Resource

Why Rectifier Transformers Need Phase-Shifted Windings

A 6 pulse bridge cannot cancel its own harmonics. Two windings 30 degrees apart put its 6th DC ripple 180 degrees out of phase, since 6 times 30 is 180. Only a winding connection makes that angle. Star and delta give 30 degrees, smaller steps need extended delta, and both change the turns ratio and current sharing. Name the vector group before the price.

If you came here from another page on our site, here is where this one sits. Our page on vector group notation already decodes the clock numbers, explains why the 30 degrees is geometry, and settles the lead versus lag argument. Our page on paralleling different vector groups covers what happens when a tie closes across two machines. Our 30 degree phase shift page is about protection relays and what the CT sees. Our comparison of 6, 12 and 24 pulse rectifiers works the harmonic math, puts each order against IEEE 519-2022 Table 2, and grades the sources behind the THD numbers. Those questions are settled and we are not going to answer them again. This page does a different job. It asks why the phase shift has to live inside the windings at all, what the standard says about it, and what the iron and copper actually pay for the angle. No harmonic limit table, no pulse number ranking, no THD arithmetic. If you are writing a specification, this is the page that tells you what to ask the factory before the winding drawing is released.

The short answer, before the table

Here is the six row shape a search engine likes to lift, with every cell holding a number you can recompute. Run it across your own submittal before the RFQ goes out.

If your question isWhat actually answers itNumber to checkWhat goes in the spec
Why two windings at allA single bridge cannot cancel its own 6th ripple6 times 30 equals 180 degreesThe pulse count and the number of isolated secondaries
Where the angle comes fromThe connection of the windings, not an external device30 degrees from star versus delta, 360 divided by pulse numberThe full vector group of every winding
What makes it officialIEC 61378-1:2011 Edition 2 added winding connections with phase displacementPublished July 2011, replaces the 1997 editionStandard number, edition, corrigendum status
What 30 degrees costsEqual line voltages need a 1 in 1.732 turns ratio0.577350, and an intermediate yoke of 51.7638 percentTwo primaries or two secondaries, with the core drawing
What a smaller angle costsExtended delta by the sine ruleAt 15 degrees, 0.366025 extension and 1.224745 output ratioTap positions and the phasor diagram, not just the angle
What proves it worksPhase displacement measurement and a current sharing test1 degree is 6.6667 percent of a 15 degree stepTest report, angle tolerance, impedance match, IPT data

Read down the second column. Not one row says cleaner, premium, low distortion or heavy duty. Each row gives you a formula, a date, a turns ratio or a test. That is the point. Most of the pages we looked at print a three column comparison with adjectives in the cells, and adjectives are exactly what disappear when a reviewer asks for the winding drawing.

Six rows instead of three definitions

The comparison table is not the problem. Buyers scan a table in ten seconds and the ranking pages are right to put one near the top. The problem is what sits inside the cells. One vendor page runs a four row table of architecture, harmonic content, cost and recommended application, then fills the last column with electrolytic aluminum, chlor alkali and green hydrogen. Those are not limits. They are sales copy, and a reviewer cannot check any of them against a test report.

Keep the shape and change the contents. Every row above is one question answered by one number you can recompute on a single line. If a quotation cannot fill the fourth column, the winding discussion is not finished. Ask again, and ask which edition of the standard the design was checked against.

Give the ranking pages their due

Several of them are genuinely useful, and we use their work.

A drive consultant writes the cleanest causal chain we found. A two secondary transformer has 30 degrees between its windings. The 6th harmonic on the DC side is shifted by 6 times 30 degrees, which is 180 degrees, so the two 6th harmonic components are in opposite phase and cancel. The DC side 6th disappears and the corresponding 5th and 7th on the AC side go with it. The same page lists all five winding connections used in this trade, says the smallest shift available from star and delta is 30 degrees, and says anything smaller needs zigzag, extended delta or polygon delta. It also explains the clock convention down to the minute and gives d11.33 for a minus 20 degree shift. That is a consultant page doing engineering, and we give it full credit.

A transformer technology tutorial goes somewhere nobody else goes. It points out that a delta secondary and a wye secondary need a turns ratio of 1 divided by the square root of three to give equal line voltages, and that a low voltage, high current secondary has so few turns that this ratio cannot be built exactly. Its solution is two primary windings, one star and one delta, with two identical secondaries. It then names the price: an intermediate yoke sized for the difference between the two limb fluxes, about 52 percent of the main limb area. It also describes why an interphase transformer saturates when the two bridge currents drift apart. That single page is the reason this article has a section on iron instead of only one on harmonics.

GE Vernova publishes a rectifier transformer brochure that states the 30 degree displacement for 12 pulse units comes from star and delta connections on the secondary, and that higher pulse counts are reached by shifting on the primary with zigzag or extended delta, or on the regulating autotransformer. It gives the business reason in one line: that arrangement allows a simple and identical design for every rectifier transformer at the same plant. It also describes the single tank and two tank options and says an interphase transformer lets each diode or thyristor conduct for 120 degrees. Brand literature with a named design consequence is rare, and we treat it as a primary source.

A standards commentary on IEEE C57.18.10-2021 and several standards database entries give us the edition history, the change lists and the scope wording. A public patent text gives one simulation with and without primary side shift. We grade every one of these in the ledger near the end.

One angle, one place it can come from

A single 6 pulse bridge draws a 120 degree block of current. Its DC output has a ripple at six times line frequency, 360 Hz on a 60 Hz supply and 300 Hz on a 50 Hz supply. That ripple is not a defect that a better core can fix. It is the waveform the bridge makes by switching six devices per cycle. One bridge, one waveform, one set of 5th, 7th, 11th and 13th harmonics on its AC line.

Now feed a second, identical bridge from a supply displaced by 30 electrical degrees. On the DC side the two 6th ripple components are 6 times 30 degrees apart, which is exactly 180 degrees, so they oppose. The combined DC bus sees 12 pulses per cycle instead of six. At 60 Hz the ripple goes from 360 Hz to 720 Hz, and the period from 2.7778 milliseconds to 1.3889 milliseconds. The AC side consequence is that the corresponding 5th and 7th contributions from the two bridges oppose at the common primary. We are using the DC side argument here because it is one multiplication and it is checkable. The AC side sequence signs are a longer discussion and belong to our pulse comparison page.

The angle between adjacent bridge groups is 360 degrees divided by the pulse number, which is the same as 60 degrees divided by the number of 6 pulse groups.

Pulse countBridge groupsAngle between adjacent groupsDC ripple at 60 HzRipple period
6160 degrees360 Hz2.7778 ms
12230 degrees720 Hz1.3889 ms
18320 degrees1080 Hz0.9259 ms
24415 degrees1440 Hz0.6944 ms
36610 degrees2160 Hz0.4630 ms
4887.5 degrees2880 Hz0.3472 ms

Why a filter cannot make the angle

The 30 degrees is a relationship between two supplies. It is not a property of one waveform. A line reactor changes the shape of the current a bridge draws by adding impedance upstream. A passive tuned filter takes harmonic current out after it has been made. An active front end replaces the bridge with switching devices that shape a near sinusoidal current. All three act on one supply, or on what one supply injects. None of them creates a second supply at a defined relative angle.

