In our research into different types of ballast to control current demand on various projects, we found that it is often useful to be able to vary the current independently of the voltage if a single power supply is to be used for multiple projects with

SATURABLE REACTORS

In our research into different types of ballast to control current demand on various projects, we found that it is often useful to be able to vary the current independently of the voltage if a single power supply is to be used for multiple projects with different V and I requirements. In the process we ran across the concept of the Saturable Reactor. The idea is simple. Introduction of a small variable DC voltage into a separate winding on an iron frame inductor will bring the core to saturation, opposing the inductance of the power winding. The closer to saturation the core becomes, the lower the inductance of the reactor and the larger the current that is allowed to flow.

We find this concept intriguing because it offers variable control of large currents by way of a low power control circuit. Large multi-tap autotransformers in series with a circuit may be used to perform the function of variable current control but their use is mechanically complicated by the necessity of multiple connections to one or more tap switches along with the fact that the switches cannot be operated under high current loads that exceed their specifications. With the saturable reactor, a simple Variac and bridge rectifier are all that should be needed to provide infinitely adjustable current control even under load.

The simplest version of the concept – a single control winding on the frame of a single winding inductor – is not very effective due the transformer effect. An AC current will be induced in the control winding that will oppose the DC introduced. This not only hampers the saturation of the frame, but can also destroy the rectifier circuit (we found this out first hand). After speaking with several knowledgeable sources on the subject, the design requirements became clear. There must be 3 windings. Either the power winding or the control winding must consist of 2 separate coils placed in series in such a manner that no AC voltage is induced into the DC control.

To test the theory, a pilot reactor was designed. We had on hand several ‘filament transformers’ removed from x-ray power supplies. Many of these are designed so that both the large and small focus filament transformers are on the same ‘figure 8’ frame. Each is ‘shell wound’ with a low current primary of fine wire underneath a higher current secondary of about #12 AWG magnet wire in about a 10:1 reduction ratio of turns. An example is pictured below.

After removing the corner bolts that hold the leaves of the top of the 20 lb. transformer in the next photo below, the top was carefully removed. The transformer winding pairs were then slid off the legs. The secondaries, which were connected in series, were then unwound/slid off of the primaries. The primaries were then reinstalled on the outer legs of the transformer and the combined secondaries were then re-wound onto the center leg. Because the center winding was much stouter, it was chosen as the inductive/power winding, making the 2 windings on the outer legs the control windings. These two control windings were placed in series by connecting together one lead from each winding (using the green and black wires shown not connected in the photo below).

The leaves of the top of the transformer frame were then meticulously re-interleaved with those of the legs and the frame re-bolted. The inductance of the center winding was measured ‘live’ by placing a voltage across it and using an ammeter in series. Inductance was calculated via the formula: L=v/ωi and found to be 278 mH. A voltmeter across the combined control windings showed a very significant voltage in them when the center winding was powered up. The series connection was broken and the lead from one control winding was swapped with the other of its pair and this then used to reconnect the windings in series (green and green wires shown connected in the photo above). Powering up the center winding again resulted in virtually no induced voltage in the control windings.

The center winding was placed in series in a circuit that is known to draw a current of 8-9 amps without ballast. An ammeter in the circuit showed a little over 800 milliamps flowing with the inductor in place. The full-wave rectified output of a Variac was introduced into the series pair of control windings and the voltage gradually increased. Over the range of 0-25 VDC control voltage, the current in the power circuit also gradually increased proportionally to its full value of a little over 8 Amps. Power was removed from both circuits after a few minutes of operation and the system checked for problems. No significant heat was apparent at the rectifier or in any of the windings. This is considered to be a successful test of the theory using 2 control windings and one power winding.

The obvious next steps seemed to be to test the theory using 2 power windings and only a single control winding, to use a larger system with much higher current, and even to test the use of toroidal cores in the design. The 180 lb. 3-phase shunt transformer pictured below was chosen to test the first two of these.

The transformer has 8 leads (taps) for each leg. The middle 6 on each leg were secured out of the way and only the start and end leads for each leg were used. The end windings, A & C, were connected in series in such a way that there was 0 Volts in the center winding, B, when the A-C series was energized. Using the A-C series as the power windings resulted in the control winding drawing a huge amount of DC current – an arrangement that would probably work but seemed impractical. For this reason, the configuration was reversed and B was now the inductor and the A-C series the control. With no DC in the control windings, the inductance of B was so high that a 10 kVA pole pig running almost to dead short (1/16 inch gap) produced no visible spark!

240 VDC at 12.5 Amps was then introduced into the A-C control winding series and the pig sprang to life, producing a Jacob’s climbing arc with plasma about 1 cm in width. The response was weak but it did appear that at 240 VDC we were at least on the magnetization curve for this brute. Since it was originally designed for and used in a 480 to 560 Volt system, it seemed that it would probably require considerably higher voltage to saturate the huge frame. We have a large autotransformer with output voltage to 700VAC with 240 Volts in that will be used to follow up on this. Up to 600 VDC were introduced into the control winding via a 160 Amp bridge rectifier. The current demand in the control winding again was huge and required ballast. A small hobby welder was used and quickly overheated due to the current. No significant change in inductance was produced.

