Does phase rotation matter for this SCR?

Status
Not open for further replies.
S

shockking

Member
Location
Sacramento, CA
Occupation
engineer
Does phase rotation matter for this SCR? Diagrams below.

Context: We have a 1,000 HP motor originally installed in 1987. It gets DC power from a three-phase silicon controlled rectifier (SCR) bridge rectifier assembly. I'm not too familiar with these so I'm doing some reading. I'm trying to confirm if phase rotation matters for this circuit. We replaced the step down transformer and we're trying to tell if it will be a problem if for some reason the phase rotation gets reversed. The documentation describes a phase sensing circuit on the firing circuit board. But I can't tell if this is sensing rotation, or what exactly.

Here's the rectifier diagram:

1675968899021.png
Detail of the hard firing circuit:
1675968935129.png

Schematic of the firing circuit board and regulator board:
1675976564265.png

Thanks in advance for any advice.
 
As long as the A, B, and C inputs to the bridge match exactly to the A, B, and C inputs to the control board, it does not matter how the firing circuit itself is powered nor what the phase rotation is.
Ideally both sets of leads are internally wire to a single set of line input terminals for the unit.
 
I worked on on VFD that was sensitive to the line side rotation....would not function if the line side rotation was incorrect. It was a long time ago, and not sure of what brand it was.
 
I suspect that the control circuits are designed for only one rotation.

The time intervals when the SCR's on a given phase (for example phase A) are conducting should be independent of the rotation sequence. It’s just the order in which the SCR’s for the other two phases are conducting when the SCR’s on the given phase (A) are also conducting that is different.

If an SCR rectifier was operating at maximum duty cycle with an A,B,C sequence, then it would duplicate the output of a full-wave rectifier as shown below (from https://www.electronics-tutorials.ws/power/three-phase-rectification.html ) The duty cycle can be reduced by appropriately delaying the triggering of the SCR’s.

Three-phase_rectification_waveforms.png
Notice that the rectifier output is line-to-line voltage VAB followed by VAC during the 120° interval when phase A is the most positive of the three phases. Reversing phases B and C would exchange the order such that VAC comes first and then VAB comes afterwords. But the period when the A phase is conducting to the rectifier output remains unchanged.


In the schematics of post #1, the three channels for the firing circuits of the phase A, B, and C SCR's are driven by voltages A-C, B-A, and C-B as a result of the connections made to the three transformer primaries.
If, for example, phases B and C from the source of power are reversed, then the firing circuit channel that triggers the phase A SCR’s will be driven with A-B instead of A-C. That’s a 60°phase shift that will also shift the time intervals when the phase A SCR’s will be conducting. But since all connections of phase A to the firing circuit and the SCR’s remain unchanged, the firing intervals of the SCR’s connected to phase A should also be unchanged but they are not.
I suspect that the line-to-line connections to the transformers on the firing circuit board were chosen to compensate for time delays in the firing circuits, especially the larger delays from the RC ramp generating circuits.

The SCR firing circuit for a DC SCR drive from Joliet Technologies at https://joliettech.com/information/model-1681-instruction-manual-firing-circuits-dc-scr-drives/ has a similar topology as the one provided by the OP, but it uses digital counters to control the SCR firing period rather than analog RC voltage ramps.

Joliet_Technologies_SCR_firing_circuit.png

However, the instructions at the link above say the following:

"3.12.1 Phase Sequence and Phase Loss

Correct phase sequence of the incoming power lines is necessary for proper thyristor firing.
Therefore, the fault logic circuits contain a phase sequence detector to prevent incorrect operation.

If the incoming phase sequence is not ABC, the phase detector will inhibit thyristor firing and no power will be supplied to the motor. Incorrect phasing is corrected by reversing any two input power lines."

Therefore, it would not be unexpected that the circuit posted by the OP would also require a specific phase sequence.
 
synchro said:
I suspect that the control circuits are designed for only one rotation.

The time intervals when the SCR's on a given phase (for example phase A) are conducting should be independent of the rotation sequence. It’s just the order in which the SCR’s for the other two phases are conducting when the SCR’s on the given phase (A) are also conducting that is different.

If an SCR rectifier was operating at maximum duty cycle with an A,B,C sequence, then it would duplicate the output of a full-wave rectifier as shown below (from https://www.electronics-tutorials.ws/power/three-phase-rectification.html ) The duty cycle can be reduced by appropriately delaying the triggering of the SCR’s.

