Start again, make sure RCA sockets are isolated from case etc.... of coarse it is wood... just fixing one channel will not help debugging, as you have to get the whole ground structure in balance...
Seperate those input and output cable bundles from each other, and get the inputs away from those capacitor cases...
Seperate those input and output cable bundles from each other, and get the inputs away from those capacitor cases...
Do you know why?
1) What is the reason behind the hum caused by using more than one rectifier board per transformer?
2) Also, why does shorter cable length help, especially for ground.
3)What should be considered a maximum length?
1) What does it mean to have the whole ground structure in balance?
2) Why does this have an impact? The input and output traces are in close proximity to each other and the capacitors on the PCB. The connections to the PCB are right near each other. How could I seperate them?
Thanks for your help and advice I'm trying to understand my mistakes so I will not make them with future projects.
In answer to question 1) there is a direct path between the grounds of each channels thought the amp board, rectifier, transformer, rectifier board and the adjacent amp board. This path is at very high resistance, int he order of megaOhms, however, it is the source of the hum.
Question 2) What I found was that when the ground wires were very short between the amp boards (i.e. chassis ground terminal) then the hum reduced as this provided a path of least resistance thus negating the hum caused by the path through the transformer.
Question 3) That depends on how much hum you'll put up with, but it will always be there without moving to either a single rectifier board or multiple transformers.
I hope this helps.
Question 2) What I found was that when the ground wires were very short between the amp boards (i.e. chassis ground terminal) then the hum reduced as this provided a path of least resistance thus negating the hum caused by the path through the transformer.
Question 3) That depends on how much hum you'll put up with, but it will always be there without moving to either a single rectifier board or multiple transformers.
I hope this helps.
2 wires paralel to each other = a capacitor.... the longer a wire is also the more of an inductor and arial it will be... a capacitor and an inductor is a simple oscilator... try to put a 10ohm resistor between the negative tab of the rca connector and the inpput ground wire leading to your PCBs... It should reduce the hum a little...on shared transformers...
The more current that flows through a wire... the stronger will the magnetic field be around the wire.... when your high current output cables are too close to the input cables the current flowing in the one will cause a voltage in the other ... like transformer windings distributing their power in a magnetic field....
Short bits of traces reasonably spaced on relatively low speed amps like this should not be too much of an issue...
with the wooden case you also have no RF shielding, so you may wan to consider a 220pf cap from input+ to input -, close to the PCB, to clean up some of the input wire's "reception".
The more current that flows through a wire... the stronger will the magnetic field be around the wire.... when your high current output cables are too close to the input cables the current flowing in the one will cause a voltage in the other ... like transformer windings distributing their power in a magnetic field....
Short bits of traces reasonably spaced on relatively low speed amps like this should not be too much of an issue...
with the wooden case you also have no RF shielding, so you may wan to consider a 220pf cap from input+ to input -, close to the PCB, to clean up some of the input wire's "reception".
It has been a few years since I have built one of these amps, but I had a similar problem with a four channel amp using the same boards. Tried a whole bunch of things, but it was a ground loop problem. What ended up solving the problem for me was making a signal ground bus that joined all of the signal grounds at the RCA inputs. My inputs were all within a few inches, so I just ran a length of 2mm copper wire in a straight line between the left and right input RCA jacks and soldered all of the ground tabs to this. I went from an almost un-listenable amp to almost dead quiet.
Hope this suggestion helps.
Chris
Hope this suggestion helps.
Chris
Well, I disconnected all of the amplifier modules except for one of them from the transformer so now it's just 1 transformer, 1 PS, 1 amp. I still get FM radio playing through the output. It is not as noticeable when there is no RCA cable connected, but gets much louder when I connect a cable to the RCA input. So, now I'm pretty certain my problems are not PS related. Any idea how to get my amp to stop picking up FM radio? =)
m0tion said:
I'm not overly familiar with BrianGT's design, but is the input (signal ground) isolated from the main ground? I use a low value resistor (50 ohms) to isolate these two grounds.
Also, looking at the pic in post # 1, I see you are using huge unshielded wires for input from the RCA jacks. This would be OK if the chassis is all metal and grounded but in this case, these leads should be shielded. I use cheap interconnect wire for this.
