My bench power supply

[Aodhan145]

Hi, I have been working on a bench power supply for a while and I was wandering if anybody could review the schematics for me. I built it around Dave's uSupply (because it is one of the best explained designs) but I wanted higher current. I realised immediately after that there are limitations with the design due to the 1 ohm current shunt meaning 1V loss for every amp. Then I realised this would not be a problem as I only need high current at low voltages. I modified it to have current up to 8A voltage to 30V and a negative voltage rail but the negative voltage rail can only do 1.5 A. Its controls are on a separate board and its controlled via RS232. The inputs are 32V rectified from a transformer and -32V from another transformer. Sorry if i have made any basic mistakes I only started electronics recently.
Thanks in advance for any help.

[dom0]

Hi, I have been working on a bench power supply for a while and I was wandering if anybody could review the schematics for me.
...
Sorry if i have made any basic mistakes I only started electronics recently.

Lab-grade supplies from scratch are not easy to design, especially not as a beginner. The µSupply design aims at a very specific niche and isn't a good general-purpose design, especially not for mains/transformer powered gear.

The generally as best conceived topology for pure linear lab supplies is the "HP-style", i.e. a NPN common-collector output stage which is regulated by error amps floating on the positive output. The schematic and theory of operation are explained in most manuals of the small single channel HP lab supplies. There is also a "modernized" version of it here (German comments).

30 V at 8 A is quite a lot and needs a big transformer (~400 VA or more). If you only need high currents at low loads this is possible with multiple taps (although tapped windings with changing current ability are only available in custom-manufactured transformers AFAIK, not doing that would add significant unused weight and cost) or a switching pre-regulator. The latter will introduce significant circuit complexity, likely not up to your abilities. Another option could be to modify an existing switching regulator, or a SMPS. Again, the latter requires care and attention to correctly analyze the circuit of the SMPS in order to find out how to control it. Also, most COTS units do not have auto-ranging ability (higher currents at lower voltages, as long as a maximum power level is not tripped), but rather fixed current limits.

From my experience it seems likely that two separate supplies would better suit you. E.g. a low voltage high current one (popular choice: 8 V / 5..10 A) and another medium voltage medium current (the popular choice being 25..30 V and 2..3 A) one. Construction will be significantly easier. But generally speaking a good lab supply as a first project is hard to do. Consider buying at least one working lab supply.

[Aodhan145]

Thanks I will take a look into the pure linear design. I started electronics around 6 months ago but I have been teaching my self so I am still a beginner I have likely missed fundamental pieces of electronics. I bought a Temna power supply recently from toploser (32V 5A dual channel) so I have one I was just wanting to build one to learn more about it and know how things work. I have a transformer that is gives 30V at 10A so I have a powerful enough transformer (took out of a power supply that has a broken variac) but it is not center tapped so I got another 32V 3A transformer which I was going to use as a negative rail. Would my design work? I spent a lot of time on it and I dont want to just scrap the design I already routed the PCB and I it was only my third PCB and my other PCB's were very simple so it took a lot of time to route this one. Would it be worth while scrapping it as I have spent 3 months on it? I was hoping on building it in  August.

[prasimix]

I spent a lot of time on it and I dont want to just scrap the design I already routed the PCB and I it was only my third PCB and my other PCB's were very simple so it took a lot of time to route this one. Would it be worth while scrapping it as I have spent 3 months on it? I was hoping on building it in  August.

I'd like to say you can count on another half a year, year if you are quick learner and can dedicate enough time for such project.  Maybe I missed something but just a quick look at PCB says that you can expect a real trouble. Such small heatsink on LT3080 without pre-regulator looks like a bad joke. I also have no idea how you are going to provide 10A with LT3080? By paralleling 6 or 8 of them?

Dom0 comments are really useful, I'd like to recommend you to read it a few more times. Also a nice source of real stuff is Liv's PSU (at the end I decided to adopt his design for post-regulator in my PSU). A good discussion can be found on void_error's thread.

Using spice simulator could spare you a lots of time but also generate a lots of trouble if you didn't make a good model. Also breadboarding is an REAL option especially if you are not working with switching (as a pre-regulator stage or as a complete solution). You can check your ideas and results from spice and spare a lots of money in ordering PCBs that can become instantly obsolete (before its even arrived from manufacturer).

Good luck! 

[Kleinstein]

A lab. supply looks easy at first, but gets quite tricky if you go into details and want it to be good. Even some of the commercial units still have some flaws - a good supply should not have. Especially the cheap ones are usually not able to withstand a short for extrended time or at high temperatures.  They just allow to much power for cost reasons.

Buiding a supply, one can learn a lot. For most of it one should start small (e.g. 30 V and 1 A) so the smoke is less expensive and not so much of it. Going to a larger unit is than relatively easy.

The LT3080 is essentially only usefull at small voltage drop: So either only very low voltage (e.g. 0-5 V) or with a preregulator.

Using some form of Spice is the way to go today. Its very hard to get it right without it.

[macboy]

I spent a lot of time on it and I dont want to just scrap the design I already routed the PCB and I it was only my third PCB and my other PCB's were very simple so it took a lot of time to route this one. Would it be worth while scrapping it as I have spent 3 months on it? I was hoping on building it in  August.

I'd like to say you can count on another half a year, year if you are quick learner and can dedicate enough time for such project.  Maybe I missed something but just a quick look at PCB says that you can expect a real trouble. Such small heatsink on LT3080 without pre-regulator looks like a bad joke. I also have no idea how you are going to provide 10A with LT3080? By paralleling 6 or 8 of them?

Dom0 comments are really useful, I'd like to recommend you to read it a few more times. Also a nice source of real stuff is Liv's PSU (at the end I decided to adopt his design for post-regulator in my PSU). A good discussion can be found on void_error's thread.

Using spice simulator could spare you a lots of time but also generate a lots of trouble if you didn't make a good model. Also breadboarding is an REAL option especially if you are not working with switching (as a pre-regulator stage or as a complete solution). You can check your ideas and results from spice and spare a lots of money in ordering PCBs that can become instantly obsolete (before its even arrived from manufacturer).

Good luck! 

You missed that the LM3080 is controlling an external PNP pass transistor that handles the grunt work. Still, one extra pass transistor is NOT going to cut it for 10 A output. I'd suggest 5 or even more. A 32 VAC transformer is going to give around 45 V at the collector of those transistors. At 10 A and low output voltage, you need to dissipate over 400 Watts. A single transistor will go up in smoke no matter how well heatsinked. I'd suggest maybe 75 W or less each, and that is with big TO-3P/TO-247 transistors. The LM3080 with that little PC board mounted heatsink may even struggle just driving those 5+ transistors without over-heating. Strongly consider the heat generated at low output voltages. It is not something to ignore. This is why all high-current lab supplies use some form of pre-regulation: a tracking SMPS, switching of multiple transformer taps, or phase control of the AC input. 

[Kleinstein]

Using a small regulator IC (LT3080 or LM317) boosted by an external PNP is prone to istability and oscillytions. This may be OK for a well behaved load, but generally not for a lab supply.  The real difficulty in designing a lab supply is having it stable / non oscillating at essentially any reasonable load.

The other not so easy part is finding parts that can take the power - here its much about reading a datasheet, ignoring the mareting part on the first page.

So sorry the circuit from above is no good at all.

Doing the layout is something for the very end, after the Spice model and maybe a simplified breadboard Version is working. Its also good idea to ask here, before doing the layout.