gary h said:
Looking at your grounding sketch and the schematic from your website: You show the signal input shield tied directly to the chassis, where the earth ground is connected. That seems wrong. I thought that the signal input shield was connected to the ground points for the pot (R301) and the input resistor (R302).
Hi Tom,
I used a balanced mini plug cable for my test lead. I cut the rca off one channel at the other end, cut back the shielding a cm or so as you suggested, and soldered the aligator clip onto the line of that channel. The other rca/channel is taped back out of the way. When holding the test lead in the air I get a nearly flat signal.
Here are some more responses:
"Is the input of the amp grounded to the amp chassis?"
I don't think so. The inputs enter immediately onto a pcb which is insulated from the steel section of chassis by a 1/16th inch rubber pad. I don't see any other leads going to the chassis except the tranny CT
"Do you get any voltage readings (DC or AC) at any of the electrolytics after it has been turned off?"
I'm not sure I understand Tom, I thought that the caps are supposed to store a charge. Perhaps you could teach me something here.
"I know that you have a DC power supply circuit for the filaments. The ground for it should be connected to the main ground."
It's the same ground track that runs around the pcb.
"Have you tried using 90% Isopropyl Alcohol and a stiff toothbrush to clean all of the solder flux residue etc off of the board?"
I used isopropyl alcohol and q-tips. To be honest, although I rubbed vigorously I notice many connections still have that shine around them.
"Do you know of anyone else with this preamp, who doesn't have a hum problem?"
I don't know anyone else with this preamp. I built it as a kit from tubes4hifi.com.
"Can you measure or see any 60Hz or 120Hz between the ground points in the center of the pcb (that eventually make their way all the way around one side or the other of the loop) and the point where the ground wire leaves the pcb, OR, for completeness' sake, between the two ends of that wire, OR, between the far end of the ground 'loop' trace (or the inner grounds that connect to it, that are actually farther from the wire's connect-point), i.e. farthest from the ground wire, and the non-PCB end of the ground wire?"
The scope measurement seen on http://home.comcast.net/~garyworld/preunplugged.jpg is typical measured allover the chassis, starground, and various places on the pcb ground track. I'll check the places you mentioned with the preamp on and on/connected to the amp.
"Have you tried plugging only the preamp in through the isolation xfmr, and everything else into a regular outlet, and other combinations like that?"
Yes. The isolation transformer hasn't made a noticeable improvment in the sound. In fact it has a slight hum itself that is barely audible from the listening position. If I am correct it removes line noise caused by fluctuations in voltage that come from your street supply. I assume these are transient noises caused by transformers under dynamic load. Someone told me that it can help remove noise caused by dimmer switches too. So far it hasn't helped noticeably.
"(By the way, what is the value of the smoothing capacitor, in your lower-voltage supply? The schematic I saw had 10 uF. That couldn't be right, could it? If that's for the heaters, would a three-terminal adjustable regulator be worth trying?)"
I don't know why that is on the schematic. Probably a previous version that was not corrected. Either way, it is not installed. If you look on the line section schematic, on the bottom there is an H+ schematic. Caps 304 and 305 are installed, they are 10k uF 16V. (Also on that H+ schematic it shows a cap 306- .10 uF 50V. This was not included nor is there a spot for it shown on the board layout diagram. Could this be a significant omission? I'm afraid a 3 terminal adjustable regulator is beyond me. Feel free to explain this if you have time.
Could you recommend resistor values for those I can use to step down the sensitivity of scope. If I blow my computer up I won't be able to harrass you with this hum problem any more.
Thanks,
gary
I used a balanced mini plug cable for my test lead. I cut the rca off one channel at the other end, cut back the shielding a cm or so as you suggested, and soldered the aligator clip onto the line of that channel. The other rca/channel is taped back out of the way. When holding the test lead in the air I get a nearly flat signal.
Here are some more responses:
"Is the input of the amp grounded to the amp chassis?"
