Trouble with LT1073 Buck converter.

Welcome all. Today I encountered problem with my LT1073 Buck converter. I am trying to build 9V to 5V Buck converter able to deliver about 70 mA. I followed schematic in datasheet. However, some parts I had to improvise with. Buck converter works with no load however it experiences voltage spikes. Real problem begins when I connect load in this case 67 Ohms to force about 75 mA. I attached pdf with schematic and two jpegs showing behaviour. Open to any suggestions as I suspect I mad some rookie mistake.

Inductance of coil is 120uH.

1.pdf (37.3 KB)



Layout has to be done really well to avoid issues with buck converters. Your schematic maybe perfect, but if the layout is not good it will not be stable. I have never built one myself, but have been involved in testing. How did you build this circuit? On a breadboard or something?

Some pictures of the circuit could help.

Yes, it is completed on breadboard all cables are laid flat long ones are parallel to expected magnetic field. New to design so perhaps it is not most optimal layout. After your post I made some improvements to tidy things up and it helped a bit. No spikes! So need to keep wires short to minimum to prevent voltage inducting in them. However I still get about 0.5V ripple with 50mA load.

You might be able to get by with point to point wiring on a bit of perfboard. I build a lot of circuits that way, but for best results you really need a printed circuit board. I generally buy pre-made modules instead of trying to do that. Once you have a pcb then you can tweak values. One thing that could be a concern is what type of caps you are using. You need to read the datasheet and make sure that your caps meet the specs. Even then, you may have to play with the values a bit to get it working really well.

However, some parts I had to improvise with.

A BZY88c zener diode instead of a schottky diode?
Seems like stretching it too far.

Now I am confused. Datasheet has this schematic for Buck mode. With this written below “D1 MUST BE A SCHOTTKY DIODE” However notation for the diode is 1N5818 and this part is Zener diode according to:


Yeah, you need a schotky diode; that design also doesn't look like the normal buck topology, is that really what the datasheet recommends?

And yeah, you can't make a DC-DC converter on breadboard. Breadboard is kryptonite for DC-DC converters.

Thank you for all the hints. Stanley enough it does. I believe I might read it incorrectly or perhaps there is a mistake here.

1N5817-1N5819 are general purpose schottky diodes.
Only max reverse voltage is different.

Google "1N5819 datasheet".

The 1N5851 you linked to is a zener diode.

Yes, it is completed on breadboard all cables are laid flat long ones are parallel to expected magnetic field. New to design so perhaps it is not most optimal layout. After your post I made some improvements to tidy things up and it helped a bit. No spikes! So need to keep wires short to minimum to prevent voltage inducting in them. However I still get about 0.5V ripple with 50mA load.

Optimum layout means the main current loop through diode/cap/inductor has to be as tight as is possible
(a few mm^2 is good) for minimum stray inductance and minimum magnetic coupling to other parts of the circuit.
This means surface mount components only I'm afraid - can't reliably be achieved on a breadboard with
through-hole components, you'll get interference with the rest of the circuit such as
voltage feedback sensing if your layout is poor, likely causing instability and other issues.

Most datasheets for switch mode converter chips give a recommended PCB layout - use it...

Thank you all for replies! Today long waited 1N5818 arrived and it went straight in to breadboard. I am trying to create 9v to 5v Buck converter. Able to deliver about 100mA to be on safe side minimum required 75mA. After assembling circuit like datasheet suggests got nice stable output voltage of about 7.2V. However, this is not voltage I am going for I hoped to achieve 5V. Pictures below show my first attempt on circuit coping datasheet: circuit_1 and datasheet_1.
Later I looked for alternative arrangement and copied second circuit with potential divider form the output feeding back to FB (Sense) pin: circuit_2 and datasheet_2.
I also attached LTspice simulation that is accurate in comparison to my measurement on breadboard. Systems seems stable and used scope to measure ripple and any spikes ect and nothing abnormal seems to be happening apart from incorrect voltage.



The datasheet-1 picture is for the LT1073-5. That is the fixed 5V variant.
Presumably, you have the LT1073, which is adjustable, so you must wire up the feedback.

In your original post, you wire the feedback pin to a trim pot.
In your circuit-2 picture, you use 2 fixed resistors.
The ratio of those 2 resistors suggest that you should get 3.05V out.
212mV * (536k + 40k) / 40k = 3.0528.
The voltage on the FB pin should be 212mV. To get that from a 5V output, you must divide by 23.6
Try 22k for R2 and 1k for R1. This divides by 23. Close enough? It should give you 0.212 * (22k + 1k) / 1k = 4.876V.
In the data sheet, they use 40.2k and 909k for R1 and R2 respectively to get 5V out.
0.212V * (909k + 40.2k) / 40.2k = 5.0057V

(To my personal uneducated taste, those resistor values are very high; I would keep the total of R1 + R2 between 10k and 100k. But if the data sheet says ~1M is OK, I guess it's OK. Maybe they assume nice short PCB tracks, not breadboard...)

If you have a trim pot of at least 10k, wire that to the feedback pin and adjust it until you measure 5V out. Then take the trim pot out and measure the resistance. the wiper should be at 1/23.6 of the total.

You can absolutely build switching DC/DC converters on breadboard, but you should put everything as tight together as possible, use as few wires as possible, and if you have to use wires, use short solid jumpers, not long thin stranded wires. I built this a few days ago. The bit on the right is a buck converter. It steps down 16V to 5V. I use an LM2576. [Edit to add] The trim pot is 20k.

@Jobi-Wan. Thank you very much for your post after reading it noticed some fundamental errors I made and adjusted my LTSpice simulation. I believe you are right that my physical chip is adjustable will confirm it tomorrow. However, after swapping LT1073-5 to LT1073 in simulation adjusting potential divider as suggested I got following result (picture below). Regarding pots I believe I have 10k variable resistor that fits nicely on breadboard so ill use that. My only concern is value for ripple as it could affect my threshold detection circuitry in later stages. However very pleased that now I have working circuitry and I can implement it with LT1044 to get +/- 5V supply.

Your output capacitor is 22µF. This seems low. The 'typical application' schematics in the data sheet all use 100µF. You can put a 100µF cap parallel to your 22µF cap. Low ESR caps are preferable.
If you need to reduce ripple more, you can try adding a LC filter stage. Alternatively, you can step down to 6V or so and add an LDO regulator, but this will cost efficiency.

@Jobi-Wan Once again thank you for pointing out wiring issues now. It turns out I have 1073-5. Today I completed 3V to 5V step up circuit along with 1044 switch cap inverter to build battery operated +/- 5V power supply and it works perfect! I remember about ESR from power electronics and I achieve something about 4.95V after cap therefore I can live with that. Increasing cap size helped a bit but it wasn’t proportional relation I mean I increased cap size by factor of 5 and ripple decreased by factor of about 2. However, I still get 100mA out of it with reasonable ripple.

You might try a couple of 0.1µf or 1.0µf ceramic capacitors across the 22 / 100µ cap.
The electrolytics are not so good at high frequencies.

@JohnRob. Thank you for advice I’ll do that tomorrow and check results of it.