I'll be interested in hearing your ultimate solution...
I have a couple LM3875 P2P IGC implementations that I've put together recently. The first has moderate hiss/hum that can be heard out to about 2' with 94db speakers. It also receives 100.7 fm clearly enough to know that is indeed 100.7 fm
Incidentally, more prominent in the left channel and strengthens when the RCA input cable is connected.
The second unit was first fired up this morning and is currently being "tested" using the same PSU (30V +/-, 15000uF per rail). Zero fm and much less hiss/hum. My (inefficient) test speakers were so quiet that I double-checked my power cord.
The differences;
unit 1 uses a 4.7uF NP electrolytic as the Ci and no caps on V+/V- at the chip. 200k/10k R set the gain. A single conductor carries signal ground to the power ground star at the PSU.
unit 2 uses a 2.2uF film cap on the input, 1000uF caps at V+/- and lower R values, but still a gain of ~20. Power ground star located on amp board with a short jumper to sig-ground
While rebuilding unit 1 to unit 2 specs will take all of about 20 min, I am certainly interested in your findings and I just might experiment a bit instead.
Good luck,
7/10
I have a couple LM3875 P2P IGC implementations that I've put together recently. The first has moderate hiss/hum that can be heard out to about 2' with 94db speakers. It also receives 100.7 fm clearly enough to know that is indeed 100.7 fm
Incidentally, more prominent in the left channel and strengthens when the RCA input cable is connected.
The second unit was first fired up this morning and is currently being "tested" using the same PSU (30V +/-, 15000uF per rail). Zero fm and much less hiss/hum. My (inefficient) test speakers were so quiet that I double-checked my power cord.
The differences;
unit 1 uses a 4.7uF NP electrolytic as the Ci and no caps on V+/V- at the chip. 200k/10k R set the gain. A single conductor carries signal ground to the power ground star at the PSU.
unit 2 uses a 2.2uF film cap on the input, 1000uF caps at V+/- and lower R values, but still a gain of ~20. Power ground star located on amp board with a short jumper to sig-ground
While rebuilding unit 1 to unit 2 specs will take all of about 20 min, I am certainly interested in your findings and I just might experiment a bit instead.
Good luck,
7/10
Construction Errors
motion,
here's a list of errors (IMO) i can find in the construction of your amp:
1. chassis: use metal case (i've never seen any commercial amplifier which uses a non-metallic casing). the reason for this is that the case (when grounded) isolates the amplifier from external noise (magnetic or RF). since you have a wooden box, try placing a metallic layer on the bottom of your case. then try mounting your PCB as close as possible to this layer. note, have this metal layer earth-grounded.
2. wiring of low-level signal input wire: from your description, it looks like the probable point of noise entry is your input wires. from your picture, you are using large gauge stranded wire (twisting it is fine, however, you are still using larger wires). replace this with a shielded signal wire. and make your wiring as short as possible.
3. wiring of your power supply lines: you are using individual supply for each amp. this is good. however, you are routing your wires the long way. instead, i suggest that you change your layout. place your junction box in between the toroid and the power supply PCBs. run your power lines as short as possible.
if these won't help, then troubleshooting your PSU and amp should follow. but i suggest try doing this tips (before putting any filter caps and possibly affecting the performance of the amp).
motion,
here's a list of errors (IMO) i can find in the construction of your amp:
1. chassis: use metal case (i've never seen any commercial amplifier which uses a non-metallic casing). the reason for this is that the case (when grounded) isolates the amplifier from external noise (magnetic or RF). since you have a wooden box, try placing a metallic layer on the bottom of your case. then try mounting your PCB as close as possible to this layer. note, have this metal layer earth-grounded.
2. wiring of low-level signal input wire: from your description, it looks like the probable point of noise entry is your input wires. from your picture, you are using large gauge stranded wire (twisting it is fine, however, you are still using larger wires). replace this with a shielded signal wire. and make your wiring as short as possible.
