How to Improve Filtering in Low Power Variable Power Supply for Synthesizer Noise?

201 14

#1 21660671 13 Jan 2012 16:03

Michael Fleischer Michael Fleischer

Anonymous
Post #1
21660671 13 Jan 2012 16:03

I just purchased and built a low power variable power supply kit to power a solid state synthesizer I built years ago from an article in Popular Electronics. The synthesizer uses 50 mils at 12-15 volts DC to produce sounds resembling those of a steam engine. It's used with a model railroad. When I use the power supply with the synthesizer it sounds very garbled and noisy. I'm wondering if the power supply has enough filtering. Any suggestions on how I can modify the attached schematic for better results?

Attachments:

forums-scan0002-1326398581.jpg (44.3 KB) You must be logged in to download this attachment.
ADVERTISEMENT
#2 21660672 13 Jan 2012 16:29

Todd Hayden Todd Hayden

Anonymous

Post #2
21660672 13 Jan 2012 16:29

You can add additional cpacitance across C2 (observe polarity). If the leads from this supply to your synthesizer are long, you can add the capacitance at the synthesizer input. Not knowing how your system is physically laid out, I would also twist the output wires from the power supply around earch other and make sure they are routed away from the power infrastructure that runs your model railroad. Also make sure the leads to the potentiometer are short and you can try a cap (0.1uF to 1uF) from the output of the potentiaoleter to ground (i.e. transistor base to ground).
#3 21660673 13 Jan 2012 16:55

Steve Lawson Steve Lawson

Anonymous

Post #3
21660673 13 Jan 2012 16:55

This power supply really isn't suited to audio applications. Not enough ripple rejection. I would replace the transistor with a 78L12 Voltage Regulator, remove the Pot and short the wiper line to ground (i.e. you want the ground pin of the 78L12 to go to ground). This will make the supply no longer variable, but it sounds like you don't need a variable supply for your application. If you do, then I would use an LM317L and rewire the pot and limit resistor (i.e. the variable voltage divider) as required for the LM317L (check the LM317L datasheet if you're not sure how to do that). With the LM317L, though, it won't adjust to 0 (more like 1.25V).

Another alternative is to try something like a 3300uf cap from the base of Q1 to ground. The size of this cap depends on the Beta of Q1 and how much ripple noise can be tolerated. Basically this turns Q1 into a capacitance amplifier. You might be able to go as low as 600uf.

Here's my reasoning: IT=CV so C=IT/V if I=Ic/Beta and the Beta of Q1 is, say, 70, then I=50ma/70=0.71ma T is approx 6ms and V is whatever maximum voltage will not produce perceptible noise in your Train Whistle Synthesizer. Lets say 1mV. So, C=0.71ma*6ms/1mv=4260uf. Round down to the next standard value and you get 3300uf. Best case might be a Beta of 100, and a ripple voltage tolerance of 5mV: C=(50ma/100)*6ms/5mV=600uf.

This is all linear approximation math, but it should get you in the ballpark.
ADVERTISEMENT
#4 21660674 14 Jan 2012 00:01

Kevin Parmenter Kevin Parmenter

Anonymous

Post #4
21660674 14 Jan 2012 00:01

exactly what Steve said - replace the transistor circuitry with an LM7812 or LM7815 and make sure you have at lease 1000 uf of capacitor for the bulk cap (output of the rectifiers - input to the regulator) and then a 0.1 uf close to the output for HF decoupling of the regulator so its stable and 470 uF or so on the output capacitor (make sure voltages are high enough on the caps) this should reduce the ripple and give you nice smooth regulated quiet supply for audio work.
The datasheets for LM78xx regulators abound with application circuit ideas and technical tips. if you measure the output with a meter it will be close to either 12 or 15 volts (depending on what you chose) and look at it with a scope ac coupled with the gain turned way up and a 1x probe and the noise should be low.

you can also buy a DC output regulated wall wart (or find one you are not using for anything at the present time and use the output of it and you get all the global safety requirements met too! I keep a box of them around for just such occasions.

