1. Home
  2. Electronics Design
  3. Possible internal bleed over causing latching ?

Possible internal bleed over causing latching ?

A while ago, I posted here problems we were having with a micro inverter drive.. We got a lot of them for a couple of projects and they work however, there was a problem with them and still is. Because we got so many in service I just thought of something today that maybe the cause of our problems.

This problem has been taken up with those that put their name on the drive. They have contacted those that wrote the software in the drive and have found that the software is not the cause of the effects we get under certain conditions but believed to be a hardware problem. Well, they told us that the hardware would not be fixable of course, in our units if it turned out to be that and that seems to be the issue..

Any way, getting to what I think may be the problem now.

These drives supply a 12 volt reference terminal to be used for the speed command on another terminal or power sensor devices that return a voltage command from it, ect.

The speed command input is rated to work 0..10 or +/- 0..10V

What happens now and then is if the input happens to get lets say over 10 volts, like up to 12 volts from the drives REF terminal, because this is coming from a pot. I think it could be causing the ADC circuit internally to cause some near by latching effect on the IO that controls the RUN command. I say this because when the defect shows, the ADC still acts but the behavior is incorrect and the RUN command terminal is completely ignored.

In other words, the drive is out of control and the RUN terminal does nothing and it stays in RUN. YOu have to remove power from the drive to correct it.

I can took my Func Gen which will produce 12 volts for me and programmed a slow Sine wave as a test. I can make it go out of control after some time but not predictable as to when. It seems very random to me.

Dropping it down around 10VP or less seems like it lives happy there however the specs states that the input should be able to handle the 12 volts from its own terminal.

I looked in the drive and it is very low end electronics to put it nicely. The ADC input and run command input terminal go directly to the main uC chip.

I wonder if I am getting an internal leakage that is causing a input near by to latch up from the ADC input being over driven? If that is actually what's happening?

I can fix this of course with a R at the top side of the pot which I should send out a memo on the idea as the problem comes along. Would like some input on this if any one here has experienced things like this ;)

Jamie

Reply to
Jamie
Loading thread data ...

e.

re

ut

I

Please supply more information as in a schematics, quite hard to quess what you are doing without more insight

You say the input is connected directly to the ADC. If so, normal practice would be to add ESD/burst/Surge protection on that input. At least some sort of clamping circuit to avoid driving current into the input of the ADC (results in erratic behavour or latchup).

Check if the problem occurs when the voltage is close to the rails, then you might have a overdrive problem as stated above...

Cheers

Klaus

Reply to
Klaus Kragelund

d

es

me.

re

ut

I
s
d

add a 1k Ohm series resistor in series with the control signals.

It is AMAZING how many problems I have solved over the years with that solution.. :-)

Mark

Reply to
Mark

We took one apart, and the input terminals have very little between them and the input to the uC.. at best, All I see mostly is caps across the inputs. If there are any ESD diodes there, I don't see them unless they are already in the uC?

I don't have a schematic, I can only tell you what I find in it and that is, it's a low end piece of electronics. And driving the input to

12 volts seems to be causing it randomly going out of control, even though the manual states that it can handle it.

A complete power restart is the only thing that will correct it.

Jamie

Reply to
Jamie

What I did for the moment, was to have a R in series with the top side of the pot from the 12 volt reference to bring it down to ~9.5 volts for the full sweep of the pot. I don't need the full range of the sweep any way. That has seem to fix that problem with the test drive and function generator test.

If that is indeed the problem, I wish they would clear that up in their manual.

P.S. If one uses a linear type output sensor, then it would make sense to have a 12 volt reference since most of them that do generate a 0..10 volt out requires a min of 12 volts to power up. How ever, this drive also has a 24 volt utility supply too!

Jamie

Reply to
Jamie

They are in the uC but those are parasitic. If you send more than a few tens of mA into one three things can happen:

a. It locks up something in the uC. That seems to happen in your case.

b. It could drive up the supply rail of the uC causing if to go ... phsss .. *PHUT*

c. It could some day cause a latch-up upon which the uC could self-destruct and pop its lid ... *BANG*

But: How can a uC take a 10V range directly? I can't remember seeing any that had supply voltages higher than 5V. Mostly 3.3V these days. Ok, except for the ones that operated with tubes :-)

So how can you fix this externally? A resistor from top of the pot to the 12V reference is what you've found works. However, potmeters have a fairly large tolerance in their total resistance, they are more meant to generate precise ratios. Your control voltage at any give potmeter setting will now vary more between units.

You could try Mark's suggestion but put the resistor on the outside. See which uC is in there and then see what sort of ADC it has. Most are high-impedance. So you could have a 1k from wiper of the pot to the screw terminal at the box. Maybe even higher. Then if you want to be extra good hang a cap from that screw terminal to ground.

[...] 
--
Regards, Joerg

http://www.analogconsultants.com/
Reply to
Joerg

As standard practice in the industry, a 5k ohm pot is used. 10k also but not as common.

I put a 1k in series from the 12v to the top side of the pot.

input reference is now = 12*(5k/6k)= 10 Volts.

I don't bother with the input loading calculations, I am actually getting closer to 9.5 volts, but you must consider other things like DMM loading and input Z to the drive speed cmd terminal.

yes, there is only a couple of R's and maybe 2 caps before it reaches the IO's of the uC.. Nothing active components I can see before that.

Jamie

Reply to
Jamie

Ok, but if you have access to it measure a new potmeter and one from a shipment earlier last year. Might show a not so pleasant surprise.

The input Z should be quite high if they allow 10k potmeters. Control becomes rather non-linear otherwise.

Ah, ok, you hadn't mentioned resistors. If they dimensioned them correctly points b) and c) should not happen. a) is often due to suboptimal IC design.

--
Regards, Joerg

http://www.analogconsultants.com/
Reply to
Joerg

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.