That is the honest answer to the keyword. A rectifier transformer needs phase shifted windings because the cancellation is a comparison between two bridges, and only the windings can set the terms of that comparison. You can buy the angle as a separate phase shifting transformer ahead of two ordinary units, and plants do that, but you have still bought windings. What you cannot do is specify an angle without specifying a winding connection.

There is a second reason, and it is about what the standard expects. A phase displacement is a rated characteristic of the transformer, measured in a voltage ratio and phase displacement test and printed on the rating plate. A filter performance figure is a system result that depends on source impedance and load. Putting the angle on the transformer puts it on a test report. Putting it on a filter puts it on a simulation.

What none of them prints

Eighteen of the pages we read say a 12 pulse front end needs 30 degrees. Fourteen say star and delta provide it. Nine mention an interphase reactor. Six explain the clock notation. Four list the special winding connections. Two give a turns ratio formula for an extended delta winding. One gives the 1 in 1.732 problem. One gives the 52 percent yoke. One is a brand brochure explaining why primary side shift makes a whole plant easier to build.

None of them does the following five things together: name the edition of the standard that added phase displacement windings, recompute the turns ratios they publish, show that the 52 percent yoke is a vector difference, say that the angle is a rated characteristic rather than a system result, and grade which of their numbers came from a standard, a brand sample, or an unsourced content page. The published extended delta table in this search has four rows, and all four fail a sine rule check. We show that arithmetic later.

We are not saying the keyword is a bad search. It is a good one. It is exactly what a buyer types after a supplier quotes a 12 pulse unit with two secondaries and the price looks high for what seems like one extra coil. The problem is that a page which answers only with harmonic cancellation has explained the benefit and skipped the cost.

The standard finally named the winding

Here is the piece of evidence that should sit at the front of any bid review. IEC 61378-1:2011 is Edition 2 of the converter transformer standard for industrial applications. We found its metadata on five independent standards databases, and all five reproduce the same list of significant technical changes from the previous edition. The first item on that list is the addition of winding connections with phase displacement.

The other items on the same list are not decoration. They are the rest of this article.

#Technical change listed for Edition 2What it means for a buyer
1Addition of winding connections with phase displacementThe angle became a design item the standard names
2Addition of transformers with more than one active part in the same tankOne tank can hold more than one electromagnetic unit
3Change of reference power definitionThe French summary says it is now based on the fundamental component of current
4Addition of a regulating transformer feeding a converter transformerVoltage regulation can live in a separate unit
5Addition of considerations about current sharing and hot spot temperature in high current windings for various winding arrangementsTwo bridges sharing a bus is a thermal question, not only a harmonic one
6Addition of transductors for DC voltage regulation with diode rectifiersFine regulation has its own hardware

Edition 2, not a new brochure

The standards databases agree on the shape of the document. It is Edition 2, it replaces IEC 61378-1:1997 Edition 1, it comes from technical committee TC 14 and its maintenance team MT 7, it sits under ICS 29.180, and the English version is 90 pages while the English and French version is 184 pages. The corrigendum of January 2012, COR1:2012, is folded into the current copy. Publication day appears as both July 25 and July 26 of 2011 across the databases, so we do not print one as verified. If a date goes on a purchase order, get it from the publisher’s page.

The scope covers power transformers and reactors integrated into semiconductor converter plants of any power rating. Typical applications named are thyristor rectifiers for electrolysis, diode rectifiers for electrolysis, thyristor rectifiers for large drives and scrap melting furnaces, and diode rectifiers feeding inverters for variable speed drives. It also covers regulating units such as step down regulating transformers and autotransformers. It does not cover ordinary industrial or public AC distribution transformers. Those live under IEC 60076.

That scope sentence matters. A distribution transformer standard does not ask what phase angle two secondaries hold relative to each other, because a distribution transformer normally has one. The converter standard asks it because the whole installation is built around the answer.

Thirty degrees or less than thirty

Here is a genuine disagreement we are not going to smooth over. The English abstracts on four databases write the first change as an addition of winding connections with phase displacement of 30 degrees. The French abstract on a fifth database writes it as an addition of winding connections, zigzag and extended triangle among them, with a displacement of less than 30 degrees. Same edition, same change item, two different angle descriptions.

We did not buy the standard, so we will not say which wording is authoritative. We will say what both wordings support. The 2011 edition formally brought phase displacement windings into the standard, and it distinguishes the plain star and delta case from the special connections used when the required angle is smaller. That distinction is the entire subject of the next section of this article. If your specification says 12 pulse, the first case applies. If it says 18, 24 or 48 pulse, the second one does, and a supplier who quotes a plain Y and D pair for a 15 degree step has quoted the wrong winding.

A standards reseller page also lists the tests and design topics the 2011 edition covers: voltage ratio and phase displacement measurements, commutating reactance, inductive voltage drop, determination of load loss under non sinusoidal converter currents, eddy current effects, tank losses, hot spot assessment for high current windings, and current sharing measurements. We report that list as a reseller summary, not as clause text. It is still enough to make one purchasing point: the angle and the current sharing are both testable items on the same unit.

What the 2021 IEEE document adds

A secondary reading of IEEE C57.18.10-2021 describes a 93 page standard for semiconductor power rectifier transformers at 100 kW and above, single phase or three phase, excluding electrostatic precipitators, HVDC converters and mixed loads with more than 10 percent non converter fundamental power. It defines commutating impedance, requires both rms kVA and fundamental kVA on the rating plate, adds electrostatic shield requirements and interphase transformer requirements, and replaces the older empirical 60 over 40 loss split with harmonic loss factors FHL-WE and FHL-OSL derived from the spectrum. We use it as a secondary account and label it that way. It is a different document from IEC 61378-1, and a submittal that cites one as if it satisfied the other has merged two standards.

Three questions the change list answers

Read those six items as three questions a bid reviewer should ask. First, is the phase displacement a rated characteristic of this transformer with a measured value on a report, or a system result somebody simulated? The first item on the list makes it the former. Second, are the two bridges going to share current without one winding running hotter than the other? The fifth item puts current sharing and hot spot temperature in the same standard as the winding connections, which means they are not two separate submittals. Third, where does the regulating equipment sit? Items 2, 4 and 6 all describe hardware that can be inside one tank, in a second tank, or in a separate autotransformer. A single tank with more than one active part is now a design the standard names, not a special arrangement a vendor has to justify.

None of those three questions is answered by a pulse number. All three are answered by a winding drawing and a test report. That is the practical reason the 2011 change list belongs in a specification meeting rather than in a footnote.

