Repairing panel volt-amp meter

[soldar]

About three or four years ago I bought on eBay some mains panel meters like this one and all have had the same problem. They seemed to work correctly at first, for some months or years, indicating voltage and current correctly, but after some time the voltage reading starts to climb slowly. While my meter indicates the mains voltage is still 230 volts, the panel meter started to climb and one of them is now indicating something like 280 volts. The current measurement reading seems to not be affected and I believe it is still reading the same as it was reading at first.

I do not think it is worth much effort so it would have to be an easy repair. The Amps reading seems good enough for me and that is really what I am most interested in. 

Any ideas?

[nsrmagazin]

This issues is probably do to a bad capacitor which gets charged to a higher voltage gradually.

Can you use a multicet to measure the voltage on different components and see on which one does the voltage rise to "280VAC", star from the input and gradually go the output.

[soldar]

There are two PCB's full of SMD's and they are parallel and very close so it is almost impossible to get in there. I assume one PCB is the display driver and the PCB in the back is the one that does the measuring, etc. I have measured and compared the resistor values I had access to and found no difference between a new, good unit and a bad one.

I guess it could be an electrolytic cap going bad but changing those would be quite a challenge.

I can buy another unit for under $10. On the other hand it seems they are not very lasting or reliable.

[CatalinaWOW]

Seems likely that the board side which is visible is just isolation from line and power development.  All of the interesting stuff is behind.  It isn't obvious from the pictures how the back board is mounted.  If the units were totally useless I would just do a destructive tear down on one to find out what is going on.  Since you are happy with current reading that option isn't as attractive.

I would look for how the reference voltage for the A/D is developed and how it might be degraded.  At this price point it is probably integrated into the A/D, but may have a couple of external components for filtering.  A leaky capacitor might pull it down, leading to a high reading.  Might even be something as simple as residual flux and contamination.

[soldar]

Seems likely that the board side which is visible is just isolation from line and power development.  All of the interesting stuff is behind.  It isn't obvious from the pictures how the back board is mounted.

It seems to me the board closer to the display does only the display driving and the upper board does all the measuring and conversion.

The boards are only connected by eight pins in two groups of four. I have circled them in the photo. Four in the upper right and four in the lower left. That is all the connections between boards.

Then there are a couple wires for the LED backlight in the lower part.

The voltage screw terminals in the upper left and the current sensor with two pin connector.

As several units are failing in the same way I suspect some bad batch of some type of components. And in my experience electrolytic caps fail more than anything else and this is twice as true if they are cheap Chinese caps.

I could try replacing them and see what happens but it is almost impossible to get to them without some major work which I am not very inclined to do.

[CatalinaWOW]

I was more speaking of the mechanical connection between boards.  Are they soldered together via standoff pins. 

I agree that bad caps are high on the list of suspects, but none of yours bulge or show other obvious signs of failure. 

I have often found cracked or otherwise high impedance joints in the input line.  If the back board is actually a single board panel meter you could get lucky and find that touching up those interboard connections solves the problem.

[soldar]

I was more speaking of the mechanical connection between boards.  Are they soldered together via standoff pins. 

The eight pins soldered on those two corners provide all electrical and mechanical connection. There are no other connections between PCB's.

I have several bad meters and one that is still good because I have not used it. I guess I could try seeing with a scope what kinds of signals I see at the pins and other places but I prefer not to try this because it is fed 230 Vac and a slip could be disastrous.

These SMD boards are so compact you can hardly get to the components.

[nsrmagazin]

I see at least 7 capacitors that you can check.

[soldar]

Yes but first I would have to separate both PCB's to gain access to the other side and chances are I would do some damage doing that or trying to remove the caps. I am not very experienced in working with such miniature SMD boards.

Since the Amp measuring seems to be OK I would rather not run the risk unless I thought the chances of success were good.

[soldar]

I had assumed the AMP current reading was good but have realized it was also reading too high.   I have three units and all have the same problem: after a while both volts and amps are reading about 25% too high.

After some experimenting i figured the current pickup toroid coil (transformer) went to a resistor on the board of about 0.1 ohm and it is easy enough to put in parallel  an adjustable resistor of 0.4 ~0.5 ohm and adjust it until the amps reading is correct.

For the volts reading I put a resistor in series with the supply in order to lower the supply voltage. About 30K seems to do the trick for now. This has the disadvantage that the backlight LED is barely visible but that's a separate problem.

Since both volts and amps readings seem to have drifted off in the same direction and amount I am thinking they may share the same ADC.

For now I will install the panel meter in a box where I have easy access to change the resistors if the meters continue to drift in the future. 

The adjustable 0.5 ohm is just a piece of nichrome wire between two screw terminals and I can adjust the value easily.

I guess I could put an adjustable resistor in series with the power supply wire. The current is very small and should not present a heat dissipation problem.

I could also put a separate LED to light it.

This work-around has turned to be easier than I thought at first.  I will continue to work on this.

[schmitt trigger]

Do not under estimate the board surface contamination creating leakage paths in high impedance nodes.

