Unstable power supply with Propeller education kit

TL

We've been using Basic stamps to control test equipment in our lab through reed relays. But after the last setup used 5 program slots, I decided it was time to get something easier to program.

I bought a Propeller Education Kit, and set out to make my next setup on the PEK breadboard. Because I was using reed relays again, I used a 2Amp 9V power supply in place of the 9V battery.

A problem has arisen, however: When I connect a reed relay to the PEK power supply, the line voltage goes UP to 4.1 volts or more. I added diodes to prevent back EMF, and added a resistor in line with the relay coil to help damp out oscillations, which seemed to solve the problem. But when microcontroller turns on two reed relays at once, the voltage again jumps up to 4 volts.

Since the circuit need to be able to close three (out of six) relays at once, I have to fix this. The relays are 250 ohms and can draw 20 mA each. (less with the resistor in line) Basic Radio Hut junk, but it seems like we're always in a hurry, and they're just a mile away.

Any ideas? I'm much better with software than electronics, and this is a bit beyond me.

-TL

T Chap

Do you have a schematic of what you are trying? The single time I had volage fluctuations higher than the 3.3V was when I left out the 10uF on the 3.3VLDO output. Since you are using a breadboard, I suspect some ground issue or check that the cap is there. With a schematic of your relay setup it might be easier to tell.

Post Edited (originator) : 3/9/2007 7:49:26 AM GMT

TL

The circuit is pretty simple:

6 propeller pins are connected to one side of six different reed relay coils, and the other side was connected to ground. This resulted in >4 volts between Vss and Vdd when one or more propeller pins were driven high, or when one of the wires was disconnected from the propeller and connected directly to Vdd.

The following changes were made: The other side of each reed relay coil was connected to ground through a 100 ohm resistor. This was determined by experimentation to be the lowest resistance that would keep the voltage at 3.3 V.

A diode is mounted "backwards" across the coil to eliminate the back EMF generated when voltage is suddenly withdrawn from the coil. (Trick learned after frying a few Basic Stamps)

The fix eliminates the problem for a single reed relay, but not for two or more. Increasing the resistor to the largest resistance that will still close the reed relay also isn't working.

There are a few other things connected to the propeller, like a 4x16 line serial LCD display and an optoisolator to monitor the state of an external circuit, but that's it.

Any ideas?
-TL

T Chap

I would try an NPN transistor on the low side of the relays first, or a Darlington. You may be trying to source too much at once off the Prop. Check the manual for the max current per pin.


** I/O Port A Can source/sink 30 mA each at 3.3 VDC.

Do not exceed 100 mA source/sink total across any group of I/O pins at once.

Meter your current per relay, see if the aggregate is exceeding the 100mA limit.

Post Edited (originator) : 3/9/2007 8:18:08 AM GMT

TL

Somebody said...
You may be trying to source too much at once off a pin.

I was doing that originally, when I was driving two relays from one pin. I fixed that, and per the manual, I should be OK now.

And the problem still occurs when I connect the high end of the coil directly to Vdd, without using the propeller at all, so I don't believe that is the problem.

One of my more electronics-savvy coworkers believes there is some sort of oscillation occurring between the coils and the power supply. If this is that case, would it really do any good to put a transistor in to switch the relay?


Also, I needed to add the 0.47 uF cap across the 3.3V output to drive the propeller even when it was running off the battery.

-TL

T Chap

Have ou tried pulling the relays off the Propeller completely, just putting 3.3v on one side and GND on the other? If you did that and still got a voltage increase, then you have some power supply issue(check for the 10uF). I would start there, meter everything out again, check all GNDs are solid on the board to the power suplply, check that the 3.3v regulator is installed correctly, as something doesnt add up. Personally I would not run relays off the pins directly, a Darlington or NPN would be typical on the low side. If you are positive you have the diode the right direction, leave it in, otherwise try without it.

Post your code if you can that you are testing with. Use the code block feature to keep the indent the same of the forum.

