# Excessive hysteresis in LM2577 step up converter

Update:

The conclusion I'm coming to at this point is that I believe the IC is damaged. My reasoning is that the device simply is not responding to the control input in a meaningful way. It turns driver/oscillator on when the control signal drops below 1.25 v, which is correct, then the inductor does its thing just fine and the output voltage starts to rise, but the IC leaves the "drive" running until the feedback input reaches about 1.5 v, which is far too high. Since the rise time for that input signal is in the region of 100 mS, it's not reasonable to attribute this to normal response latency.

I will say that this is my first foray into working with switching regulators, so I don't know what I don't know, but it's not my first foray into "feedback control systems" and this feels just flat out wrong to me. Anyway, I'm still open to other suggestions if anyone has one, but for the time being I shall put this on the shelf, design a PCB so that "layout issues" and long wires associated with breadboard may be reduced, and then I'll try again with a fresh chip.

The one thing that is still bugging me about this belief is that the point at which the IC turns off the oscillator/driver varies with inductance, and with input voltage. But I can't see any rhyme or reason to that. If that is suggestive, I'd love to know how/why!

Body of original post

I'm trying to build this step-up converter to generate 48 volts for phantom power to a microphone, and having difficulty making it stable. This is the circuit I started from. Note below a couple of changes:

The original circuit image is here https://sound-au.com/p193-f1.gif and there's a whole article describing it here: https://sound-au.com/project193.htm The data sheet for the LM2577 is here: LM2577-ADJ

I built (most of) the project on a breadboard, and it kinda-sorta works but the output varies by a total of 16 volts on a slow sawtooth. Let me fill in the gaps:

• I have used inductor values of 150 uH and 300 uH
• I left out the diode on the input side. This is not shown anywhere in the data sheet and as best I can interpret, this is in protection against reverse input voltage. Since I didn't have a suitable diode handy that would fit the holes in the breadboard I left it off.
• I place a 4k7 resistor as a load on the output.
• I didn't have space for the 100 nF capacitor on the output so I left that off too (for now).
• I have run this on a 9 V PP3 battery, and on a 12 V 10 A PSU
• I get 16 v sawtooth hysteresis off the PP3 and about 5 v when running off the 12 v supply.

This is the input and output waveforms on the PP3 (Yellow-output, blue-input)

Notice it's bouncing around about 16 volts peak to peak at just under 2 Hz.

This (blue) is the signal at pin 2 (the feedback pin). It almost seems that the 2577 isn't responding properly as the swing here seems far wider than it should be, and yet the device doesn't seem to be changing from drive to idle or back when it should be.

This shows the signal at the switch pin. Clearly, the device is "driving" when the sense voltage is "low", and the output capacitor is being charged up, similarly, it stops driving when the sense voltage is "high". Yet the hysteresis is far too great.

One note here is that the hysteresis is much better, but still grossly unacceptable at about 5 volts, when running off 12 v. I need this to run of a 9 v battery.

Breadboard layout is probably terrible, see below. Since current suggestions are that this matters and could be relevant, I think I shall have no choice but to put a PCB together before going much further.

Update: I have changed the output capacitor, and the peak-peak voltage swing does not change, but the slope of the discharge behaves exactly as one would expect for a capacitor discharging (double the capacitance, double the "decay" time). Similarly, removing the load resistor increases the decay time pretty much as one would expect.

In the meantime, can anyone shed any more light, or make other suggestions that I might be able to ?

EDITS:

• Added the battery voltage waveform as requested.
• Added trace of signal at feedback input pin.
• Added traces of output and input voltage when running of 12 V 10 amp supply.
• I have an idea forming: This device has "soft start", and it seems like there's a huge drop in the input power rail (looking at the scope trace, it's down to about 4.5 V which is the minimum input for this device). Is it possible that I'm triggering the soft start behavior every time that happens, so it always charges slowly?
• Added scope trace of the switch pin

TIA for any input, Toby

• Comments are not for extended discussion; this conversation has been moved to chat. Feb 13, 2021 at 21:18

Normally, the frequency of the ripple should be 52kHz, which is the internal switching frequency of the converter. But since it's around 2Hz, I think that there are some stability problems. Because the converter's response is way too slow.

The compensation network can affect stability:

• Check the values of the 2k2 resistor and the 220n capacitor.
• Make sure they are properly connected and placed near to the 2577 chip.

Also, remember that keeping the feedback path long can cause some stability problems. Make sure VR1, R2, and R3 are placed close to the 2577 chip.

• I'll poke around--though I have to admit I'm not sure what I'm looking for. But I'll check connection integrity (of course, getting short traces is challenging on a breadboard, perhaps I need to lay out a PCB). Feb 12, 2021 at 13:40
• Those parts seem to be connected properly, I've added a note about the (beneficial) effect of running on a 12 V high current supply. Feb 12, 2021 at 14:58

The LM1577 datasheet, on page 16, section B, contains an equation for the minimum inductance needed for stability of a boost regulator based upon that chip. The equation is

$$\L_{min} = \frac{6.4*(V_{in(min)}-0.6V)(2D_{max}-1 )}{1-D_{max}}\$$

Plugging in:

$$\V_{in(min)}=9V\$$

$$\D_{max} = 82\%\$$

gives

$$\L_{min} = \frac{6.4*8.4*0.64}{0.18}=191\$$ uH

So, you need to increase your inductance to at least 200 uH. Since you want to run your circuit on minimal current, I would suggest even greater inductance, such as 1 mH, if you can do that.

Now, even though your inductance of 100 uH is below what is required, I am not sure that this is the only problem with your circuit. The 16 volt ripple at 2 Hz, suggests something else, though at the moment, I know not what. I would at least fix your inductance problem, and see where that gets you.

Edit:

I have an idea forming: This device has "soft start", and it seems like there's a huge drop in the input power rail (looking at the scope trace, it's down to about 4.5 V which is the minimum input for this device). Is it possible that I'm triggering the soft start behavior every time that happens, so it always charges slowly?

If the input voltage falls to 4.5 V, your $$\\frac{V_{out}}{V_{in}}\$$ will increase to $$\\frac{48}{4.5} = 10.67\$$. The duty cycle will need to increase correspondingly (to over 90% in CCM). Definitely pushing towards the limits of the device/circuit capabilities.

Edit: It appears from your scope shot that the rise time of your output sawtooth is 48mSec. If C4, which is nominally 100uF charges 16V in 48mSec, that corresponds to an average 33mA that it is drawing. Since this is a boost converter from 9 or 12V to 48V, that means that the average current through your inductor during this time would be somewhere around 5 times that, or 150mA. The peak current could easily be twice that, or 300mA. Do you know the maximum current rating for your inductor?

• Ah, thanks, I had not done the math. I'll order more inductor values so I can raise this.That'll take a couple of days, of course. Feb 12, 2021 at 13:38
• Behavior changes significantly on a 12 V supply, as updates in question, I'm ordering a selection of inductors to try out. Feb 12, 2021 at 15:00
• I realized I had two of my original 150 uH inductors, and tried them in series. That did not change the ripple/hysteresis at all :( Feb 13, 2021 at 19:12
• If you change C4 to, say, half its value, does it affect the amplitude or frequency of the ripple? Even a small amount? Feb 13, 2021 at 22:30
• Interesting, I'll mess with that. I notice, on your prompting, that the datasheet calls for 680 uF in all the samples it shows, so perhaps there's an error in the project's schematic. Feb 14, 2021 at 16:44