What's that high-pitched squeal

KenFordham

I just got a Propeller Proto board. Naturally, my first program is a LED blinker. After I downloaded it to the Propeller, it worked fine the first time. But I noticed that there is a very high-pitched squeal that comes and goes with each blink. Has anyone else experienced this? Any ideas on how to prevent it?

Mike Green

Where is the squeal coming from? It doesn't make a lot of sense. What kind of power supply are you using? If it's a switching supply, it might be be due to the increased load when the LED turns on although the increased load is very small.

If your Propeller is running with an 80MHz clock, it draws about 10mA per cog. If it's running with the default internal clock (RCFAST), it would draw maybe 2mA. The pilot LED on the Protoboard draws more than that and the blinking LED wouldn't make that much difference (maybe another 10-15mA).

KenFordham

Thanks for the quick reply, Mike.

I really can't tell where the squeal is coming from.

The power comes from a wall wart that outputs 8.4 volts DC. This is then plugged into the power input on the proto board where the power supply on the board knocks it down to 5v and 3.3v. I unplugged the wall wart to look at it and then plugged it back in. Wouldn't you know it, the squeal is now gone.

Thanks for the help.

Leon

The wall wart might be a switching PS which is operating at a frequency within the audible range. The noise might have come from the inductor which is used in a switcher. It could be varying with the load. Unplugging it and plugging it in again probably displaced the inductor slightly stopping the noise.

Leon

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Amateur radio callsign: G1HSM
Suzuki SV1000S motorcycle

Post Edited (Leon) : 7/7/2008 8:20:17 AM GMT

Paul Baker

Hi Ken, is this an ordinary Protoboard or a Protoboard USB? If it's the later is there a through hole capacitor connected right below the surface mount capacitor for the 5V regulator?· Also take a metal object (I use my wire wrap tool) and place it on the 5V regulator (the one on the left side) and place your ear up to the other end. Next place it on the 3.3V regulator and do the same. Which one·sounds·louder? I know what's causing the noise, I just want more information about your board before I give you a remedy (because this issue should have been addressed in house, and I need to track down how a singing unit got out the door).

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Paul Baker
Propeller Applications Engineer

Parallax, Inc.

Post Edited (Paul Baker (Parallax)) : 7/7/2008 8:30:58 PM GMT

Ken Peterson

Singing voltage regulators....that's a new one to me!

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔

KenFordham

Paul,

The board is an ordinary protoboard (non-USB). I listened to both regulators as you described and couldn't tell any difference in volume level for either one. Both sounded equally loud as I put my ear to them. All I can really tell is that it's coming from the board and not the wall wart.

Ken

Post Edited (KenFordham) : 7/10/2008 2:12:24 AM GMT

Paul Baker

Thanks for the information, it explains everything. I am nearly certain it's the 5V regulator. The solution is simple: solder a 10uF tantalum cap between the 5V and GND observing the correct polarity by placing the red bar pin in 5V. There are two adjacent holes right under the 5V regulator that make this very convenient to do. We can replace your board if you want with a non-singing one, but it would take the time of shipping back and forth.

The reason this is happening is some voltage regulators require the output capacitor to have some amount of ESR (equivalent series resistance) to achieve output stability. Ceramic capacitors are among the lowest ESR capacitors availible. At some point either a more sensitive regulator or an ultra low ESR lot of caps entered the picture, and the output of the 5V regulator began oscillating. Since silicon substrate is piezoelectric, and the output mosfet was designed to be a well regulated DC voltage and not designed to output AC, the substrate vibrates causing the sound heard.

The permanent fix is to replace the ceramics with tantalums, the temporary fix on booards already produced is to add a parallel tantalum across the output of the 5V regulator as I described above.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Paul Baker
Propeller Applications Engineer

Parallax, Inc.

Post Edited (Paul Baker (Parallax)) : 7/10/2008 3:55:59 AM GMT

Phil Pilgrim (PhiPi)

Paul,

Have you ever run into this issue with the LM2937 3.3V regulators? They spec a 0.01 to 3 ohm ESR range. (I was planning to replace the tantalums in a new board design with ceramics to save room.)

-Phil

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
'Still some PropSTICK Kit bare PCBs left!

