1. Home
  2. Electronics Design
  3. Ideas for Smart SCR Control Board

Ideas for Smart SCR Control Board

I have built my first prototype of a PIC-based SCR controller board, and I have started preliminary assembly coding and testing. Most of the functions are fairly simple, but I would appreciate comments on the best way to do the required functions and also some which are future enhancements or special purpose variants.

This is a single phase switch used to control loads which range from several hundred milliamps to several hundred amps, at voltages from several volts to about 600 VAC. The load is a transformer which steps down to about

12 VAC at currents up to 50,000 amperes with an inductive load (circuit breaker) which may interrupt the flow of current at any point on the waveform.

To minimize DC offset, the initial phase angle is delayed about 60 to 90 degrees. It is adjusted so that all subsequent current peaks are about equal, to produce consistent true RMS readings and breaker operation.

Some problems we have experienced with previous designs are:

(1) Loss of gate triggering to one SCR, causing huge DC currents (2) Large initial current surges due to remanent magnetism (3) Need for remote programmable initial phase angle due to load variations

My new design uses a PIC18F242 which has many advanced features. I have added an RS232 port for possible diagnostics and remote programming, but the main control inputs are INIT, ENABLE, and PULSE. There is a BCD switch to preset a number of cycles in PULSE mode, and a pot to preset nominal phase angle. I also have a CT to measure the SCR current, and optos to provide external phase reference and to sense when either SCR has a voltage on it. I also added a thermistor input for optional temperature sensing.

To solve problem #1, I will first make sure there is voltage of both polarities across the SCRs before allowing initiation. Once gate voltage has been applied, I will check that there is no voltage on the corresponding SCR. An ERROR shutdown will be set if there is a problem, and shut down before the next half-cycle. This should minimize the fault current. No current should flow if the first gate did not cause conduction. If the second gate is faulty, the first half cycle will still cause a large DC surge, although it will be limited to much less than that produced by a second half-cycle.

For problem #2, I may monitor the number of half-cycles that pass through the SCRs, by reading current. An even number of half-cycles should not produce much remanent magnetism, but an odd number will. If that occurs, I might be able to initiate a brief "degaussing" cycle by phase firing both SCRs at a reduced level. This should only occur when the load has tripped. If the load is still connected, I will simply leave the gates on for an additional half-cycle to make an even number.

For #3, the easiest way is to use the RS232 connection to program the phase angle from the remote controller, which also monitors the current waveform and can detect excessive DC offset that needs correction. However, the existing systems have only the INIT signal available, used to turn the SCR on or off. What I plan to do is send a pulse train on the INIT line that will be interpreted as a programming command to preset the phase angle. It would only need to be set from 60 to 90 degrees, and to a precision of 3 degrees, so a string of 1 to 10 pulses could simply be counted and used directly. That may be easier than implementing an eight bit data structure of marks and spaces that would need to be done in software, rather than a USART, although possibly the existing USART could be parallelled to the INIT line and used normally. An algorithm to send data in this format is probably much simpler than to receive it. This will be a future enhancement.

I will be using assembly code for most, if not all, of this project, although I do have a C18 compiler that might be useful for more complex enhancements. I am using a 14.7456 MHz crystal which divides down nicely for standard baud rates and a 7200 per second timer interrupt which provides 3 degrees per tick for phase angle adjustment.

My purpose in posting this is to see if anyone has any suggestions for better ways to implement the required functions, and also if there may be other uses for this board. It could probably be adapted easily as a variable phase-fired controller with voltage or current feedback, and remotely programmed via RS232. It may be useful for welding applications, where a preset number of cycles (or time) may be programmed. The components are not very expensive, perhaps $60 plus $25 for the board, and much of that is the DC-DC converters I chose to use instead of larger but cheaper transformer supplies.

Thanks,

Paul

formatting link

Reply to
Paul E. Schoen
Loading thread data ...

I can't help you much, but from the Microchip website: "Not Recommended for new design The intent of this guide is to provide you with recommendations of newer products that may be more cost effective and/or have more features than some of our more mature products. This is not an end of life notice for these products. Alternative Device Please consider using device PIC18F2420"

I don't think you should be using a part that's being phased out for a new design.

Reply to
a7yvm109gf5d1

Thank you very much for the observation. I have ordered samples of the new part. It is still listed as preliminary, and I think it is fully compatible for my purposes. It is frustrating when parts are phased out so quickly. I had originally selected this part about two years ago when it was fairly new.

Paul

Reply to
Paul E. Schoen

Interesting project, to make a $60 programmable SCR controller! (this is coming from a user of a $700 SCR controller in a homwbrew welder)

two random thoughts.

Can you control your magnetism issue by controlling initial phase angle (ie, alowing a tiny bit of pulse as the first pulse)

Also, can you have an analog, idiot proof way of detecting DC component and shutting everything down in such case?

i
Reply to
Ignoramus15962

"Paul E. Schoen" skrev i en meddelelse news:44b4af97$0$3630$ snipped-for-privacy@news.coretel.net...

Use a triple-winding pulse transformer and a single driver so that either both SCR's are triggered or none are (the one that is reverse biased will not turn on anyway). You *will* use a pulse-train for the entire conduction period rather than bet the farm on a single pulse, right??

You need to switch off at zero current instead of at zero voltage (where the current is largest). Another way is to ramp the conduction period up over quite a few cycles. Probably up, in your case, coming from a trip.