Good luck
ADVERTISEMENT
#5 21660675 14 Jan 2012 10:31

Michael Fleischer Michael Fleischer

Anonymous

Post #5
21660675 14 Jan 2012 10:31

Thanks, Todd. I'll try this first before I move on to Steve's and Kevin's suggestions. I sure appreciate the help from the three of you - I've always been facinated by electronics and can read schematics and solder but that's about it.
#6 21660676 14 Jan 2012 10:45

Michael Fleischer Michael Fleischer

Anonymous

Post #6
21660676 14 Jan 2012 10:45

Thanks for the great suggestions, Steve. These forums are fantastic. As I mentioned to Todd, although I'm facinated by electronics, I'm only familiar with the very basics. I'll give Todd's ideas a try before replacing the transistor.
#7 21660677 14 Jan 2012 10:51

Michael Fleischer Michael Fleischer

Anonymous

Post #7
21660677 14 Jan 2012 10:51

Thanks for fast reply and detailed approach, Kevin. Hopefully I won't have to do too much rework but it's nice knowing what to change if I have to. It's reassuring how much knowledge is available in these forums and how willing you guys are to help - it's very much appreciated!!!
#8 21660678 15 Feb 2012 10:32

Samuel Sloan Samuel Sloan

Anonymous

Post #8
21660678 15 Feb 2012 10:32

I also have a similar problem. My synthesizer is also not working properly and I think it is due to my old generator which I use for powering the synthesizer. I guess I will "check a power rental(check a power rental)":http://www.aggreko.com/northamerica/products---services/power-generation-solutions.aspx and then will rent a new generator in order to see if this will help. Hopefully it will.
#9 21660679 15 Feb 2012 12:21

DAVID CUTHBERT DAVID CUTHBERT

Anonymous

Post #9
21660679 15 Feb 2012 12:21

Place a capacitor from the wiper of the POT (base of the transistor) to the negative rail. 220 uF will cut the 120 Hz ripple by a factor of 40.
#10 21660680 15 Feb 2012 16:13

Michael Fleischer Michael Fleischer

Anonymous

Post #10
21660680 15 Feb 2012 16:13

Thanks, David. As a matter of fact, I haven't had a chance to date to implement the changes the others suggested so I thought I'd try yours since it seemed to require a minimum of effort. I used a 220uF cap and it works fine. Thanks!! Again, this is a great forum - thanks to you all!
#11 21660681 16 Feb 2012 11:36

Mark Harrington Mark Harrington

Anonymous

Post #11
21660681 16 Feb 2012 11:36

Have a look at these links as well All really standard now for variable power supplies and easy to construct

If you want higher Amperage then consider using an LM317K National semiconductor devices with a series pass transistor

If you want simple overvoltage protection or crow protection consider using this in your design as well
http://www.ieeta.pt/~alex/docs/DataSheets/MC3423.pdf

http://www.diy-electronic-projects.com/p57-LM317-VARIABLE-POWER-SUPPLY
http://www.eleccircuit.com/power-supply-regulatable-0-28v-20amp-by-lm317-and-2n3055/
http://www.electronics-project-design.com/UniversalPowerSupply.html
http://www.k7mem.com/Electronic_Notebook/power_supplies/lm317.html
ADVERTISEMENT
#12 21660682 16 Feb 2012 11:46

Mark Harrington Mark Harrington

Anonymous

Post #12
21660682 16 Feb 2012 11:46

Now you know why all these so called mains adaptors that you buy for your radios and other battery appliances are rubbish

Simply take one apart and all is revealed You might be lucky enough to see a regulator inside but in general they comprise a full wave bridge rectifier No over volts protection at all No Over current protection at all and maybe if its truly wonderful you might just get what it say’s on the super shiny label on the cheap plastic casing Total cost when it goes wrong , One radio , One keyboard and one horrible bill

But not to worry as my dear friend say’s Its cheaper In China!! hmmmmm Europe can manufacture everything for us We are so brilliant that we don’t need to be able to distinguish a bad desin from another After all It’s a throw age “Yeah OK , what ever !! “ Im sure you know best ha , ha ha !!

Of course this doesent quite end here This will no doubt be the same person who walks into your bussiness and say's "Oh please , please can you fix this for me " Aaaah !!

No worries after all we dont need any designers any manufacturing We just get everyone else to do it for us " Say no more !

You have to laugh don’t you at this
#13 21660683 15 Mar 2012 03:48

Deejay Thomas Deejay Thomas

Anonymous

Post #13
21660683 15 Mar 2012 03:48

Hi Todd,
can you help me out in few questions? they are pretty basic .
question: the effect on the 5V supply of doubling the value of the 4700UF capacitor is to :
a. Increse the RMS ripple volatge when a current of 100 mA is drawn.
b.have no effect on the ripple voltage
c. increase the output volatge

can you tel me which one is the right answer?
#14 21660684 15 Mar 2012 08:16

Todd Hayden Todd Hayden

Anonymous

Post #14
21660684 15 Mar 2012 08:16

I'm not sure what circuit you are referring to. It sounds like a general question with regard to the output capacitor of a regulated power supply. Regulators charge the output capacitor as well as supply the load. Current flow in the capacitor changes creates a change in voltage across the capacitor i = C * dV/dt, or rearranging terms, dV = i * dt / C. So for a given current, a larger C will result in less dV. Charging the capacitor creates positive dV and the load draws current out of the capacitor creating negative dV. The two sum together to give peak-peak ripple voltage.