I don't think so. The inputs enter immediately onto a pcb which is insulated from the steel section of chassis by a 1/16th inch rubber pad. I don't see any other leads going to the chassis except the tranny CT
"Do you get any voltage readings (DC or AC) at any of the electrolytics after it has been turned off?"
I'm not sure I understand Tom, I thought that the caps are supposed to store a charge. Perhaps you could teach me something here.
"I know that you have a DC power supply circuit for the filaments. The ground for it should be connected to the main ground."
It's the same ground track that runs around the pcb.
"Have you tried using 90% Isopropyl Alcohol and a stiff toothbrush to clean all of the solder flux residue etc off of the board?"
I used isopropyl alcohol and q-tips. To be honest, although I rubbed vigorously I notice many connections still have that shine around them.
"Do you know of anyone else with this preamp, who doesn't have a hum problem?"
I don't know anyone else with this preamp. I built it as a kit from tubes4hifi.com.
"Can you measure or see any 60Hz or 120Hz between the ground points in the center of the pcb (that eventually make their way all the way around one side or the other of the loop) and the point where the ground wire leaves the pcb, OR, for completeness' sake, between the two ends of that wire, OR, between the far end of the ground 'loop' trace (or the inner grounds that connect to it, that are actually farther from the wire's connect-point), i.e. farthest from the ground wire, and the non-PCB end of the ground wire?"
The scope measurement seen on http://home.comcast.net/~garyworld/preunplugged.jpg is typical measured allover the chassis, starground, and various places on the pcb ground track. I'll check the places you mentioned with the preamp on and on/connected to the amp.
"Have you tried plugging only the preamp in through the isolation xfmr, and everything else into a regular outlet, and other combinations like that?"
Yes. The isolation transformer hasn't made a noticeable improvment in the sound. In fact it has a slight hum itself that is barely audible from the listening position. If I am correct it removes line noise caused by fluctuations in voltage that come from your street supply. I assume these are transient noises caused by transformers under dynamic load. Someone told me that it can help remove noise caused by dimmer switches too. So far it hasn't helped noticeably.
"(By the way, what is the value of the smoothing capacitor, in your lower-voltage supply? The schematic I saw had 10 uF. That couldn't be right, could it? If that's for the heaters, would a three-terminal adjustable regulator be worth trying?)"
I don't know why that is on the schematic. Probably a previous version that was not corrected. Either way, it is not installed. If you look on the line section schematic, on the bottom there is an H+ schematic. Caps 304 and 305 are installed, they are 10k uF 16V. (Also on that H+ schematic it shows a cap 306- .10 uF 50V. This was not included nor is there a spot for it shown on the board layout diagram. Could this be a significant omission? I'm afraid a 3 terminal adjustable regulator is beyond me. Feel free to explain this if you have time.
Could you recommend resistor values for those I can use to step down the sensitivity of scope. If I blow my computer up I won't be able to harrass you with this hum problem any more.
Thanks,
gary
gootee said:
It's not a very good drawing. The input does go first to a volume pot. I didn't include it for simplicity. Then to the PCB ground track. This ground track is connected (from inbetween the filter caps section) to the disconnect network. So the signal grounds are connected to the chassis but via the disconnect network (R/Caps/Diodes.) This network jumps the insulated standoff inbetween the signal grounds and the earth ground, tranny CT, and chassis. So power grounds are connected to chassis, input/pcb/output grounds go to the disconnect which connects to the chassis.
I might have this all wrong but after reading that information you referenced I thought a star ground has everything connect to the chassis eventually. I'm going to redraw the illustration so it is not so confusing.
gary
Tom,
there's a new ground scheme diagram here:
http://home.comcast.net/~garyworld/groundscheme1.jpg
Let me know if this makes more sense.
gary
there's a new ground scheme diagram here:
http://home.comcast.net/~garyworld/groundscheme1.jpg
Let me know if this makes more sense.
gary
gary h said:
Sounds good.
