| Electronics > Projects, Designs, and Technical Stuff |
| My Lab PSU project - open source! |
| << < (3/4) > >> |
| nemail2:
--- Quote from: Kleinstein on November 17, 2018, 03:12:52 pm ---It should be possible to get a faster current sense amplifier with a separate PNP (or P FET) and OP. --- End quote --- do you have an example for this? --- Quote from: Kleinstein on November 17, 2018, 03:12:52 pm ---However besides the current sense amplifier, there is also the compensation at the OP that can take quite some time, as initially the OPs output would be all the way up and the 1 nF capacitor correspondingly charged quite a lot. There are a few possible ways to improve on this: A diode in parallel to R7. The current read point should move to the other side of R7 anyway. --- End quote --- what would the diode do? this is copied 1:1 from daves µsupply and I guess he put R7 in to slow things down a bit. what do you mean by "the current read point should move to the other side of R7 anyway" and how would that be achieved using the parallel diode? --- Quote from: Kleinstein on November 17, 2018, 03:12:52 pm ---It may also be possible to move C5 to the other side of D1 - this might need some extra small capacitor directly at the OP. --- End quote --- huh? how would that look like? --- Quote from: Kleinstein on November 17, 2018, 03:12:52 pm ---Getting much better current precision without a larger change is difficult, as the minimum load current can not be separated at the collector side. The more normal way would be to have the shunt on the low side and than change the reference part a little. This might still have the reference current flowing through the shunt, but this current is at least more constant. --- End quote --- understood. the issue is that this is my first "more complex" PSU design and I'm happy that I kind of understood most of the aspects here. Low side current shunt measuring is something I have absolutely no idea of and I'd lack of any point to start from. --- Quote from: Kleinstein on November 17, 2018, 03:12:52 pm ---For testing a very large (e.g. 5000 µF) low ESR electrolytic is also a difficult case for the regulator. Even some commercial supplies don't like this very much. --- End quote --- What about a 5F supercap? :-D --- Quote from: Kleinstein on November 17, 2018, 03:12:52 pm ---The transistor T4 might help to get a slightly higher maximum output voltage as it allows extra current from the OP, when at a high voltage. With the parallel resistor there should be no problem with the transistor if the R1 value fits (e.g. R1 > R24 * Hfe). With R24 that small one might have to make sure the OP for the current limit can bring the voltage all the way down and has enough driving power, as the 1 K resistor might deliver up to about 15 mA. So it might take a higher value for R24 when the voltage gets higher. --- End quote --- So far all good, getting about 13V output voltage, even under full load, thats even over spec. I have sized R24 accordingly so enough power remains to bring the voltage down in CC mode. --- Quote from: Kleinstein on November 17, 2018, 03:12:52 pm ---For better load sharing one could combine T3 and T7 to a single transistor and only have the TIP3055 in parallel. Getting the transistors from the same batch might be good enough. --- End quote --- I wasn't sure about this so I decided to put two Darlingtons in parallel. I could leave both BD139 footprints in and connect one of them to both of the TIP3055 and put in a jumper to disconnect one of the TIP3055 from the BD139 in my next iteration of the board so I could try both variants without having to reproduce the board if it doesn't work well. You won't hear any objections from me if I can save one of those expensive puppies ;-) |
| Kleinstein:
The BD139 should not be so expensive. With not so much power a BD139 should be sufficient to drive 2 of the TIP3055. A diode parallel to R7 would speed up the phase when die current is significant higher than normal. The compensation would be back to normal, slow, when closer. So this should not have much effect on stability. It it more like a simple bodge type modification to speed up the reaction in case of a short, it won't save a LED, but it's only a diode (e.g. 1N4148) to add. Reading the actual current should be at the 1 K resistor and thus the other end of R7 anyway, even in the otherwise unmodified circuit. At the OPs input the current reading would be off during transients. The diode would change that error, but the current reading is not very accurate anyway. With only 15 V supply for the OP, T4 might not be needed. It might give some 0.5 V higher maximum output voltage at high current. Going to low side current measurement (e.g. a shunt on the negative terminal )would be a major change in the circuit. The max4080 is more like a low power version. One can build the max4080 internal circuit with a P-FET and OP with a higher speed. Something like 3-5 times the speed should be relatively easy. |
