| Electronics > Projects, Designs, and Technical Stuff |
| Reliable Low Cost Method to Convert 24VAC to 24VDC |
| (1/3) > >> |
| GeorgeIoak:
I'm working on a redesign of a HVAC controller board that was designed maybe 20 years ago. Due to cost (and potentially reliability) I'd like to use a LDO and the ST Micro L7824CDT-TR is a nice low cost fixed 24VDC output. * To convert the 24VAC to DC I plan on using a bridge rectifier with smoothing capacitor so I'm expecting about 34VDC (24 * sqrt(2) = 34V) going into the LDO. * The 24VDC rail will be powering a bank of relays being driven by a ULN2803 and a few LEDs for status. Only 3 relays will be powered at the same time and I'm estimating that to draw ~15mA each * Adding current up we're looking at ~50mA for the relays and less than that for the LEDs but round up to 100mA * After the LDO with be another LDO for a 3.3V rail powering a STM32 and some optoisolators on the input. Best guess right now might be another 100mA So I'm looking at the power dissipation for the LDO to be at least 1W (10V * 100mA) which might not be too bad for a D2PAK but ambient in a home's attic can get fairly hot and I'm looking for long term reliability. Switching to a TO-220 with heat sink would solve the "problem" but that drives up cost. I'm thinking of perhaps dropping the input voltage to LDO with maybe a diode. I've seen designs which also use a TVS on the input so perhaps one of those could be used which is rated at 28V? I haven't used a TVS before so perhaps it's not the right part for what I'm trying to achieve. I hoping that someone here has done industrial designs and can help me out with ideas. Part II of my question is on the protection across the relays (which drive 24VAC motors which pull ~385mA). The reference board I have used a Panasonic ERZV07D560 which is a 56V Varistor but in a TH package. I'm looking for a protection device to place across the relays that is a SMD part. I'm not so sure on how to determine the appropriate specifications to look for. Thanks in Advance! |
| TimNJ:
You don't want to use a TVS in parallel with the input. A TVS (or a regular zener) used as a clamp needs to "work against" the input impedance of the source. Think about it: If the source resistance is 0, then no matter how much current the TVS "pulls" it cannot change the voltage across itself. In general, I wouldn't use a TVS in parallel for continuous operation. The TVS will have to pull a lot of current to maintain 28V (for example) across itself, by dropping voltage across the series input resistance. On the other hand, using a TVS in series may work okay. (A normal zener might have a more tightly controlled Zener knee voltage). I suppose you could use something like a 4.3V or 5.6V zener. Maybe an SMA size. Have you experimented with PCB copper heatsinking? Double sided board with thermal vias? |
| GeorgeIoak:
I was thinking of using a diode to get a drop before the input to the LDO. The RS1M has a 1.3V Vf at 1A. I was thinking of a Zener as well but I'm not sure why you stated a 4.3V or 5.6V. Wouldn't I want something more like a 28V that's placed right before the LDO. The Zener would then bring the ~34VDC from the diode bridge down to 28V and the LDO would only have to drop 28V down to 24V. I saw something like this in 1 design where they took 1 leg of the 24VAC, ran that through a fuse wih a 56V Varistor in parallel across the fuse. After that was a TVS before the input to the regulator. Something like this: |
| ajb:
TVS diodes are designed to handle brief surges (hence the Transient in the name), so while they boast high peak power ratings, the acceptable continuous power dissipation is going to be much much lower (and you might not even see a rating for it in the datasheet). This is related to the fact that they tend to come in packages with fairly poor junction to ambient thermal resistance. A better bet is probably to use a second DPAK LDO in series to spread out the power dissipation. You could also use split the load into two separate +24VDC rails with their own LDOs, which would accomplish the same thing, but it may not be realistic depending on your load characteristics. Alternatively, how close to 24VDC do you actually need your relay supply to be? Relays tend to tolerate a fairly wide range of coil voltages, so even if you can't go unregulated, you may be able to bump up the rail to 28V or something--obviously the coil current will increase according to its DC resistance, but this should still result in a net reduction in LDO dissipation (coil current goes up by 17%, but LDO series voltage drops by 40%). You could also implement peak-and-hold driving by simply PWMing the coil after it pulls in to reduce the steady state current consumption, but that's probably not necessary if the coil is only pulling 15mA. |
| TimNJ:
--- Quote from: GeorgeIoak on April 17, 2019, 09:16:23 pm ---I was thinking of using a diode to get a drop before the input to the LDO. The RS1M has a 1.3V Vf at 1A. I was thinking of a Zener as well but I'm not sure why you stated a 4.3V or 5.6V. Wouldn't I want something more like a 28V that's placed right before the LDO. The Zener would then bring the ~34VDC from the diode bridge down to 28V and the LDO would only have to drop 28V down to 24V. I saw something like this in 1 design where they took 1 leg of the 24VAC, ran that through a fuse wih a 56V Varistor in parallel across the fuse. After that was a TVS before the input to the regulator. Something like this: --- End quote --- Typically, a TVS diode is used to clamp a momentary spike. All voltage sources have some series output resistance. Think Thevenin model. A TVS can't work unless there is ~something~ in between the TVS and the source. That something is a resistor. If the TVS and source were connected directly in parallel (perfect conductor), then how could the TVS do any clamping? The voltage across the TVS would have to be exactly the voltage of the source. |
| Navigation |
| Message Index |
| Next page |