Hailteflon and maybe others asked about this, first, IIRC. (By the way, caps store energy, not charge. They 'store charge' only in the sense that they can store a difference between the charges on each of their two plates. But the total charge in them is always the same. That's my understanding, anyway.) In many or maybe even most pieces of electronic equipment that use smoothing capacitors in the power supply, the power supply caps have 'bleeder resistors' installed in parallel with them, so that the caps get discharged after the unit is turned off. Most caps will eventually self-discharge, through internal leakage. But that might take quite a while, depending on the cap. If yours are staying charged after power-off, they might still be energizing something, such that the hum problem is caused or contributed-to. That's the theory, anyway, or something like it.
You can use something like a small screwdriver, to scrape it from between adjacent pads. It might make your PCB ugly, though. I guess you could also try a stiff brush, instead of Qtips.
It probably wouldn't make much difference for the low frequency hum. But you could always solder one across a larger cap's terminals. It shouldn't hurt anything, unless it's a film or other very-low-ESR 0.1uF cap and happens to set off some unwanted HF resonance. An X7R or worse (i.e. not NPO or C0G) ceramic type should be OK.
You'd need to know the max allowable input voltage, for your sound card. I guess we could assume 1 volt, to be safe. But that might be somewhat inconvenient, resistor-wise. It would be nice if it were more like 10v. At any rate, I'd probably use 1 Meg (or maybe 100k) for the 'base' value (i.e. across which you connect the sound card input. Then it's just a simple voltage-divider type of calculation. If Rtip is the resistance between your sound card connection and the measurement point, and Rbase is between your sound card input and ground, and Vtip is the voltage to be measured (i.e. from probe tip to ground) and Vbase is the voltage across Rbase, which is the voltage that your sound card input would see, then Vbase = Vtip [ Rbase / (Rbase + Rtip) ]. So, to give the soundcard 1/10th of the tip voltage, Rtip would be 9X Rbase. To give the soundcard 1/100th of the tip voltage, Rtip would be 99x Rbase. As you can see, the resistor values can get large, for Rtip. But, they are actually easy to find, even up to 100 Meg or more. However, for safety, you might want to use multiple series resistors for Vtip, anyway. For the 10X type of probe (i.e. 1/10 of actual voltage gets seen by soundcard input), it would make sense to buy 10 resistors of the same value, say 1 Meg, and use 9 of them for Rtip and 1 for Rbase. But, for a 100X version, you probably wouldn't want 99 series resistors for Rtip. And you might actually need a 300X or more probe, if the soundcard can only handle 1 volt input. So you might want to try something like a 100K Rbase and three 10 Meg resistors in series for Rtip. (Check my math.) That would actually give you a 301X probe. Now that I think about it, you'll probably want something more like 500X, so you have a little safety margin. At any rate, check the soundcard's input specs, and try to use about 2/3's of that range for your B+. i.e. maybe make the probe so it will be safe for your soundcard input with 150% of your B+ voltage.
gary h said:
(I might get sucked-into saying something stupid, again, but...)
Well.... In your new grounding diagram, it LOOKS like the signal input shield/ground connects to the chassis, at the point of entry.
If that is the case, your problem might be solved by completely insulating the input connector(s) from the chassis. i.e. I think that the signal input shield/ground should ONLY connect directly to the pot and the input resistor.
I would insulate *ALL* i/o connectors, pots, switches, and everything else, from the chassis, so that the circuit's ground connection to the chassis is only at one point, and one point only. (I thought that that was standard practice.)
As is, it looks like some ground loops are created, too. For example, the input shield connects to the chassis where it enters. But then it also connects to the chassis through the PCB grounding. (Remember that any enclosed loop will have current induced in it by any changing electromagnetic fields. The bigger the loop's enclosed area, the worse it can be.)
The Nylon standoff is confusing, to me. How are you connecting grounds to nylon?
gary h said:
Wait. It sounds like you have only one connection, there. You should have a circuit ground connection at the probe end, too.
Typically, you'd have an alligator clip for the ground/shield of the probe, and the center conductor wire protruding to act as the tip of the probe.