There are other subtle aspects of doubling the output capacitor. You have to watch the ESR rating of the capacitor, usually increasing value in the same size case causes an increase in ESR. Current flowing in and out of the cap generate additional ripple voltage due to the ESR. Also changing the capacitor affects the loop bandwidth of the regulator and depending on how your load draws current, ripple voltage can increase due to the regulators inability to keep up with rapid changes in load on its output (outside the loop bandwidth).

Hope that helps.
#15 21660685 15 Mar 2012 15:20

Steve Lawson Steve Lawson

Anonymous

Post #15
21660685 15 Mar 2012 15:20

One trick for dealing with the increase in ESR (or doubling the filter capacitance) is to use two capacitors in parallel. The ESRs of each cap will combine like parallel resistors (if each the same ESR, then total ESR = ESR/2). So, instead of one 4700uF cap, use two 2400uF caps.
Create an account, log in here. You will receive points by participating in discussions.
Join this discussion.

Install Elektroda application

Didn't find an answer? Ask Artificial Intelligence

*I agree to send the question to OpenAI, Anthropic PBC, Perplexity AI, Inc., Kagi Inc., Google LLC - owners of language models in order to prepare the best response. The companies may monitor and log information entered into the form.

*I agree to publicly display my question and answer. The question and answer will be publicly available to everyone. The process may take a few minutes. Upon completion, you will be redirected to the page with the answer.

Wait...(2min)

Topic summary

✨ A low power variable power supply kit used to power a solid state synthesizer for model railroading produces noisy, garbled audio due to insufficient filtering and ripple rejection. Suggested improvements include adding additional capacitance across existing capacitors (e.g., C2), placing capacitors at the synthesizer input if leads are long, twisting output wires to reduce interference, and minimizing potentiometer lead length. Replacing the transistor-based regulator with a fixed voltage regulator such as the LM7812 or LM7815, or a variable regulator like the LM317L, can significantly improve ripple rejection and output stability. Bulk capacitance of at least 1000 µF at the rectifier output and smaller high-frequency decoupling capacitors (0.1 µF) near the regulator output are recommended. Adding a capacitor (220 µF to 3300 µF) from the transistor base or potentiometer wiper to ground can reduce ripple by filtering 120 Hz noise. Using parallel capacitors can reduce equivalent series resistance (ESR) and improve filtering performance. Alternative power sources such as regulated DC wall adapters or higher current LM317K with pass transistors are also suggested. Datasheets and application notes for LM78xx and LM317 series provide valuable design guidance. The discussion emphasizes the importance of proper filtering, regulator choice, and wiring layout to achieve a clean, stable power supply suitable for audio synthesizer applications.
Generated by the language model.

FAQ

TL;DR: For quiet audio, replace the pass transistor with an LM317L or add ~3300 µF from Q1 base to ground; this “turns Q1 into a capacitance amplifier.” Statistic: 3300 µF often hits the target at ~50 mA loads. [Elektroda, Steve Lawson, post #21660673]

Why it matters: Cleaner DC reduces 120 Hz hum and garbling in low‑power synthesizer rails.

Quick Facts

Typical load cited: ~50 mA at 12–15 VDC for the steam‑engine synth; noise appeared on a new kit PSU. [Elektroda, Michael Fleischer, post #21660671]
Drop‑in fix: LM7812/LM7815 with ≥1000 µF bulk, 0.1 µF HF bypass, and ~470 µF output for low ripple. [Elektroda, Kevin Parmenter, post #21660674]
Minimal‑rework tweak: 220 µF from pot wiper/base to ground can cut 120 Hz ripple ≈40×. [Elektroda, DAVID CUTHBERT, post #21660679]
Layout helps: Twist PSU output leads, keep pot leads short, and add 0.1–1 µF from base to ground. [Elektroda, Todd Hayden, post #21660672]
Lower ESR trick: Use two caps in parallel (e.g., 2×2400 µF) instead of one 4700 µF. [Elektroda, Steve Lawson, post #21660685]

How do I quickly cut the hum without redesigning the supply?

Add a 220 µF capacitor from the pot wiper (transistor base) to the negative rail. Users reported it slashes 120 Hz ripple by about 40× and restored clean audio. Keep leads short and observe polarity. “220 µF will cut the 120 Hz ripple by a factor of 40.” [Elektroda, DAVID CUTHBERT, post #21660679]

What’s the best regulator swap for an audio‑quiet 12–15 V rail?