To pursue the question of how to apply this concept to toroidal systems, we disassembled the 125 kV, 300 m.a. Bennett T-835 x-ray high voltage generator pictured above. The HV transformers are mounted on opposite sides of a 70lb. toroidal core. This assembly was removed and the secondaries were cut away to leave only the primaries intact.

The direction of the magnetic flux in a toroidal core is either clockwise or counterclockwise, depending on the winding direction of the coil producing it and which half-cycle of the AC is present in it. There are two ways to connect the two primaries above in series – connecting one set of leads from the same end of each coil or connecting the leads from opposite ends. One method will cause the magnetic fluxes from each of the coils to oppose each other and the other will result in the fluxes aiding each other. Inductance of one hookup measured only 16 mH while the opposite hookup measured 2.7 Henries! Next, a third winding is added on one of the free sides such that no voltage is developed in it when the seriesed primaries are energized.

200 turns of #22 AWG magnet wire were wound around the core and used for the control winding. Again, a very large current demand was observed in the control winding which began to overheat almost immediately. Neither the use of a second oppositely wound control nor switching to 50 turns of #8 AWG magnet wire resulted in controllable saturation.

The failures were unexpected, signaling the need for further research. To this end, a second authority was contacted. The retired engineer who built saturable reactors for use in military equipment explained that SR’s operate on the principle of ‘Amp-turns.’ This means for example, that if the power winding has X number of turns and is passing 100 Amps, a control winding with the same number of turns (such as the experiment above) would also take 100 Amps to control. Placing 100 times the number of turns in the control winding as are in the power winding, would reduce the current required to control the reactor to 1 Amp. The control winding should also be wound of wire with about 0.01 of the cross-sectional area as the wire used in the power winding.

Although this did explain why the first experiment above worked so well and the other two failed, significant design drawbacks were immediately apparent. The winding of 15,000 turns to control an inductor of 150 turns was daunting to say the least. Also, the 1:100 turns ratio could easily result in a dangerous high voltage transformer if a mistake in design resulted in the primary fluxes not being cancelled out. Consideration of this dilemma brought to mind an interesting possibility. X-ray transformers typically produce voltage step up on the order of 1:500. There are often dual 1:250 transformers used that are out of phase with each other to produce the 1:500 step up. The secondaries thereof are often dual windings each of which are in around 1:100 or so winding ratio with respect to their primaries. The ratio of cross sectional area of the wire used in the primaries versus the secondaries is certainly at least 100X. It seems we may have been a bit hasty in having secondaries cut off of the Bennett transformer pictured above. Fortunately, we have a supply of these transformers with which to experiment. Using an identical Bennett T-835 unit, we began a more detailed experimental analysis of the transformation and inductance properties of the windings as follows:

1. Secondaries connected in series for the control winding and the outputs connected to HV side of NST the low side of which is connected to a DMM

2. Primaries connected in series for the power winding

3. 10-40 volts applied to primary while monitoring the DMM. A reading of 0 VAC reflects successful wiring scheme. 25-100VAC indicates HV output and requiring one lead in the series connection of the primaries to be reversed.

4. Series ammeter testing is done on primary with secondaries open to determine baseline inductance.

5. DC output of a Variac is then cautiously applied across the control winding and the voltage range required to increase the current reading on the ammeter is noted.

6. Place the secondaries in series opposing and test again

7. Place the secondaries in parallel and test again

8. Place the secondaries in anti-parallel and test again

PRIMARIES	SECONDARIES	INPUT	OUTPUT	AT 250 VAC
S-OPPOSE	S-AIDING	25	184	1840
S-OPPOSE	S-OPPOSE	25	2163	21630
S-OPPOSE	PARALLEL	25	1375	13750
S-OPPOSE	ANTI-PARALLEL	25	388	3880
S-AIDING	S-AIDING	25	2375	23750
S-AIDING	S-OPPOSE	25	1688	16880
S-AIDING	PARALLEL	25	193	1930
S-AIDING	ANTI-PARALLEL	25	1500	15000

Secondaries have the 4 possible connection configurations shown below. The dots represent connections to the Federal HV receptacles.

SERIES

PARALLEL

For the sake of completeness and to determine where we are on the magnetization curve when using 25 VAC as the test voltage, the other 8 possible configurations of the windings were tested:

PRIMARIES	SECONDARIES	INPUT	OUTPUT	AT 250 VAC
PARALLEL	S-AIDING	25	0	0
PARALLEL	S-OPPOSE	25	13	1625
PARALLEL	PARALLEL	25	19	2375
PARALLEL	ANTI-PARALLEL	25	0.5	62.5
ANTI-PARALLEL	S-AIDING	25	20	2500
ANTI-PARALLEL	S-OPPOSE	25	0	0
ANTI-PARALLEL	PARALLEL	25	0	0
ANTI-PARALLEL	ANTI-PARALLEL	25	20	2500

* Indicates a huge current draw that tripped the 15 Amp breaker even at 25 VAC input even though no output voltage registered.