If you only have the tip of the probe, and no circuit ground connection at the probe end of the cable, then your measurments would include all of the ground voltage differentials between your measurement point and the soundcard's input jack's ground. :-o
Even WITH the probe's ground connection, you still have to be careful about where you connect it, especially when looking for a problem like you're having, now. 'Just any ground point' won't do, when connecting the probe ground. You need to always think about exactly what voltage-difference you are measuring, between the probe tip and the probe ground.
For 'regular' display of voltages, you would normally want to use the closest directly-relevant ground point, i.e. the correct ground reference for the signal you want to display. For example, if you just want to see the actual voltage at an output jack, you would want the probe ground connected directly to the output jack's shield/ground. In that case, connecting the probe ground to some random ground point in the circuit or even the main power supply ground, instead, while it might seem from a schematic to be technically correct, would corrupt your waveform and measurments due to voltage differences between the chosen ground point and the ground to which the signal should actually be referenced (e.g. the pesky voltages which can be induced by currents flowing through the parasitic impedances of the ground conductors).
gootee said:I would insulate *ALL* i/o connectors, pots, switches, and everything else, from the chassis, so that the circuit's ground connection to the chassis is only at one point, and one point only. (I thought that that was standard practice.)
The Nylon standoff is confusing, to me. How are you connecting grounds to nylon?
Tom,
both i/o jacks are insulated from the chassis on a non-conductive strip. Rat Shack 4-pack. So I hope this eliminates the ground loop question. The nylon standoff is has a screw in each end that helps bind the ring connectors of each lead together. One end connects output, pcb, and one end of the disconnect network. The other end connects earth, CT and the other end of the network to the chassis. I use the standoff because I couldn't think of another neat and secure way of connecting the first three relatively close to the starground and be able to quickly disconnect them when experimenting with different arrangements. Here's a pic: http://home.comcast.net/~garyworld/starground2.jpg taken before the current arrangement. (You can also see the inputs in this photo.) Does this make sense? I think it is important that I get on the right track with this starground so let me know if it still seems wrong with these corrections.
gary
gootee said:
I'm glad you pointed this out. It is confusing to me considering it is the ground scheme that is in question here. So if I connect the probe ground clip to the nearest point (of measure) on the pcb ground track for signal measurements, and to the chassis at the star ground for power supply measurements, does this sound reasonable?
gary
gary h said:
Hi Gary,
Sorry. I had thought that your last ground-scheme sketch, at http://home.comcast.net/~garyworld/groundscheme1.jpg , showed the shield connected to the chassis at the point of entry. But, instead, it actually seems to show the signal input connected to the chassis, there.
)I guess I still don't have my head quite all the way around the star grounding for this preamp, yet. And since the hum happens even with the power off, the non-star ground might not even be the root cause. But it looks like you'd have to start cutting your PCB's ground track loop, to do the star grounding better. And I don't think I want to be the one to advise that.
Look especially at the "Notes" that accompany the last two diagrams, here: http://www.geofex.com/Article_Folders/stargnd/stargnd.htm . For example, each filter cap's ground should have a separate wire to the star ground, as should the transformer's center tap if possible. i.e. They should connect to _nothing_ else. Etc Etc.
DON'T do the following, but, in your preamp's case, to do minimal modification, maybe you would cut the ground track to the left of the filter caps, and make the CT's PCB attach point be a 'sort of' star ground point, and connect to the chassis from that point only. But it might be better to cut on either side of both sets of filter caps, jumper the ground track around them, and add new ground wires from them to the star point. But, if you were going to go that far, you would probably want to just do the rest of them separately, too, with a few more trace cuts and separate return wires. And, actually, the signal inputs' and outputs' ground reference points might be the first ones I'd cut and run separately, probably even before doing the filter caps. And I'd probably think about optionally including 'ground separator' resistors, maybe 1 Ohm or somewhat more, in series with the inputs' ground return wires, although they shouldn't be needed if the grounding scheme is ideal.