What 30 degrees costs in copper and iron

Getting 30 degrees out of star and delta looks free on a single line diagram. On the winding drawing it is not free at all.

A delta secondary has a line voltage equal to its phase voltage. A wye secondary has a line voltage equal to the square root of three times its phase voltage, which is 1.732051 times. To feed two identical 6 pulse bridges from secondaries with the same line voltage, the wye phase turns must be the delta phase turns divided by the square root of three, which is 0.577350. That is the 1 in 1.732 ratio.

The turns ratio that will not fit

A low voltage, high current rectifier secondary may have only a handful of turns, sometimes fewer than ten. Turns are integers. If the delta winding has ten turns, the wye winding needs 5.77350 turns. If the delta has twenty, the wye needs 11.54701. There is no integer count that gives exactly 0.577350, and a rounding error of one turn on a twenty turn winding is a 5 percent voltage error between the two bridges. That is far outside the matching tolerance anyone would want for current sharing.

The tutorial we cited gives the practical fix: put two primary windings on the transformer, one star and one delta, and make the two secondaries identical, both star or both delta. The 30 degree displacement now comes from the primaries, where the voltage is higher and the turn count is large enough to absorb the ratio, and the secondaries can be wound with exactly the same number of turns. That is why a 12 pulse rectifier transformer often has four winding systems rather than the three a single line diagram suggests.

Two primaries and an extra yoke

Two primaries displaced by 30 degrees produce two limb fluxes that are not in phase. Where the two magnetic circuits meet, the core has to carry the difference between them. The tutorial says an intermediate yoke is needed and gives its area as about 52 percent of the main limb area.

That number is a vector difference, and we can recompute it. Take two flux phasors of equal magnitude separated by 30 degrees. The magnitude of their difference is 2 times the sine of half the separation, so 2 times the sine of 15 degrees, which is 0.517638, or 51.7638 percent. The tutorial’s rounded 52 percent is 0.2362 percentage points above our arithmetic. Two independent routes, one answer, and the extra iron is a real cost.

What the yoke number is and is not

It is a geometry result for two equal phasors 30 degrees apart. It is not a general rule for all pulse counts, not a loss figure, not a temperature rise, and not a clause value from the standard. We computed it, and the tutorial printed a rounded version without showing the derivation. If a supplier’s core drawing shows a yoke sized near half the main limb, this is why. If it shows no intermediate yoke on a design with two displaced primaries, ask where the flux difference goes.

The same reasoning explains why higher pulse counts get expensive fast. At 24 pulse the adjacent step is 15 degrees, and at 48 pulse it is 7.5 degrees. A smaller angle does not mean a smaller magnetic problem. It means more winding systems, more terminals, more tap positions, tighter manufacturing tolerance and more core detail to keep the phases where the drawing says they are. The 30 degree case is the cheap one, and it is the only one star and delta can give you.

Smaller angles need a different winding

Star and delta give you 30 degrees and nothing finer, because a clock position is 30 degrees and a change of connection moves you by one position. For 18 pulse you need 20 degrees, for 24 pulse 15 degrees, for 48 pulse 7.5 degrees. None of those is a whole number of clock hours. You have to build the angle into the winding itself.

The five connections named in the trade are star, delta, zigzag, extended delta and polygon delta. The first two give 30 degrees by geometry. The last three add a tap or an extra section to each phase so the output line voltage sits at a chosen angle to the input. One consultant page puts the practical limit plainly: an arbitrarily small shift is possible in theory, but manufacturing tolerance, cost and complexity set the limit in the factory. Another page says for higher pulse operation extended delta tends to beat zigzag because it produces lower eddy losses and lower short circuit forces. We report that as a technical tutorial’s judgement, not as a standard requirement.

The sine rule and a second route to the same number

For an extended delta winding, the main section forms one leg of the delta and an extension section, wound on the adjacent phase, is added in series to reach the new terminal. Those two phasors meet at 60 degrees inside the phasor triangle. Apply the sine rule and the design relations for a wanted displacement of delta degrees are: the extension to main turns ratio is the sine of delta divided by the sine of 60 minus delta, and the output to main voltage ratio is the sine of 60 divided by the sine of 60 minus delta.

We checked those two relations a second way instead of taking them on faith. Put the original line voltage at one on the real axis, and add an extension of size e along the direction of the adjacent delta side. The new voltage is one plus half e on the real axis and the square root of three over two times e on the imaginary axis. Solve for e from the tangent of the wanted angle. For every angle in the table below, the two routes agree to six decimals and the resulting phasor angle comes back exactly at the wanted value. That is the kind of agreement worth putting in a submittal.

Wanted displacementExtension to main turns ratioOutput to main voltage ratioVector route for extensionAngle returned by the vector route
3.75 degrees0.0786601.0415600.0786603.7500 degrees
7.5 degrees0.1645251.0916010.1645257.5000 degrees
10 degrees0.2266821.1305160.22668210.0000 degrees
15 degrees0.3660251.2247450.36602515.0000 degrees
20 degrees0.5320891.3472960.53208920.0000 degrees
22.5 degrees0.6286261.4226020.62862622.5000 degrees
26.25 degrees0.7960991.5588050.79609926.2500 degrees

Notice the 15 degree row. The extension is 0.366025 of the main section and the resulting line voltage is 1.224745 times the main section voltage. A supplier cannot simply wind the extension and leave the rest alone, or the two secondaries will not have equal line voltages. The main section turns have to be reduced by the matching factor to hold the rated secondary voltage. That is a winding drawing question, not a footnote.

Four rows that do not survive a check

One content page in this search publishes a four row table of displacement against extension ratio and output ratio. It is the only table of its kind we found, which is exactly why it needs checking. Here are its numbers beside our two independent calculations.

DisplacementPublished extension ratioCorrect extension ratioDifferencePublished output ratioCorrect output ratioDifference
7.5 degrees0.1490.1645250.0155251.0771.0916010.014601
10 degrees0.2040.2266820.0226821.1041.1305160.026516
15 degrees0.3180.3660250.0480251.1551.2247450.069745
20 degrees0.4400.5320890.0920891.2221.3472960.125296

All eight published values are low, and the gap grows with the angle: from 0.015525 at 7.5 degrees to 0.092089 at 20 degrees on the extension ratio. If a winding were built to the published 20 degree row, the actual displacement would not be 20 degrees.

There is a nice piece of internal evidence here. The same site publishes a separate design note for a 7.5 degree pseudo 24 pulse transformer. That note uses the vector method, works through tan of 7.5 degrees, and arrives at an extension fraction of 0.1646, or 16.46 percent. That is our 0.164525 to four decimals. So the site’s own worked derivation agrees with the sine rule and disagrees with its own summary table. We are not accusing anyone of anything. We are pointing out that a buyer who takes the table on faith will get a different winding from a buyer who takes the derivation.