We have also found that it is possible to use 2 transformers to create an SR. MOT’s seemed ideal with turns ratios around 20:1. We were able to find 2 identical heavy duty units built by Advance and 2 others by YEC:

For each pair, the primaries were wired together in parallel. The secondaries were placed in series by connecting the HV tab of each unit and connecting a wire to the frame of each by means of a bolt run through one of the mounting hotels in the frame. These output wires were connected to the HV side of a 125:1 NST to which a DMM was connected to the LV side. 0-145 VAC was introduced into the parallel MOT primaries while monitoring the DMM for voltage. If no voltage registered, the DMM was moved to the HV side of the NST and the procedure was repeated. A value of 30 Volts or less indicated a successful series connection in the ‘opposing’ sense and confirmed that the transformers chosen were close enough to identical to proceed. If the first test had indicated significant high voltage output, one pair of wires in the parallel primary connection was swapped and the test repeated to confirm that the seriesed secondaries no longer registered significant voltage.

Direct measurement of the inductance of the paralleled primaries was then performed with an ammeter in series with the input supply circuit set at 35 VAC. The ammeter registered about ½ Amp, indicating a baseline inductive reactance of around 60 Ohms. The ends of the seriesed secondary circuit were the wires attached to the frame of each transformer. This series forms the DC control winding. These wires were attached to the rectified output of a small Variac. The introduction of 0-82 VDC into the control caused the reading on the ammeter to increase smoothly over the range to a final value of 16.9 Amps. We did not push this further due to the 20 Amp limitation of the ammeter, but this corresponds to an inductive reactance of slightly over 2 Ohms, making the test a resounding success. With cooling, this pair could reasonably be expected to handle 40 or 50 Amps as ballast and the other pair gave a very similar test result.

The question then became whether the two pairs could be successfully paralleled for higher current handling capability. To this end, shunt wires were run to connect two sets of paralleled primaries. Then, the two sets of seriesed secondaries were connected in parallel with respect to each other. A brief power test was performed just to insure that no voltage was induced into the control. At this point, the inductance/saturation testing was repeated on the combination of all 4 MOTS. The testing was also very successful and the results very similar to those from the tests of the individual pairs with a couple of exceptions, which are as follows. First, the baseline reactance was reduced to about ½ of the value measured on the individual pairs – 30 Ohms instead of 60. This was to be expected pursuant to the law of parallel inductors. Second and more surprising, there was only required a total of 28 VDC in the control to reduce this value to 2 Ohms. It would seem to follow that more pairs could be added with a corresponding increase in current capability and decrease in baseline reactance. The high end reactance drop should not resent a problem since the useful range of inductive reactance for most of our project work is about 2-8 Ohms.

Taking this to its logical conclusion, we built the 8 MOT reactor pictured below.

This concept appears to be amazingly versatile with a large number of possible configurations to address the particular V and I parameters of its intended application. We have verified that the control windings may be placed in series rather than parallel to give a higher and wider control voltage range without ill affect. In fact, with the 8 pack above, we placed the control windings of all 4 pairs of MOT's in one continuous series, resulting in a control range of about 0-100 VDC.

Placing the additional 2 pairs in parallel with the first 2 pairs did, as expected, drop the high end reactance to 15 Ohms. The low end remained 2 Ohms.

We also tested the idea that the primaries should be placed in 'straight' parallel - that is the left input tab of one MOT connected to the left tab of its pair partner and the right to the right of the other. PLEASE NOTE that this results in nearly 4000 Volts in the control winding. It appears that if identical transformers are used, the primaries must be wired in 'cross/inverse/anti' (pick a term) parallel - that is the LEFT input tab of one MOT is connected to the RIGHT input tab of the other MOT in the pair and visa versa in order to have low or no voltage in the control. Of course, this is predicated on the use of the HV tabs to connect the secondaries in series, which as you can see is what we are using.

There also appeared to be no reason that the primaries of each pair may not be connected in series (making sure that they are wired such that no voltage is induced in the secondaries) and then the pairs connected to each other in parallel. This configuration may be more suitable for heavy current work in the 200-300 VAC range.

To this end, we used 2 YEC MOT’s, weighing about 11 lbs. each that are of the same model (though slightly different form factor) for testing. Placing a test voltage across either primary resulted in exactly the same voltage on the corresponding secondary. The primaries were then connected in straight series (aiding) and the secondaries were connected in series opposing. Introducing the output of a Variac into the primary series resulted in no more than 1 Volt induced in the secondary series at 143 VAC input as can be seen below.

For use in 240 VAC circuits, this will now be scaled up to 4 pairs of MOT’s each wired in this manner. The pairs will then be connected in parallel with respect to each other on the primary side (power circuit) and in series with respect to each other on the secondary side (control circuit).

Another interesting project will be to use the transformers inside of a pair pole pigs for a similar experiment:

The LV side is often composed of 2 separate windings that could be placed in series opposing to hopefully give 0 Volts on the HV side. The 60:1 turns ratio would presumably allow control with a relatively low current into the HV windings. The critical point is going to be whether a pig LV winding has enough inductance to be useful.