a battery for a moment onto the primary: if say the primary end with the dot is to battery positive, a moving coil voltmeter on the secondary, with positive to dot end will flick forward (positive volts). You make sure the battery will not deliver a damaging current into the primary resistance and set the meter on a high range initially, of course.

67model

I also drew the diodes the wrong way round to get a negative polarity output!!

And the last point in post #30 about an extra diode is wrong. What is needed is to change D3 to a zener diode (BZX79C16 or similar), protecting D2 from high reverse voltage.

67model

Sorry forgot to answer your question regarding the speed indicator.

The turbine/alternator does not have a speed indicator. The only indicator is the Hz meter. I have checked and confirm the alternator rotor speed to be 1500 RPM at 50 Hz with a mechanical rev counter. /sude

If you consider it usefull, I have some tachometer circuits which would work off the "25 volt" winding and drive a moving coil meter.

67model

Hello 67model,

Thank you for all your comments and explanations over the past few days. Have been very busy with my work load and haven't had much time to "develop" the circuit board etc. Will read all you have said and digest it at leisure and come back to you as soon as time permits.

One interesting thing I did. I put a big coil actually one half of a 5kva generator rotor winding in series with the bulb set up I had. When the variac is turned low the bulb lights up ok but when it is advanced - increase of sense voltage - the bulb starts to flicker and actually can get it to pulse with a frequency of about one per second. Having seen that I was wondering if the system would pulse at the regulating voltage. Any thoughts?

Sude

Hi Sude,

We are in the business of manufacturing LT Alternators for Mini Hydro Application. Our Alternators have two systems for controlling the terminal Voltage.. The Terminal Volatge is raised above the Rated Voltage with the help of Over Excited Compunding Transformer and then is lowered with the help of AVR in auto mode of operation. When the AVR fails to operate we have one Potentiometer and couple of Bridge Rectifiers to manually adjust the Voltage required.

If you could upload the photograph of the Terminal Box of this alternator, we can comment and suggest about how to get the Rated Voltage..

Secondly I do not think that manufacturing AVR with little knowledge about electronics is a good Idea.. Do not go for it..

I know one person from Rajkot, Gujarat, India who is in this business of manufacturing AVRs, he can help you out.. His details are Mr Asif Barmal M/s Electrotechnique, Rajkot -Phone removed - CR4 Admin.

thanks

Hello 67model,

My apologies for being silent the past few days. In response to your post #30, yes there was a three phase step down transformer providing the necessary low voltages. Alas I do not have any of the original AVR/components. I think I got rather fed up with trying to get the original to work without any success and threw the whole thing away! BIG mistake. I will not attempt to see the firing pulses as the scope I have is not a storage type. Important thing is the circuit works and that is good enough for me.

I have done a circuit board for the AVR and have included a photo of it alongside the breadboard circuit. Have not included the SCR and 0.1uF cap for clarity. Now need to populate the board and test it. Having done that I hope to take it up to the estate and try it out on site. Will report back once that is done.

With reference to my post # 34. What effect would the pulsing have on the out put voltage of the generator? If it were to happen what should I do to over come the issue. I will appreciate a feedback.

Regarding the speed indicator, I will appreciate a simple alarm circuit if you have which will detect over speed to keep my operators on their toes! Something that will get activated at say, 1650 RPM of the generator shaft.

Thank you and best regards/Sude

Hello 67model,

Last week I was up at the estate and tried out the AVR. Regrettably it did not function as hoped. I was able to adjust the out put voltage to 230 volts 50Hz, however as the water was opened further the voltage would rise and vice versa.

What I noted originally when doing the bench test was the voltage variation across the load (auto lamp) was about +/- 1.5 volts for an input between 200 and 230 volts AC. As mentioned in one of my earlier posts the sense voltage is derived via a 230/17 volt step down transformer.

Perhaps I have done some thin wrong which I can not see. Any ideas and suggestions to overcome/rectify the problem will be greatly appreciated.

Thank you and best regards/Sude

Sude,

What are you using to monitor and control the frequency? As you know, for a fixed excitation level, as the speed of an alternator increases so does the voltage (up to a point). Also, does your AVR have any frequency sensitive components/circuits in the feedback loop, and are there any provisions for adjusting its gain and time constant? Are your tests being performed with a load on the terminals?