The output voltage grows too

The second calculated column deserves its own paragraph. At 20 degrees the output voltage is 1.347296 times the main section, a 34.7296 percent rise before the main section turns are corrected. At 26.25 degrees, the largest angle in the 48 pulse set, it is 1.558805, a 55.8805 percent rise. The extension section does not just rotate the phasor. It lengthens it. That is why the correction to the main winding is not optional, and why a nameplate turns ratio on an extended delta design cannot be checked against a plain star or delta ratio.

A Chinese technical page gives a worked 3.75 degree example and states that the ratio of delta voltage to extension voltage is 11.713, then takes 64 turns for the main section, 59 for the delta side and 5 for the extension. Our extension ratio for 3.75 degrees is 0.078660, whose reciprocal is 12.7130, not 11.713. We could not reconcile the difference and we are not using that page’s turn counts. A one unit difference in a reciprocal at this scale is a whole turn on a small winding, which is precisely the problem the 1 in 1.732 discussion described.

Three more angles worth checking

The extended delta relations hold until the wanted angle reaches 60 degrees, because the interior angles of the phasor triangle sum to 180 degrees and the construction cannot go past that. Every practical angle in this trade sits well below it. The 24 pulse designs use 7.5, 15, 22.5 or 26.25 degrees depending on the phasor set chosen. The 48 pulse set runs at minus 26.25, minus 18.75, minus 11.25, minus 3.75, plus 3.75, plus 11.25, plus 18.75 and plus 26.25 degrees, a 7.5 degree step. At the smallest of those, the extension section is 0.078660 of the main, about one turn in thirteen, and a one turn manufacturing slip is a large fraction of the intended angle.

That is the real reason cost climbs with pulse count. It is not only more windings. It is that each winding has a smaller extension section to build accurately and a larger voltage correction to absorb.

Where the 60 degree ceiling comes from

The sine rule relations stop working at 60 degrees, and the reason is in the denominator rather than in any factory limit. Both ratios divide by the sine of 60 minus the wanted angle. As that angle approaches 60, the denominator approaches zero and both the extension ratio and the output voltage ratio grow without bound. At exactly 60 degrees the construction degenerates and there is no finite winding that satisfies it.

Every practical angle in this trade stays well below that ceiling: 7.5, 10, 15, 20, 22.5 and 26.25 degrees in our table, with 26.25 degrees already needing an extension of 0.796099 and a voltage correction to 1.558805. A supplier quoting a single extended delta winding for a 60 degree shift has either used a different geometry or made an arithmetic error. Ask for the phasor diagram. The 30 degree case never needs this construction at all, because star and delta give it directly.

The clock number on the drawing

A vector group is a compact way to write the connection of each winding and the phase displacement between them. One clock hour is 30 degrees, from 360 divided by 12, and the hour is further divided into 60 minutes, so one clock minute is 0.5 degrees. A consultant page gives minus 20 degrees written as d11.33 or d minus 20 degrees. Recompute it: 11.333333 times 30 is 340 degrees, which is minus 20 degrees on a 360 degree circle. Written as the rounded 11.33, it gives minus 20.1 degrees. The notation closes.

Four Siemens groups

A standards commentary reproduces four three winding vector groups attributed to Siemens converter documentation, all giving the required 30 degrees between secondaries.

Vector groupSecondary 1Secondary 2Displacement between secondariesShort circuit voltage
Dd0Dy11delta, 0 degreeswye, 330 degrees lag or 30 degrees lead30 degreesuk 4 to 6 percent
Dd6Dy5delta, 180 degreeswye, 150 degrees lag30 degreesuk 4 to 6 percent
Yy0Yd11wye, 0 degreesdelta, 330 degrees lag30 degreesuk 4 to 6 percent
Yy6Yd5wye, 180 degreesdelta, 150 degrees lag30 degreesuk 4 to 6 percent

The same source adds a sequencing rule that is easy to miss: the three phase current system of the slave bridge must lag the master bridge by 30 degrees. It also says that on high power installations two separate two winding transformers can replace one three winding unit, as long as their vector groups still produce the 30 degrees. And it gives a clean reason why pure star or pure delta pairs cannot do the job: between windings of the same connection type the clock only allows 0 or 180 degrees, so Dd0 and Dd6, or Yy0 and Yy6. Anything else between same type windings needs a zigzag or a phase shifting winding, and that breaks the equal voltage and equal reactance symmetry the two bridges need to share current.

One more practical point from the same page: a delta primary gives triplen harmonic currents a circulating path inside the delta and handles unbalanced secondary loads better. The primary connection does not change rectifier operation as long as the two secondaries are 30 degrees apart with equal voltage. It is chosen for the supply network and its harmonic behaviour.

Two impedance ranges that do not agree

The Siemens transcript gives uk of 4 to 6 percent for all four groups. A different content page in this search gives uk of 6 to 8 percent as typical for the same 12 pulse application. A third, a Chinese engineering blog, says it asks the factory for an impedance deviation report between the two secondary windings and holds it to plus or minus 5 percent.

Those three statements are about different things. The first two are absolute short circuit voltage ranges for the windings. The third is a matching tolerance between two windings. We report all three, we do not average them, and we do not pick a winner. If your specification quotes an absolute uk without saying which source it came from, a reviewer has no way to tell whether it is a converter duty figure or a general VFD transformer figure. Put the absolute value and the allowed deviation between the two secondaries on two separate lines.

One formula we could not verify

A drive configuration page gives the displacement as 30 degrees times the absolute difference of the two clock numbers, taken modulo 12, then says the 30 degree condition is satisfied by combinations differing by 1, 5, 7 or 11 clock positions. Recompute those four: 1 gives 30 degrees, 5 gives 150, 7 gives 210, and 11 gives 330. The first is the 30 degrees we want. We could not find a derivation showing why 150, 210 or 330 degrees would produce the same 6th ripple opposition as 30 degrees, and the page does not give one. We are printing the four values and marking the equivalence claim as unverified rather than repeating it as a rule. A difference of 6 clock positions is 180 degrees, and that one is explicitly described elsewhere as an opposing pair that causes circulating current, not a valid 12 pulse arrangement.

What the same type rule costs you

The rule that same type windings allow only 0 or 180 degrees has a consequence that reaches back into the iron. If 12 pulse needs 30 degrees, and 30 degrees only comes from pairing a delta secondary with a wye secondary, then the two secondaries cannot be identical. One has a line voltage equal to its phase voltage and the other has 1.732051 times its phase voltage. That is the 1 in 1.732 turns problem from the first part of this article, and it is not a coincidence. It is the same geometric fact seen from the winding schedule instead of the clock face.

This is also why the two primaries solution is attractive. Move the displacement to the primary, where there are enough turns to build the ratio, and both secondaries can be wound identically with the same number of turns, the same impedance and the same current density. Equal secondaries are exactly what two bridges sharing one DC bus need. The clock rule, the turns ratio and the current sharing are one problem with three names.

