DC Switching linear regular design question

Thread Starter

jcarlson023

Joined Dec 19, 2020
4
Hello,

I am fairly new to electronics design (I have a mech e background), and am designing a board that requires dropping 12v to 5v with a large current requirement (up to 8A@5V). In a bid to save on BOM, I took a shot at creating my own design vs buying one off the shelf, using the MAX20008AFOC/VY+. I copied the recommended design and have had boards made, but it is not operating as expected. The circuit is outputting a voltage that is jumping between ~3V to ~5V irregularly. There is a hum too, which I assume might have something to do with the switching frequency and the inductor? Below is my circuit design, can anyone see anything glaringly wrong? Thank you!

max20008_circuit.PNG
max20008_pcb.PNG
 

ronsimpson

Joined Oct 7, 2019
1,343
Give the part number for L1. Any data will help.
You might push down on parts with finger. Mechanically stressing the solder joints might help a bad spot.
 

Thread Starter

jcarlson023

Joined Dec 19, 2020
4
Here's the link to the inductor. I had a few boards made, and they all performed the same when I hooked them up to a power supply so I'm fairly confident it's not a mechanical issue (although the pads for the MAX20008 are dang small...). Let me know any other information I could provide, I appreciate the help
 

Papabravo

Joined Feb 24, 2006
16,512
Short of a wiring error or inadequate cross sectional area in traces and ground returns, I expect that output capacitor is unable to hold the output up when the switch is off.
Isn't there supposed to be a diode in a buck converter to supply the output capacitor when the switch is off? Edit: Never mind, it's inside the chip
8A out of an IC pin is a ridiculous amount of current for an IC leadframe.
Your schematic is backwards with the input on the right and the output on the left. Drawing the schematic as if you were doing a layout is not very helpful in understanding the circuit. There are certain things we look for that are just very difficult to see on a drawing that we cannot zoom in on.
 
Last edited:

Thread Starter

jcarlson023

Joined Dec 19, 2020
4
@crutschow fair play, typo on my part!

@Papabravo I agree 8A sounds crazy high, but this is what the datasheet claims and I copied their suggested schematic from the datasheet. Perhaps this was a naïve move on my part. Apologies for any... unconventional... methods in the schematic/pcb drawing, I am self taught from my bedroom. Would attaching the KiCad files be useful?
 

Papabravo

Joined Feb 24, 2006
16,512
@crutschow fair play, typo on my part!

@Papabravo I agree 8A sounds crazy high, but this is what the datasheet claims and I copied their suggested schematic from the datasheet. Perhaps this was a naïve move on my part. Apologies for any... unconventional... methods in the schematic/pcb drawing, I am self taught from my bedroom. Would attaching the KiCad files be useful?
I have to take a closer look at the package and it's thermal resistance. In any case you need to use special layout and fabrication techniques that are less than obvious from the datasheet. The Kicad files will be useful to other members that can open and examine them. One important feature that you may have overlooked is the use of thicker copper layers. For SMPS we used what is referred to as 3 oz. copper. I think it refers to using that amount of copper per unit area, but at the moment I don't remember what that area was. (Edit: it is ounces of copper per square foot) A typical PCB uses 1 oz. copper. 3 oz. copper provides a dramatic increase in the cross sectional area of the traces and is helpful in reducing inductance and dissipating heat.
 

Wolframore

Joined Jan 21, 2019
2,318
You're missing a pullup to VCC for your reset pin, try adding one to see if it stabilizes. It would vacillate from 3.3V to 5V since you've tired FB to BIAS. Hopefully there are no bad trace issues which again we can't see in your image.

1608404055027.png

1608404076897.png
 

LowQCab

Joined Nov 6, 2012
798
Always read every word on the Data Sheet.
I can virtually guarantee that there is a statement similar to .......
"It is the customers responsibility to verify the suitability and specifications of the following suggested circuits",
XYZ manufacturer does not guarantee the specifications of these circuits,
and assumes no responsibility for the performance of these circuits in any particular application ............
This device is not suitable for Medical, Military, Aerospace, or any Life-Support application ...........

You must verify the Math using every formula provided in the Data Sheet.
Lets start with a simple one ....... Lets assume that some non-ideal condition causes the circuit to
operate at a relatively low efficiency of around ~75%, for a short period of time.
12V X 8A = 96 Watts X ~25% = ~24 Watts of Heat Dissipation,
that's a serious amount of Heat to get rid of from ~3 square-inches of PC Board.
So, unless you have a couple of Fans blowing on each side of the Board,
the internal Over-Temp-Protection-Circuitry is going to shut-down the chip,
hopefully before the Blue Smoke leaks out.
How many hours are your Capacitors going to survive running at 125C ????
How much Output Voltage Drift can you expect at those temperatures ????
How many other Component Specifications are going to drift all over the place at those temperatures ???
Do you have an Infa-Red Temp Sensor Gun ???
This is just one of the Testing Tools that you will need to verify the viability of your design,
and to be able to properly trouble-shoot any problems,
along with a decent Oscilloscope, of course.

Quoting the Data-Sheet .........
--------------------------------------------------------------------------------------------------
"" Thermal Considerations:
The devices are available in 4A, 6A, or 8A versions; however,
the average output-current capability is dependent on several factors.
Some of the key factors include the maximum ambient temperature (TA(MAX)),
switching frequency (fSW),
and the number of layers and the size of the PCB.
See the Typical Operating Characteristics for a guideline. ""
---------------------------------------------------------------------------------------------------

Since you are starting from scratch, I would go with the lowest available
Switching-Frequency ( 220khz according to the Data-Sheet ),
( since EVERYTHING gets closer to CRITICAL as the frequency gets higher ),
and over-size the Output L-C Filter by at least ~20%.
Then, figure out a Heat-Sink solution that is at least ~2-X the calculated C/W capacity recommended.
After you're done with that, don't expect a full 8-Amps Output.
Load-up the Output hard to document it's performance under CONTINUOUS worst-case conditions.

I like to DOUBLE the maximum current capacity needed right from the start,
it makes most things so much less critical,
and generally means that the efficiency will go UP, rather than DOWN, under heavy loads / adverse conditions.
Notice that this chips efficiency is highest at around ~3 Amps, (12Vin / 5Vout),
so yeah, it has "the capability" to handle ~8 Amps, before the Blue-Smoke leaks out,
but it is the most "comfortable" running at ~3 Amps.

You may go through ~3 or ~4 PC Board re-designs before you get it right.
I'd be willing to bet that you'll have to have a proper Aluminum Heat-Sink before you're done.
.
.
 
Top