"motor boating" is specifically low frequency, typcially was on the slope of the DC blocking high pass - there are feedback loop frequency/phase stability requirements at low frequency too - don't close a DC loop around a AC coupled amp
op amp's internal protection circuitry can cause some of them to cycle on/off in thermal limit if loaded too heavily - can will also be at very low low frequency
just possibly high frequency oscillation could cause power dissapation that also cycles the thermal protection
if the circuit allows add a ~100 Ohm R in series with the op amp output, outside of the feedback loop - decouples excessive Cload that may cause oscillation - and can make a easy place to measure short circuit current if there is a problem
I sometimes run bare bus wire or Cu tape on the top side for power, gnd distribution, keeps loop area low - though the bare power traces increase chance shorts with probes, tools while debugging
and a 'scope is a really handy tool when things aren't working - multimeters don't cut it for dynamic errors, oscillation
op amp's internal protection circuitry can cause some of them to cycle on/off in thermal limit if loaded too heavily - can will also be at very low low frequency
just possibly high frequency oscillation could cause power dissapation that also cycles the thermal protection
if the circuit allows add a ~100 Ohm R in series with the op amp output, outside of the feedback loop - decouples excessive Cload that may cause oscillation - and can make a easy place to measure short circuit current if there is a problem
I sometimes run bare bus wire or Cu tape on the top side for power, gnd distribution, keeps loop area low - though the bare power traces increase chance shorts with probes, tools while debugging
and a 'scope is a really handy tool when things aren't working - multimeters don't cut it for dynamic errors, oscillation
Last edited:
Bad soldering covers a lot of possibilities but if everything connects where it should, then however it "looks", well it should still work OK.
Something wrong with the PSU. That's quite possible but its impossible to diagnose from pictures alone.
The best advice I can give is this... the cost of parts is minimal so why not make another board but this time build it up and test it outside of the preamp. I would also use a far far simpler PSU. For an opamp a very simple resistor and zener supply will work well, probably better than all those regs. Much smaller, much easier.
Something wrong with the PSU. That's quite possible but its impossible to diagnose from pictures alone.
The best advice I can give is this... the cost of parts is minimal so why not make another board but this time build it up and test it outside of the preamp. I would also use a far far simpler PSU. For an opamp a very simple resistor and zener supply will work well, probably better than all those regs. Much smaller, much easier.
Thanks for your inputs. 
I think I know where the problem lies. As you have expected, something on the old PCB is probably turned on. I measured the resistance across the empty fuse holder, it is not infinite (about 100+kΩ), so some current is flowing through who knows what.
Earlier on I tried connecting the output to the power amp without going through the RCA jacks on the PCB. Turned out that is not enough, I also need to disconnect the inputs. Now the noise is much lower, there are still some high frequency whirring if I put my ears to the tweeters, but that is probably my dirty mains. I thought 2 stages of regulation would do it, but I guess LM317/337 were never that great for high frequency noise.
I'd hate to have to remove all the components on the old boards. I think I am going to remove all the active components and see what happens.
I think I know where the problem lies. As you have expected, something on the old PCB is probably turned on. I measured the resistance across the empty fuse holder, it is not infinite (about 100+kΩ), so some current is flowing through who knows what.
Earlier on I tried connecting the output to the power amp without going through the RCA jacks on the PCB. Turned out that is not enough, I also need to disconnect the inputs. Now the noise is much lower, there are still some high frequency whirring if I put my ears to the tweeters, but that is probably my dirty mains. I thought 2 stages of regulation would do it, but I guess LM317/337 were never that great for high frequency noise.
I'd hate to have to remove all the components on the old boards. I think I am going to remove all the active components and see what happens.
Any noise from a regulator or noise that gets past a regulator is minimal and many orders of magnitude below what would be audible from an opamp. Linear supplies are pretty clean anyway, the only real "noise" is line frequency (mains freq) related and even if there were a couple of volts ripple on the rails the opamp would reject it anyway and not pass it to the output.
Its great that you are working on this and looking for the reason
Its great that you are working on this and looking for the reason
In the past, I used to have all my audio equipments plugged directly to the mains. If I put my ears to the tweeters, I can hear some kind of whirring noise. Then I started using line filters for my sources (the standard electronic part kind, not the fancy "audiophile" kind). This noise is attenuated significantly. Since then I always plug my sources and line stages behind line filters.
So far everything I built needs this, except a headphone amp with choke filtered PSU. So that was my conclusion. If you look at the LM317 datasheet, the ripple rejection goes down with increasing frequency. Although I agree with you that it would take an awful amount of noise to make it past the opamp. I have no idea though.
I removed the rectifiers and the transistors, that put the resistance across the fuse holder more than what the DMM can measure. But the noise is still there. Then I removed the transistors on the relay circuit and the first stage opamp. The noise is still there.