Two 24 pulse angle sets that are the same set

Two pages in this search give different looking phase sets for a 24 pulse transformer. One gives 0, 15, 30 and 45 degrees. Another gives minus 22.5, minus 7.5, plus 7.5 and plus 22.5 degrees. Subtract 22.5 degrees from the first set and you get minus 22.5, minus 7.5, plus 7.5 and plus 22.5. They are the same four angles with a different zero reference, and both have a 15 degree step between adjacent groups.

That is worth checking before sending an RFQ that names one set as if the other were a mistake. What matters on the drawing is the step between adjacent groups and the symmetry about zero, not where the manufacturer chooses to put the reference arrow. A 15 degree step from 360 divided by 24, and four groups from 24 divided by 6. Both sets satisfy it.

Why the shift sometimes moves to the primary

GE Vernova’s brochure describes a design choice that no other page in this search explains. For 12 pulse units, the 30 degrees comes from star and delta connections on the secondaries. For higher pulse counts, the phase shifting between rectiformers is done either on the primary of the rectifier transformer with zigzag or extended delta connections, or on the regulating autotransformer. The stated benefit is a simple and identical design for every rectifier transformer at the same plant.

Read that as a procurement decision, not just an electrical one. If the phase angles are created upstream, every rectifier transformer downstream can be wound the same way, with the same drawings, the same spares, the same test procedure and the same impedance. A plant with eight rectiformers gets eight interchangeable units instead of eight different winding schedules. That is worth more than a small difference in copper weight, and it only appears in a brand engineering brochure.

The same brochure describes two physical arrangements. In a single tank solution all active parts are grouped in one tank, which reduces footprint and installation work. In a two tank solution, the regulating autotransformer with its on load or off circuit tapchanger sits in one tank and the rectifier transformer assembly in the other. It also describes an interphase reactor placed between the two neutral points of a double star low voltage winding, with its DC midpoint connected to the anode of the load, so that each diode or thyristor conducts for 120 degrees.

Where the regulation lives

One more reason the shift and the winding drawing are inseparable: a rectifier transformer for electrolysis often needs a very wide secondary voltage range. The brochure says aluminium pot lines may need up to 100 percent variation to start from cold, split between off circuit and on load tapchangers or a multi coarse tapchanger, while copper, zinc and chlorine lines need less. Taps on the low voltage secondary are impractical because that winding has very few turns and very high current. So taps go on the primary or in a separate regulating autotransformer. Once you have a regulating unit ahead of the rectifier transformer, putting the phase shift there as well is one design decision rather than two.

That is the short version of why this keyword has a long answer. The angle, the turns ratio, the core, the tapchanger and the interphase reactor are all the same conversation. A quotation that lists a pulse number and a price has not had that conversation yet.

The other half: two bridges sharing one DC bus

Most pages stop at the angle. Two secondaries, two bridges, harmonics cancel, done. In a real plant the two bridges usually feed one common DC bus, and two voltage sources that are not instantaneously equal cannot simply be paralleled. Something has to absorb the difference. That something is an interphase reactor, also called an interphase transformer or a balancing reactor, and it is the other half of the reason a rectifier transformer is built the way it is.

A technology tutorial explains the magnetic balance. Under balanced conditions the average currents in the two windings of the interphase transformer are equal, so the net ampere turns in the window are zero and the flux lines in the closed magnetic path are absent. A slight imbalance between the two bridge currents produces a nonzero ampere turn set, which can drive the core into saturation and makes the interphase transformer draw a high excitation current. That is one reason, alongside core loss, to keep the operating flux density low. The same page says that with controlled rectifiers the difference in voltage the reactor has to absorb grows with firing angle, so the saturation risk rises, and that with diode rectifiers the difference stays constant for a given load, which is why a five leg rectifier transformer can be used instead.

GE’s brochure gives the functional version: the interphase reactor connects between the two neutral points of the double star low voltage windings, its DC midpoint goes to the load, and it lets the two elementary three phase systems work independently so that each diode or thyristor conducts for 120 degrees. A Chinese technical encyclopedia entry adds that the reactor limits circulating current between the two three phase half wave groups, and that using opposite polarity on the two secondaries cancels the DC magnetomotive force in the core so DC magnetization is avoided.

Matching tolerances and their evidence grade

One content page publishes a compact table of matching requirements for a 12 pulse arrangement. It is the only such table we found. It has no named source, no test method and no unit basis, so we grade it L3 and report it as a claim.

ParameterPublished toleranceConsequence stated on the pageOur grade and note
Open circuit voltage between secondariesplus or minus 1 percentCurrent imbalance and weaker harmonic cancellationL3, no measurement method given
Leakage impedanceplus or minus 5 percentUnequal voltage drops under loadL3, matches a separate engineering blog’s request
Winding resistanceplus or minus 3 percentCurrent sharing errorsL3, no temperature basis stated
Interphase reactor inductance0.5 to 2.0 mHDepends on load currentL3, no kVA or current basis
Interphase reactor isolation2.5 kV minimumIndustrial systemsL3, no test voltage duration
Core saturation margin20 to 30 percentMargin above operating fluxL3, no flux density figure
Bridge current sharing10 percent maximum deviation, ideal 50/50Heavier loaded group runs hotterL3, from a separate fault finding page

What makes the table useful is not the numbers. It is the list of parameters. Voltage, impedance, resistance, reactor inductance, isolation, saturation margin and measured current split are seven lines that belong on a factory test report. What makes it dangerous is that a buyer might paste the numbers into a specification as if they were standard values. Do not. Ask the supplier to state their own values, the test method behind each, and the edition of the standard the test follows.

Notice that IEC 61378-1:2011 lists, among its Edition 2 changes, the addition of considerations about current sharing and hot spot temperature in high current windings for various winding arrangements. That is a standards database transcription, not clause text, but it places the same seven parameters inside the scope of the converter transformer standard. The angle and the current sharing are not two separate purchases.

One degree of angle error at three pulse counts

An angle tolerance written as plus or minus one degree sounds identical on every data sheet. It is not. As a fraction of the intended step it triples as the pulse count doubles.

Design stepOne degree of error as a share of the stepWhat that does
30 degrees, 12 pulse3.3333 percentSmallest relative error of the three
15 degrees, 24 pulse6.6667 percentTwice the 12 pulse figure
7.5 degrees, 48 pulse13.3333 percentFour times the 12 pulse figure

This is pure arithmetic and it belongs next to every angle tolerance a supplier quotes. The same absolute manufacturing capability produces four times the relative error at 48 pulse. A consultant page says manufacturing tolerances are one of the things that set the practical limit on how small a shift can be built. Our three rows show why in numbers. It also explains why the 24 pulse angle sets in the previous section are worth checking for step consistency rather than just for the absolute angle values.

There is a related field symptom. A fault finding page says the two bridge currents should split 50/50, that winding impedance differences, unequal firing angles and different cable lengths cause imbalance, and that a deviation above 10 percent shows up as higher temperature in the heavier loaded winding and its rectifier devices. We grade that as L3, but the causal chain is the same one the interphase reactor saturation discussion describes. Angle error, voltage mismatch and impedance mismatch all land in the same place: one bridge works harder than the other and the cancellation degrades.

