| Electronics > Projects, Designs, and Technical Stuff |
| GDT surge life question. |
| << < (4/4) |
| riccardo.pittini:
I designed the surge protection for outdoor /street light. We had a test running continuously and it took 5kV/2.5kA (1.2/50us standard surge) about 2500 times (then actually we stopped the test but it could have taken more-not any sign of visible degradation) It was embedded as part of the EMC filter based on a big MOV + GDT +small MOV. Degradation is mostly related to energy and current. As you have 500A, and degradation is not linear. I think it's feasible. It takes a detailed analysis (simulations and test) of your input circuitry. |
| Electr0nicus:
Yes it is a ship based system. Actually the hardware is mounted on deck. I can not disclose the exact application as this falls under a NDA with the customer. But the electronics, which is powered, is a heating application with integrated temperature regulation. So the temperature regulation is the sensitive hardware. The power intake is 4000W from a delta 115VAC system. I would consider using a multiple stage overvoltage protection, consisting of GTDs, MOVs and TVSs and add a bit of series resistance after each stage. The power components (rectifiers, MOSFETs etc.) are designed to withstand a minimum of 650VDC. So the protection circuitry has some room to fluctuate, it doesn't need to clamp the overvoltage to a exact narrow band value. Cheers. |
| joeqsmith:
I would assume the electronics that you need to protect draws very little current. The drivers are in series with a resistive load and the load itself could very well handle the surge. It may not be that big of a deal. Do they use the standard 1 minute interval? |
| SiliconWizard:
--- Quote from: Electr0nicus on October 26, 2018, 08:26:28 am ---Yes it is a ship based system. --- End quote --- I was thinking of either that or an offshore platform. --- Quote from: Electr0nicus on October 26, 2018, 08:26:28 am ---I would consider using a multiple stage overvoltage protection, consisting of GTDs, MOVs and TVSs and add a bit of series resistance after each stage. --- End quote --- Probably your best bet. I would tend to avoid MOVs, they are great but I don't like their failure modes for very long-term, maintenance-free devices. >:D Hopefully you have means of testing this in your lab so you can iterate and don't have to wait till your design is fully completed and the client tests it, and it fails. |
| T3sl4co1l:
Is this device subject to possible direct strikes, or is it shielded and supplied in such a way that the customer can guarantee the surges are maximum (not typical) the amplitude and duration given? Example: conventional mains distribution combines lightning arrestors on distribution lines, with transformers, and nominal air gaps, that ensures end-point surge (at 120/240V depending) has a maximum around 5kV, with the traditional 1.5/2.5kV surge levels being most common. Contrary example: telecom lines are typically not clamped periodically, so any induced or direct lightning surges are carried on the lines. Upside, the lines don't carry much power, the resistance is high (typically 20 ohms is used in the test). The lack of clamps means the surge is longer, hence the 10/1000us waveform. The catch is, you can construct a system that degrades more or less gracefully with increasing duration or amplitude. There may be useful gains in reliability or size/cost for a less graceful system. Example, the series-inductor, diode-shunt-capacitor method I mentioned above: ~unlimited life within ratings, but the clamp voltage will greatly exceed the capacitor's if the surge is too long, and the diode (and clamp voltage) ratings will be exceeded if the amplitude is too high. Or both. In short, the distribution of surge amplitude/duration matters on this level! Tim |
| Navigation |
| Message Index |
| Previous page |