Continue to Site

Welcome to our site!

Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

  • Welcome to our site! Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

Low Current TL494 buck

Status
Not open for further replies.

Mosaic

Well-Known Member
Is it reasonable to step down 36V to 12V for perhaps 300mA max draw using a TL494 and no external switching transistor?

I hope to rely on the internal 200mA x 2 transistors for the switching.
 
dont care about anything else? cost, footprint, EMI, ripple, efficiency ?

Lots of better choices
 
It's a BOM thing. Don't want another unique part.

Right now I have a LM317 dropping 24V which makes 8 W of dissipation required.

As I only need about 300mA, perhaps a TL494 driving 3 paralleled mmbt3906 PNPs (SOT-23) with 0.35 DCR ohm current leveling ferrite beads on their emitters might do the job. (no parts unique)
The 3906s should drop around 0.4V Vce @100mA and dissipate around 40mW average.

I'd have to add an inductor though. But losing 8W of heat should be worth it. I can clock the Tl494 around 150Khz.
 
ok good buck it.
 
PNP needs 60mA drive to saturate 600mA out peak worst case for 300mA average, not more PNP, thus quasi PNP pair is more efficient if you have no Mosfets

shown on the right..

Darlingtons_small.gif
 
Last edited:
The TL494 NPN driver transistors can supply 200mA of PNP base current.
I am driving 3 parallel mmbt3906 via 2.7K base resistors from the TL494 NPN, common emitter. The Vcc is 36 with a 1.3Vce from the NPN driver, so PNP base drive: (36-1.3-1 [Vbe])/2700 =12.5mA, which is good for 125mA per transistor or 375mA net. There is a 0.35 Ω DCR ferrite in series with each PNP emitter for load balancing. Sat. Vce on the PNPs should be about 0.5V max giving 65mW dissipation per PNP.
There is a 2 ohm current sense (hi side) with a PNP current limiting foldback at around 0.7V drop or 350mA.

Here is a pic of the schema so far....PCB area is about 1.5 sq in overall using a DIP TL494 and this inductor:
https://www.newark.com/bourns/sru1038-331y/inductor-shielded-330uh-520ma/dp/09J2978?ost=09j2978
prereg.png


PS: I am aiming for 15+ V to feed into the LM317 to obtain regulated 12V.
 
Last edited:
3 PNP's dont help as I said before. THey behave as one unless overheating.

Change R192 to 10k~20K and R188 to 330 Ohm and then drop 2 transistors.
 
Thats a little better. the 3 transistor have more base drive. The TL494 uses 70 of 250mA max and three 2904s use up to 350mA of 600mA max for an effective gain of 5 max and effective net Rce of 2/3Ω in addition to DCR(L)=0.35Ω for a total of 1Ω where the feedback lowers the Zout to improve load regulation error, but then the current sense R adds 2Ω to degrade it.

1) Can you redirect the feedback from output , instead of before Isense drop of 0~0.7V?
2) Will there be a capacitive load that may false trip foldback?
3) Your clock RC is floating and should be grounded.
4) If you wanted faster startup clock, you would use 1nF to gnd and 1nF to Vref so it reaches Vref/2 on powerup and starts clocking immediately.
 
Thats a little better. the 3 transistor have more base drive. The TL494 uses 70 of 250mA max and three 2904s use up to 350mA of 600mA max for an effective gain of 5 max and effective net Rce of 2/3Ω in addition to DCR(L)=0.35Ω for a total of 1Ω where the feedback lowers the Zout to improve load regulation error, but then the current sense R adds 2Ω to degrade it.

1) Can you redirect the feedback from output , instead of before Isense drop of 0~0.7V?
2) Will there be a capacitive load that may false trip foldback?
3) Your clock RC is floating and should be grounded.
4) If you wanted faster startup clock, you would use 1nF to gnd and 1nF to Vref so it reaches Vref/2 on powerup and starts clocking immediately.

1) I could, but it's there to drop the voltage (0.7 x .35 = 1/4W) to the LM317 when under full load to reduce linear reg dissipation. I am limited by BOM resistor selection & space for setting the voltage precisely.
2) No, only a 10uF cap after the LM317. Do u suggets a cap on the foldback Tr base to add stability?
3) It is grounded via the ground pad47
4) Ok
 
2) Cap on foldback must consider latency , false trigger margin , normal operating margin and Rja thermal resistance, max Tj rise as well as Operator intervention on latchup foldback. DO you want an alert signal or a timer to reset the foldback? These are all your choices. If your only load is the LM317, then it may be overkill.

So far looks ok,, but real world tests determine the reliability. LM317 are pretty old school with a big drop but work.