Post Edited (originator) : 3/15/2007 4:42:27 AM GMT

TL

I'm certain the code isn't the issue, because I can make the problem happen while bypassing the propeller pins entirely, with the generic Parallax LCD test program running in the background. One the other hand, I haven't tried completely extracting the propeller to actually prove that there wasn't some interaction between the chip and the board. Good idea - I'll have to try that tomorrow.

I know the diodes across the relays are in the right way, because if they're backwards, the current goes straight to ground without going through the relay coil! [noparse]:)[/noparse] And all it does is damp the voltage spike when the coil is turned off. The propeller was resetting repeatedly before we added the diodes, and now it is fine. Either way, it doesn't really affect the overvoltage problem when they're in or out.

The reed relays are supposed to be fairly low power devices, safe for such applications. And they're braindead simple to wire up. We've been running them with Stamps for quite a while now.

-TL

T Chap

I would not continue testing with the Prop in the circuit with voltages higher than 3.3v, you are cruising for a dead Prop. You should test the 9v battery > 5.0 > 3.3v Reg and see what happens on the relays regarding the volage rise. Recheck your connections on the 5v ref into the 3.3. Meter GND on both the middle pins on the regulators, meter GND on the Prop. Start from scratch with your power connections, the 5 volt output should be looking at the in of the 3.3v, with the large value caps + side on each regulator output. Try to duplicate the problem and meter the 5v reg output, then the 3.3v reg. The reed could be causing some strange behavior on the regulators. Someone else may have better ideas in the daylight.

If you have a camera that can take a clear picture of your board, maybe something will be more obvious.

Verify you are doing this with the electro caps, insure the caps polarity is right. Add some .01 and .1 caps near each regulator out, at the Vcc near the relays, at all Vcc pins on the Prop, at the Vin to the regulators, across all power busses (4 busses that I can see). Test all 4 power busses for Vcc and GND.

www.national.com/pf/LM/LM2940.html

www.national.com/pf/LM/LM2937.html

Post Edited (originator) : 3/9/2007 5:56:23 PM GMT

Graham Stabler

If the regulators are linear then I don't see the fact the relay is inductive as a factor in increasing the voltage in the supply when you just place it on the rails.

Could it be your 9v supply that is causing the problems?

Another thing to check is how good a connection the regulator components are making with the breadboard, a loose capacitor might cause trouble as the current draw goes up.

Its rather mysterious to me.

Graham

Peter Jakacki

Forgive me if I have come on this thread a little late but I am guessing that the inductive load across the 3.3V supply is causing the increase in voltage?

I took a look at the PEK schematics and noticed they were using an lm2937 with a surprise surprise 1,000uF cap (just to make sure). The LM2937 uses a PNP pass element which can exhibit stability problems if not properly decoupled. add a 10uF tantalum as close to the output and ground pins to the regulator as you can manage and your problems will disappear. The regulator is actually trying to compensate but overshoots and then undershoots etc. so it is oscillating badly and the average DC value is what you are reading.

I have no idea why they have such a large cap (and it's probably an electrolytic!!) on the output, the capacitors ESR and inductance will be too high to enable the regulator to be stable when dealing with load transients which inductive loads will certainly give you.

*Peter*

OzStamp

Hi

You nailed that one Peter.. for sure.
We have had an issue here also with one set that went to a customer.
A 10 uf and a 0.1uf fixed it ... yes they are electrolytic ones...big 1000 uf suckers...
We locally inlcude these parts (10 + 0.1) to solve that ..

Hopefully somebody at Parallax will see this thread and fix this issue.

Ronald Nollet (Parallax OZ disty)

TL

Well, I've got the 0.47 uF cap directly acrosss the 3.3 V outputs. It's an electrolytic cap that comes with the board.

The instructions say you won't need it with the 9V battery, but you will need it with an power supply to eliminate noise. Well, I couldn't get the prop to boot off the 9V battery until I added that cap!

It's my day off, but I'll drop by the lab this afternoon and try replacing the 0.47uF with a 10 uF cap. I'll see if I've got any in the parts barrel, otherwise I've have to drop by Radio Hut.