| nemail2:
ok just a quick update plus photos :-) i have to work on the suggestions and try different things, in the meantime enjoy a photo of the actually not so bad measurement and control performance :-) please be aware that I didn't have the correct high precision low ppm resistors for all the adc voltage dividers and for the opamp feedback loops so i had to put in what i found at home (paralleling and serialing resistor values) so precision should be better with the actual high precision parts. and the second photo, a picture of the board, in the middle of soldering. takes me 3 hours to solder that thing every time. i have to get a fume how do you call it ... purifier? https://owncloud.ne-mail.net/s/S3N7zyTjbtzSXcf https://owncloud.ne-mail.net/s/Gxq6jTW28g2wwTN |
| wasyoungonce:
never mind... |
| nemail2:
Hi ok, got some time again now. --- Quote from: Kleinstein on November 17, 2018, 10:37:04 pm ---The BD139 should not be so expensive. With not so much power a BD139 should be sufficient to drive 2 of the TIP3055. --- End quote --- True, it is about EUR 0,50. I had in mind it was more expensive. I will leave both footprints for both BD139 in but connect both emitters through a jumper so I can try it out without having to redo the PCB if it doesn't work. --- Quote from: Kleinstein on November 17, 2018, 10:37:04 pm ---A diode parallel to R7 would speed up the phase when die current is significant higher than normal. The compensation would be back to normal, slow, when closer. So this should not have much effect on stability. It it more like a simple bodge type modification to speed up the reaction in case of a short, it won't save a LED, but it's only a diode (e.g. 1N4148) to add. --- End quote --- I think I'm getting it now.. maybe... As soon as the forward voltage is reached, the diode conducts without any resistance (in opposite to R7 which will always have 1k), am I correct? Forward voltage is about 0.7V so that would be about 1.4A with the MAX4080F @ 0.1 Ohm shunt where the diode starts conducting and speeds up regulation - right? --- Quote from: Kleinstein on November 17, 2018, 10:37:04 pm ---Reading the actual current should be at the 1 K resistor and thus the other end of R7 anyway, even in the otherwise unmodified circuit. At the OPs input the current reading would be off during transients. The diode would change that error, but the current reading is not very accurate anyway. --- End quote --- It doesn't seem to be too bad currently, or am I wrong here (referring to the photo in my last post)? You mean the opamps input should be connected directly to the MAX4080 output, by saying "the other end of R7 anyway"? --- Quote from: Kleinstein on November 17, 2018, 10:37:04 pm ---With only 15 V supply for the OP, T4 might not be needed. It might give some 0.5 V higher maximum output voltage at high current. --- End quote --- I still have the footprint in the board so I can always populate T4. --- Quote from: Kleinstein on November 17, 2018, 10:37:04 pm ---Going to low side current measurement (e.g. a shunt on the negative terminal )would be a major change in the circuit. --- End quote --- Guess it would make more sense to recreate the whole thing if going for low side current measurement... --- Quote from: Kleinstein on November 17, 2018, 10:37:04 pm ---The max4080 is more like a low power version. One can build the max4080 internal circuit with a P-FET and OP with a higher speed. Something like 3-5 times the speed should be relatively easy. --- End quote --- You mean like this: https://www.analog.com/-/media/analog/en/landing-pages/technical-articles/robust-high-voltage-over-the-top-op-amps-maintain-high-input-impedance-with-inputs-driven-apart-or-w/32025.png?la=en&h=300 or this: https://www.analog.com/-/media/analog/en/landing-pages/technical-articles/robust-high-voltage-over-the-top-op-amps-maintain-high-input-impedance-with-inputs-driven-apart-or-w/32024.png?la=en&h=300 ? |
| Navigation |
| Message Index |
| Next page |
| Previous page |