In future work towards the DC-DC types used on MOBO' for CPU's
Same area and cost in volume but >30x the current.
 
Last edited:
(EDIT: note post #23 solved this heating issue!) Ok, I had to use a TIP30C as the switcher. Ganging the 3906s' worked but the little buggers overheat under full load, switching losses I suppose @ 180Khz. Paralleling more was too cumbersome.
Foldback seems ok without a cap.
Added a 4.7 Ω, 2W, R.emitter to limit startup transient spikes. 3906s died instantly without it. Didn't test the TIP30C without it since circuit works fine with it. A 300mA load uses 1/6 duty cycle. Had to parallel 5, 2.7k 0805 units to handle the thermals for the base drive. Estimate the switching saturates at 800mA peak @ a 15x hfe minimum per the spec. sheet. At 300mA load 15.3V hits the LM317, 16.2V with a 16mA 1K load.

Energy conservation calcs =>
Voltages: 15.3 V.out with a 36V supply (less PTC and rectifier drop external to this cct) for a 35Vin.
Currents: 300mA out with a 1/6 duty => 1800mA x 15.3/35 => 785 mA pulses which matches up fairly well with the 800mA estimate above.

I used this tool to figure the inductor and capacitor parameters.
https://www.daycounter.com/Calculators/Switching-Converter-Calculator.phtml

In summary, I achieved an 8X reduction in LM317 heating, but I had to include 2 new BOM parts - the inductor and the TIP30C. Hopefully it should balance out with enhanced reliability.

Based on this tool:
https://www.efunda.com/formulae/hea..._forced/calc_lamflow_isothermalplate.cfm#calc
The 7W saved amounts to 62% of the heat dissipation capability of the 4" x 6 " x 3/8" flat Al. plate @ 2m/s air flow @80°C, that I am using.


Here's the final schematic.
prereg.png
 
Last edited:
Now you understand why parallel tiny BJT's wont work.
Since GP transistors are 10:1 for current and R ratios , you have it almost right.
Since a Darlington TIP is two the ratio is 100:1 or 470 Ohms max for Rb.
For less loss, reduce Re and Rb but keep same ratio.

What you had before was like 5 weedwackers trying to power a motorboat.
 
I am looking at this improved spec. sheet. It has some useful PCB layout tips for EMI suppression.
https://www.ti.com/lit/ds/symlink/tl494.pdf

DTC soft start seems simple to add at 180Khz as it only requires a 0.1uF cap/ 2.7K combo for 50 cycles start.

I have noticed that most TL494 sample circuits driving a PNP switch include a pullup base resistor at about 1/3rd the value of the base drive resistor. Have a look at figure 15. They mention discharging carriers for proper switching. Does this mean that the transistor is hovering a long time in linear mode without that kind of pullup?
 
I am looking at this improved spec. sheet. It has some useful PCB layout tips for EMI suppression.
https://www.ti.com/lit/ds/symlink/tl494.pdf

DTC soft start seems simple to add at 180Khz as it only requires a 0.1uF cap/ 2.7K combo for 50 cycles start.

I have noticed that most TL494 sample circuits driving a PNP switch include a pullup base resistor at about 1/3rd the value of the base drive resistor. Have a look at figure 15. They mention discharging carriers for proper switching. Does this mean that the transistor is hovering a long time in linear mode without that kind of pullup?
Yes,
Just makes it turn off faster which could be important as fast as you are running it.
 
I did some tests on the switching and the TIP30 is hard saturated and takes a while to switch off.
By adding a 100 ohm pullup I can speed up the switch off so that the PWM moves from 1uS on to 1.6 uS on.

Result is the TIP30 drops about 24 °C (to 42°C) for the same 150mA load, no heat sink.

ChrisP58 , that part MC33063 looks quite adequate as a 1.5A part with an integrated switch and current limiting. It looks workable with the same inductor I have at a 100Khz clock. Makes sense if I have to add a new BOM PNP switch I may as well get a more integrated solution, cost being reasonable.
Tony was on the right track all along ...I needed a more modern part.
 
The 33063 isn't really a modern part, I've been using it for over 20 years, but it's a good performer.

And at ~$0.20 US in production quantities, it's a good value when it fits the need.
 
Can't seem to find anything to compete with it in the 40V range although I see it has heating issues past 500mA. My app is < 500mA though.
Also, an over sized inductor with low load current draw might lead to discontinuous mode operation, although I suppose a limiting resistor can fix that or just clocking it slower.
 
OK now compare that with this .
upload_2015-5-15_12-4-46.png

Tony
 
Status
Not open for further replies.

Latest threads

New Articles From Microcontroller Tips

Back
Top