My system has a Pico W and a AHCS-7123 30A current sensor with the output of the sensor going to an AD654 V/F converter.
They are all mounted on a board which also houses 2 x Buck converters from Jaycar.XC-4514 which unfortunately has no usable specs.
The converter in question supplies 5vdc to the Pico through a Schottky Diode.
The same 5vdc supplies the AHCS-7123 which at zero current puts out 2.5V that is fed to the AD654. I am not using the usb when doing the tests.
Although everything works and my counted pulses for one minute are within 5 or 6, the problem is that when I look at the 5vdc with my Picoscope 2000 , there are spikes of 40mv to 80mv on the 5v rail randomly.
Unplugging the Pico reduces the amount from every 5-10ms to every 100ms.
I suspect the buck converter is not a synchronous converter and it struggles to satisfy the current spikes from the Pico. I have added a 4700uF cap to the 5v rail which does reduce the spike amplitude, but suspect I would need to add 10 or 20,000 uF to do any good.
So, the questions:
Which of the converters supplied by Core would be most suitable to get the 5v to maybe a couple of millivolt ripple.? Is that even possible.
Should I add a small cap on the Vsys pin (after the diode)?
Did You look here:
Found under the "downloads heading. By the way, that is where all Jaycar’s data sheets are hiding.It is not super but gives all the info needed to get up and running. Even a reminder about the multi turn pot that has had quite a few confused insofar they did not think the device was working.
I did get that sheet, but was expecting graphs of load, temp stability, drift ripple etc.
ie: Like the AHCS-7123 spec sheet.
I think if the manufacturers (or anybody else for that matter) put this through these types of tests and produced that sort of documentation you would be paying quite a few more dollars. It’s called getting to market quickly at a competitive price.
In other words you get what you pay for. Not completely satisfactory but a fact of life these days.
Think of government, military or other like organisation. Want a piece of equipment, robust, reliable, latest technology and anything else you can think of. Then think of all the tests you could do including environmental and they will want it…BUT, the customer will pay many thousands of dollars for this wrapped up in the per each (or total contract) price and are happy to do so. This complete testing might only happen to a pilot production run or a sample from first production and a random selection thereafter but is done.
That doesn’t help me pick a better 5v converter ?
AHCS-7123 is a component. The equivalent datasheet for the module you are using would be:
Thanks Jeff, that gave me the detail to conclude it is not a “synchronous converter”.
My reading of the different types is that the synchronous converter has a much faster response to current surges.
If no one can recommend a particular one , then I will have to pick one and try it.
Still though, would like to know what the best ripple/spike voltage is that I can expect.
Very true. Just trying to point out why this sort of information is not readily available for the type of device aimed at the hobby/maker community. But as Jeff pointed out most of the detail is there and you have to get down to component level to find some.
What is your voltage input to that converter and how important is a ripple free 5V.
If your input is somewhere in the region of 7.5V to 9V maybe a linear regulator could be consideredlikw a 7805 or 1117-5V (I think it is 111) linear regulator would be OK. Not as efficient maybe as a switch mode device but that one you have is only 80% efficient so will not be far behind. You will be blessed with a ripple free output.
Simple to use, VIN, Gnd, VOUT. Sometimes a capacitor is used on input and/or output but I have used the 7805 many times successfully without. They are good for around 1A output. The down side is that if your input voltage is higher (12V???) and current up it will get hot and might require a heat sink. For instance input 12V, output 5V current 1A voltage drop across the device is 7V and will have to dissipate 7W. But if your input is down to 7.5V or 9V and current at about 500mA you should not have a heat problem. The input voltage has to be a bit above output volts, refer data sheet to ascertain by how much.
PS: Horror, this is getting into the analog world but don’t despair the 2 sides of electronics often have to overlap.
The source voltage will be 13.2 v that needs to be 5v for the ACHS-7123.
The voltage output from the ACHS-7123 converts to counts per second and spikes in the voltage will cause erroneous counts.
I am tackling this from 2 fronts. The ACHS-7123 has a filter capacitor input and I am hoping that by setting it to 0.1uF it will ignore the fast rising spikes.
