The Modular Bench Power Supply ++ - Power Output Stages

That sound great. I look forward to the diagrams and your video.

Thanks John

Well the new op amps arrived today and they work a treat, biggest problem is getting rid of all the heat, I have 20 ohm 10W loads ish, but it still got hot enough to burn my bench...hot hot hot

anyway, these new amps even seem to now allow the 0V to be achieved on the output which is nice and will work up to 44V supply

one additional conformation is that the FET never gets the gate source voltage much above 3V with a 1A load, which I alluded to in the discussion on the video where i described it as a voltage controlled resistor, at 3V it sets the current to an amp and of course with 20 ohm load that's 20V out, i cant remember the exact value but it was about 3V with a 10 ohm load too which was expected (I will have to go back and measure). Of course it wont be exactly the same value but close enough to use as a rough guide

So now to add the current control to the circuit but with a low shunt resistor of about 50mOhms, I will get 500mV for 10Amp load so in order to have a display read correctly I will need a 5:1 resistor divider (Display has 200mV FSD) or change the resistor for my 10mOhm one. This is where the kelvin resistors from Vishay will be used, so the display will represent 20A FSD but it also means I need to have the current op amp fed with a 0 - 200mV signal representing the 0 - 20A current (Worse case put in a 100mOhm resistor and take a hit on 0 - 2A giving 0 - 200mV drop across the sense resistor for full scale on the digital display, my other displays will be able to scale to what ever i need, but he 2 basic ones are stuck with 200mV full scale, ah well

without kelvin measurements, the 200mV across the shunt will cause errors in the output of the power supply as it elevates the negative rail

Ill draw all this up for the weekend and post he details (And make the movie of course)

Hi Peter, It seems the North American late night crew has arrived. I agree I was not seeing a phase reversal on the ISL Op Amp. Perhaps I made too much out of the anomaly that I saw. Part of my problem was definitely the measuring instruments that I was using. I saw a pretty linear 1:5.8 ratio between input voltage and output voltage except as I got low on the input voltage it went to 1:7.8 and finally 1:10.

It is amazing to me how complicated the pursuit of perfection in a power supply is. When you are done with these lessons we will all have a new appreciation for what goes into planning and building such a device. Like you mentioned even wire gauge will have an impact.

Tonight I am looking over dental equipment control circuit boards for interesting chips or interesting circuit arrangements. I am getting so I recognize the chips and their applications from their numbers. From previous learning experiences I have learned that if one keeps concentrating on the details eventually a pattern emerges. The other day I pulled a couple L298 from a board. They are dual Full-Bridge Drivers capable of 2 A. Once free of the board I wired 1/2 of one of them up to a bread board and made a small motor run CW and CCW. I like to collect applications like this and catalogue them. Then, when I do have an application, they are like building blocks that can be used.

Stay warm.  minus 20 C here right now.

John

Been doing more digging and what I was observing was defiantly phase reversal, your chip selected should not be seeing any such affect so your issues may be simply down to a single turn pot proving difficult to set at the low end

I was using a 10 turn pot and as soon as I went below 700mV the output would reverse, I have confirmed this to be phase reversal, it is easy to avoid but does highlight a feature to be avoided in OP amps for this kind of application

If your using a single supply rail and you want to have the output go to zero, you need an op amp with a Common mode input that goes to zero or just below, most will not go the the + supply rail without reversal but then we don't need it too so that's not an issue. However having an op amp that also has an output that goes rail to rail or as close as possible is also desirable

the Intersil you have will drive its output to within 10mV of each rail and its common mode input can go to 0.5V below the negative supply rail and will work up to 40V so perfect selection on those counts

I have these coming from TI TLE2142 and TLE2144, there the closest I can find in a TI chip (So Far), should be here tomorrow with luck

Another interesting observation was when trying to get the output to zero, the pot would not actually allow me to reach 0 volts on the input to the op amp. part of this I'm sure has to do with using a breadboard with thin wires, I was also loading the circuit with a 10ohm load so far exceeding your few hundred milliamps, though I was for the most part keeping the high currents off the bread board, never the less, with no soldered connections, there will be an affect

Again, reason for a slight negative rail to allow a solid 0V on the output, or precision control in the mV ranges, this negative rail may be achieved with a couple of diodes placed at just the right spot, time for more investigating

well be ready for the next instalment in a few days I think, there is also a serious need for an instalment on wire gauge and the importance of appropriate circuit design to keep the power part away from the control side, it does not take much current to generate a few mV drop in the wrong place and affect the output / feedback

Did I mention Kelvin Connections and precision power current sense resistors, I am planning a great video on current sense resistors and have a bunch compliments of Vishay that would be perfect for the current sensing in the power supply, the DC load and the power meter

Regards

Peter

I continued my experiment and got a rough approximation of your design to work with Darlingtons and MOSFETs in both the regulated voltage and regulated current modes. The problem at the low end is definitely related to the Op Amp. I think that while the little Op Amps work quite well, expecting them to operate 0 to 40 volts and still maintain linearity all the way to the rail is perhaps expecting too much. This design is a nice simple structure and as such has points in its favor but if you really want to go as precise as possible your idea to operate the control system in the 12 to 15 volt range and then use transistors and other architecture to leverage the output voltage to 40 volts probably has a lot of merit. You also mentioned that some of the expensive supplies use relays to maintain a closer delta V between Output voltage and Input voltage so the outputs don't overheat as much. Perhaps the relays could switch control structure as well to better maintain linearity too. Thanks for taking the time to look into my observations. I eagerly await your next presentation.

John