Hi Sheldon,
Good work. I hope you've found the problem.
When you pulled the cap, low frequency gain dropped to about unity. That's why the AC component dropped.
Now to cure those grounding issues. Try and isolate tehtransformer case from the chassis temporarily. Measure the transformer to ground current across a 1K resistor in case it's leakage.
-Chris
Good work. I hope you've found the problem.
When you pulled the cap, low frequency gain dropped to about unity. That's why the AC component dropped.
Now to cure those grounding issues. Try and isolate tehtransformer case from the chassis temporarily. Measure the transformer to ground current across a 1K resistor in case it's leakage.
-Chris
anatech said:
Found the problem, maybe. But not the solution yet. Because of my somewhat unconventional construction practices, I had to separate the board and heatsink (where chassis earth is connected) from the rest of the chassis, so the transformer is not connected in any way physically to the amp, except via the power rails. So, the hum remains. Here's what intrigues me: I completely disconnected in power entry module and connected the AC to a freestanding connector. I leave the earth disconnected (or not, it makes no difference), and connect live and neutral, I get the usual hum. If I disconnect neutral, hum remains until the caps discharge. If I disconnect live, the amp goes quiet immediately. So all it takes is to have stored power and the live connected (no complete circuit to mains, either through neutral or ground) and I get hum.
Sheldon
gotta run off for a couple hours, but I'll be back at it later.
Hi Sheldon,
Ground loop.
Connect your AC ground directly to your chassis. Connect your circuit ground point to your AC ground through a 10 ohm resistor. This should break the loop. I've seen 100 ohms in this position also. Make sure no part of your circuit has a connection to chassis ground other than through this resistor.
-Chris
Ground loop.
Connect your AC ground directly to your chassis. Connect your circuit ground point to your AC ground through a 10 ohm resistor. This should break the loop. I've seen 100 ohms in this position also. Make sure no part of your circuit has a connection to chassis ground other than through this resistor.
-Chris
anatech said:
I think I understand. This is separate and after the 10R from signal input to ground as per Mike below.
MikeB said:
Maybe easier to explain my grounds. I the have board layed out pretty much like the schematic. If you look left to right, feedback, feedback caps, and signal ground are connected together. At that point I have a 10R that is then connected at the point all to way to the right of the schema where all the power grounds come together. The other circuit grounds (decoupling caps, CCS, current mirror, etc.) are connected to a 1.25mm thick copper wire that terminates at that same point. They occur more or less in the same order, left to right as on the schematic, except the rail decoupling caps, which are close to the terminal end. That end is connected to speaker ground, chassis and earth.
I have tried no resistor between the star and earth, as well as a 150R resistor. No difference whatever.
Sheldon
Hi Sheldon,
After re-reading your ground description, I think I found something. Not sure, but here goes.
Your speaker return should go directly to the main supply caps on the circuit side. Your other "dirty" grounds connect here also, by themselves until they reach this point. Your chassis should not connect here except through a resistance (10 ~ 100 ohms). The transfomer center tap connects to the capacitor commons by a different path so your ground does not have current pulses in it.
Treat each ground return as if it was a signal cable.
-Chris
After re-reading your ground description, I think I found something. Not sure, but here goes.
Your speaker return should go directly to the main supply caps on the circuit side. Your other "dirty" grounds connect here also, by themselves until they reach this point. Your chassis should not connect here except through a resistance (10 ~ 100 ohms). The transfomer center tap connects to the capacitor commons by a different path so your ground does not have current pulses in it.
Treat each ground return as if it was a signal cable.
-Chris
Well, I tried moving grounds around with no change, so I left them as described a couple of posts back. What was odd was that the amp seemed sensitive to handling the transformer CT, and sometimes the rail wires, but it was random. Also, more noise without a shorting plug on the input. I dunno, oscillation somewhere?
So I dug in and replaced the MPSA inputs (wrapped together face to face and shrinkwrapped as Chris suggested), the CCS transistors, and replaced the 20pf feedback comp cap, with a 10pf cap. It's now quiet. DC offset is about 1mV, and AC is about 0.1mV with the input shorted. It's just a bit more AC with the input open. It's actually quieter than the other amp with the input shorted, but a bit noisier with it open. Probably picking up a bit of inductive action due to the cramped spaces.
So, maybe it was a bad transistor or bad joint somewhere?
Anyway, thanks for all the help.
Sheldon
edit: Before I do any more mods, I think I'll wait to see AAK's boards. Looks like it would fit in my case nicely.
So I dug in and replaced the MPSA inputs (wrapped together face to face and shrinkwrapped as Chris suggested), the CCS transistors, and replaced the 20pf feedback comp cap, with a 10pf cap. It's now quiet. DC offset is about 1mV, and AC is about 0.1mV with the input shorted. It's just a bit more AC with the input open. It's actually quieter than the other amp with the input shorted, but a bit noisier with it open. Probably picking up a bit of inductive action due to the cramped spaces.