It will cost you only a couple of Euros to purchase a bottle of flux cleaner or IPA. And then thoroughly clean the boards. By thoroughly, I mean cleaning, allowing the board to dry and then cleaning again.

[soldar]

Thanks, I had not thought of that in spite of being something I have encountered in the past. I might look into it although separating the boards is problematic.

For now I will implement my work-around on the unit I have working and will play with the spares.

The other problem now is that the LED does not give enough light. Fortunately I have access to the LED terminals so I could implement an "external" supply for the LED. Maybe just a capacitor in series with the 230 V supply followed by a rectifier bridge. I need to think about this.

[Mark Hennessy]

I reverse-engineered one of these these a while back: http://www.markhennessy.co.uk/mdu/ - E&OE

I think I'd start by checking there is the right voltage across the zener diode...

[soldar]

WOW Mark, I am impressed! That helps a lot! It seems to be exactly the same as mine except on mine the current transformer coil is not on-board and there is a connector for it.  I will study that page carefully and, indeed, the rest of your site. Thanks so much! I wish I had found it sooner.

BTW, these units with current transformer coils are supposed to go up to something like 100 A but they sell pretty much the same device with straight current measuring and they are supposed to go up to something like 10 or 20 Amps.

In your schematic the current sensing resistor is 0.68 ohm but I measured mine at something like 0.1. While the design could be different I am thinking it is also possible the value increased over time. I have found several instances in past repairs where SMD resistors had changed value or blown, probably because they were dissipating more than they could. It does not seem to be the case here with the 0.68 ohm resistor though. I will have to measure and compare the values of all components on the back board.

With your schematic I can proceed with testing much more. I was not quite prepared to spend the time to reverse engineer the thing. I have reverse engineered quite a few things, mainly SMPS, and it takes me many hours. Too much work.

Thanks again for that valuable info. I will keep this thread updated as I make progress.

[stj]

check the red polyester cap,

[soldar]

check the red polyester cap,

Yes, after looking at Mark's schematic it looks like a prime suspect.  Mine is marked 470 nF / 400V, CBB22. Searching for CBB22 it says "Metallized Polypropylene film". Why would such a capacitor suffer any changes? It is not electrolytic.

At any rate, I measured one of them and sure enough, marked as 470nF it measures as 75 nF. What would cause this?

So now I need to replace it. Mark Hennessy says to replace it with a safer "X2" type. Can anyone explain the difference?

While I like the safety part I also want something that will not need replacing in a couple years.

How about this: https://www.ebay.com/itm/112240022614

[stj]

i posted that before i even looked at the link, i have seen them fail in led lights.
i suspect they cant handle constant current flow - there are ones specifically designed for high current handling.

[Armadillo]

I am puzzled, I don't know really, as to how a supposed failed ac voltage divider Poly cap 470nF -> 75nF be related to your voltage creeping up problem from 230V -> 280VAC.
 

[soldar]

I think it may be that the reduced capacitance results in lower voltage than the 9V from the zener and the lower supply voltage to the ADC results in false readings.

I am going to try to replace the cap and see if that fixes everything ... or not.

Since eBay does not show specific safety caps I guess I will just go with the higher 630V and hope that works and does not degrade.

[Armadillo]

I think it may be that the reduced capacitance results in lower voltage than the 9V from the zener and the lower supply voltage to the ADC results in false readings.

I am going to try to replace the cap and see if that fixes everything ... or not.

Since eBay does not show specific safety caps I guess I will just go with the higher 630V and hope that works and does not degrade.

Possibly, but I supposed the first insight would be the erratic back lighting or display rather than ADC readout which consumed a relative lower current than the former. Nevertheless appreciate the explanation.

Also, no offense to Mark, but specifically aim at retaining the true meaning of "reverse engineering", there must be a distinction between Function Emulation and Reverse Engineering of a board.

[schmitt trigger]

check the red polyester cap,

Yes, after looking at Mark's schematic it looks like a prime suspect.  Mine is marked

So now I need to replace it. Mark Hennessy says to replace it with a safer "X2" type. Can anyone explain the difference?

Have a read; this is the first thing that came up in Google, there are many more:

https://www.hunker.com/13408990/differences-in-x1-x2-y1-and-y2-capacitors

[soldar]

Yeah, thanks, after searching and reading a few pages I know class X and class Y are better and safer than no class (or whatever the others are) but I cannot find those classes on eBay and I am not about to pay more for the capacitor than I paid for the meter so I think I will just go with regular, normal caps from eBay. After all, there is really no security aspect here. I guess I can put a 0.1 A fuse in series in case the whole thing shorts and blows up.

[Armadillo]

Well, it is a divider with other element in series hence it is not "Across the Line" or "Line to Earth" to be precise.

Scare Voodoos I suppose.   

[soldar]

I replaced the 470K cap and it seems the readings are back in range so many thanks to those who pointed me in the right direction.

The replaced cap's capacity had dropped to 75K. I do not understand why or how.

In the unit I repaired the backlight LED is still very dim so I will have to do some further testing to see why. Maybe the series resistor has changed value. We shall see.

[stj]

if the backlight is phosphor doped then the led is dying.