I still have 2 more opamps and 2 more transistors on board (it's hard desoldering opamps). But it got me thinking, what if I am going down the right path? Is it possible that there is some kind of ground loop through the passive components? I really don't know enough on this. Can I assume that high resistance makes ground loop impossible?
So far everything I built needs this, except a headphone amp with choke filtered PSU. So that was my conclusion. If you look at the LM317 datasheet, the ripple rejection goes down with increasing frequency. Although I agree with you that it would take an awful amount of noise to make it past the opamp. I have no idea though.
I removed the rectifiers and the transistors, that put the resistance across the fuse holder more than what the DMM can measure. But the noise is still there. Then I removed the transistors on the relay circuit and the first stage opamp. The noise is still there.
I still have 2 more opamps and 2 more transistors on board (it's hard desoldering opamps). But it got me thinking, what if I am going down the right path? Is it possible that there is some kind of ground loop through the passive components? I really don't know enough on this. Can I assume that high resistance makes ground loop impossible?
Because you get increased noise when you add a bypass cap from pins 4 and 8 to ground it suggests to me that your ground connection is floating or high resistance and that this is feeding noise into the negative feedback divider on the op amp.
I would check continuity of the ground connections from the op amp back through the power supply regs and through to chassis earth.
I would check continuity of the ground connections from the op amp back through the power supply regs and through to chassis earth.
Sometimes adding block capacitors over existing electrolytes can cause problems if the ground layout wrong connected. Since you use the existing power supply unit I guess you take ground from your device. To repair this problem i suggest that you take only the positive and negative voltage and ground made using two transistors on your new board. New ground connect it with the remains of device.
Sometimes adding one capacitor between positive and negative power supply may help
Sometimes adding one capacitor between positive and negative power supply may help
You guys aren't quite following, but thanks for your input.
- When adding a 4.7uF capacitor over pin 4 and 8, I did not get an increase in noise. It made no difference.
- The increase in noise came from adding a 47nF film cap at pin 4 to ground and pin 8 to ground. I don't know if the capacitor is faulty, it does not happen with 100nF MLCC and 1000uF electrolytic.
- I did not use the existing power supply, only the transformer and its centre tap (ground).
- The problem seems to go away when I don't make signal ground connection through the existing board. My current hypothesis is that AC signal somehow interacts with some of the existing active components, or there is a ground loop (I don't really understand this topic).
- When adding a 4.7uF capacitor over pin 4 and 8, I did not get an increase in noise. It made no difference.
- The increase in noise came from adding a 47nF film cap at pin 4 to ground and pin 8 to ground. I don't know if the capacitor is faulty, it does not happen with 100nF MLCC and 1000uF electrolytic.
- I did not use the existing power supply, only the transformer and its centre tap (ground).
- The problem seems to go away when I don't make signal ground connection through the existing board. My current hypothesis is that AC signal somehow interacts with some of the existing active components, or there is a ground loop (I don't really understand this topic).
That points to something amiss with the grounds or what you "think" are the grounds.
Without seeing it all in context its impossible to say more. Once you reach a point like this then it needs a scope to help understand what is going wrong.
Without a scope all you can do is rebuild (or at least test) what you have in stages. That means isolating it from the PCB and testing it with battery power as a stand alone preamp. Just isolate the reg outputs and use two 9 volt batteries. Then power it from the tranny. And so on...
Earthing ... ground loops
1. if you put large capacitors after the regulated PSU, the regulations will not work.
2. In general use small caps of same type and value after the regulated psu, and as close to the ic/opamp as possible
3. the picture show a resistor and some wire going from the bord maybe to earth the case? ..... are you sure you don't have more loops to the case .... try cutting this ....
See also this ... very informative series on decoupling
Decoupling Capacitors 1 - YouTube
1. if you put large capacitors after the regulated PSU, the regulations will not work.
2. In general use small caps of same type and value after the regulated psu, and as close to the ic/opamp as possible
3. the picture show a resistor and some wire going from the bord maybe to earth the case? ..... are you sure you don't have more loops to the case .... try cutting this ....
See also this ... very informative series on decoupling
Decoupling Capacitors 1 - YouTube
Attachments
Last edited:
That is the LED indicator, earlier on I suspected that but unplugging that doesn't make any difference too.
I went on to remove all active components, as you guys probably predicted, that doesn't do a thing.
Then I proceeded to work on the ground connection. I mutilated the PCB and separated the ground (or at least what I think it is) into 3 parts: input, output, and the rest. I then connect these grounds to my "star ground" on the new board. Although I am not able to totally separate the input ground and the rest of the ground (there are still a couple hundred thousand ohms between them), it is now very quiet!
Thanks again guys.
Then I proceeded to work on the ground connection. I mutilated the PCB and separated the ground (or at least what I think it is) into 3 parts: input, output, and the rest. I then connect these grounds to my "star ground" on the new board. Although I am not able to totally separate the input ground and the rest of the ground (there are still a couple hundred thousand ohms between them), it is now very quiet!
Thanks again guys.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.