Ozstamp: when you say 10 + 0.1, I assume those would both be installed in parallel across the outputs?

If that doesn't work, I'll disassemble the power supply monday and rebuild it. I'm also ordering some solid state relays and optoisolators for future use. I think I'll make a big hurking driver panel of SSRs or optoisolators (whichever ends up be easier to use and more stable for our oddball aps) and just jumper to the microcontroller the driver panel to run whatever junk we need to to use at the moment. I really don't want to go back to the Stamps and their clunky bank switched EEPROM.


-TL

T Chap

The links above that I posted for the regulators clearly show a .47 on the INPUT of the LM2940 and a 22uF on the OUT. The LM2937 shows a .1 on the IN and a 10uF on the OUT. Without the larges caps on the outs the regulators will be wild, but again this was mentioned in my first response which I supposed was already installed.

"Well, I've got the 0.47 uF cap directly acrosss the 3.3 V outputs" >>> will not work.

Post Edited (originator) : 3/9/2007 9:05:15 PM GMT

Propability

I'm with originator on this·after having the same issue with my PE . The .47 goes to the input of the 5v reg IIRC . Also things get a little tite in that area .I moved the power hookup away from that area so as not to disturb anything. When I first set it up ( before hooking power to the Prop) I read the 3.3V ·at 6-7 V and wondered what the heck.·Rechecked and·went backwards ( removed the 3.3 side leaving the 5 side) . the 5 side was OK then redid the 3.3 and it read OK so there was a loose fit on my first try . If you just had the .47 on the output of the 3.3 it might allow the prop to run but not at high current draw ( just a guess).

Pete

Andy Lindsay (Parallax)

OzStamp said...
We locally inlcude these parts (10 + 0.1) to solve that ..

Hopefully somebody at Parallax will see this thread and fix this issue.

Indeed we have seen this thread.· Thanks for all the input;·we'll review and revise the regulator design.·

I haven't seen these symptoms even though I've used a variety of supplies and load characteristics with a scope hooked up.··Is there a certain combination of supply and load (and/or supply line length) that I can use to reproduce the problem?·

Peter Jakacki...
I have no idea why they have such a large cap (and it's probably an electrolytic!!) on the output, the capacitors ESR and inductance will be too high to enable the regulator to be stable when dealing with load transients which inductive loads will certainly give you.

Why do we have such a large electrolytic?· The output capacitor was originally selected for the LM2940·5 V regulator on the Board of Education carrier board to solve a problem with transients introduced by servos.· It's been in circulation for quite a few years with no complaints.··The cap's ESR·is also compatible with the LM2940 datasheet's stated·output requirements.··The PE Kit was also tested·with a variety of projected educational applications, including groups of servos and motors with no hiccups.· The assumption was that if it got used for industrial apps, separate and/or isolated supplies would be added to the board.

None of that history changes the fact that your solution fixes a reported problem.· We'll take a closer look.··With the PE Kit, there's no board rev or waiting for a shipment.· Just add a couple of caps, and revise the docs, and it's done.·

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Andy Lindsay

Education Department
Parallax, Inc.

Post Edited (Andy Lindsay (Parallax)) : 3/10/2007 4:49:17 AM GMT

TL

I was only able to drop by the lab for a few minutes Friday. I checked the 0.47uf, and it was in the right place after all. Didn't help any, tho. I'll change the other caps Monday to see how that works out.

Originator is correct about the inductive loads. Using resistors with the same resistance as the reed relay coils causes no troubles at all.

I'd stick a scope on it, but I'm a mechanical engineer and software guy. Not a sparky. Even if I managed to hook up the scope without letting the magic blue smoke out of everything, I wouldn't know how to flatten the waveforms I saw. It would really be nice if someone would post a schematic of the entire revised power supply circuit to assist .

-TL

Peter Jakacki

Again I had a rather detailed reply but lost it when the browser crashed. This version is a lot simpler. I'll just quote the datasheet and let you draw your own conclusions. BTW, the input cap of 0.1uF is only required if the input filter cap is more than 3" from the regulator.

*Peter*

LM2937 datasheet said...