Alan

Hi RAMConsult,

Many thanks for your post. The frequency is monitored by the Hertz meter and nothing else. With the present setup of fixed excitation the speed of the turbine is manually adjusted by opening or closing the water valve. The load is mostly the factory machinery. Any excess power is shed into the automatic load bank which is a set of water heaters. The heaters are arranged in four banks per phase of 1,2,3 and 4 2Kw heaters per bank.

67Model is the best person to answer your question frequency sensitive components/circuits in the AVR. My recollections of the original AVR did not have any however there was another trim pot which perhaps was a gain control. All tests were done with the load bank in operation

Many thanks for your interest

Sude

Hello again Sude,

I was in Konstanz, Deutchland for a week, so did not see your posts.

Re- your post #34. Just to be sure - you did have D2 and D3 fitted. Without them, the charge on C1 could build up over many cycles [while error small] because C1 is not reset to zero volts during every negative half cycle of AC.

Re - your post #36. You have obviously used single phase sensing with one isolating transfo 230/17V. For this, R1A + R1B would need to be about 12 kilohms - see drawing 3phsense.png.

More significantly, 3 phase sensing (as my drawing) delivers 95% peak voltage as a mean value, has a peak-peak fluctuation [ripple] of ~16% peak of sine wave on the mean value. This ripple is 300Hz.

Compare full wave rectified single phase. The mean value is 0.636 x peak of sine wave and the ripple is 100% peak-peak with 100 Hz frequency.

This has a considerable effect on the "smoothing out" of the ripple due to C3. See the graph of rectified waveforms below (bottom is zero volts; top is peak of sine wave), compare single phase [marked blue] with 3 phase [marked red].

A simple comparison is that with C3 of 1 microfarad, single phase will give 100%/16% = 6 times the ripple voltage and C3 will only be 100Hz/300Hz = 1/3 as effective as a filter. For the same ripple, with single phase, C3 needs to be increased 6 times because of extra voltage ripple and 3 times because of reduced frequency - say 18 times.

A suitable standard value for C3 would be 22 microfarads, using single phase full-wave sensing of generator output.

You will recall on "bench test", that with DC (from a variable supply) as the "feedback" of the generator voltage, SCR conduction went from minimum to maximum for a few tenths of a volt, measured at TR2 base. Ripple voltage greater than this small control band will greatly decrease the AVR sensitivity to voltage changes.

Need for speed alarm noted for later.

Hope this helps,

67model

Hello 67model,

Glad to hear from you once again. Your excellent explanations of the theory behind the design the AVR circuits makes me believe that you are Professor in Electrical Engineering. Konstanz being a University city I believe you must be a visiting Professor of the University of Konstanz.

I did not realize three phase against single phase sensing would have such an impact on the operation of the AVR. Your explanation of the effect of this is excellent. I did use a small cap of (2 mfd bi-polar) on the -12v sensing line which was half wave rectified. Do you think it would be better for me to go for full wave rectification of the single phase sensing voltage? I think this is an obvious question.

Thank you and best regards/Sude

Hello Sude,

I am glad the explanation was understood.

You are correct that full-wave (bridge) rectification would be the way to go. The reduction in component numbers compared to the performance disadvantage is not good for half-wave rectification.

I will get to the circuit operation description later.

67model

Hello 67model,

Forgot to answer one of your queries. Yes I did have D2 and D3 fitted. In your post # 33 you have recommended to fit a 16v zener diode in place of D3 which I did, but the AVR did not work with it. Perhaps I fitted it the wrong way round. Will do again and get back to you.

I have one small question. That is the transformer I intend using which came from my "larder" has two windings 1) 17v and 2) 7v. Which voltage should I use for the sensing input?

Hope to do the "corrections" and redo a new circuit board and take the AVR to try out early April when I am up on the estate once again.

Many thanks for your help. Much appreciatd

Best Regards/sude

Hello 67model,

I tried out the AVR with a full wave bridge rectifier and smoothing cap on the sense voltage. I believe the system worked and held the voltage steady. Unfortunately I could not test it fully as I detected a fault in the ELC. Many heaters in the load bank were burnt as such when the water valve was opened after a certain point the turbine speed would rise indicating the power generated was not taken up by the load bank. However I noted that the voltage would held. I would be able to test the AVR fully once I attend to the replacements of the heaters of the load bank.

Will keep you posted.

Thank you and best regards/Sude

Hello 67modelI'm interested in the AVR circuit 100 kv, and there are things in the circuit is not well understood, you can upload a nicer image, else as my language is Spanish use google traslator to understand, forgive me if some things do not well they understood, but as you understand me google translate.Ariel