Some photos of the assembled circuit
Thanks again for any help or suggestions.
I have the amp home with me for possible modification, however, the problem only seems to be present when at my office (desired location for this amp), and I won't be back there until Monday.
Post #18 of this thread has a link to the build log of the project if you want more photos.
Last edited:
I just realized my drawn schematic has an error. The output jack ground is connected after the 1k resistor, not before.
Some photos of the assembled circuit
Thanks again for any help or suggestions.
I have the amp home with me for possible modification, however, the problem only seems to be present when at my office (desired location for this amp), and I won't be back there until Monday.
Post #18 of this thread has a link to the build log of the project if you want more photos.
This looks like the kind of build that would work splendidly at audio frequencies but fall apart spectacularly at RF. Parasitic inductance all over the place. RF is not kept out of either the input or the output. (The 33 pF cap at the input looks rather big for one - ceramic NP0 would be better - and also just goes to the insulated connector's ground tab. You should also be having a cap from ground tab to chassis right there. Output wise, a common-mode choke and some capacitance may be called for.)
I take it the chassis is just that, without any further metal case parts? There is a reason why audio gear tends to come in fully enclosed metal cases. (Tube audio amps used to have big metal cages over the top.) Then Mr. Faraday will be helping you out at least, assuming all the case panels are making good contact (not always a given), and you can concentrate on keeping the RF out of your connections (all of them). Plan B is having circuit boards with a nice ground plane, but your P2P build is about the antithesis to that.
You'll have a lot to read up on, like this.
I take it the chassis is just that, without any further metal case parts? There is a reason why audio gear tends to come in fully enclosed metal cases. (Tube audio amps used to have big metal cages over the top.) Then Mr. Faraday will be helping you out at least, assuming all the case panels are making good contact (not always a given), and you can concentrate on keeping the RF out of your connections (all of them). Plan B is having circuit boards with a nice ground plane, but your P2P build is about the antithesis to that.
You'll have a lot to read up on, like this.
Last edited:
I would suggest adding at the headphone jack, an internal series resistor of 1R to 10R in each hot lead,
and a 0.01uF ceramic COG capacitor (with short leads) from each hot phone jack terminal to chassis ground
(not to the phone jack common). Also add a similar capacitor from the jack common to chassis ground.
If the strongest pickup is from the headphone cable, as seems to be the case from your earlier description,
this could help.
and a 0.01uF ceramic COG capacitor (with short leads) from each hot phone jack terminal to chassis ground
(not to the phone jack common). Also add a similar capacitor from the jack common to chassis ground.
If the strongest pickup is from the headphone cable, as seems to be the case from your earlier description,
this could help.
Last edited:
I assume by "internal", you mean inline with the signal for each of the stereo (R, L) signals? I will try this.
Thanks for the linked information. I'll give it a read later today. The input cap is 33pF, but a film cap. It is what I had laying around.
Hint - needless fullquotes (at the bottom no less) have a tendency of turning a thread into a cluttered unreadable mess and hence are best avoided. It is advised to quote only the relevant part(s) (you can also insert an endquote tag and copypasta the existing quote tag for splitting things up).
You can call on moderation (who btw happen to agree with this view but seem to be busy right now) to clean up if posts are past the 30 minute edit limit.
This is giving me flashbacks to the days of the Outlook Express invasion on Usenet... a rather quirky piece of software indeed. Gosh, I haven't thought of that in absolute ages. Around 20 years ago now probably...
You can call on moderation (who btw happen to agree with this view but seem to be busy right now) to clean up if posts are past the 30 minute edit limit.
This is giving me flashbacks to the days of the Outlook Express invasion on Usenet... a rather quirky piece of software indeed. Gosh, I haven't thought of that in absolute ages. Around 20 years ago now probably...
Good point @sgrossklass, what the forum needs is a one click to reply with an @ username. Not sure why, but the reply box no show on my mobile phone
Last edited:
Yes, be sure to use COG/NPO types, others won't work nearly as well.
Last edited:
Are you aware the phone jack you used is "switching", the two sides are NOT the same? Turn it over and look what touches what. Or at least jumper the two "Common" lugs together. (Since you apparently do have working hot terminals.)
In a high-RF environment I bet you should just bond the RCA jacks HARD to the chassis. Take those plastic washers out. (Or jump the RCA grounds to the nearest solid chassis ground.)
In a high-RF environment I bet you should just bond the RCA jacks HARD to the chassis. Take those plastic washers out. (Or jump the RCA grounds to the nearest solid chassis ground.)
If these other worthy suggestions don't give full satisfaction, this might be a good assignment for a ferrite bead in each signal lead to the headphone jack (where you put the 1 to 10 ohm resistors).
You may even need one of those cute little metal cans for your valve -- it is sticking out the top, isn't it -- and it'd probably appreciate the bit of mechanical protection.
You may even need one of those cute little metal cans for your valve -- it is sticking out the top, isn't it -- and it'd probably appreciate the bit of mechanical protection.
Thanks for all the helpful suggestions. I ordered some parts to give @rayma 's suggested fix a try. Now I wait.
