Electronics Tutorial - Multi-Output SMPS

[@oriole8789]

Excellent comprehensive video! To contribute some info - there are quite a few relatively inexpensive commercial isolated DC-DC converters with (I suspect) similar topologies. My personal preference for building custom electronics, is to buy those off the shelf parts and then simply filter the outputs as I need. It's a tradeoff of my time vs cost for clients, and in most cases the time savings are worthwhile. Some examples - for higher power, Meanwell DKA30B-15 (18-36V in, isolated reg. ±15V out, at 1A per output, 30W total, $29 USD). For medium power, Meanwell NSD10-12D15 (9.8-36V in, isolated reg. ±15V out, 330mA per output, 10W total, $12.24 USD). For low power (opamps etc), RECOM RFM-0505S (5V in, isolated unreg. 5V out, 200mA, $1.40 USD), RECOM ROE-0515S (5V in, isolated unreg. ±15V out, 66mA per output, $2.92 USD). These RECOM units are tiny SIP-4 packages with 2.54mm pin spacings. The prices are from Mouser in single units. Of course you can do better cost-wise if you build them yourself, but your time is worth something too. Higher-power designs require a very careful part selection and real world testing for stable operation over a wide range of conditions. Lower-power designs are easier to build from scratch, but the commercial parts of this kind are getting attractively cheap, depending on which market you build your products for. For extreme power density, it's hard to beat Vicor. For example: 384V to isolated 12V at 1500W (!) in a 6cm x 2.5cm package, 97.4% peak efficiency, $278. Expensive but dat power density haha. Thanks for the video! SMPS design in general is really fun. -Nick

[@FesZElectronics]

Hello Nick! Well I have to agree, if the supply specs are right (input/output voltage and current), it is way less time consuming to just buy a ready made circuit than to design one. And considering the prices of some of the parts out there, if you only build a few circuits, the complete supply module will be cheaper than buying the components to build the same thing from scratch in most cases. Regarding the last example you mentioned (the 1500W 97.4% 6x2.5 cm supply) I would be very skeptical about this performance claim. Based on the efficiency, which is very good though, you still have to dissipate 40W somehow. And 97.4% is the peak efficiency, not necessarily the efficiency at full load, that is usually less. I mean it will definitely work for 5s but not for long periods of time unless it also uses a huge radiator strapped to it, so in the end, with the complete cooling unit, you still end up with something quite big, so paying a lot for the tiny package might not be worth it...

[@oriole8789]

@@FesZElectronics Hi FesZ, I focused on the off the shelf aspect in my reply just because it's kind of interesting how this market is developing. Understanding the theory behind all this stuff is incredibly important and you're one of the few people here on KZbin who goes as far as you do to build a comprehensive picture. LTSpice ftw haha. Regarding the Vicor converter, take a look at a datasheet about the type of part I was talking about. Google "BCM6123 pdf". That's a 1.7kW part (continuous). Same tiny size. 97.3% efficiency AT FULL LOAD (35A) haha. These are industrial parts meant for large scale deployment where efficiency is everything. Heatsinking isn't an issue because they're often connected to various enclosures or metal blocks/chassis that house other switching/protection/management devices. There are other very interesting design considerations for that part.. according to the datasheet, external filtering isn't necessary for these, but rather: "To take full advantage of the BCM’s dynamic response, the impedance presented to its primary terminals must be low from DC to approximately 5MHz ... Given the wide bandwidth of the module, the source response is generally the limiting factor in the overall system response." That would be pretty non-trivial to implement haha. The datasheet provides a helpful section on calculating thermals as well, you can take a look. Admittedly I'm a fan of Vicor and have used their products countless times.. really clever designs and probably the best engineering support I've ever received. They've killed pretty much all of their less expensive product lines to focus on this "lego" approach of building custom power solutions using various blocks. I'll give you a random example though - I recently needed a compact, silent (can't have fan) and isolated 120V AC > 24V 400W PSU for a project, and I was able to grab a cheap Vicor unit on the used market in a 23.5x12.5x3.5cm package. 80-90% efficiency, 180mV ripple. They don't make these anymore but I loved them haha. The main DC-DC blocks that they use are always unserviceable (being epoxied parts), but they've been a lot more reliable to me compared to various industrial DIN-rail mounted PSUs, a lot of which have surprisingly high failure rates. Typically non-isolated as well. I hate unserviceable parts but I'll give Vicor a pass for what they do.

[@oriole8789]

To kind of add something about these types of DC-DC modules - they do have external capacitance limits, which is potentially an issue. If you're driving a highly capacitive load, you'd require soft-start circuits. For that 1.7kW module, the output capacitance limit is only 100uF, which is like, nothing haha. That's somewhat mitigated with parallel operation, but still problematic for some applications. That said, the extremely high efficiency when the system is running, is worth the extra cost of a soft-start circuit if needed. Plus, their primary function is basically pre-regulation, so that a 48V to regulated PSU can follow this particular part. 380V DC distribution is becoming common industrially due to low loss.. which is important when there is large physical space to cover. Anyhow I really hope you keep exploring this topic. Very interesting and relevant at this point in time!