Make sure that you only update settings when your switch is in a safe state!

Maybe you want RS422 or RS485 current loop instead? Noise is an issue, I bet!

Don't - building something does not imply that one should re-invent the stone axe and fire before moving on to crafting the nails and inventing geometry ;-) With C at least you got some *chance* to reuse the design on some other chip later when they obsolete it because it did not make enough sales.

Reply to
Frithiof Andreas Jensen

Ha. Our Microchip rep has been telling me for the past two years that Microchip has only ever discontinued one chip. I guess that is now two.

Reply to
zwsdotcom

Instead of pulses, I have DC available for both gates via DC-DC converters, and trigger a current-regulated signal which maintains gate current continuously. There is also a voltage sensor to make sure the triggered SCR does not have a voltage on it. The gates on this design are separately controlled by two PIC GPIOs.

An SCR by its very nature as a current latch will turn off at (near) zero current. However, if this occurs on an odd number of half-cycles, there will be net DC current into the transformer and it will be magnetized. This causes a high current surge next time voltage is applied. I think this happens when the voltage is applied in the same polarity as the current which caused the magnetization. Possibly applying an initial voltage of the opposite polarity may eliminate the surge.

When the breaker trips on the secondary of the transformer, it interrupts the current flow, although there will be some arcing, even at low voltage. The primary will still have voltage applied, but the current will drop sharply, and possibly even "ring" due to interaction of the primary inductance and the SCR snubber. There may also be a power factor correction capacitor in the circuit, possibly across the output transformer primary, or across the source ahead of the SCR switch, which may also cause some ringing. This might not cause as much magnetization as turning off the SCR after an odd-numbered half-cycle at full current, but I think the magnetization current is independent of the primary current due to secondary current. Thus, counting the voltage half-cycles and maintaining an even number under all conditions may be sufficient.

For circuit breaker testing, it is necessary to have the initial current half-cycle and all subsequent half-cycles as identical as possible. A high initial surge introduces DC offset which will cause premature tripping, and a low initial surge will be below the operating setpoint of the breaker and cause timing errors. Part of the problem is how the measurement circuitry determines the current and time, and it is much easier and more consistent when all the pulses are equal.

Noise is an issue once the SCRs have been triggered and until the breaker has tripped and all transients have gone away. It should not be a problem before that happens. However, I will probably be using the initiate signal to program the phase angle, as it is already connected to a programmable monitor and control instrument. Previously we had implemented phase adjustment there, by eliminating the zero crossing and phase delay circuitry on the SCR board for immediate triggering, but this also removed the inherent noise filtering on the initiate signal, and we saw frequent spurious triggering problems.

Some of the PIC code will need to be in assembler for fast and efficient response to interrupts. Most of the assembly code is at least upward compatible to higher level PICs, although the PIC18F242 (or PIC18F2420) is already one of the highest instruction sets outside of the dsPICs, which are way overkill. Even this may be, but I am familiar with this PIC, and I have a C18 compiler I can use for it. Some of the less time-critical functions I may implement with C, but I already have a lot of PIC code for lower level PICs that should port easily enough. This is for internal use, mostly, and hopefully Microchip will not obsolete a part without offering a compatible upgraded part (as in this case).

Thanks for your input. I'll post test results once I build a simulator and get the tricky details of the code worked out and running.

Paul

Reply to
Paul E. Schoen

See more detailed post elsewhere, but it is possible to make an SCR controller for about $60 or less. I am talking about just the trigger board, and not the hockey puks, of course. However, we purchase a complete assembly with heat sinks and two 1200 A, 1800 V SCRs for something less than $400.

For the gate drives, all you need is an isolated source of about 5-15 VDC at 200 mA or so for each gate, and an optoisolator to turn on a simple current-regulated supply for triggering. I have found this more reliable than single pulses through transformers, and a lot simpler than pulse trains. I am using DC-DC converters that cost about $10-$20 each, but the old design used $5 PCB transformers. The other components were a couple dollars.

The circuitry to provide the phase-delayed turn-on was implemented with a

555 timer that produced an adjustable time delay from each zero crossing, which then turned on the optoisolators to drive both gates. This is for a single phase AC switch for inductive loads, not a phase controller.

For a phase-control application, I also included circuitry for a ramp generator which is reset on each zero crossing, with a comparator to turn on the SCRs at an adjustable phase delay. It was added to the boards, but never fully tested or implemented. We will probably be replacing these boards with the newer design, and the parts are probably not worth salvaging, so you might be able to get a few of them for the cost of shipping. For a three phase application, you would probably need two of these boards and four SCRs, or possibly three and six. The critical thing for phase firing into a transformer is to be sure the phase angles for both positive and negative are identical. Otherwise there will be net DC which will saturate the transformer and cause excess current.

It is not as much a problem if the SCRs are on the secondary, and feeding the load directly. I had the dubious pleasure recently of working on a huge DC breaker test set that had 24 water-cooled 4000 ampere SCR hockey puks which were phase fired to produce up to 35,000 amperes into trip devices. The phase firing was not a good idea for adjusting the current to the lower levels they needed, however, because it consisted of a series of small spikes which were somewhat smoothed by reactors on the output, but they also caused a time delay in current to the load. We wound up using a huge

480 VAC 140 ampere three phase motorized powerstat to adjust the current, and we used the SCRs only to turn the output current on and off.

Paul

Reply to
Paul E. Schoen

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.