Tracy Allen

I hadn't "heard" of squealing--that's a weird one! I did just last week have a problem with an LM2940-5, in a circuit where it was supposed to supply a Burr-Brown 1 watt DC-DC converter. It was a new batch of regulators and DC-DC converters on a PCB that had worked fine with earlier batches. The new regulator would often go into a high current lockup state with oscillations. Putting a 2 ohm resistor in series with the output capacitor did make it work, but I ended up patching in a Linear Tech regulator that did not exhibit the problem. More recent regulator designs are less sensitive to ESR. It had something to do with the switching frequency of the DC-DC.

ESR of some ceramic capacitors is not good at all. A high value Y5V 10 volt capacitor may have a dissipation factor at 1000hz of 0.125, that is, 1/8 of the reactive power gets burned up as heat in the capacitor. The capacitive reactance of 1µF at 1000 Hz is 1/(2piFC) = 159 ohms, and 1/8 of that is 20 ohms. Of course X5R and X7R dielectrics are much better, with dissipation factors down to 2%, which is comparable with tantalum capacitors. Putting a tantalum in parallel with a ceramic lowers the overall ESR.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Tracy Allen
www.emesystems.com

Paul Baker

I am quite familiar with the stability issue of some regulators, because I've read the warnings in enough datasheets. However this is the first time I experienced it first hand. It making a noise was a suprise, but a blessing in disguise because it alerted us to the problem. I haven't seen a 3.3V regulator do this either with US or China sourcing, and the 5V regulator oscillation has only been seen through China sourcing. That doesn't mean US sourced parts wont do the same, but they will typically raise the issue in the datasheet if it is a potential problem. (China parts don't have datsheets, rather you find parts by doing a low-tech version of cross referencing, then evaluating it to see if it matches the required specs).

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Paul Baker
Propeller Applications Engineer

Parallax, Inc.

Post Edited (Paul Baker (Parallax)) : 7/10/2008 5:20:24 PM GMT

Paul Baker

Here is the portion of our internal repair doc directly concerned with the installation of the cap, for those who have received one of these boards and wants·to do the repair thierself.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Paul Baker
Propeller Applications Engineer

Parallax, Inc.

Timothy D. Swieter

I can second that fact about China sourcing as being sort of a "low-tech" cross reference!

This was an interesting thread to read and to learn from others experiences.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Timothy D. Swieter, E.I.

www.brilldea.com·- Prop Blade, LED Painter, RGB LEDs, uOLED-IOC
www.sxmicro.com - a blog·exploring the SX micro
www.tdswieter.com

KenFordham

Hey Paul,

Would it be possible to get you to send me one of those 10uf caps? I don't really want to exchange boards since I've already soldered some stuff to it.

Thanks,
Ken

Paul Baker

Sure Ken, call tech support (I'll let them know to expect you) and we'll ship some to you.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Paul Baker
Propeller Applications Engineer

Parallax, Inc.

Rayman

So, was the issue not enough capacitance downstream of the 5V regulator?

Phil Pilgrim (PhiPi)

I have to admit, I'm a bit confused about this myself, given the following assertions:

Paul Baker said...
At some point either a more sensitive regulator or an ultra low ESR lot of caps entered the picture, and the output of the 5V regulator began oscillating.

Paul Baker said...
... the temporary fix on boards already produced is to add a parallel tantalum across the output of the 5V regulator ...

Tracy Allen said...
Putting a tantalum in parallel with a ceramic lowers the overall ESR.

-Phil

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
'Still some PropSTICK Kit bare PCBs left!

Paul Baker

We use X5R (low voltage), X7R (higher voltage) or c0g (temperature stable) ceramics which explains the discrepance between Tracy's and my comments. Also Tracy is referencing switching regulators which require low ESR, but I am refering to linear, which depending on the part, may require a minimum ESR (ie higher ESR)

Rayman, no it was not an issue of value but of ESR.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Paul Baker
Propeller Applications Engineer

Parallax, Inc.

Post Edited (Paul Baker (Parallax)) : 7/15/2008 7:49:52 PM GMT

Phil Pilgrim (PhiPi)

Paul,

I see what you mean, sorta. In Tracy's example, the ESR of the ceramics is too high, and adding a tantalum helps to lower it. On the Proto board, the ESR is too low to start with and outside the regulator's stable region. I'm not sure I see, though, how adding a tantalum helps to raise the overall ESR if the ceramic cap is still in place. My intuition tells me the ESR would be lower still (but not by much, since the ceramic cap still dominates):

····ESR_net = ESR_cer || ESR_tan = ESR_cer x ESR_tan / (ESR_cer + ESR_tan)

This assumes, of course, that ESR calculations are equivalent to simple resistance calculations. Perhaps it's not as simple as that.