@PRR I impressed with your noticing that I had the headphone jack wired incorrectly. I have since fixed that issue, but the error still shows in some of the photos.
The input jacks are tied to ground. If the output RC doesn't fix the issue, I will try your suggestion too.
@Rick PA Stadel, Thanks for the suggestion. I have thought about adding some metal tape to the inner sides of the wooden chassis. I'm just not sure it would help, and the downside is that it would be visible from my cool acrylic bottom plate. As far as tube shield goes, I looked around, with no luck. Any suggestions on a retailer? Maybe I'll check the swap section, someone may have an old one in a parts stash.
Have a great weekend all!
@PRR I impressed with your noticing that I had the headphone jack wired incorrectly. I have since fixed that issue, but the error still shows in some of the photos.
The input jacks are tied to ground. If the output RC doesn't fix the issue, I will try your suggestion too.
@Rick PA Stadel, Thanks for the suggestion. I have thought about adding some metal tape to the inner sides of the wooden chassis. I'm just not sure it would help, and the downside is that it would be visible from my cool acrylic bottom plate. As far as tube shield goes, I looked around, with no luck. Any suggestions on a retailer? Maybe I'll check the swap section, someone may have an old one in a parts stash.
Have a great weekend all!
> The input jacks are tied to ground.
Through a very long shield. While that would work for power/audio frequency, evidence is you have a 39KW FM station right near the office. FM waves are just a few feet long. A foot of shield is NOT a short-circuit to FM waves. And by not tying RCA jacks to chassis right away, you basically have a foot of high-RF antenna looping over everything in the box.
Through a very long shield. While that would work for power/audio frequency, evidence is you have a 39KW FM station right near the office. FM waves are just a few feet long. A foot of shield is NOT a short-circuit to FM waves. And by not tying RCA jacks to chassis right away, you basically have a foot of high-RF antenna looping over everything in the box.
@PRR
Ok, I understand what you are saying now.
Would RFI on the inputs be amplified by the circuit, because it is unaffected by volume control?
Ok, I understand what you are saying now.
Chassis is powder coated, removing the washers would only be first step. Cleaning the powder coating off would be a chore.
So would I keep the shield and ground for the audio circuit, and run an additional wire from both RCA inputs to chassis ground?
Would RFI on the inputs be amplified by the circuit, because it is unaffected by volume control?
I made some modifications to the circuit this weekend.
Thanks to those of you that suggested these fixes for my RFI problem.
I won't know if the changes will be effective until I get back to the office tomorrow, but I am hopeful, and the circuit still sounds great.
I added the 10ohm resistors and .01uF caps at the output. As far as I gather, this is a low pass filter to keep the RFI from entering the circuit via output.
I also took the time to Dremel off the powder coating below the input jacks. I tested the amp after reinstalling without the isolation washers. It had a hum. I think, probably, because the back plate and top plate(chassis ground) are not electricaly connected.
So, I put the isolation washers back on the inputs, and ran another ground to the common chassis ground.
I also changed the output caps, mostly because I prefer the Elna audio caps to the common Nichicon ones. I also swapped the input film caps for C0G/NP0 types.
Thanks to those of you that suggested these fixes for my RFI problem.
I won't know if the changes will be effective until I get back to the office tomorrow, but I am hopeful, and the circuit still sounds great.
I added the 10ohm resistors and .01uF caps at the output. As far as I gather, this is a low pass filter to keep the RFI from entering the circuit via output.
I also took the time to Dremel off the powder coating below the input jacks. I tested the amp after reinstalling without the isolation washers. It had a hum. I think, probably, because the back plate and top plate(chassis ground) are not electricaly connected.
So, I put the isolation washers back on the inputs, and ran another ground to the common chassis ground.
I also changed the output caps, mostly because I prefer the Elna audio caps to the common Nichicon ones. I also swapped the input film caps for C0G/NP0 types.
Last edited:
Or because they in fact are and you introduced a ground loop. You still have to watch out for those!
A chassis without a solid electrical connection between all parts would be a bit silly (not impossible, but silly). If they put any thought into this at all they would have screwed it together first and powdercoated it then; sure looks that way to me.
I built the chassis, and had it powder coated. It was 5 separate pieces, all coated individually. I am learning a lot from this build. I made a lot of mistakes, choosing what would be visually appealing without a lot of knowledge about how I was affecting things.
I remembered to bring the amp into work today, and the RFI is lessened. I can no longer hear the actual words to songs, but I do hear the tick tick tick of the beat in the background.
Things I can still think to try:
- ferrrite beads on the output
- removing the ground connection of the inputs at the volume pot, that will remove a long run of shield from the circuit.
- Also, I don't think the heatsinks are connected to the chassis electrically either. Since they are just super glued in, I bet I can heat up the connection for removal, add connect them to chassis too.
I'm going to live with it for the week at work, and decide on Friday what my weekend project will be.
Things I can still think to try:
- ferrrite beads on the output
- removing the ground connection of the inputs at the volume pot, that will remove a long run of shield from the circuit.