What a patent simulation adds

A public patent text describes a modular multi pulse transformer rectifier for a medium voltage drive. It states that providing phase shift in the primary windings as well as the secondary windings cancels more harmonics, and gives a simulated input current THD of about 4.5 percent with primary side shifting against about 7.1 percent without it, for a 4160 volt inverter feeding a 1000 horsepower motor. That is a reduction of 36.6197 percent, a ratio of 0.633803. It also describes an equivalent 36 pulse arrangement built from two 18 pulse transformers with a 20 degree primary shift and a 10 degree secondary shift.

We report those as simulation figures from a patent document, not measurements and not a guarantee. They are useful here for one reason: they are an example of the angle being placed on the primary on purpose, which is the same design choice GE describes for making all the rectifier transformers at a plant identical. Two independent sources, one commercial and one patent, point at the same architecture.

Five ways the angle stops working

These five came out of the field and commissioning pages we read. Each one is checkable before energization, and each one leaves the nameplate looking correct.

FailureWhat happens electricallyWhat catches it
One bridge loses a fuse or a deviceThe cancellation loses one of its two contributors and the 5th and 7th returnPer bridge current measurement and a harmonic reading at the PCC
Phase rotation reversed on one bridgeThe 30 degrees becomes the wrong direction even though the transformer is correctPhase sequence check on both secondaries before closing
Both secondaries built with the same vector groupTwo Yy0 windings give 0 degrees, so the plant behaves like two paralleled 6 pulse bridgesVector group and phase displacement measurement on the finished unit
Two secondaries 180 degrees apartYy0 with Yy6 gives an opposing pair and circulating currentSame displacement measurement, plus an open circuit voltage comparison
A turn to turn short in an extended delta sectionThe angle drifts by a fraction of a turn, which is a large share of a 7.5 degree stepTurns ratio test or frequency response, since DC resistance barely moves

A drive page puts the first one well: a blown fuse on one bridge, an unbalanced transformer or a device failure degrades the cancellation and lets the fifth and seventh harmonics reappear, quietly undoing the benefit the drive was bought for. A converter configuration page adds that if both secondaries are star with no displacement, the system behaves like two paralleled 6 pulse bridges with large 5th and 7th harmonics, DC link beating, current imbalance and line side faults at load. Those two descriptions are the same failure seen from the harmonic side and the drive side.

The fifth row is specific to small angle windings. A fault finding page says a turn to turn short in a phase shifting section is hard to find with a normal DC resistance test because the section has few turns and the resistance change is small, so a turns ratio tester or a low frequency response method is needed. That is a commissioning line, not a design line, and it belongs in the RFQ.

The order that catches them

Commissioning checks are cheapest in the order that finds a wiring error before it finds a design error. Start with the vector group and the phase displacement measurement on the finished unit, because a wrong clock number or a 180 degree pair makes everything downstream meaningless. Then compare the two open circuit secondary voltages, then the two short circuit impedances, then the phase rotation on each bridge. Only after those four pass is it worth energizing at reduced load to measure the current split. Leave the turns ratio or frequency response test for last, since it is the only one that finds a shorted turn inside a small extended delta section.

That order is also the order a factory test report should follow. If a supplier sends a report that begins with a harmonic measurement at full load and never prints the phase displacement result, the angle on the drawing has not been verified. A harmonic number can look acceptable on a stiff supply while the winding underneath it is a 0 degree pair.

The alternatives, and what they do not replace

A buyer asking why the transformer needs phase shifted windings is often really asking whether there is a cheaper box that does the same job. There are alternatives, and they solve different problems.

OptionWhere it sitsWhat it changesWhat it does not doEvidence grade
Line reactorAhead of one 6 pulse bridgeRounds the current block and reduces peak harmonicsDoes not create a second supply at a defined angleL2, qualitative
Passive tuned filterAhead of the drive or at the busTakes harmonic current out after it is madeDoes not change the ripple pulse number of the DC busL2, qualitative
Active front endReplaces the bridgeShapes near sinusoidal input current and can correct displacement factorIs a different converter, not a transformer optionL2, qualitative
Separate phase shifting transformerAhead of two ordinary transformersCreates the angle in its own windingsStill buys windings, plus another tank and another lossL1, GE brochure
Phase shifted windings in the rectifier transformerInside the unitSets a rated, measurable displacement between secondariesNeeds matched voltage, impedance, current sharing and a reactorL1, IEC metadata and GE brochure

A drive guide says multipulse front ends prevent the dominant harmonics from being produced while reactors and filters attenuate them afterwards, and that multipulse tends to suit large, continuously running drives where the alternative filter would be just as substantial. A sustainability aggregator says the approach is passive, needs no control electronics after installation, and therefore offers high electromagnetic immunity. A UPS page says the phase shifting transformer is large, heavy and expensive, which is why 12 pulse is usually reserved for units above 300 kVA. Those are three reasonable statements, but none of them comes with a dated quote, a bill of materials or a defined kVA, so we are not using their cost or capacity figures as requirements.

The one thing all five rows share is that a phase angle is a relationship between supplies. Electronics can replace the bridge. A filter can remove what the bridge injects. Neither one produces a rated 30 degree displacement between two secondaries. If the specification needs that displacement, it needs windings.

Sources, graded

Here is the evidence ledger for this article. L1 means we could read a primary document, a standards database entry or a brand’s own publication. L2 means a named secondary source with a traceable derivation. L3 means a number or claim we report but do not rely on.

#SourceGradeWhat we took from itWhat we did not take
1Five standards databases on IEC 61378-1:2011 Edition 2L1Edition 2, replaces 1997 ED1, TC 14 / MT 7, six technical changes, 90 and 184 page versions, COR1:2012Any clause text, any exact publication day
2French language abstract of the same standardL1Zigzag and extended triangle named for a displacement under 30 degreesA ruling on which abstract is authoritative
3GE Vernova industrial rectifier transformer brochureL130 degrees by secondary star and delta, primary or autotransformer shifting for higher pulses, identical plant units, single and two tank options, IPT and 120 degree conductionNo prices, no delivery data
4MB Drive Services multi winding VFD transformer designL16 times 30 equals 180 degrees, five winding connections, 30 degree clock minimum, d11.33 notationNo turns ratio formula
5Public patent text on modular multi pulse rectifiersL1Simulated THD 7.1 to 4.5 percent, 20 degree primary and 10 degree secondary shiftTreated as a guarantee or a measurement
6Transformer technology tutorial on rectifier transformersL21 in 1.732 turns problem, two primaries, 52 percent intermediate yoke, IPT saturation, firing angle effect, extended delta versus zigzagThe rounded 52 percent as exact
7Siemens vector group table reproduced on a converter pageL2Dd0Dy11, Dd6Dy5, Yy0Yd11, Yy6Yd5, uk 4 to 6 percent, slave bridge lag ruleTreated as Siemens clause text
8Extended delta sine rule design guideL2The two sine rule relations and the phasor triangle geometryIts four row numeric table, which fails the check
97.5 degree pseudo 24 pulse design note on the same siteL2Vector derivation giving an extension fraction of 0.1646The 87.6 and 12.4 percent tap convention, which is a separate geometry
10IEEE C57.18.10-2021 secondary readingL293 pages, 100 kW scope, rms and fundamental kVA, commutating impedance, FHL factorsAny clause wording or test criterion
11Matching tolerance, fault finding and UPS pagesL3Seven matching parameters, 10 percent current split, 300 kVA thresholdEvery number as a requirement
12Sine rule, vector route, yoke vector difference, clock arithmetic, angle error and date intervals in this articleOur arithmetic0.164525, 0.366025, 0.517638, minus 20.0000 degrees, 3.3333 to 13.3333 percent and every interval in the date tableNot a measurement, not a standard value