View attachment 2563969
Notice that the rectifier output is line-to-line voltage VAB followed by VAC during the 120° interval when phase A is the most positive of the three phases. Reversing phases B and C would exchange the order such that VAC comes first and then VAB comes afterwords. But the period when the A phase is conducting to the rectifier output remains unchanged.


In the schematics of post #1, the three channels for the firing circuits of the phase A, B, and C SCR's are driven by voltages A-C, B-A, and C-B as a result of the connections made to the three transformer primaries.
If, for example, phases B and C from the source of power are reversed, then the firing circuit channel that triggers the phase A SCR’s will be driven with A-B instead of A-C. That’s a 60°phase shift that will also shift the time intervals when the phase A SCR’s will be conducting. But since all connections of phase A to the firing circuit and the SCR’s remain unchanged, the firing intervals of the SCR’s connected to phase A should also be unchanged but they are not.
I suspect that the line-to-line connections to the transformers on the firing circuit board were chosen to compensate for time delays in the firing circuits, especially the larger delays from the RC ramp generating circuits.

The SCR firing circuit for a DC SCR drive from Joliet Technologies at https://joliettech.com/information/model-1681-instruction-manual-firing-circuits-dc-scr-drives/ has a similar topology as the one provided by the OP, but it uses digital counters to control the SCR firing period rather than analog RC voltage ramps.

View attachment 2563968

However, the instructions at the link above say the following:

"3.12.1 Phase Sequence and Phase Loss

Correct phase sequence of the incoming power lines is necessary for proper thyristor firing.
Therefore, the fault logic circuits contain a phase sequence detector to prevent incorrect operation.

If the incoming phase sequence is not ABC, the phase detector will inhibit thyristor firing and no power will be supplied to the motor. Incorrect phasing is corrected by reversing any two input power lines."

Therefore, it would not be unexpected that the circuit posted by the OP would also require a specific phase sequence.
Click to expand...

The OP’s schematic appears to show the SCRs paired with common firing circuits and internally connected to common phase inputs. In that case, I don’t think phase sequence matters.
Granted, it’s been 35 years since I dealt with these things!
 
retirede said:
The OP’s schematic appears to show the SCRs paired with common firing circuits and internally connected to common phase inputs. In that case, I don’t think phase sequence matters.
Granted, it’s been 35 years since I dealt with these things!
Click to expand...

If it was that easy, I don't think that patented approaches to make SCR firing circuits insensitive to phase sequence would have been pursued, such as the following:

https://patents.google.com/patent/US3593105

https://patents.google.com/patent/US3883791A/en
 
synchro said:
I suspect that the control circuits are designed for only one rotation.

The time intervals when the SCR's on a given phase (for example phase A) are conducting should be independent of the rotation sequence. It’s just the order in which the SCR’s for the other two phases are conducting when the SCR’s on the given phase (A) are also conducting that is different.

If an SCR rectifier was operating at maximum duty cycle with an A,B,C sequence, then it would duplicate the output of a full-wave rectifier as shown below (from https://www.electronics-tutorials.ws/power/three-phase-rectification.html ) The duty cycle can be reduced by appropriately delaying the triggering of the SCR’s.

View attachment 2563969
Notice that the rectifier output is line-to-line voltage VAB followed by VAC during the 120° interval when phase A is the most positive of the three phases. Reversing phases B and C would exchange the order such that VAC comes first and then VAB comes afterwords. But the period when the A phase is conducting to the rectifier output remains unchanged.


In the schematics of post #1, the three channels for the firing circuits of the phase A, B, and C SCR's are driven by voltages A-C, B-A, and C-B as a result of the connections made to the three transformer primaries.
If, for example, phases B and C from the source of power are reversed, then the firing circuit channel that triggers the phase A SCR’s will be driven with A-B instead of A-C. That’s a 60°phase shift that will also shift the time intervals when the phase A SCR’s will be conducting. But since all connections of phase A to the firing circuit and the SCR’s remain unchanged, the firing intervals of the SCR’s connected to phase A should also be unchanged but they are not.
I suspect that the line-to-line connections to the transformers on the firing circuit board were chosen to compensate for time delays in the firing circuits, especially the larger delays from the RC ramp generating circuits.

The SCR firing circuit for a DC SCR drive from Joliet Technologies at https://joliettech.com/information/model-1681-instruction-manual-firing-circuits-dc-scr-drives/ has a similar topology as the one provided by the OP, but it uses digital counters to control the SCR firing period rather than analog RC voltage ramps.