-Phil

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
'Still some PropSTICK Kit bare PCBs left!

Paul Baker

Im a little fuzzy on the specifics because it's two R-C in parallel when modeling it, so theres a frequency component to calculating the Effective Resistance seen. Also with parts that have no datasheets and no immeadiate means to verify the actual ESR values of the caps it's more of qualitative exercise rather than quantitative.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Paul Baker
Propeller Applications Engineer

Parallax, Inc.

Post Edited (Paul Baker (Parallax)) : 7/15/2008 8:41:29 PM GMT

Philldapill

I'm confused about how ADDING another capacitor in parallel is supposed to INCREASE the ESR. If you look at the initial state when the device turns on, the capacitors are at 0V. At this point, you can model the two parallel capacitors as simply two resistors(each ESR) in parallel and model the actual capacitor as a short. With this, the total ESR is now lower than each individual ESR of each cap. I understand the dynamics change during steady state, but still, you are lowering the ESR, not increasing it...

Maybe I'm wrong in my thinking that the regulators go nuts because there is too LITTLE resistance into the capacitor.

Philldapill

Oh, in addition, when you're factoring in the impedance of the cap at higher frequencies, again it will be LOWER since your total capacitance is higher(Z = 1/(2*pi*C)).

Could it be that the impedance (resistance and imaginary) could have been to HIGH, not too low?

Again, maybe I'm way off. I'm only a junior studying EE, so there is probably more to it than just simple circuit analysis of a complex regulator circuit. Interesting thread, nevertheless.

Paul Baker

Ok, let me try to explain further. When you are talking about dynamic systems static equations do not describe the entire story. Linear regulators which require a minimum ESR, that ESR is only necessary upon start up. Once you have stabilized the output, you can make the ESR = 0 and it will still work. We are talking about stability in a feedback loop, which has it's own theory and set of equations.·Emperically I have noticed that in many cases an ultra-low ESR cap can be combined with a not so low ESR cap and it will achieve stability (don't ask me to explain why this is the case). It all has to do with the step response of the circuit as a whole, you cannot look at a single element and predict the circuit's behavior off of it. You must view the circuit·holisticly as a system in order to predict it's dynamic behavior. If you have studied Nyquist's Stability Theorem, this is the what applies to the analysis of the system. My ability·to paint with broad brush strokes ends here.·Further discussion requires not so easy to follow mathmatics that I don't have the time to go into. Those still curious I invite to research the subject thierself, the information is easily retrieved from wiki and other sources, you just have to be pretty good with your Laplace equations to understand it fully.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Paul Baker
Propeller Applications Engineer

Parallax, Inc.

Post Edited (Paul Baker (Parallax)) : 7/15/2008 10:43:53 PM GMT

Tracy Allen

Are you sure it is the regulator that is singing, and not the filter capacitor? Hi-C ceramic capacitors are notoriously piezoelectric and there are many reports of them singing when an oscillation is applied. The oscillation would come from the feedback with the regulator and be seen on a 'scope as an oscillation in the voltage.

Is it still an LM1086 regulator on the protoboard? The LM1086 data sheet recommends a 10uF tantalum, or higher capacitance without limit. If a larger capacitance is used (think capacitors in parallel) the ESR does go down (which might lead one to think, less stability), but so does the overall break frequency in the feedback loop. The ESR restriction has to be coupled with the value of capacitance. I'd guess that a second 10uF ceramic in parallel with the first might also stabilize the output, by simply swamping the output with sufficient capacitance. However, ceramics do have strong resonant frequencies, in the kHz or mHz range, and there are strong phase shifts associated with the resonance. Varies by construction and brand. Tantalums do not have that kind of strong resonance. And tanalums are not piezoelectric and can't sing!

The LM1086 is a quasi-LDO regulator, as described in the following Q&A from the National Semi knowledgebase:

<<
Why is there a difference in the bandwidth and stability of the LDO, Quasi-LDO, and standard NPN regulators?

The standard NPN linear regulator tends to have wide bandwidth and be insensitive to output capacitive loading. The LDO (PNP) regulator tends to have lower bandwidth and require specific output capacitors to maintain stability. A Quasi-LDO regulator has characteristics somewhere between the NPN and LDO.