A copy paste RFQ block

Fifteen lines. Paste them into the enquiry and make the supplier answer each one. The first six settle what is being built, the next five settle what is measured, and the last four settle what proves it.

  1. State the pulse count, the number of isolated secondaries, the number of bridge groups and the DC bus arrangement, series or parallel.
  2. State the full vector group of every winding, including the clock number of each secondary, not just the angle between them.
  3. State the required displacement between adjacent groups in degrees and the allowed tolerance in degrees, on a separate line.
  4. State which connection produces each angle: star and delta, zigzag, extended delta, polygon delta, or a separate phase shifting transformer.
  5. For any extended delta winding, give the extension to main turns ratio, the output to main voltage ratio and the phasor diagram used to derive them.
  6. State whether the phase shift is on the secondary, on the primary, or on a regulating autotransformer, and give the reason for that choice.
  7. State the rated line voltage of every secondary and the measured difference between them at no load, with the test method.
  8. State the short circuit voltage of each secondary and the permitted deviation between the two, on two separate lines.
  9. State the interphase reactor or balancing reactor data: type, inductance, current rating, isolation level, saturation margin and whether it sits inside the tank.
  10. State the guaranteed current split between bridges at rated load and at 25 percent load, with the test method and the instrument.
  11. State the standard and edition used for the design, the loss calculation and the tests, including corrigendum status.
  12. Give the phase displacement measurement result on the finished unit and the turns ratio measurement for each winding section.
  13. Give the harmonic loss factors used for each winding, the spectrum behind them, and the guaranteed load loss in watts at rated duty.
  14. Give the hot spot assessment for the high current windings and the current sharing test report number, laboratory and date.
  15. State what the quoted scope includes: tapchanger, regulating autotransformer, interphase reactor, enclosure, commissioning phase sequence verification and a turns ratio or frequency response test for small angle sections.

Ten dates that belong in different columns

A standard number without an edition is not a citation. A rule without a date is not a requirement. Three dates that look similar do three different jobs: the day a document was published, the day it was adopted somewhere else, and the day a rule starts to bind. Keep them apart.

#EventDateIntervalWhat it is not
1IEC 61378-1:1997 Edition 1September 1997, month only5,076 days before the Edition 2 date used hereNot the current edition
2IEC 61378-1:2011 Edition 2 publishedJuly 2011, day differs across sources5,589 days before this articleNot the EN adoption date
3Corrigendum COR1:2012January 2012, month only5,430 days before this articleNot a new edition
4IEEE C57.18.10-1998 edition1998, year only23 years by edition label before the 2021 editionNot an IEC document
5IEEE C57.18.10-2021 edition2021, year only2,142 days before this articleNot a compliance deadline
6EU 548/2014 Tier 2 appliesJuly 1, 2021Carried over from earlier work, not re verifiedNot a winding rule
7IEEE 519-2022 publishedAugust 2022, day not verifiedCarried over from earlier work, not re verifiedNot a phase displacement standard
8Date of this articleNovember 13, 20265,589 days after the Edition 2 date used hereNot any of the above
9US federal efficiency compliance dateApril 23, 2029892 days after this articleCarried over from earlier work, not re verified
10Edition label gap 1997 to 201114 years by edition yearA label gap, not a day countNot a compliance window

Every interval is computed from the dates printed in the table, using September 1, 1997, July 26, 2011, January 1, 2012, January 1, 2021 and November 13, 2026 as arithmetic stand ins. From the 1997 edition to the 2011 edition is 5,076 days, from the 2011 edition to this article 5,589 days, from the corrigendum to this article 5,430 days, from the 2021 IEEE edition to this article 2,142 days, and from this article to the federal date 892 days. We used July 26 because it appears on three of the five standards databases, but July 25 appears on two, so the day is not verified here. If a date sits on a purchase order, get it from the publisher’s current page.

Rows 6, 7 and 9 are in this table for one reason only. All three are rules that show up in the same specification sections as converter transformers. None of them changes a phase displacement requirement. A submittal that lists Tier 2, IEEE 519 and a vector group as if they were three rungs of one ladder has merged three documents.

Bring this to the specification meeting

The whole answer in one place

If you have five minutes in a design review, say this. A single 6 pulse bridge cannot cancel its own harmonics, so the cancellation needs two bridges with a defined angle between their supplies. At 30 degrees the 6th DC ripple lands 180 degrees out of phase, because 6 times 30 is 180, and the corresponding 5th and 7th on the AC side oppose at the common primary. Only a winding connection can set that angle, which is why IEC 61378-1:2011 Edition 2 lists the addition of winding connections with phase displacement as its first technical change. Star and delta give 30 degrees for free on a diagram, but equal line voltages need a turns ratio of 1 in 1.732, or 0.577350, and a low voltage high current secondary has too few turns to build it exactly. The common fix is two primaries, one star and one delta, with identical secondaries, and the flux difference between those primaries needs an intermediate yoke of 2 times the sine of 15 degrees, which is 51.7638 percent of the main limb. Smaller angles need extended delta: the sine rule gives 0.164525 at 7.5 degrees, 0.366025 at 15 degrees and 0.532089 at 20 degrees, and the corresponding output voltage ratios are 1.091601, 1.224745 and 1.347296. Two bridges on one DC bus also need an interphase reactor to force current sharing, and one degree of angle error is 3.3333 percent of a 30 degree step but 13.3333 percent of a 7.5 degree step. Put the full vector group, the angle tolerance, the turns ratio, the impedance match and the current sharing test on five separate lines of the RFQ.

That is the whole article in one paragraph. Everything else here is the evidence for one of those sentences.