View attachment 2563968

However, the instructions at the link above say the following:

"3.12.1 Phase Sequence and Phase Loss

Correct phase sequence of the incoming power lines is necessary for proper thyristor firing.
Therefore, the fault logic circuits contain a phase sequence detector to prevent incorrect operation.

If the incoming phase sequence is not ABC, the phase detector will inhibit thyristor firing and no power will be supplied to the motor. Incorrect phasing is corrected by reversing any two input power lines."

Therefore, it would not be unexpected that the circuit posted by the OP would also require a specific phase sequence.
Click to expand...
The fact that the (line -) and (line +) SCRs for a given line conductor are driven from the same analog reference signal in the OP's circuit does indeed suggest that the phase rotation is important. I did not take that into account. The current in line conductor A will be balanced by a phased combination of line conductor B and line conductor C.
 
SCR firing schemes vary. In all cases, you need to know the “zero cross” point of each half of each sine wave every time in order to affect the proper timing of when the SCR is gated (turned on). The problem is, you cannot measure zero, so what is done is that you measure the point where a sine wave is descending, and the next one that is ascending crosses it, meaning those two values are equal. Then you use that crossover point to calculate where the zero cross is going to be on the next cycle. Some schemes do this for each SCR individually, but that requires more hardware in the firing board. So to keep everything smaller (and cheaper) some do it by measuring one crossover point between phases, then calculating all 6 zero cross points from that one measurement. One down side of that method is that phase rotation matters, because you are assuming an entire sequence of events critical to the gating scheme.

In the earliest days of soft starter development in the late 60s early 70s, the first two soft starters in the market were a company in Florida called Vectrol, who used the more expensive method, and a company from South Africa called Saftronics (South African Electronics), who used the cheaper method. Virtually all soft start firing designs are based on those two initial ones. So if they are phase rotation sensitive, they are using the Saftronics firing scheme, if not, they are using the Vectrol. Only a few soft starts on the market still use the Vectrol scheme because it is slightly more expensive and more difficult to miniaturize. So most soft starters on the market are phase rotation sensitive.
 
Jraef said:
SCR firing schemes vary. In all cases, you need to know the “zero cross” point of each half of each sine wave every time in order to affect the proper timing of when the SCR is gated (turned on). The problem is, you cannot measure zero, so what is done is that you measure the point where a sine wave is descending, and the next one that is ascending crosses it, meaning those two values are equal. Then you use that crossover point to calculate where the zero cross is going to be on the next cycle. Some schemes do this for each SCR individually, but that requires more hardware in the firing board. So to keep everything smaller (and cheaper) some do it by measuring one crossover point between phases, then calculating all 6 zero cross points from that one measurement. One down side of that method is that phase rotation matters, because you are assuming an entire sequence of events critical to the gating scheme.

In the earliest days of soft starter development in the late 60s early 70s, the first two soft starters in the market were a company in Florida called Vectrol, who used the more expensive method, and a company from South Africa called Saftronics (South African Electronics), who used the cheaper method. Virtually all soft start firing designs are based on those two initial ones. So if they are phase rotation sensitive, they are using the Saftronics firing scheme, if not, they are using the Vectrol. Only a few soft starts on the market still use the Vectrol scheme because it is slightly more expensive and more difficult to miniaturize. So most soft starters on the market are phase rotation sensitive.
Click to expand...
I disagree. We designed and manufactured SCRs. Phase rotation is fixed. But what do I know.....
 
Jraef said:
SCR firing schemes vary. In all cases, you need to know the “zero cross” point of each half of each sine wave every time in order to affect the proper timing of when the SCR is gated (turned on). The problem is, you cannot measure zero, so what is done is that you measure the point where a sine wave is descending, and the next one that is ascending crosses it, meaning those two values are equal. Then you use that crossover point to calculate where the zero cross is going to be on the next cycle. Some schemes do this for each SCR individually, but that requires more hardware in the firing board. So to keep everything smaller (and cheaper) some do it by measuring one crossover point between phases, then calculating all 6 zero cross points from that one measurement. One down side of that method is that phase rotation matters, because you are assuming an entire sequence of events critical to the gating scheme.

In the earliest days of soft starter development in the late 60s early 70s, the first two soft starters in the market were a company in Florida called Vectrol, who used the more expensive method, and a company from South Africa called Saftronics (South African Electronics), who used the cheaper method. Virtually all soft start firing designs are based on those two initial ones. So if they are phase rotation sensitive, they are using the Saftronics firing scheme, if not, they are using the Vectrol. Only a few soft starts on the market still use the Vectrol scheme because it is slightly more expensive and more difficult to miniaturize. So most soft starters on the market are phase rotation sensitive.
Click to expand...