EXTERNAL CAPACITORS
The output capacitor is critical to maintaining regulator stability, and must meet the required conditions for both ESR (Equivalent Series Resistance) and minimum amount of capacitance.

MINIMUM CAPACITANCE:
The minimum output capacitance required to maintain stability is 10 μF (this value may be increased without limit). Larger values of output capacitance will give improved transient response.

ESR LIMITS:
The ESR of the output capacitor will cause loop instability if it is too high or too low. The acceptable range of ESR plotted versus load current is shown in the graph below. It is essential that the output capacitor meet these requirements, or oscillations can result.

It is important to note that for most capacitors, ESR is specified only at room temperature. However, the designer must ensure that the ESR will stay inside the limits shown over the entire operating temperature range for the design. For aluminum electrolytic capacitors, ESR will increase by about 30X as the temperature is reduced from 25°C to −40°C. This type of capacitor is not well-suited for low temperature operation. Solid tantalum capacitors have a more stable ESR over temperature, but are more expensive than aluminum electrolytics. A cost-effective approach sometimes used is to parallel an aluminum electrolytic with a solid Tantalum, with the total capacitance split about 75/25% with the Aluminum being the larger value. If two capacitors are paralleled, the effective ESR is the parallel of the two individual values. The “flatter” ESR of the Tantalum will keep the effective ESR from rising as quickly at low temperatures.

TL said...
I was only able to drop by the lab for a few minutes Friday. I checked the 0.47uf, and it was in the right place after all. Didn't help any, tho. I'll change the other caps Monday to see how that works out.

Originator is correct about the inductive loads. Using resistors with the same resistance as the reed relay coils causes no troubles at all.

I'd stick a scope on it, but I'm a mechanical engineer and software guy. Not a sparky. Even if I managed to hook up the scope without letting the magic blue smoke out of everything, I wouldn't know how to flatten the waveforms I saw. It would really be nice if someone would post a schematic of the entire revised power supply circuit to assist .

-TL

Andy Lindsay (Parallax)

TL,

Let's get your project down the road via a tech support call. Below is a link to Parallax' toll free tech support number. Ask the receptionist for Andy when you call.

If you are outside the US, please email me your phone number and a good time to call. (Use the email button to the left of this post.)

Parallax Telephone and FAX Numbers
http://www.parallax.com/html_pages/tech/support/phone_fax_email.asp

Thanks, Andy

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Andy Lindsay

Education Department
Parallax, Inc.

Post Edited (Andy Lindsay (Parallax)) : 3/12/2007 5:11:18 PM GMT

Andy Lindsay (Parallax)

TL,

Is this the relay you're using?

· OMR-112H

Please let me know; I'd like to test it on my PE Kit.

Thanks, Andy

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Andy Lindsay

Education Department
Parallax, Inc.

Post Edited (Andy Lindsay (Parallax)) : 3/13/2007 5:28:21 AM GMT

Andy Lindsay (Parallax)

Peter,

The datasheet quote still doesn't explain the oscillations. The ESR of the capacitor we are using is .36 ohms at room temperature, which is within the LM2940 datasheet's 0.1 to 1.0 ohm requirement. Further, the datasheet says that the output cap's·capacitance doesn't need a ceiling. It does point out that the ESR of an electrolytic cap can get too large when it gets really cold. Again, that doesn't account for the oscillations unless TL's prototype is in a refrigerator.

The kit parts were selected assuming classroom/lab use (room temperature). This in mind, the tantalum cap is one of the items I excluded to keep the cost as low as possible for students. I apologize for not having documented that in the Setup and Testing lab. I will update the docs with an explanation about the expected temperature range and how to modify the circuit for cold weather, noisy inductive loads, and whatever else we can come up with.

Andy

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Andy Lindsay

Education Department
Parallax, Inc.