What actually costs money is what sits between the lines. One standard edition added phase displacement windings as a named design item and nobody in the search results quoted it. One published turns ratio table has four rows that all fail a sine rule check, while the same site’s worked derivation agrees with the sine rule. One rounded 52 percent yoke turns out to be a 51.7638 percent vector difference. Two sources give conflicting short circuit voltage ranges, 4 to 6 percent and 6 to 8 percent, for the same 12 pulse duty. One clock formula claims that four different clock differences all give the same result, and we could not verify three of them. One patent simulation puts THD at 7.1 percent without primary side shifting and 4.5 percent with it. Six of these are document failures, not engineering failures, and every one is catchable with a single question before the winding drawing is released.

Twelve numbers worth keeping

NumberWhat it decides
6 times 30 equals 180 degreesWhy two bridges need a 30 degree displacement
2.7778 and 1.3889 millisecondsDC ripple period at 6 and 12 pulse on a 60 Hz supply
30, 20, 15, 10 and 7.5 degreesThe 360 divided by p ladder from 12 to 48 pulse
0.577350 and 1.732051The wye to delta turns ratio and its reciprocal
0.517638 and 51.7638 percentThe intermediate yoke as a flux vector difference
0.164525 and 1.091601Extended delta at 7.5 degrees, turns and voltage
0.366025 and 1.224745Extended delta at 15 degrees, turns and voltage
0.532089 and 1.347296Extended delta at 20 degrees, turns and voltage
0.078660 and 12.7130The 3.75 degree extension and its reciprocal
0.5 degrees and minus 20.0000 degreesOne clock minute, and d11.333333 resolved
3.3333, 6.6667 and 13.3333 percentOne degree of error at 30, 15 and 7.5 degree steps
36.6197 percent and 0.633803Simulated THD reduction from 7.1 to 4.5 percent

Keep the first two for the quotation conversation, because they answer why there are two secondaries at all. Keep the angle ladder for the winding drawing and the turns ratio pair for the core drawing. Keep 51.7638 percent for the moment someone asks why the yoke is half the size of the limb. Keep the three extended delta pairs for the submittal review, and remember that the second number in each pair is the voltage correction, not a bonus. Keep 0.078660 and its reciprocal for the 48 pulse conversation, where one turn is a large share of the angle. Keep the clock arithmetic for the notation on the nameplate. Keep the three error percentages for the tolerance line, and the two patent figures for the primary side shift decision, remembering that they are a simulation.

Those numbers fit on one page of a bid review, and each one can be recomputed from the figures on the sheet in front of you. If you have a single line diagram, a draft specification, two quotations and a winding drawing, send them over. We will tell you which connection produces each angle, whether the turns ratio and the output voltage correction agree with the sine rule, whether the two secondaries can actually share current, and which of the fifteen RFQ lines still need an answer before the order is released. Start here.

FAQ

Why can’t one 6 pulse bridge cancel its own harmonics?

Because cancellation needs two waveforms in opposition. One bridge draws a 120 degree current block with a 6th ripple at 360 Hz on a 60 Hz supply. A second bridge fed 30 degrees away puts that 6th ripple at 6 times 30, or 180 degrees, out of phase. The DC ripple period halves from 2.7778 to 1.3889 milliseconds.

Why is the angle exactly 30 degrees for 12 pulse?

It is 360 divided by 12, the same as 60 divided by 2 bridge groups. At 18 pulse it is 20 degrees, at 24 pulse 15 degrees, at 36 pulse 10 degrees and at 48 pulse 7.5 degrees. Star and delta can only produce the 30 degree case.

Can a filter or an active front end replace phase shifted windings?

Not for the angle. A line reactor rounds one waveform, a passive filter removes harmonics after they are made, and an active front end replaces the bridge. None of the 3 creates a second supply 30 degrees away. A separate phase shifting transformer does, but that still buys windings and one more tank.

What did IEC 61378-1:2011 add?

Edition 2 lists six significant technical changes, the first being the addition of winding connections with phase displacement. It also added transformers with more than one active part in one tank, changed the reference power definition, added regulating transformers feeding converter transformers, added current sharing and hot spot considerations for high current windings, and added transductors for DC regulation.

Why can’t star and delta give 15 degrees?

A clock hour is 30 degrees, from 360 divided by 12, and a change of connection moves one hour. Same type windings only allow 0 or 180 degrees, so Dd0 and Dd6 or Yy0 and Yy6. A 15 degree step needs zigzag, extended delta or polygon delta, with a tap or an extra section on each phase.

What is the 1 in 1.732 turns problem?

A delta line voltage equals its phase voltage, while a wye line voltage is 1.732051 times its phase voltage. Equal line voltages across the two secondaries need a wye to delta turns ratio of 0.577350. A high current secondary may have fewer than 10 turns, and no integer count gives that ratio exactly.

How big is the intermediate yoke?

Two equal limb fluxes 30 degrees apart differ by 2 times the sine of 15 degrees, which is 0.517638 or 51.7638 percent of one flux. A technical tutorial prints a rounded 52 percent, 0.2362 percentage points above our arithmetic. It is geometry, not a clause value, and only for this two primary arrangement.

What is the correct extended delta extension ratio at 15 degrees?

0.366025 of the main section, with an output to main voltage ratio of 1.224745. Both come from the sine rule in a phasor triangle with a 60 degree interior angle, and a separate vector derivation returns the same numbers and exactly 15.0000 degrees. A published table gives 0.318 and 1.155 instead.

What does d11.33 mean on a winding drawing?

A clock number of 11.333333 times 30 degrees is 340 degrees, which is minus 20 degrees on a 360 degree circle. One clock hour is 30 degrees and one clock minute is 0.5 degrees. Written as the rounded 11.33 it gives minus 20.1 degrees, so the exact form is 11.333333.

Which vector groups give 12 pulse?

A Siemens table reproduced in a converter guide lists Dd0Dy11, Dd6Dy5, Yy0Yd11 and Yy6Yd5, all with uk 4 to 6 percent. Two Yy0 windings give 0 degrees and behave like paralleled 6 pulse bridges. Yy0 with Yy6 gives 180 degrees and circulating current. A separate page quotes uk 6 to 8 percent.

What is an interphase reactor for?

Two bridges on one DC bus are not instantaneously equal, so the reactor absorbs the voltage difference and forces current sharing. GE says it lets each diode or thyristor conduct for 120 degrees. A tutorial says balanced currents give zero net ampere turns, and a slight imbalance can saturate the core; with thyristors the voltage difference grows with firing angle.

How much does one degree of angle error matter?

It is 3.3333 percent of a 30 degree step, 6.6667 percent of a 15 degree step and 13.3333 percent of a 7.5 degree step. The same manufacturing capability produces four times the relative error at 48 pulse. Angle error, voltage mismatch and impedance mismatch all end in the same place: one bridge carries more current.

What should the RFQ say?

15 lines: pulse count and bus arrangement, full vector groups, angle and tolerance, connection type, extended delta ratios and phasor diagram, shift location, secondary voltages and their difference, short circuit voltages and deviation, reactor data, current split at 2 loads, standard and edition, displacement and turns ratio test results, loss factors and guaranteed watts, hot spot and sharing report, and scope.