Must touch power cable to stop noise?!

OK, you said "I built", how did you built it ??

PCB board, peg board, point to point wiring ??

How did you build it ??

Please post a pic somewhere.

hamilton

Here you go:

The other circuits on the board are:

- Hsync & Vsync to Composite Sync converter

- V-USB MAME panel

- NE556 based circuit for high frequency Hsync rejection

Thanks for your help.

WOW, what a mess !!

1. have you done this type of project before ?
2. do you plan on rebuilding this in a cleaner package ?
3. will you make a PCB from the schematics ?

When you say "touch the power cable" where is this power cable located in the pic ?

Your "power cable" is transmitting noise everywhere, it needs to be stopped there, where ever there is.

Your circuit the way it is built, may not be fixable.

Your audio is picking up noise for the rest of the circuit.

Touching anything is inducing noise everywhere, I bet they can "see" your circuit next door.

Good luck

hamilton

:-P

No.

I would rebuild the audio amp. If I had to rebuild everything, I'd rather buy some commercial product (there are a couple around).

No.

It's the one at top right, starting with the white plug. It goes into the PC. The power reaches the audio amp through the yellow and black wires that start from there. The audio amp is the circuit at lower right.

I noticed you placed it on top a CRT... Is the CRT on when it happens? They _do_ generate a little noise.

Sometimes it's on, sometimes it's off. It doesn't make a difference. It also does that with a totally different PC in another room.

That chip does not have much ripple rejection at all- which tells you how power supply noise affects the output. You can kill it by using say a 5V regulator 7805 to power the chip. Bring your 12V into the

7805 and then 5V to the amplifier. Be sure to bypass the 7805 at its input and the 100uF at its output- also put it close to the amp. If you need more power output then use a 7809 or LM317 configured as 9V. See if that improves things.

That chip does not have much ripple rejection at all- which tells you how power supply noise affects the output. You can kill it by using say a 5V regulator 7805 to power the chip. Bring your 12V into the

7805 and then 5V to the amplifier. Be sure to bypass the 7805 at its input and the 100uF at its output- also put it close to the amp. If you need more power output then use a 7809 or LM317 configured as 9V. See if that improves things.

Is there local bypassing of power lead?

I always use flea clips when building boards like this. Make putting components on and off easier.

Greg

Could also assemble a battery and power it with pure DC to see if the problem is being injected externally or not.

snipped-for-privacy@dico.no (Slater) wrote in news: snipped-for-privacy@nntp.aioe.org:

Time to check, it sounds like a classic case of oscilating electronics. A scope, to check, or if that is not possible, touching different parts of the electronics, while connecting a volt or amp meter here and there, will show fluctuations, when you touch part of the oscillating circuit.

I cant see the cct diag, link wont open. But normally such things are down to either lack of enough supply decoupling (add more capacitance across the rails) or oscillation due to inadequate stablity margin. The latter can often be cured by using nfb to reduce the bandwidth a bit, but it can be more complex.

NT

Do as Fred stated, use a separate regulator and one of these:

It seems I made the problem disappear by cutting the red wire (+5V) in the power cable, thus shutting down the sync converter and the NE556. I'm trying to add better decoupling and bypass and will made further tests afterwards.

Actually, I forgot to put any decoupling in those!

1) Integrate your finger into the power cable. OR 2) Put PI CLC or CRC filter(s) at the amp power input point(s) AND add a good 0.1uF or 1.0uF cap across chip supply pins.

The link is OK. Save it and open it with a PDF reader.

--
You can't have a sense of humor, if you have no sense.

Thank you. I'll put a PI CLC filter. Would you please suggest the values for the components? The amplifier chip is powered from a PC PSU's 12V line, while the other two chips get 5V from the same source and they process signals of 15 - 16 KHz.

Fortunately, the problem disappears when the whole device is used: I was making tests with the input to the sync processing chips disconnected. If I connect everything, no problem.

You have to be really careful with any type of amplifier.

Good rules of thumb.

1) Filter the PC power supply: buy a ferrite torroid or bead and put around the cable coming from the PC power supply. Those are switching power supplied that generate noise in both the common mode and differential mode Just snake your cable through the torroid for a few turns and that will help knock down the power supplies radiated noise. Next, right were the power enters your board add some bypass caps (100 - 1000uf electrolytic with .1uF ceramic in parallel. The reason is that the electrolytics are not very good at filtering very high frequencies (MHz range). The .1uF ceramics are effective at those frequencies and will take over where the electrolytics peter out.

2) Amplifier power leads: Bypass the power supply leads on the amplifier directly across the Vcc (V+) and Vss (GND) leads with an electrolytic and a .1uF ceramic capacitor in parallel. The leads must be very short between the caps and the power leads of the chip or radiated noise can enter the loop and get on our power rails. Also any excessive lead length can introduce inductance when inhibits the filtering action of the capacitors. Run the supply (V+) and return (gnd) wires next to each other (twisted is best) back to the power plug bypass caps. You want to distribute power to the high power parts in a "star" arrangement where each chip has in independent pair of power wires to the power supply entry to the board. Do NOT daisy chain ground or power or that will allow feedback to occur. When they talk of "ground loop noise" daisychaining is a big cause if that.

3) limit the bandwidth of the frequencies going into your inputs. You can put a small RC filter with a cutoff frequency on the 40K - 100KHz range something like a 1k ohm in series and a 3300pF from the input pin to ground. This will prevent any RF noise that gets picked up on the wires leading to the input from getting into your amplifier. You might have to play with the values for the 1K resistor since your amp has in input impedance spec of your amp says is has a 100K input impedance, the 1K in series with that you reduce the gain by ~ 1%. The 10K call out should be before the 1K so the circuit should be signal in -> 10K to ground -> 1K in series -> C to ground -> then chip input pin.

4) Tighten up your wiring: You don't want output signals crossing input signals. You also don't want input signals close to power supply rails. If you cross them do so a as close to 90 degree angles as possible. Also eliminate any excess wire loops, make the wiring as close to point to point as possible.

5) Does the back of your perf board have a ground plane on it? if so make sure it is grounded. Leaving unconnected copper is a good way to couple unwanted signals.