Use a 7812 or 7815 linear regulator. Provide ≥1000 µF at the rectifier output, 0.1 µF close to the regulator for stability, and ~470 µF on the output. This combination delivers a smooth, low‑noise rail suitable for audio loads around tens of milliamps. [Elektroda, Kevin Parmenter, post #21660674]

Can I keep it variable? What about LM317L?

Yes. Rewire the pot as the LM317L’s adjust divider. Note the output won’t go to 0 V; minimum is about 1.25 V. If you need near‑zero volts, add a negative reference or a different topology. “With the LM317L, it won’t adjust to 0 (more like 1.25 V).” [Elektroda, Steve Lawson, post #21660673]

How big should the extra smoothing capacitor be if I stay with the transistor?

Place a large electrolytic, often around 3300 µF, from the base of Q1 to ground. This forms a capacitance multiplier effect and improves ripple rejection at low current. Steve’s linear estimate targeted 600–3300 µF depending on transistor gain and tolerated ripple. [Elektroda, Steve Lawson, post #21660673]

What wiring and decoupling practices reduce audible noise?

Twist the PSU output pair, route it away from mains and motor wiring, keep potentiometer leads short, and add a 0.1–1 µF cap from the transistor base to ground. These steps lower pickup and stabilize control impedance. [Elektroda, Todd Hayden, post #21660672]

Is my noisy wall‑wart or generator the culprit?

Unregulated adapters and aging generators often have high ripple and spikes. A regulated DC source or a retrofit regulator can eliminate garbling. Consider a regulated supply before chasing synthesizer faults. [Elektroda, Kevin Parmenter, post #21660674]

Does doubling a 4700 µF output capacitor reduce ripple?

Yes. Ripple scales with 1/C, so increasing capacitance reduces ΔV. It does not raise output voltage. Watch ESR and loop stability when changing values. “dV = i·dt/C,” so larger C means less ripple for the same current. [Elektroda, Todd Hayden, post #21660684]

How can I lower ESR without oversized capacitors?

Place two electrolytics in parallel, such as 2×2400 µF instead of one 4700 µF. Parallel ESR combines like resistors, decreasing effective ESR and ripple. This trick maintains value and improves dynamic performance. [Elektroda, Steve Lawson, post #21660685]

What’s a capacitance multiplier, in simple terms?

Feeding the base of a pass transistor from a large, smoothed node makes the emitter see an effectively much larger capacitance. It multiplies smoothing by the transistor’s current gain, improving ripple rejection for small loads. [Elektroda, Steve Lawson, post #21660673]

Any fast 3‑step How‑To for the 7812/7815 retrofit?

Add ≥1000 µF after the rectifier; wire IN to the regulator.
Tie GND to supply ground; place a 0.1 µF ceramic at OUT‑to‑GND near the regulator.
Add ~470 µF from OUT‑to‑GND; verify 12 V or 15 V under load. [Elektroda, Kevin Parmenter, post #21660674]

What minimum load and current should I expect in this synth scenario?

The referenced steam‑engine synthesizer draws about 50 mA at 12–15 VDC. That low current favors simple linear regulation or a capacitance multiplier for quiet operation. Ensure headroom for line sag and startup. [Elektroda, Michael Fleischer, post #21660671]

What ripple period should I design around on 60 Hz mains?

Full‑wave rectification yields 120 Hz ripple. In the thread’s sizing math, a time slice of about 6 ms was used for estimates when picking base capacitors. Design filters around that interval. [Elektroda, Steve Lawson, post #21660673]

Can overvoltage destroy my synth? How do I add protection?

Yes. Add simple overvoltage crowbar protection to short the rail if it rises dangerously, tripping a fuse. Designers often pair crowbars with adjustable regulators in variable supplies. See the MC3423 approach linked. [Elektroda, Mark Harrington, post #21660681]

Will twisting wires and moving them actually make an audible difference?

Yes. Twisting the supply pair reduces loop area and cancels magnetic pickup from nearby power wiring or motors. Short, tidy pot leads also reduce injected noise. Add a small 0.1–1 µF base bypass if needed. [Elektroda, Todd Hayden, post #21660672]

Any edge cases I should know before choosing LM317L?

LM317L cannot output below about 1.25 V, so it won’t work for 0 V testing. Plan for that limit or use a different regulator architecture if you need true 0–12 V adjustment. [Elektroda, Steve Lawson, post #21660673]

Generated by the language model.