So, maybe it was a bad transistor or bad joint somewhere?
Anyway, thanks for all the help.
Sheldon
edit: Before I do any more mods, I think I'll wait to see AAK's boards. Looks like it would fit in my case nicely.
Hi Sheldon,
There is no need to have the transistors face to face really. Back to front will work fine. The important thing is to couple them thermally (grease) and isolate them a little from the enviroment (heat shrink).
I have some pictures of my boards somewhere "back there" in these threads.
-Chris
There is no need to have the transistors face to face really. Back to front will work fine. The important thing is to couple them thermally (grease) and isolate them a little from the enviroment (heat shrink).
I have some pictures of my boards somewhere "back there" in these threads.
-Chris
anatech said:
Thanks Chris, good to know. Actually though, face to face was the easy. I put them face to face, with the leads pointing in opposite directions. I banded the bodies with shrink wrap the same length as the transistor body. Then I bent the leads 90 degrees so that the pair formed an upside down U, with the coupled transistor bodies at the top and the leads facing down. This way, the lead orientation, and spacing is the same for both transistors as it was before and I could solder the pair right in where the old pair came out.
Sheldon
Copy away.
If I may paraphrase an old saying; necessity is the mother of engineering. My tendency to try and stuff as much as I can into compact space (and to forgo proper planning and wing it), often forces me into these contortions. The downside is that my "designs" would be the service tech's nightmare.
Sheldon
If I may paraphrase an old saying; necessity is the mother of engineering. My tendency to try and stuff as much as I can into compact space (and to forgo proper planning and wing it), often forces me into these contortions. The downside is that my "designs" would be the service tech's nightmare.
Sheldon
Hi thanh,
Yes you can. But understand that discontinuities in speaker motion or poor connections will show a falsely high distortion reading.
You would also need to measure the speaker impedance at each test frequency. We don't test that way because the data is corrupted by outside influences.
-Chris
Yes you can. But understand that discontinuities in speaker motion or poor connections will show a falsely high distortion reading.
You would also need to measure the speaker impedance at each test frequency. We don't test that way because the data is corrupted by outside influences.
-Chris
Hi Mike,
Yes, within reason.
But measuring distortion normally means that you are measuring a quantity that is tiny in comparison to the fundamental signal. It doesn't take much to add a lot of noise into the residual. I just had a dirty contact that played havoc with my measurements. Cleaning did the trick. I was using speakers as a load just for chuckles about two weeks ago. Thumping on the cone showed up big time as well.
To make a meaningful measurement, you really need to use a known, stable load. That will mean some kind of dummy load with no switch contacts involved, unless they are well maintained contacts.
In all fairness, the defective switch is on the order of thirty years old. I will probably replace it.
-Chris
Yes, within reason.
But measuring distortion normally means that you are measuring a quantity that is tiny in comparison to the fundamental signal. It doesn't take much to add a lot of noise into the residual. I just had a dirty contact that played havoc with my measurements. Cleaning did the trick. I was using speakers as a load just for chuckles about two weeks ago. Thumping on the cone showed up big time as well.
To make a meaningful measurement, you really need to use a known, stable load. That will mean some kind of dummy load with no switch contacts involved, unless they are well maintained contacts.
In all fairness, the defective switch is on the order of thirty years old. I will probably replace it.
-Chris
Feeback or Ground problems
I have two stereo symasyms in my system. All four channels were made on identical boards and, as far as I can tell are all wired identically. One channel is slightly noisier than the others - maybe 0.5mV hum, vs. 0.3 or less for the others. Here's the odd thing. I can induce significantly more hum by moving my head near the speaker phase plug (metal), or by manipulating the speaker wires. There is no continuity between the wires and the speaker frame or phase plug - so it must be an induced effect. If I actually touch the phase plug or touch one of the speaker wire ends, the extra hum dissapears. I've tried switching thing all around and it only does it with this channel. Something in the feedback circuit, or a bad solder joint somewhere? Any ideas to help narrow the search?
Sheldon
I have two stereo symasyms in my system. All four channels were made on identical boards and, as far as I can tell are all wired identically. One channel is slightly noisier than the others - maybe 0.5mV hum, vs. 0.3 or less for the others. Here's the odd thing. I can induce significantly more hum by moving my head near the speaker phase plug (metal), or by manipulating the speaker wires. There is no continuity between the wires and the speaker frame or phase plug - so it must be an induced effect. If I actually touch the phase plug or touch one of the speaker wire ends, the extra hum dissapears. I've tried switching thing all around and it only does it with this channel. Something in the feedback circuit, or a bad solder joint somewhere? Any ideas to help narrow the search?
Sheldon
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.