[@FesZElectronics]

@whatlions To be honest, I haven't worked with isolated supplies that much, I mean I built 1 or 2 but that's it. I mostly worked with non-isolated circuits. But you do have quite a lot of interesting insight on this topic, and I will look into this in more detail in the future, hopefully. Regarding the output capacitance and softstart; from my experience, the output capacitor value has an influence on the power-supplies stability, and needs to be taken into account when choosing the compensation components; very large capacitors can cause the supply to be unstable, so soft start might hide an instability but that doesn't mean it can't appear. Soft start on the other hand is useful to limit inrush current, the current drawn when starting up the supply. This current is of curse larger if the output capacitor is large but the SMPS should have extra over-current limit circuitry to prevent any damage. Also soft start is usually built into SMPS IC's, by proving an external capacitor with which to set the startup time. Though if the supply is sold as a module, you probably don't have access to this. I did observe capacitor problems with supply's powered from the grid but on the input capacitors. I mean if the input capacitor is very large, the inrush when plugging it in will be huge, and can damage the input rectifier or input filters. Its the reason why you might notice a spark happening when you plug a laptop charger or any other high power supply into the power grid. Here an input soft start might be helpfull.

[@oriole8789]

​@@FesZElectronics "so soft start might hide an instability but that doesn't mean it can't appear" - very good point! High output capacitance will cause lag between load demand and PSU response. It's kind of like introducing a delay in the feedback of any control loop (like in a PID controller). It will lead to instability. There are other issues too, all of which you're likely aware of since you sound knowledgeable on this subject. If the output capacitance is very large, it would interfere with the SMPS's overcurrent protection circuitry since you could have some intermittent sparking short circuit which would draw massive current (from the capacitors) in bursts, but the feed from the SMPS may never reach OCP limits, causing havoc for downstream components and user safety. To add to the "sparking when plugging in laptop" thing, you'll notice that the sparking isn't consistent. And that's of course dependent on what the phase of the sine wave happens to be. If you plug it in close to the zero-crossing point, then no spark haha. Nice bit of lottery. XD The modern trends in SMPS designs favor high bandwidth designs, which reduces the need for huge capacitive buffers. A lot of output capacitors will naturally reduce the overall efficiency of the system, since you would waste energy if the system needs to be frequently power cycled. I try to use the minimum necessary output capacitance in my designs, which is often necessary since I use a lot of OCP components like AUIR3315's which lets PSU outputs be short-circuited without even a spark, since OCP reacts within a few microseconds. If you're dealing with super high-current designs, soft-start systems become VERY expensive. NTC thermistors aren't an option for very high current, and often you can't even use regular wirewound resistors, since the extreme inrush current can cause the internal wires to simply break or the resistors to crack. Instead, you have to use "ceramic composition" resistors, in which the actual ceramic is mixed with conductive materials, which makes the entire body of the resistor conduct energy instead of just a wound wire. Those types of resistors do well under extreme pulsed loads, and also have uses for high frequency systems (snubber networks in high kilowatt/megawatt variable frequency drives etc), due to having extremely low inductance (wirewound resistors being the opposite). And of course they are very expensive haha... I'm kind of a fan of high power electronics, even though I don't really work with it directly. I just think there's a lot of good knowledge in that area which could be very applicable to lower power designs. High power stuff really exposes the underlying physics - the stuff we typically ignore "because we can" haha.

[@2bit661]

great lesson!

[@FesZElectronics]

I'm happy you enjoyed it!

[@merden102]

Hi. Thanks for Ltspice videos. I have a question. How do I measure phase differance(Current and Voltage) so power factor in LTSPice?

[@FesZElectronics]

Hello! Well the easiest way is to perform AC simulations rather than transient. There you can plot any parameter in the circuit in reference to the input signal, and you will get both the phase difference and amplitude ratio in reference to the input. On the other hand if you really want to perform measurements on transient simulation signals, it involves a bit more complex mathematics - I presented an example like this in the Loop measurements episode ( kzbin.info/www/bejne/jmXcknWtlrRknc0 ) where the amplitude ratio and phase difference between 2 signals was calculated, maybe that can be adapted to help here, but I'm not sure.

[@merden102]

​@@FesZElectronics thanks for solution proposal but I wonder Have a Ltspice function for a power factor measurement?

[@FesZElectronics]

Unfortunately, I do not know if there is a dedicated function for this. If I do find something though, I will let you know.

[@thezodiace7399]

Can we get a boosted differential power supply just by using a CP configuration?

[@FesZElectronics]

Well, the main advantage of using the dedicated boost stage is that it can provide a precise output voltage at any ratio compared to the input (the CP (charge pump) will only provide integer multiples of the input minus the voltage drops on the diodes) and you only need 1 switch (the low side transistor) - for the CP you need a push-pull stage. You should be able to build the differential supply only using charge pumps but the same issues as mentioned above are present - you can't obtain any voltage.

[@thezodiace7399]

@@FesZElectronics Thank you for the explanation.