The reason for these differences has to do with the orientation and type of the output transistor. In an NPN regulator, the input voltage is applied to the collector and the load is placed on the emitter. In an LDO, or PNP-based design, the input voltage is applied to the emitter and the load is applied to the collector. If you mentally turn the circuit on its side, you will see that the standard, or NPN-based regulator forms an emitter follower. The overall circuit generally has a gain of less than 1. With the PNP LDO circuit, the load impedance acts as a collector resistance. The LDO circuit creates a common collector amplifier, with the load impedance acting as a collector gain resistor. At some combination of load current (re) and load impedance (Rc), you will create an amplifier out of the output stage - and a poor quality one at that. This extra gain and phase shift by the output stage creates the condidtions for severe oscillations.

The capacitor on the output is required to shunt the load impedance (Rc) and keep the gain below 1. This requires the output capacitor to have very low series resistance. Since we are only talking about a few ohms, standard off-the-shelf electrolytics cannot be used because of their high internal resistance (ESR). Tantalum, Ceramic or special low ESR output capacitors must be used. Because not every regulator design is the same, some regulators require certain types of capacitors with certain ESR ranges. The regulator manufacturer will usually specify the type of capacitor that should be used, or the range of acceptable ESR for the capacitor.

Heed their warnings and use the recommended capacitor types. When replacing a "Standard" regulator with a LDO device in an exisiting design, make sure the exisiting output capacitors meet the regulator manufacturer''s requirements.
>>

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Tracy Allen
www.emesystems.com

KenFordham

Paul,

I just got the caps and soldered one in. That solved the problem of the "singing" proto board. Thanks very much for your help.

Ken

JoshuaLewis

Hello,
I was having the same issue with my 5Volt Regulator singing, it has been doing it for probably about 10 hours of use. would this have damaged the regulator?

Also after using a 10uF cap on the terminals of the Regulator, i am still noticing the singing, but it is significantly quieter. would a larger cap make the noise completely disappear?

I noticed the pitch of the noise changes depending on how much is being driven. also that my speakers situated maybe 5 meters from the propeller board are also picking up the high frequency noise and reproducing it,...

The Brand of singing regulator that i used in my board was a National Semiconductor, LM1086CS-5.0/NOPB if others have this issue again it could be related to this chip itself.

any help is greatly appreciated,

Josh

Post Edited (JoshuaLewis) : 9/1/2008 11:46:48 AM GMT

Leon

You probably need a SM tantalum capacitor, as close to the regulator pins as possible. Check the regulator data sheet. The oscillation won't damage the regulator but will stop it working properly. A friend of mine once built a PS and didn't include capacitors; the regulators were oscillating at VHF and didn't produce any output. Just putting a meter probe on the output stopped the oscillation and they started working, which was a bit confusing at first.

Leon

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Amateur radio callsign: G1HSM
Suzuki SV1000S motorcycle

Post Edited (Leon) : 9/1/2008 12:09:33 PM GMT

Paul Baker

We have seen no indication of the oscillation causing damage to the regulator, however if you connect sensitive components to a oscillating regulator, it is possible to damage those. The type of capacitor is important, electrolytics will not work. It's not the value thats affecting the regulator operation but the equivalent series resistance.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Paul Baker
Propeller Applications Engineer

Parallax, Inc.

JoshuaLewis

awesome, thank you very much this has answered my question, the data sheet explained to lower ESR.

It is desirable to have large output capacitance for applications that entail large changes in load current (microprocessors for example). The higher the capacitance, the larger the available charge per demand. It is also desirable to provide low ESR to reduce the change in output voltage: ΔV = ΔI x ESR It is common practice to use several tantalum and ceramic capacitors in parallel to reduce this change in the output voltage by reducing the overall ESR. Output capacitance can be increased indefinitely to improve transient response and stability.

I am going to try 2 10uF tantalum capacitors on the output lines. this should minimize the singing further.

Thank you

--Josh

johnfl68

Are there any tantalum capacitors on the Propeller Demo Board?

John

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Inigo Montoya: Let me 'splain. (pause)
Inigo Montoya: No, there is too much. Let me sum up.
The Princess Bride (1987)

Paul Baker

No, why do you ask?

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Paul Baker
Propeller Applications Engineer

Parallax, Inc.