3. wiring of your power supply lines: you are using individual supply for each amp. this is good. however, you are routing your wires the long way. instead, i suggest that you change your layout. place your junction box in between the toroid and the power supply PCBs. run your power lines as short as possible.
if these won't help, then troubleshooting your PSU and amp should follow. but i suggest try doing this tips (before putting any filter caps and possibly affecting the performance of the amp).
I agree with the recommendation of running shielded smaller gauge for the input, only grounded at one end, and trying to separate the power supply wires as far as possible from the input wires.
On a side note, I had a friend who lived in an apartment, and had a hell of a time diagnosing a hum in his amplifier, and he brought it over to my place. We had found that the amp sounded just fine at my place. He went back home, and plugged it in at a different part of his house and found that it was just fine there too. It ended up being a result of old wiring in his house, with half of it that was replaced when it was renovated. Are you having any issues with any other amplifiers in your house?
--
Brian
On a side note, I had a friend who lived in an apartment, and had a hell of a time diagnosing a hum in his amplifier, and he brought it over to my place. We had found that the amp sounded just fine at my place. He went back home, and plugged it in at a different part of his house and found that it was just fine there too. It ended up being a result of old wiring in his house, with half of it that was replaced when it was renovated. Are you having any issues with any other amplifiers in your house?
--
Brian
seventenths said:I'll be interested in hearing your ultimate solution...
I have a couple LM3875 P2P IGC implementations that I've put together recently. The first has moderate hiss/hum that can be heard out to about 2' with 94db speakers. It also receives 100.7 fm clearly enough to know that is indeed 100.7 fm
Incidentally, more prominent in the left channel and strengthens when the RCA input cable is connected.
The second unit was first fired up this morning and is currently being "tested" using the same PSU (30V +/-, 15000uF per rail). Zero fm and much less hiss/hum. My (inefficient) test speakers were so quiet that I double-checked my power cord.
The differences;
unit 1 uses a 4.7uF NP electrolytic as the Ci and no caps on V+/V- at the chip. 200k/10k R set the gain. A single conductor carries signal ground to the power ground star at the PSU.
unit 2 uses a 2.2uF film cap on the input, 1000uF caps at V+/- and lower R values, but still a gain of ~20. Power ground star located on amp board with a short jumper to sig-ground
While rebuilding unit 1 to unit 2 specs will take all of about 20 min, I am certainly interested in your findings and I just might experiment a bit instead.
Good luck,
7/10
You should really have 100nF and a larger (I use just 100uF) bypass cap(s) close to the chip to keep it from oscillating, though I have got away with accidentally forgetting the 100nF before, I prefer to have them there regardless. The large hiss you hear and the fact that it's picking up radio may well be from HF oscillation.
If you don't have a scope to check with then you could try a ~100W light bulb in series with the mains live. Upon power on it should flash bright for a second or two while caps initially charge and then fade out to very dim / not lit up at all. If it's constantly glowing any brighter than that with no input signal then this may indicate oscillation is present. Another indicator would be excessive heating.
Motion
I did not read the thread completely but did you got rid of your fm-signal?
If not; is you input impedance low enough (20-50k)? If it is to high then your input can works as a radio.
Picking up radiosignals almost allways is the combination of high gain and high impedance. That is why high-impedance systems are shielded and low-impedance not. Other problem is a bad/corroded connection working as a dioderadiodetecter, but this is very(!) rare.
Please forgive me if I did not read everything correctly, just trying to help...
I did not read the thread completely but did you got rid of your fm-signal?
If not; is you input impedance low enough (20-50k)? If it is to high then your input can works as a radio.
Picking up radiosignals almost allways is the combination of high gain and high impedance. That is why high-impedance systems are shielded and low-impedance not. Other problem is a bad/corroded connection working as a dioderadiodetecter, but this is very(!) rare.
Please forgive me if I did not read everything correctly, just trying to help...
I just killed my hum completely and did a writeup here:
http://www.diyaudio.com/forums/showthread.php?s=&threadid=118735
Not sure if that would apply in your situation as you are using PCB but maybe same concept, different sizes of cable and binary grounding.
http://www.diyaudio.com/forums/showthread.php?s=&threadid=118735
Not sure if that would apply in your situation as you are using PCB but maybe same concept, different sizes of cable and binary grounding.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.