But for something less destructive, to try right now, maybe you should try moving your star ground point to where the CT enters the PCB, and connect that point to the chassis. I'd try it without the disconnect network, too. You might be able to do it with minimal harm to the original setup by installing a bolt from under the chassis, right next to where the CT is soldered to the PCB, so you can connect your star ground returns to the bolt. You'd have to add a wire from where the CT is soldered, with a contact that could slide onto the bolt. Alternatively, and possibly better, you could place the bolt next to the ground for the first filter capacitor. As a compromise, and even less destructive, you could use the existing PCB mounting-screw hole, that's between the filter caps and the CT connection, near the corner of the PCB.
I don't really remember how you did your other 'star ground' wires, either. Do you have a slightly larger photo that shows where they come from? Or you could just describe them.
However, I still don't really know if it's even worth worrying TOO much more than that about trying to make a 'real' star ground, at this point, other than isolating everything from the chassis and connecting to it at only one point, with minimal shared ground conductor before the star point so the filter caps and CT induce minimal ripple voltages into the other ground returns.
Interestingly, maybe, it looks like he (in that linked-to article) has the center-tap connected only when the power switch is on. And it looks like you do have an unused pole on your power switch (I guess you didn't implement a 'standby' mode, which might have helped to save your solid-state diodes from frying.
.Also, it seems like you should have the two red transformer wires twisted together, or braided, along with the red/yellow (CT) wire. And your umbilical kind of worries me, since it might increase some 'enclosed loop area', and might also provide coupling between whatever wires are paralleled in there.
gary h said:
Hi Gary,
That sounds more-or-less reasonable. That's the correct general idea, anyway. I'm just not sure about the power supply measurement statement. Maybe either the point where the center-tap connects to the PCB, or else where the filter caps connect to the ground track, would be the best reference for that. (But seeing if there are differences between using those three ground references for measurements might give you some insights.)
Tom, Gary: I haven't gotten any emails about posts since Friday. I'll have to catch up to see what is going on.
I did get one email about an old conversation concerning counterfeit caps.Fake Hitachi DIY thread
I'll try to catch up. Mark
I did get one email about an old conversation concerning counterfeit caps.Fake Hitachi DIY thread
I'll try to catch up. Mark
I see a problem with your star grounding - output signal and its return doesn't follow each other. This will pick up magnetic fields - even when turned off. I'd use the screen of the output lead connected in both ends.
Transformer center tap must go directly to the first filter cap (C101) too. Chassis is not right, only earth ground lead from power cord should connect to chassis side of the ground isolator network. The original connection point isn't optimal either, center tap currents both flow down to the caps and all the way round the board polluting signals. You should cut the ground trace around the peremiter just above (in picture preamp2) where the center tap is supposed to connect.
Depending on how the filter caps for the heaters are connected the upper side of the cut might be best jumpered to a point just to the left of C102 or maybe best left floating. The ground loop breaker can probably be connected either to both shields of the output RCA:s OR to a point just to the left of C102. I'd probably connect it to the shields of all RCA:s strapped together. In that case the loop area between input and output should be minimized. Your wiring looks okay, just move the output leads and input leads closer together.
If after doing this there is hum with shorting plugs on the input which is at its strongest when volume is mid-range then shielding may be needed for the connection between volume pot and amp input.
That line fuse on board looks scary, not enough creepage distance and the mains leads look like they might come off without too much force. I'd use a chassis mounted fuse holder (screw-in type) on the back panel. Good for hum pickup reasons too keeping the leads away from the circuit.
Transformer center tap must go directly to the first filter cap (C101) too. Chassis is not right, only earth ground lead from power cord should connect to chassis side of the ground isolator network. The original connection point isn't optimal either, center tap currents both flow down to the caps and all the way round the board polluting signals. You should cut the ground trace around the peremiter just above (in picture preamp2) where the center tap is supposed to connect.
Depending on how the filter caps for the heaters are connected the upper side of the cut might be best jumpered to a point just to the left of C102 or maybe best left floating. The ground loop breaker can probably be connected either to both shields of the output RCA:s OR to a point just to the left of C102. I'd probably connect it to the shields of all RCA:s strapped together. In that case the loop area between input and output should be minimized. Your wiring looks okay, just move the output leads and input leads closer together.