Because I have a 9v Down converter on the same board (for the AD654) I can use it
as a supply for the 7805 avoiding the heatsink issue.
Either that or replace both buck converters with 7809 and 7805.
At 13.2 volt the entire board (including the Pico) consumes less than 100mA.
Interestingly, the pico doesn’t seem to worry about the spikes it creates.
Staying with analog and not switching at any sort of frequency has to be better.
Will get some bits and test rig it to prove stability.
At 100mA I could not see too much of a heating problem.
Use the 7809 first then the 7805 regulating from 9V to 5V. This is common practise even with switching regulators. Doing this I think your concerns about ripple etc might go away. At about $1.85 each it is not going to cost an arm and a leg to find out.
I use those buck converters in a project that requires 2 x 5V supplies. The buck converter is set to 7V, the output goes through a 100µH inductor to a 470µF capacitor then to 2 * LM1085 low dropout regulators (each with a 47µF capacitor on output). It can take 9V to 18V input and runs cool. I did a bit of research before coming up with that scheme, concluded that the output of a buck converter was always going to be noisy. If the inductor is small it delivers current in pulses. If it is large, the pulses are much smaller but the response to changing load is sluggish (which may be why you are getting spikes).
Thanks for that reply Alan.
I tried a 100uH on the output of the buck converter, but only succeed in getting ringing.
What you have done there with the LM1085 is more efficient than going straight to 5v with a 7805. In fact I wouldn’t mind betting that you could adjust the converters lower than 7v .
In the end I simply tested a 7809 and 7805 in series (as I needed both those).
Would have liked to drop the 13.2 v to the 7809 by some other (non heat wasting) method, but couldn’t find any. Either way the board now draws 82mA wheras before it was 50mA. So burning off 30 mA is the cost of getting no spikes and practically no ripple.
You need to be careful what you do here. The regulator (even low drop out) needs a little bit of head room. If you are smoothing out a supply with known ripple the low point of the ripple must be above the minimum regulator input. Otherwise when the low point goes below this minimum voltage it will be ineffective for this period of time.
Not sure exactly what happens if this occurs as I have never had occasion to try stand always had enough input to get a clean result.
There are trade offs quite often in this business. It often comes down to choosing the lesser of two evils. One thing to remember you can never get out more than you put in. In other words you get nothing free.
A higher value inductor would give better results with a low current like 100mA. Maybe 1mH. I tried to use what is described as Mini360 DC-DC Adjustable Buck converter which I tested at 1.5A and was quite satisfactory. At 100mA it was hopeless. I replaced the inductor 10µH with 100µH (out of interest) and it was happy. Those Mini360 have no filter capacitors so they need to be added. The converter you used does have the capacitors which actually worked out cheaper (in my case).
The converter could be adjusted lower, but not much. It is 7V into the 100µH inductor which has ohmic losses so comes out about 6.7V. I wanted to give the LM1085 a bit of headroom as I believe the regulation is better (but that may be false). A bit of wet finger in the air guesstimation (that works).
I think is true. When the specs are published for a linear regulator there are usually 2 regulation figures. 1) Load regulation, That is the regulation with varying load currents, and
2) Line regulation. The regulation performance with a varying supply.
I know some cut it a bit fine but I like to have about 3V above regulated output which seems to make them happy and has minimal problems. Nice to keep these sorts of things happy and they seem to go forever.
I usually have a few LM317s around which are very handy as they are adjustable very easily with the addition of 1 resistor. A 120Ω resistor is requires between the output and “Adj” pin at all times to maintain a mandatory load current for adjustment purposes. I think this type is supplemented these days with more modern versions but they all work pretty much the same.
At a pinch you can do something similar with a transistor, 1 resistor and a zener diode or some other reference. Works pretty well.
Perhaps I’m a bit late to the party here - I agree with the assessment that linear regulators are a low-fuss way to get a stable power supply.
For higher efficiency switchmode options, i’d look into anything that Pololu has made in our Regulators category. Their designs are well load tested, graphs supplied. You can also easily look up the device used on board to check out its specs too.