Does that also apply to DC variable speed drives?
 
Sorry for de-railing this phase rotation thread slightly - but thought this technical discussion above was quite interesting and I also thought some would be interested in related zero crossing story.

About a dozen years ago I was involved in deploying a large number of relatively sophisticated sorting machines. Much of the automation was done with Lenze Servo Drives. Like any good German company, they do some really interesting things in the background that you never notice until something goes wrong to showcase them.

Early on in deployment, we had some 3 phase AC Mains quality issues (an Ohio POCO iirc, and in summertime to boot). If certain parts of the machine were running (with large motors), others subsystems would have trouble re-starting if you power cycled the Lenze Power Supply (AC -> DC) for the Servo Drives.

Long story short, it turns out that on a cold boot, rather than connecting the 6 diode rectifier bridge to the AC Mains Supply straight away in the Power Supply - they performed a very quick test to try and detect obvious problems on the devices connected downstream on the DC bus, which were the Servo Drives and the wiring to and between them. What they were doing was looking at zero crossings of the AC Mains and firing a set of thyristors just at those crossings and leaving them "on" for just a short time/angle. This would charge the DC bus to ~25 or 30 volts - which they would monitor for in the power supply and if all was good and that voltage was detected, then they would close a set of relays to let the whole shebang into the rectifier bridge. And Awaaay We Go as Jackie Gleason would say.

However, if they did NOT detect the proper voltage, they would not close those relays and throw a specific fault. It worked very well, and I certainly saw it detect a few "Sunday wired to Monday" problems where somebody swapped DC+ and DC- on a connector. Especially on new systems.

But... they would also throw a fault if your incoming AC mains contained too much distortion for them to accurately predict zero crossings. If there was enough phase shift - which could happen if other large motors were running, they could not predict (to their satisfaction) the right "time/angle" to fire the thyristor. And if they gated them at the wrong time, and there was a downstream problem on the DC bus - then the "Smoking Lamp was lit" as they used to say.

Again - really interesting stuff. We got them to relent, after much discussion, and to relax the window of tolerance for their zero crossing tests. They really really wanted us to fix our crappy American AC power (and we could certainly see their argument). But we knew exactly what the problem was and it could be solved with just a little more flexibility on their end. And we were deploying so many systems that we got them to release a new firmware version (for free even!!) to solve this issue.

So again, to the OP (Shockking), sorry this story was not related to your phase rotation question. But I hope it was close enough to be interesting to some here.
 
shockking said:
Sounds like the consensus is yes. Thanks for all the feedback, much appreciated.
Click to expand...
Even if it doesn't matter about rotation if something goes wrong and they're looking for something to blame I'll bet it will......
 
MD Automation said:
...

Long story short, it turns out that on a cold boot, rather than connecting the 6 diode rectifier bridge to the AC Mains Supply straight away in the Power Supply - they performed a very quick test to try and detect obvious problems on the devices connected downstream on the DC bus, which were the Servo Drives and the wiring to and between them. What they were doing was looking at zero crossings of the AC Mains and firing a set of thyristors just at those crossings and leaving them "on" for just a short time/angle. This would charge the DC bus to ~25 or 30 volts - which they would monitor for in the power supply and if all was good and that voltage was detected, then they would close a set of relays to let the whole shebang into the rectifier bridge. And Awaaay We Go as Jackie Gleason would say.

However, if they did NOT detect the proper voltage, they would not close those relays and throw a specific fault. It worked very well, and I certainly saw it detect a few "Sunday wired to Monday" problems where somebody swapped DC+ and DC- on a connector. Especially on new systems.

...
Click to expand...
Two different things there. One is the "pre-charge" circuit, used to prevent damage to the DC bus capacitors in energizing them. Some pre-charge circuits just put a resistor in series with the DC bus for a second, then short around it. Others, as this sounds, use an SCR based rectifier system (instead of all diodes) and "ramp" the rectified voltage into the caps. Lots of large VFDs do that too.

The other circuit you are describing is called a "toe-in-the-water" circuit; testing out the connected devices before fully energizing them. I find those to be VERY useful, I wish more manufacturers employed them. But it takes more hardware, so more engineering, more real estate on the PC boards and ultimately more cost.
 
Status
Not open for further replies.
Top