voltage reg running under voltage

Thread Starter

eng_stu

Joined Jan 29, 2012
13
Hi all,

I have a circuit running at 3.3V from a LM9036 voltage regulator. This is supplied from a 4-20mA loop (loop powered) from a 24V supply. For those of you unfamiliar with a 4-20mA circuit, there is a device (transmitter) which regulates the current from 4mA (0%) to 20mA (100%) as an analog output of a measured parameter (pressure, temperature, etc). Any indicator devices are in series in the same loop to measure the current and display the value and/or be powered by the 4-20mA signal. My board is powered by the 4-20mA signal (loop powered).

In effect my voltage reg is being supplied by a constant current supply. My circuit is designed to use less than 4mA (I think it's about 3.2mA). The rest of the current is sunk by a 5.2V zener in parallel with the voltage reg. The idea is that when the voltage reg is getting too much current it will raise its input voltage to resist it and then the zener will kick in.

The problem is that although the voltage reg is a 3.3V one it often sits at 2.6-2.7V. If i run the supply up to 18-20mA then it will go to 3.3V and stay there even if I come back down to 4mA. My circuit still works at the lower voltage but I feel its going to bite me later on and either die or become unreliable. Plus I would like to understand why so I can build better circuits in the future.

I'm guessing it's because the voltage reg is being starved or it's in an unstable state. On the DSO the voltage is stable and it starts up ok (quick rise, slight overshoot then settles). I have the minimum recommended capacitance but adding more doesn't help. The LM9036 datasheet says the ESR should be between 0.1 and 8Ω but the capacitor datasheet doesn't have ESR only tanδ. Tanδ is 0.2 at 120Hz. I worked out the ESR was about 6.7Ω at 1kHz but I'm not confident in that value. I'm not even sure if it's relevant.

Anyway, has anyone come across this sort of power supply issue before? I couldn't find anything on any of the forums or the web in general.

Thanks for reading,
Stuart
 

ifixit

Joined Nov 20, 2008
652
You should post a schematic so we can see if if is connected correctly.

When it is in the failure mode, what is the voltage at the regulator Vin? It should be 5.2 Volts, and it needs to be at least 3.5V, or the reg out (3.3) will drop.

You are right to be concerned about this anomally. It needs to be understood and fixed.

Regards,
Ifixit
 

t_n_k

Joined Mar 6, 2009
5,455
Perhaps try placing a suitable resistor in parallel with the zener diode or maybe a forward biased series connected diode string in place of the zener - say 6 diodes [~4V total forward voltage] with a continuous current rating greater than 20mA. You could probably leave the zener in place anyway - in parallel with the diode string.
 

bountyhunter

Joined Sep 7, 2009
2,512
The problem is that although the voltage reg is a 3.3V one it often sits at 2.6-2.7V.
Here's why:

1) Input voltage too low for the regulated output to stay at 3.3V

2) Device in thermal shutdown, output collapsed

3) Drawing too much current and limiting circuits activated

4) Regulator oscillating from various causes
 

Thread Starter

eng_stu

Joined Jan 29, 2012
13
When at 2.6-2.7V the Vin is always about 0.2V higher than Vout, as per the spec sheet and stays that way until restored. That voltage does climb as the current increases (it might be 2.9V at 12mA and 3.1V at 16mA). I originally used a 3.9V zener but the breakdown voltage is really horrible at higher currents (like 20mA) on the smaller value zeners, it could get over 7V and I can't afford that much voltage drop in the loop. The 5.2V zener has a much flatter response.

It's not thermal shutdown because the reg is rated to 50mA and the supply is current limited to a maximum of 20mA.

It must be too much current draw for the supply to handle. It just seems strange that the voltage wont recover until several times more current is supplied and that it exhibits the hysteresis effect where it will be fine after increasing then decreasing the current back to the same value.

I will measure the current again when I get a chance.

@bountyhunter: if the regulator is oscillating would I see it on the osciloscope or would the capacitor smooth it sufficiently to become not visible? I didn't see any oscillations on my 50MHz/1GS/s DSO. Would/could it be a faster frequency than I can measure? What causes oscillations? Is it the reg being pushed to its limits in voltage, current draw, thermal effects, previously ESDed, EM effects etc?

@t_n_k: Are you thinking my zener isn't responding as well as I need it to, in terms of sinking the appropriate amount of current with the appropriate time response?
 

crutschow

Joined Mar 14, 2008
34,282
If the input voltage is only 0.2V over the output voltage of 2.6-2.7V and you are drawing less then 4mA, then where is the extra current going from the 4-20mA loop? Something is wrong there.

Can you post your circuit?
 

bountyhunter

Joined Sep 7, 2009
2,512
@bountyhunter: if the regulator is oscillating would I see it on the osciloscope or would the capacitor smooth it sufficiently to become not visible? I didn't see any oscillations on my 50MHz/1GS/s DSO. Would/could it be a faster frequency than I can measure? What causes oscillations? Is it the reg being pushed to its limits in voltage, current draw, thermal effects, previously ESDed, EM effects etc?

