Electronics > Projects, Designs, and Technical Stuff
Bench Power Supply Design - Idea Stage
(1/1)
Gandalf_Sr:
I'm thinking of a triple, isolated, linear, output design like the Rigol DP832 (although the DP832 shares a ground for Ch2&3).  It will have a small color LCD screen and keypad run from a microcontroller, probably a PsoC.  If we start off with a transformer with isolated 32V 5A secondaries, we can get the main supply and, ideally, I'd have a 3rd secondary winding that gave 8V at 3A.  There could be a small 5V isolated supply module to steal power for the microcontroller & display from the 3rd supply; it would generate the signals to set voltage and current via digital isolators; also via digital isolators, V & A readings would pass from separate output section PCBs.

The bit that I'm looking for ideas on is the power section PCBs. I need the ability to set voltage at mV levels. I think that I can use a 16 bit DAC to generate the target voltage and the level could be set by serial (I2C) data passed over the digital isolator.  Anybody got ideas or come across anything that would serve as the per-channel  control board?  It will likely need its own microcontroller to implement limits and so on.
excitedbox:
Have you seen dave's series on designing a power supply?

AliExpress has some good transformer suppliers that will even make custom transformers for really cheap. There is also a really good video on making your own toroidal transformer on youtube if you are interested in going that route I will look for it.

ST makes a few great micros for power supplies and also diodes and rectifiers. I recently got some samples sent to me for free.
I got 5 STTH8R03DJF-TR - 300 V, 8 A PowerFLAT High Efficiency Ultrafast Diodes and a STM32F303ZET6 Micro. Some BLDC controllers with integrated STM32 some ADCs and L6472HTR motor drivers. About $30 worth of parts in all which was really nice of them. Their micros are def one of the best bang for buck and it also gets you access to their GUI library if you don“t want to use LittleVGL or buying a license for one.

I am really more interested in the cheap switch mode supplies than linears because they are so cheap but the controls are all still as if they were designed in the 80s with knobs and 7 segment displays but micros and LCDs have gotten so cheap that you can replace the 7 segment displays with a touch screen for about $2-3 more. There are plenty of linear supplies in the $200-300 range with modern LCD UIs but I only know of 1 cheap $60-80 programmable switch mode supply with LCD and it looks kinda low quality.

I would love a power supply with knobs for quick voltage and current adjustments and a touchscreen when you need to use the programmable features. It would be cool to have an expandable/modular one like that mini unit but in the normal form factor. Like a little box with a screen that has 1 channel and you clip extra channels onto it. That would also allow you to clip different types of modules on such as signal gen or load. If you used an ESP32 they could even be placed around the lab and be controlled over bluetooth or wifi.

I drew a quick sketch.  :-DD
Gandalf_Sr:
Thanks for the suggestions, particularly the tip about toroidial transformers from AliExpress.

I've worked with STM32s before and they are OK but the PSoCs are even easier to work with once you get used to the PSoC Creator interface although Cypress - now bought by Infineon. As an aside, I believe that Cypress has shot itself in the foot by moving to an IDE called MODUS that is much harder to work with.

From a PSU topology perspective, I started thinking the standard transformer, bridge rectifier, followed by linear regulator setup but, although linear is the lowest noise, there are a number of issues with this from the efficiency viewpoint which means large thermal stresses, heatsinks, and fans.

I own an Exetech switched mode supply which is small but the output is so noisy that some of my circuits wouldn't run properly on it and it's now relegated to run my under shelf LED lighting.  Linear Technology have a reference design that puts a switching supply in front of paralleled LDOs with a neat setup to keep the SMPS output 1.7V higher than the LDO output but it must've been the result of a competition to see how many $$$ LTC parts they could get in one design plus it needs a fixed 40V input.

I still like the idea of a SMPS front end with a Linear final stage, the question is, can I achieve the tracking I need and drop from 115 down to as low as 3V with an isolated mains switcher?  Probably not at the currents I want.  So my compromise idea is that I have 1 (or 2) big SMPS(s) that drop down to a noisy 50V 12A supply, and then use it to feed small isolated switchers on each output that pre-regulate down to the desired output voltage plus 1 or 2 volts; this is my present strategy, anyone want to comment on that?
H713:
Try Antek for a toroidal transformer. Antek doesn't fill the center of their toroids with epoxy, so it is relatively trivial to add that additional 8V winding you were interested in.

You have a few options to combat your efficiency concerns. You never specified how much current you want to source, nor a maximum voltage. Assuming a 4A output with a 40V DC bus, at 5V you'd have about 150W of dissipation. That's a lot, but it can be managed with a reasonably large heatsink and cooling fans.

If you implemented tap switching on the transformer, that would improve the efficiency at low output voltages. If you don't want to wind a custom transformer, you can still implement this system. An AS-2218 has a pair of 120V windings on the primary and a pair of 18V windings on the secondary- this means that with some relay switching it can be configured to output 9, 18 or 36 VAC. This is very simple to implement, however, you will need a transformer for each of your 3 power supplies. If you're shooting for a 3A output, you may be able to use a 100VA transformer depending on the duty cycle. 

There's another way that might be very interesting, however. The primary of the power transformer could be controlled by a variac driven by a stepper motor. You could set it to auto-track with the output to minimize the drop across the series pass transistors. I have seen this used in extremely large magnet power supplies and it works quite well. It can be quite efficient. The advantage of this over a buck converter is that you don't have to worry about it introducing high-frequency noise.

There are a few other ways to do this. First, you could use a thyristor-based "dimmer" circuit on the primary of the transformer. More than a few large power supplies use this technique, though I don't see much advantage of using it over a buck converter.

You could use a buck converter as a pre-regulator, but those are a few interesting options you may not have considered.

Lastly, and while this is generally frowned upon, you could simply accept the idea of lousy efficiency, perhaps implement the most basic of transformer tap switching, and live with 60 watts of dissipation. A few TIP35C or (TIP142 if you want a darlington) will be able to handle it without too much difficulty, and the heatsink doesn't have to be all that large if forced air cooling is employed. Efficiency isn't usually that critical with a bench power supply, especially since you probably won't be pushing it hard for long periods of time. Cooling fans can be kept off until the heatsink reaches a certain temperature (say, maybe 60 degrees celsius).



Navigation
Message Index
There was an error while thanking
Thanking...

Go to full version
Powered by SMFPacks Advanced Attachments Uploader Mod