OzStamp

Hi Andy

A combination of a very low load ( few milli-amps) and the load being inductive has a bearing on PSU circuitry.
It is common on an industrial powersupply that for example drives several stepper motors and
possibly some relay's to shunt it with a resistor ( so that there is a minimum pure resistive load)
Highly specialized power supplies like we sell also to industrial customers when used with stepper motors
that do super fast indexing (start and stoppping rapidly with lots of Back EMF) have a built in "BRAKE RESISTOR"
This effectively shunts away high voltages ...
Now a simply led + resistor drawing 10-20mA may well be worth testing on your lab kit ...( as a permanent load)
This would also indicate if there is actually power there... low cost ..
Just a thought for you.

Ronald Nollet

Peter Jakacki

Andy Lindsay (Parallax) said...
Peter,

The datasheet quote still doesn't explain the oscillations. The ESR of the capacitor we are using is .36 ohms at room temperature, which is within the LM2940 datasheet's 0.1 to 1.0 ohm requirement. Further, the datasheet says that the output cap's capacitance doesn't need a ceiling. It does point out that the ESR of an electrolytic cap can get too large when it gets really cold. Again, that doesn't account for the oscillations unless TL's prototype is in a refrigerator.

The kit parts were selected assuming classroom/lab use (room temperature). This in mind, the tantalum cap is one of the items I excluded to keep the cost as low as possible for students. I apologize for not having documented that in the Setup and Testing lab. I will update the docs with an explanation about the expected temperature range and how to modify the circuit for cold weather, noisy inductive loads, and whatever else we can come up with.

Andy

Hi Andy,
I understand where you're coming from. Personally I found it unusual that such a capacitor is used in this manner. However considering that it is a low-cost education kit then obviously there are good reasons for using such a capacitor. But someone always manages to make a working design go bad somehow. I would never of thought of using 3.3V reed relays off the same supply as the processor. It is normally standard practice for me to give any inductive load another supply. But it's called an education kit for more than one reason. We all learn.

Oh, and Ron's suggestion of a resistive load or power led is a good one for many reasons as you know. Sometimes I combine the power and a 'heartbeat' led into one indicator by only supplying a little current to the led from the supply and also feeding the led from another lower value resistor attached to a port pin. That way if you have a dim led then you know there is power but no activity. Of course, the cpu must toggle the port pin to have it flash steadily.

*Peter*

Beau Schwabe

TL,

Is there anyway we can get a visual of your circuit design?· There are only a few occasions that I have seen the output of a voltage
regulator increase with load.· The +%20 voltage increase that you are observing seems to be extreme.


Here is a regulator design that has been used on several occasions where the power supply has been less than favorable in terms
of noise.· The stable 5V output can cascode into a 3.3V regulator similar to the PE Kit design.

Note: the schematic shows a 7805 regulator... This could just as easily be the LM2940······


Circuit operation:

The diode connected to Vin of the regulator in combination with the 1000uF capacitor form the first stage diode-capacitor sample
and hold style of filter.· A second stage filter is connected to the Vout of the regulator and the 10uF capacitor similar to the
first stage.· Because a diode will drop the voltage across it's PN junction by roughly .6V, it is necessary to offset the voltage
regulator's GND connection to compensate for the second stage diode-capacitor filter in this design.· Thus the reason for the diode
on the GND connection of the regulator.· The 100uF capacitor functions as a standard output stage regulator filter cap.








▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Beau Schwabe

IC Layout Engineer
Parallax, Inc.

bambino

Beau,

I've placed a diode between Supply and the 5V regulator simply
because I've tried to hook up the 9V battery in reverse at 3:00 in the morning
before. Is the diode to GND needed in that situation, or just when a second stage
is used. So far so good, just wondering if not haveing one on GND will come back
·to haunt me

Graham Stabler

Am I being thick but if he gets a voltage increase when the coil is just put straight accross the regulator then its just going to act as a resistor as the current is DC?

I understand potential problems when switching but he said:

"And the problem still occurs when I connect the high end of the coil directly to Vdd, without using the propeller at all"

Graham

Beau Schwabe

bambino,



The GND diode is only necessary if you implement the second stage diode-capacitor filter...



