Energetic discussions have sprung up right here in editor Aalyia Shaukat’s Design Concepts concerning the constraints and quirks of, and design tips for, the present management topologies proven in Determine 1.
Determine 1 The right way to management amps of Iout with mA of Ic utilizing legacy voltage regulators as present regulators the place Iout = (Vadj – IcRc)/Rs.
Wow the engineering world together with your distinctive design: Design Concepts Submission Information
Reader Ashutosh Sapre contributed a disturbing commentary in regards to the seemingly impact on regulator reference accuracy of temperature rise from self-heating, as illustrated in Determine 2.
Determine 2 LM317 reference variation with junction temperature as seen in web page 5 of LM317 datasheet.
As proven in Determine 2, the temperature stability of those legacy units is pretty good. However, there are conditions the place the tempco may be problematic.
For instance, contemplate a situation that begins with programming for 100% of full-scale output present (e.g., 1 A) in order that regulator warmth dissipation is excessive. Assume it’s maintained lengthy sufficient for the regulator’s junction temperature to rise from 25oC to 125oC. Determine 2 predicts that this massive temperature swing will trigger Vref to float from 1.25 V to 1.2375 V, inflicting the output present to say no by about 1% of full scale.
This 1% corresponds to 10 mA out of 1000 mA and is considerably lower than 3 LSB of an 8-bit setting. That’s maybe not nice, nevertheless it’s not horrible both. However what if the output is then reprogrammed for 10% of full scale (e.g., 100 mA) whereas the regulator remains to be sizzling?
Then that 1% of full-scale error would develop into 10% of setting is what. It’s going to manifest as a really prolonged thermal settling tail lasting many seconds as junction temperature step by step cools from 125oC, permitting Vref to (slowly) return to its preliminary 1.25 V and output present to settle on the appropriate 100 mA. It’s going to occur finally, however the time required will likely be objectionable. It might be unacceptable.
Thankfully, Ashutosh additionally contributed a easy and sensible answer to the issue within the type of an auxiliary present shunt transistor. The shunt would enable a lot of the output present and, consequently, a lot of the self-heating to bypass the regulator solely. This would go away its junction unheated and its Vref undrifted. Drawback solved!
Or is it? Ashutosh additionally identified that the bypass transistor, whereas handily fixing the thermal drawback, would sadly additionally bypass different issues. Particularly, the nifty fault safety options (e.g., automated present limiting and overheating shutdown) constructed into LM317 and LM337 chips could be misplaced. Whereas these belongings might probably be added to the transistor shunt, that might lose a lot of the simplicity that made it enticing within the first place.
So, I questioned if Ashutosh’s shunt concept may very well be carried out in a means that might inherently retain the fascinating 317/337 options whereas staying easy. The apparent factor (I like apparent!) is perhaps to simply make the shunt out of one other LM3xx. Determine 3 exhibits simply that: A design that cross-connects complementary regulators utilizing U1’s 317 for management and U2’s 337 for shunt. Management and shunt currents are then summed again collectively earlier than passing by Rs to offer suggestions to U1 the place Iout = (I2 + I3) = (Vadj_U1 – IcRc)/Rs and I3 >> I2. Discover how the shunt will get turned “the wrong way up.”
Determine 3 Cross connection reduces self-heating error as a result of shunt regulator U2 carries a lot of the present, getting comparatively sizzling, whereas U1, whose Vref is in management, stays comparatively cool and correct.
Determine 3’s U1 is linked principally per Determine 1, apart from Rx. The sign developed by Rx * I2 feeds U2’s ADJ pin in order that when U1 enter present I2 rises above about 10 mA, U2’s ADJ pin will drop sufficient to make it begin conducting. This causes the I3 present part to rise and finally comprise the vast majority of whole present I1 = I2 + I3. Thus, U2 dissipates a lot of the self-heating Watts, making certain that U1 stays comparatively cool and its Vref stays correct.
The 1N4001 in parallel with Rx protects Rx and U2’s ADJ pin if U2’s over-temp or over-current shutdown characteristic kicks in. That would go away U1 making an attempt to shoulder the entire load, dropping sufficient voltage throughout Rx to seemingly harm U2 and fry the resistor. The diode prevents that.
Determine 4 exhibits the concept working as a detrimental present supply.
Determine 4 If the 317 and 337 swap locations and the diodes reverse, Determine 3’s circuit can work for detrimental present, too.
If extra present functionality is required, extra U2 shunts and better capability diodes may be added (Determine 5).
Determine 5 Increase present dealing with capability with beefier diodes and extra U2s.
Determine 6 integrates this concept into an entire PWM managed detrimental present supply as detailed in: “A detrimental present supply with PWM enter and LM337 output.”
Determine 6 Damaging present supply circuit incorporates means for compensating part tolerances, together with these of U1 and Z1 references. Observe Rs = 1.1 Ω and ought to be rated for greater than 1 W.
The one-pass adjustment sequence is:
- Set Df = 100%
- Regulate CAL pot for 1 amp output present
- Set Df = 0%
- Regulate ZERO pot for zero output present.
Accomplished. Iout = 1.1 Df /Rs, the place Df = PWM obligation issue.
In closing, thanks go (once more) to savvy reader Ashutosh for his solutions and (likewise once more) to editor Aalyia for the fertile DI setting she created, which makes this type of teamwork workable.
Stephen Woodward’s relationship with EDN’s DI column goes again fairly a great distance. Over 100 submissions have been accepted since his first contribution again in 1974.
Associated Content material
- 1 A, 20V PWM DAC present supply with monitoring preregulator
- PWM-programmed LM317 fixed present supply
- A detrimental present supply with PWM enter and LM337 output
- 1-A, 20-V, PWM-controlled present supply
The put up Cross join complementary present sources to scale back self-heating error appeared first on EDN.
