Notifications
Clear all
Topic starter
2022-11-20 5:23 pm
Hi all,
I have a question about a circuit for a power supply. I'm going to build a power supply based upon an ATX supply, similar to the one in the DroneBot Workshop [1]. There are a few differences, however, and I have a few questions about that.
- I am going to include a VoltAmmeter, similar to the one in the mentioned project (DSN-VC288 or an equivalent one, such as the one used in [2]. I want to use a rotary switch to select the output rail that it will measure. I could also include a separate VAmeter on each output but I think that would be overkill, and also would take too much space.
The use of a rotary switch of course isn't essential. But one thing that bothers me is that these meters measure the current in the negative (ground) line, which means that the ground of the output is not the same as the ground of the power supply, as mentioned in project [2] above. And especially it isn't equal to the ground of the other outputs. I think this is bad in case you have a circuit that needs different voltages with a common ground. Then you will essentially short the shunt resistor that is used to measure the current. Another solution would be to put the shunt in the common ground line, but then you would measure the total current of all outputs, which isn't what I want.
This problem was also described in a stackexchange post without really receiving a solution. - I also want to add a variable power supply, by feeding the 12 Volt to a Buck-Boost converter, and use the output of that as an additional line. Kind of like the power supply in [3] but that one is a Buck only converter. The one that I have seen in a couple of projects is the XYS3580 Buck-Boost converter. This one has a built-in VAmeter, and as far as I can see this uses the same principle of measuring the current with a shunt in the ground line. And this causes the same problem as described above.
Now I have been looking for a solution to this, which would be measuring the current in the positive rail of each output. But this would require a modification to the VAmeters.
One answer to another topic on stackexchange mentions the INA210 family of current mirror ICs, and although the solution there is for use with a microprocessor, not for use on a VAmeter, I thought I could use it for my case also. Basically these chips multiply the voltage difference across a shunt, and output a Voltage relative to ground that is proportional to the current through the shunt. The shunt can be placed either in the ground or the positive rail with these ICs. There are also INA138, INA22x and INA26x versions, although at the moment I don't know which of these are preferable.
In the same post also a chip ZDS1009 is mentioned, but it seems no longer to be available and replaced by ZXCT1009, but these seem to be difficult to get in small quantities.
I'm afraid this post is getting too long, so I'll stick to the essential question now. The basic setup of the VAmeter as intended is as the following diagram:
I want to remove the shunt, and then put a shunt with a current mirror in the positive rail. Fortunately, some of the VAmeters have external shunts, so no desoldering would be necessary. However, with the XYS3580 the shunt is on the PCB, so it must be desoldered. From what I have seen (on video) that shouldn't be too difficult. And no new connections have to be soldered, as all connections are on screw terminals.
So the new setup would be:
and for the XYS3580:
Now my questions:
1) Is this a feasable solution?
2) Does anybody have experience with this kind of setup?
3) Is there a better solution?
4) What are possible pitfalls?
5) As these current mirrors have a considerable gain (could be 50) the shunt must be much smaller than the original one, and/or the output must be applied through a voltage divider. Are there any caveats with this?
2022-11-21 4:46 pm
Hi @pietervo,
I have had similar 'concept' thoughts about putting a power supply together, and wanting to measure the current in the 'live' rail, rather than in the ground return path. Sadly, it is still largely in the 'to do' list with lots of things in front, so I can't offer much in the way of 'hands on' experience.
Current mirrors are certainly used and supported by purpose designed chips, though how easy they are to procure in the present trading conditions I don't know.
---------------
You mention rotary switches for measuring different lines. My last personal experience of these is many decades ago when they were used in radios, etc. In those applications, even though the current flows were modest, many devices failed prematurely with high contact resistance, arcing and so on. The trend in recent years has been to reducing voltage and increasing current, which would increase such problems. Hence, unless you can find some specifically designed for higher current ratings, I would advise avoiding them in a circuit to carry the current flow associated with the output of a power supply rated at more than 1A or so.
Of course, they could be used in the low current areas, such as to select between different 'sensors' or to sample the voltage across shunts. But, electronic multiplexing methods to achieve this result might be better.
Alternately, it might be cheaper and simpler to use multiple ammeters, in spite of the 'overkill'!
---------------
An obvious alternative is to look at the Hall effect devices by companies like Allegro, which include a low resistance path through the package that is electrically isolated from the sensor itself
... and devices in this class also seem to available in the 'cut price' bazaars if that is suitable for your purposes.
e.g. https://www.aliexpress.com/item/1005001798580897.html
(Please note, I only know what I can see in the advert .. I haven't actually used any.)
----------
Another possibility, which was one I was going to try first, is to try to fool one of the digital ammeter modules that measures current in its ground rail by powering it via small DC to DC isolated converter. I haven't tried it yet, so it might have an 'obvious' flaw. Also, some care is needed to ensure the voltage across the shunt is of the correct polarity.
Just wild speculations for you to consider ... no responsibility for any mistakes oversights.
Best wishes,
Dave
Inst-Tech reacted
Topic starter
2022-11-21 5:00 pm
Hi Dave,
My idea was to switch only the measuring lines, not the power lines, so these would have small currents.
Like this:
Inst-Tech reacted
2022-11-21 5:43 pm
Hi @pietervo,
As I said, "they could be used in the low current areas, such as to select between different 'sensors' or to sample the voltage across shunts", which is what you are showing in your new diagram.
Sorry, I don't have any 'first-hand' practical experience of current mirrors to offer, but obviously your conceptual circuit should work.
Make sure your rotary switches are 'break before make' not 'make before break' .. in ancient history, both types were made.
Also, assuming very little current is expected to flow to the current mirror, you might consider putting a resistor into each wire from the shunt resistor to the switch, with the resistor physically close to the shunt. Then, should a fault develop, such as the switch shorting to ground, only a small current will flow. Choose a resistor value which will limit the short circuit current to a safe value, but only introduce an insignificant voltage drop in normal operation.
Best wishes, Dave
Inst-Tech reacted