- Also, I don't think the heatsinks are connected to the chassis electrically either. Since they are just super glued in, I bet I can heat up the connection for removal, add connect them to chassis too.
I'm going to live with it for the week at work, and decide on Friday what my weekend project will be.
Last edited:
I do think the ferrite beads would be effective. But let's sort out a couple things first.
I don't remember the pinout of that brand of headphone jack. Is the right row (under the purple arrow to the right) just being used as a tie-point? And is that red wire going to the eye terminal (yellow arrow) the one providing the ground for the 0,01 uF?
It looks like the resistor and capacitor might be connected in series. Better effect would result wiring the capacitor in shunt (parallel with the capacitance of the shielded cable) and the resistor in series. I know one of the wise contributors earlier advised shunting the capacitors to the chassis nearby, but I'd like to suggest one additional step. Put a larger value, maybe 0,1 uF (ceramic or other high-frequency type) from the jack ground to the chassis (assuming the jack ground is NOT connected to the chassis). Then place the two 0,01's directly across the cable to its ground. The resistor (with the ferrite bead on it, if placed) then connects that node to the headphone signal.
Then a couple general things:
- Solder is for the electrical connection; the mechanical should be made first. Your soldering looks first-rate but it would go faster and easier with little hooks in the wire ends. See Construction Tips, or Best Practices in one of the ARRL Handbooks or some such.
- Always leave what the books call a 'service loop'. The way you have the 2,2 ohm resistors across the jack may seem reasonable and an efficient use of space, but it greatly aggravates any changes that might need to be made. Bend the resistor leads in a short horseshoe shape (you'll usually need to cut off almost half of the lead), then form the hook in the end to secure it mechanically to the terminal.
- High frequencies require shorter leads (than the red wire and long capacitor leads to chassis ground). When you have a 100 MHz, multi-kilowatt source a block away, it sees any 5 to 8 cm wire length as a coil! It may only be a few tens of nano-Henry but that's enough to allow a few microvolts of signal, which will be audible in a headphone amp.
Now for your points (post #38).
To the second -- please don't do that. It won't help.
Grounding the heat sinks won't help the incoming RFI problem, but it is a good idea. Then the top plate can help with the heat dissipation, too. Also, as you consider using glue in the future, keep two things in mind -- the thermal cycling will tend to loosen glue (especially a 'rigid' formulation like super glue), and what might get shorted when that bit of gear breaks loose.
That is definitely a beautiful chassis; maybe next time save money and only have it powder-coated on the outside. Sure is purdy, though ..
Regards
Oh, and an inside star lockwasher, and some ThreadLocker, would be great under that grounding screw terminal.
And, oh - oh, almost forgot -- similar treatment may be applied to the power inlet jack, too.
I don't remember the pinout of that brand of headphone jack. Is the right row (under the purple arrow to the right) just being used as a tie-point? And is that red wire going to the eye terminal (yellow arrow) the one providing the ground for the 0,01 uF?
It looks like the resistor and capacitor might be connected in series. Better effect would result wiring the capacitor in shunt (parallel with the capacitance of the shielded cable) and the resistor in series. I know one of the wise contributors earlier advised shunting the capacitors to the chassis nearby, but I'd like to suggest one additional step. Put a larger value, maybe 0,1 uF (ceramic or other high-frequency type) from the jack ground to the chassis (assuming the jack ground is NOT connected to the chassis). Then place the two 0,01's directly across the cable to its ground. The resistor (with the ferrite bead on it, if placed) then connects that node to the headphone signal.
Then a couple general things:
- Solder is for the electrical connection; the mechanical should be made first. Your soldering looks first-rate but it would go faster and easier with little hooks in the wire ends. See Construction Tips, or Best Practices in one of the ARRL Handbooks or some such.
- Always leave what the books call a 'service loop'. The way you have the 2,2 ohm resistors across the jack may seem reasonable and an efficient use of space, but it greatly aggravates any changes that might need to be made. Bend the resistor leads in a short horseshoe shape (you'll usually need to cut off almost half of the lead), then form the hook in the end to secure it mechanically to the terminal.
- High frequencies require shorter leads (than the red wire and long capacitor leads to chassis ground). When you have a 100 MHz, multi-kilowatt source a block away, it sees any 5 to 8 cm wire length as a coil! It may only be a few tens of nano-Henry but that's enough to allow a few microvolts of signal, which will be audible in a headphone amp.
Now for your points (post #38).
To the second -- please don't do that. It won't help.
Grounding the heat sinks won't help the incoming RFI problem, but it is a good idea. Then the top plate can help with the heat dissipation, too. Also, as you consider using glue in the future, keep two things in mind -- the thermal cycling will tend to loosen glue (especially a 'rigid' formulation like super glue), and what might get shorted when that bit of gear breaks loose.
That is definitely a beautiful chassis; maybe next time save money and only have it powder-coated on the outside. Sure is purdy, though ..
Regards
Oh, and an inside star lockwasher, and some ThreadLocker, would be great under that grounding screw terminal.
And, oh - oh, almost forgot -- similar treatment may be applied to the power inlet jack, too.
Last edited: