| Electronics > Beginners |
| TL431 linear power supply |
| << < (25/32) > >> |
| spec:
--- Quote from: blackdog on December 21, 2018, 10:10:06 am ---Hi Spec, Thank you for the insight of my psyche, I didn't know that yet, keep up the good work! :-DD --- End quote --- No sweat and thanks for your informative and helpful post. :-+ |
| spec:
--- Quote from: not1xor1 on December 21, 2018, 08:19:02 am --- --- Quote from: spec on December 21, 2018, 05:24:24 am ---not1xor1 I had missed this simulation that you did for the pre reply #89 circuit. Thanks a lot- very interesting. :) But you have omitted one fundamentally important component: R19 (10K) What is the battery on the non inverting input of the opamp doing there (the reference voltage is the variable). Also in #89 revised schematic: R13, R20, and R21 are 56R. All small signal transistors are BC337/327 D1 (D5) has been removed and replaced by a trace R4 has changed from 1k to 56R --- End quote --- You're right. I had overlooked the fact that the reference voltage was on the inverting input. Of course the local feedback is affected by the resistor in series with that input, but there is little difference between 10k and 11k (i.e. the pot at its minimum) of source resistance. I also changed the transistors and the resistors, anyway the bode plot is still ugly :D. I tried to compensate the circuit in a different way and it improved a little (phase margin in the 50-70° range according to the load). But I've no idea if that might work in the real world. Later I'll run a transient load simulation. --- End quote --- not1xor1 Breakfast was nice, but I choked on the sim results. So much for my initial slug approach to get the circuit going. Your compensation looks good for the initial configuration so, if it is OK with you, I will embody it in the schematic of reply #89. Taking BlackDog's comment about the high source resistance for the LM358 inputs. I only increased the value from the original 1k to 10k so that mike-mike would be able to use a 1uF capacitor rather than the original 10uF. Is there any way that we can use your compensation scheme and reduce the source resistance back to 1k. So would: R16=24K R17=1k R6= 1k R7= 100k C2= 47n work? |
| Kleinstein:
For low noise the feedback divider should not divide much more than really needed to stay inside the OPs working range and the practical reference range. So maybe up to 5 V after the divider. For a 30 V maximum this would be something like 51 K and 10 K. It helps to have the divide by 6 or similar stage, as in this case the simple capacitor in parallel to the upper resistor can be well used in the frequency compensation. Even the 240 K and 10 K divider is not bad because of high impedance, but more due to the high divider ratio. So 24 K and 1 K would be nearly as bad. With a TL431 or similar reference one would still have quite some noise from the reference itself. So only at small voltages (e.g. < 2 V) the noise from the LM358 would be really relevant. The extra transistor stage with gain makes the compensation quite complicated. So unless really needed because the voltage goes higher than the OPs output range, I would avoid the extra gain stage and directly control the Darlington output stage from the OP. To get a good regulation over a wide range of loads, including low ESR capacitive loads, it really helps if there is some output capacitance with ESR. There is a reason that nearly all lab supplies have some electrolytic capacitor at the output - though one may not need as much as often found. This is not just for the capacitance but also for the ESR of the electrolytic cap. The output ESR is kind of part of the compensation scheme. As a loss to the capacitor it still has an effect if there is an external load that includes a capacitor with less ESR. For the phase margin one should have a really good phase margin with a well behaved resistive load. With a tricky load like low ESR capacitor plus current sink the phase margin will go down. This is acceptable and essentially unavoidable. Still a good phase margin with a resistive load does guarantee stability with a difficult load. So it might help to adjust the compensation for the difficult case first and only than check the easy case. |
| blackdog:
Hi, Some extra remarks.... The LM358 is a slow opamp and with a well-designed circuit it should provide a reasonable power supply with the 2N3055 transistors. The TIP41 and the 2N3055 are both "slow" transistors and they do not help to make the circuit stable. Hint, make the base emittor resistor a little lower say 56 or 68 Ohm, which helps with the paracit capacitors of the 2N3055. And also this, there are so may different 2N3055 transitoren, Ft 0f 0,8Mhz, Ft of 3Mhz etc, So if someone present here a schematic with 2N3055 transistors, then you will actually have to take into account the worst version when it comes to loop compensation. A solution for better phase margin Better is to use faster power transistors like the cheap 2CS5200 (the real one) Almost all 2SC5200 transistors have a Hfe of > 90 and if you use a BD139 as a driver and some coling, then you have power stage that does not affect the phase margin of the LM358 much. That doesn't mean that Q6 is a good idea, Q6 means trouble :-DD Can it be done with Q6, look at several HP power supply schematics, do not forget the ferriet bead. An example Suppose we take 4x a 2SC5200 and a max 5a power suply. If we then assume an Hfe of 90 (usually it is more, especially if the transistor gets warm) The base current in then 1.25A/90 = ~14mA per 2SC5200 which is then about 60mA of total base current for four transistors, there is then a little extra for the emitter base resistor, say total 70mA. Drive this 70ma with a BD139-16, Hfe > 100x, that makes the base current of the BD139 below 1mA. Make a 5mA current source and a diode from the opamp to de base of the BD139 to sink the current. This makes a fast power stage that does not affect the phase behaviour of the opamp very much. Testing output capacitors This is a setup how i tested, to make a good RC section over a wide frequency for the design i was working on. And a nother test. Prefered schematic/design I prefer the Harison setup with a floating power supply for the opamps, the extra transistor is olmost never necessary. www.bramcam.nl/NA/NA-01-PSU/Harris-Tech-Letter-02.pdf www.bramcam.nl/NA/NA-01-PSU/Harris-Tech-Letter-03.pdf www.bramcam.nl/NA/NA-01-PSU/Harris-Tech-Letter-04.pdf www.bramcam.nl/NA/NA-01-PSU/Harris-Tech-Letter-07.pdf www.bramcam.nl/NA/NA-01-PSU/HP-AN90A.pdf And then there's this one, very important :) www.bramcam.nl/NA/NA-01-PSU/Calex-Power_Impedance-Decoupling.pdf How to test phase and gain margin AN-1889 How to Measure the Loop Transfer Function ofPower Supplies From TI www.bramcam.nl/NA/NA-01-PSU/snva364a.pdf More than enough information for the holidays. :-DD Kind regards, Bram |
| Kleinstein:
The slow 2N3055 were the more robust old style transistors. One hardly finds them any more. However the once popular 2N3773 are usually still slow. For a high performance power supply it really helps if the transistors are fast, but for a simple low cost supply one can still use the normal 2N3055 or TIP35. It still helps if the second transistor in the Darlington configuration is fast. A low power the BD139 is a good choice, at higher power something like D44H11. A slower supply would need a larger output capacitor though. @blackdog: the links provided are interesting, but have a trailing / too much. |
| Navigation |
| Message Index |
| Next page |
| Previous page |