?
You can see the oscillations on a scope.

Please post schematic.
 

t_n_k

Joined Mar 6, 2009
5,455
I was sufficiently intrigued in the end to pull out some parts and do an experiment.

I have a 24V loop powered 4-20mA controller I built some years back. My DC source is actually 22.8V and I decided to try a bog standard LM7805 5V regulator with a shunt zener diode of 10V. Added a shunt capacitance of 330uF as well.

I placed a 3.9kΩ load on the regulator output. The loop controller was set to give 10mA loop current. I also included an additional loop resistance of nominal 330Ω [324.2Ω actual].

The schematic is shown in the attachment. My measurements were taken with a DMM.

It all works fine as one might expect. The regulator gives a stable 5.04V DC output across the 3.9kΩ load.

The question arose in my mind - "What would happen if I remove the zener?"

Removing the zener had an interesting effect. The loop controller could no longer force the loop current to 10mA - rather the loop current settled at ~6.36mA with the regulator output sitting nicely at 5V DC. This makes some sense. The LM7805 has a nominal quiescent current of 5mA and with a load of 3.9kΩ this accounts for the additional current of ~1.3mA.

I suspect the LM7805 is therefore (not surprisingly) behaving in itself as a current controller of sorts. So without the zener to absorb the current difference, the loop controller [set to 10mA] loses the fight with the LM7805 which requires only ~6.3mA to meet its obligations.

If I tried to set the loop controller to give something near the 6.36mA current required by the LM7805 the current loop stability was very flukey. Warming the LM7805 with my fingers increased the instability.

Looks like you need the shunt zener to be in conduction all the time.
 

Attachments

Thread Starter

eng_stu

Joined Jan 29, 2012
13
Sorry for the slow response but I finally got around to doing more tests. At 20mA the voltage is correct at 3.3V and the current in to the reg is 2.7mA. It seems the problem is the startup conditions. I have attached the startup traces for reg voltage and reg current at 3 conditions: loop current at 4mA, 12mA and 20mA. I'm supplying the loop from a battery powered loop injector. I'm measuring the current across a 10 ohm resistor on the input side of the reg, hence on the graph 10mV = 1mA.

Interestingly, on the 20mA current graph the reg draws about 18mA once it stabilizes and then after a bit suddenly drops down to about 2mA. It seems to me that something needs to charge up before the voltage can make it to 3.3V and that something needs 18mA. I thought it might be capacitor in-rush current but I bumped up the capacitance from 20uF to 67uF and also down to 2.2uF but the voltage still sits at 2.63V although the current trace becomes quicker to stabilize and has less oscillations at lower uF's.

Any thoughts? Could it be something to do with the zener breakdown time?
 

Attachments

Thread Starter

eng_stu

Joined Jan 29, 2012
13
I had a think about the zener kicking in and decided to run some more tests. The zener doesn't conduct when the loop current is below 16.3mA at startup. Also the time for the zener to start conducting is shortest at 20mA and longest at 16.3mA. I have attached some graphs showing reg voltage (yellow CH1) vs zener current (blue CH2, 10mV=1mA) at 16.3mA, 16.5mA and 20mA loop current.

However, the zener will conduct below 16.3mA if I go up to 20mA and then back down again.

I also disconnected everything that the reg supplies and replaced it with a 2.7k resistor (the cap is still there though). Thus I have a resistive load with 1.2mA draw. It still behaves the same though.
 

Attachments

Thread Starter

eng_stu

Joined Jan 29, 2012
13
I tried swapping the LM9036 out for the LM2936 because they are quite similar and the LM2936 datasheet has more info in it but they both do the same thing. The LM2936 work a little bit better in that it will reach 3.3V at 16mA instead of 18mA. However, it looks like there is no way around it: If you starve it in startup it will not work properly.

I tried t_n_k's experiment with the 7805 and was getting similar results to the LM9036 but then I realised I was using a 6.2V zener and the 7805 needs at least 7V.
 

Thread Starter

eng_stu

Joined Jan 29, 2012
13
Sorry I got distracted. I have attached a schematic. D1 zener is 6.2V. I have detached C7 and C6 and the problem still occurs. Off Vcc there is a micro and the modem but I replaced them with a 2.7k resistor and the problem still occurs.
 

Attachments

Thread Starter

eng_stu

Joined Jan 29, 2012
13
I solved the problem: throw it out and buy a new one. I'm switching to the TL431 shunt regulator which is ideally suited to this application. I only found out yesterday that shunt regulators existed. This one works like a dream. I can run my circuit right down to 3.5mA and the voltage is very flat. Plus I get to ditch the zeners and my loop voltage drop is nearly half.
 
Top