Because of the diode in the second stage filter, it drops the regulator output by 0.6V. To compensate

for this voltage drop, placing a diode on the regulators gnd terminal will effectively increase the regulators

output terminal to 5.6V with reference to GND.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Beau Schwabe

IC Layout Engineer
Parallax, Inc.

Post Edited (Beau Schwabe (Parallax)) : 3/13/2007 5:44:43 PM GMT

bambino

Thanks Beau,

Good Eye Graham!

TL

Here's the existing circuit, showing only one of the 6 reed relays.

Version 1 had only the coil, LED, and LED stepping resistor.
Version 2 added the diode across the coil. This solved a few problems we had with voltage spikes on the BS2px when the power was removed from the coil. Didn't help any here.
Version 3 added the 100 ohm resistor in series with the coil. I tried increasing this to just below the resistance that would prevent the relay from closing, to no effect.

The voltage goes up when the pin connected to the propeller is driven high, either by the propeller itself or by disconnecting the wire from the propeller and connecting it to the Vdd directly. Connect two relays and the voltage goes up dangerously.

And yes, Andy, OMR-112H is indeed the relay we're using. The circuit the relays are driving is totally isolated from the propellor. I could tell you what that circuit does, but then I'd have to kill you.

Turns out I won't be able to rewire the power supply until Thurs or Friday.

-TL

T Chap

You need 10uF on the outs of the regulators. Review the datasheets posted days ago for those regulators, one of them shows a 22uF, one shows a 10uF. I am curious where you got that schematic from?

Post Edited (originator) : 3/14/2007 4:28:09 AM GMT

Peter Jakacki

Graham Stabler said...
Am I being thick but if he gets a voltage increase when the coil is just put straight accross the regulator then its just going to act as a resistor as the current is DC?

I understand potential problems when switching but he said:

"And the problem still occurs when I connect the high end of the coil directly to Vdd, without using the propeller at all"

Graham

A regulator ain't a battery and if it were a battery then it wouldn't be a problem. The regulator attempts to match the output voltage with it's reference and when the load current increases (senses a voltage drop) it needs to decrease it's series resistance by means of the pnp pass element to keep the voltage at the same level. Trouble is that inductors aren't resistors and the current lags behind the voltage just as with capacitors the voltage lags behind the current. If you place a voltage across an inductor, initially it won't draw much current but then that rises exponentially to the level of the coils dc resistance (saturation). Of course remember too that the energy is stored in it's magnetic field and when the voltage is removed the field collapses quickly and can generate a high-voltage spike unless that energy is shunted (diodes/caps etc).

Now the pnp style regulators are a little slower than npn or pmos and so their attempt at compensation is a little out of synch with the load which is already playing tricks with the current. Remember, the regulator is sensing the voltage, not the current, because if it did sense the current it would be more in synch with the load. Ok, so this explanation could get complicated and I am trying to keep it short so I'll leave it at that and just remember to make allowances for regulators and various loads.

*Peter*

Peter Jakacki

TL said...
Here's the existing circuit, showing only one of the 6 reed relays.

Version 1 had only the coil, LED, and LED stepping resistor.
Version 2 added the diode across the coil. This solved a few problems we had with voltage spikes on the BS2px when the power was removed from the coil. Didn't help any here.
Version 3 added the 100 ohm resistor in series with the coil. I tried increasing this to just below the resistance that would prevent the relay from closing, to no effect.

The voltage goes up when the pin connected to the propeller is driven high, either by the propeller itself or by disconnecting the wire from the propeller and connecting it to the Vdd directly. Connect two relays and the voltage goes up dangerously.

And yes, Andy, OMR-112H is indeed the relay we're using. The circuit the relays are driving is totally isolated from the propellor. I could tell you what that circuit does, but then I'd have to kill you.

Turns out I won't be able to rewire the power supply until Thurs or Friday.

-TL

Hi TL,
I noticed the led in the circuit which comes on when the coil is off. If you place the led in series (1.6V drop) with the 100R and tie this to the +5V line then you will isolate the inductive load from the 3V3 and you will find that the circuit will be a lot more stable. BTW, the led will then come on when the coil is on. That is the way that I would do it.

*Peter*