If after doing this there is hum with shorting plugs on the input which is at its strongest when volume is mid-range then shielding may be needed for the connection between volume pot and amp input.
That line fuse on board looks scary, not enough creepage distance and the mains leads look like they might come off without too much force. I'd use a chassis mounted fuse holder (screw-in type) on the back panel. Good for hum pickup reasons too keeping the leads away from the circuit.
Hi Tom: This email problem has persisted from the beginning. I don’t know if it is a DIY software problem or if I have checked a box somewhere that limits it. Over a year ago I turned off some emails to certain conversations and there seems to be some "phantom" trace of it in the system.
I cancelled a Newark catalog in the late 80s and their system still blocks a catalog request.
Did Gary ever try the new caps that he ordered? I didn’t see any mention of it.
Also, did he measure the B+ voltage after turn-off? There seems to be a total emphasis on a grounding problem that may not be the cause of the hum.
I agree with you that the spikes in the scope trace are indicative of an unshielded probe. Mark
I cancelled a Newark catalog in the late 80s and their system still blocks a catalog request.
Did Gary ever try the new caps that he ordered? I didn’t see any mention of it.
Also, did he measure the B+ voltage after turn-off? There seems to be a total emphasis on a grounding problem that may not be the cause of the hum.
I agree with you that the spikes in the scope trace are indicative of an unshielded probe. Mark
Also, the one place in the ground system that could be the source of this hum is the connection between the DC heater and the B-.
I was looking over some forums and/or tutorials about DC heater circuits and there seems to be a school that chooses to raise the potential of the DC heater ground to about 50V volts over the B- (system ground). This indicates that the DC heater ground is often a source of noise/hum. I didn’t read much of it, just scanned over it.
If it is so sensitive as to require raising the heater ground to an unterminated potential (not a closed circuit) of 50 volts for example then possibly a cold solder joint in the heater ground could give rise to this hum. Mark
I was looking over some forums and/or tutorials about DC heater circuits and there seems to be a school that chooses to raise the potential of the DC heater ground to about 50V volts over the B- (system ground). This indicates that the DC heater ground is often a source of noise/hum. I didn’t read much of it, just scanned over it.
If it is so sensitive as to require raising the heater ground to an unterminated potential (not a closed circuit) of 50 volts for example then possibly a cold solder joint in the heater ground could give rise to this hum. Mark
What rectifier tube is used now? I'm a bit worried that h-k voltage might be a bit high for it. The same for the upper tube in the SRPP circuit. Raising the heater ground voltage is usually done for that reason - to not exceed tube ratings. I checked a 12AU7 datasheet I found - max is 180V. 145V as stated in the schematic is close but should work. Don't know about the rectifier though as I don't know what is used.
hailteflon said:
Maybe the Administrator could fix that for you.
Interesting. I don't use Newark, very much, preferring mouser.com, digikey.com, and alliedelec.com. The Allied and Mouser paper catalogs, especially, are great information sources. Allied's 2,208-page 2008 catalog even has most photos in color. Mouser's quarterly catalog is up to 2,110 pages.
I don't think so.
Not that I know of.
I agree, and mentioned something like that, recently. But, even though it is a commercial product, there do seem to be some major errors in the grounding, which *could* be the cause. And it's difficult, for me at least, to not start troubleshooting by focusing on already-known possible causes of the problem.
I did ask megajocke to look at this thread and try to offer suggestions, having seen his excellent advice on grounding (and many other issues) in another thread. (Thanks for dropping by, megajocke!) Let's hope that his fresh set of eyes and expertise can help Gary make some progress.
I think it turned out that he was using the cable 'probe' with no circuit ground connection, which made the ground reference be at the soundcard's input jack(!), and had the probe-cable shielding unconnected (except at soundcard end).
Gary, I hope that you will try getting some more scope shots, with the probe's circuit-ground connection in use, to see how they compare. Once that seems to be working better, at least you'll be able to show a scope display, if someone needs that to help diagnose your problem.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.