|
|
Post by deepdark on Aug 11, 2015 10:40:04 GMT -6
Hi there
I know Svart has made one (seen the thread over Groupdiy) so maybe you will be able to help me out, or anybody 
I linked the schematic used. I drawn the psu I made (RC filter) and the other thing is that R3 and R4 are 47K, like the original user manual schematic instead of the 12K, appearing on teh schematic laying Under original AM864.
Input and output transformer are Edcor, as well as power transformer.
So far, it passes audio. But there is some little thing that seems to be innehrent to some build.
1. Hiss and hum. There is a little. More of hiss, thought. it don't cover the audio, but it is there. Heater is balanced and grounded at transformer center tap. There is basicaly 2 ground point: one at the IEC and taking all PT center taps and psu ground. The other one is near the main board and take R8, V1/2 pins 1-3-5, C1/R1 and R9-R13-R14. So far, I think about it and maybe V1 and V2 aren't well balanced, maybe R8 (current control) should be grounded at the IEC. I'm pretty sure I have enought filtering, I calculated around 150db of ripple reduction, which should be plenty enough. Maybe the fact that input transformer isn't shielded it could picked up some noise/parasyte?? Input and ouput xlr are shielded cable and pin 1 to case. Maybe disconnecting the output pin 1 from ground? The noise is there with no signal passing through. The more I turn the attenuation knob, the more hiss it has, but I should say it's more around the middle position through the full cw that hiss and hum is present. So, of course, we could just drive the input with a strong signal and just use the first half of the attenuation, but I really want to get it to be clean.
2. Motorboating. It appear about the 3/4 position of thershold and sometimes while I turn the current control, too. I'm pretty sure it has to do with V1 and V2, maybe they aren't well balanced so bias is not optimal??
3. The lack of low end. I mean, a real lack. Coupling caps are stock to the original (0.1uf). I really don't know where in the circuit it could be lost. A bad 6sk7 tube? Maybe I could try to pull out the 6SQ7 tube, so basically the compression side and see if my low end came back. Since I used the same Edcor transformer than other folks, it shouldN,t be the culprit.
Thanks guys Attachments:AM864 SCHEMATIC.pdf (447.94 KB)
|
|
|
|
Post by svart on Aug 11, 2015 11:20:43 GMT -6
Hey, I never actually finished it. I traded the unit for some other things years ago!
I'll have to look at the schematic again to remember everything.
|
|
|
|
Post by deepdark on Aug 11, 2015 11:38:12 GMT -6
Thanks. I'm thinking about C1 that could be bad??? Maybe it is Noisy and/or isn't fully passing frequencies??
|
|
|
|
Post by EmRR on Aug 11, 2015 11:48:06 GMT -6
Current imbalance in the output will erase low end. The originals have no low end anyway. I'm sure you've seen my plots at GroupDIY. Pull the 6q7 like you said and see, it may motorboat more with B+ rise.
Motorboating is usually layout related assuming substituting tubes changes nothing. I'd suspect the pots in question or anything around the 6sq7 layout.
Hum is layout/shielding/proximity, based on everything else you've said about power supply. Hum is also not great in the original, somewhat masked by compression.
I don't remember the total gain, can you measure the hiss? May be normal.
|
|
|
|
Post by deepdark on Aug 11, 2015 12:45:55 GMT -6
Thanks EmRR. The current balance pot is 2W instead of 1W, so I'm pretty sure trouble isn't there.
When hooking the unit through my apogee (which has an output of 100ohm and input of 10K, so basically I load the unit at 10K), I can see on the tracking meter of my freq. analyser plugin (melda) a noise floor of about -51db, no signal, attenuator fully counterclockwise.
So my grounding would probably looks right, then? I know I could try grid stopper and other tricks, though. Maybe current control ground is better off to be strapped at the IEC bolt?? Other than that, power transformer is bolted out the chassis, psu is Inside at one extremity and the rest of the audio is at the other extremity, so a good 15inches far from the psu. Tubes have their pin 1 (shield) to ground. One thing, maybe the fact that shielding and signal ground met themselve at the same ground connection, noise and interferance flow along signal ground and pollute it? maybe if I strap these pin 1 straight to the tube socket mounting bolt it will help?
|
|
|
|
Post by deepdark on Aug 11, 2015 12:47:07 GMT -6
oh and by current imbalance in the output, do you mean inbalance between both 6sk7 sucking out my lows?
|
|
|
|
Post by EmRR on Aug 11, 2015 13:53:45 GMT -6
I mean imbalance in the 6SN7 through the output transformer. There's no balance pot there so good tube balance is required. 6SK7 won't have much to do with lows/imbalance.
36dB max according to manual. Fully counterclockwise is more like 25dB noise contribution from amplification, since input transformer is remove from equation. I assume that's -51dBFS, peak reading. RMS reading can be 10dB or so less from memory. Still not great. I have 80 year old high gain limiters with much lower noise floors. You can study my power induced noise plots in the DaveP 26C thread and also his shield observations in the recent U73B thread. Suffice to say there's a lot of what would appear to be voodoo, and it takes both measurement and hands on experimentation to solve. You can make some things better, while having others get worse. The 26C in the end required use of outboard PSU to get quiet, and even then I tried every possible variation of what went in which box to find the lowest result. If you have a stand alone bench supply that can power this, hook it up rather than powering from internal and see what the result is. That is the only way to be certain where it's coming from.
You should load the output with at least 1K, 680R if you feel 'by the book'. Response will be better and noise may seem so too. As is you are not using the output near it's defined impedance rating.
C1 doesn't pass audio, it's DC bias timing.
It would make sense for current control to be grounded at same point as B-. I imagine the original has a ground bus bar, haven't looked at one in awhile.
|
|
|
|
Post by deepdark on Aug 11, 2015 14:37:54 GMT -6
So mine has better noise rejection than an original one, reading at those spec, isn't it? I'll move current ground to psu ground. So my trouble could be the 6SN7 tube itself, or the circuit around? I know I'm not using it as I should with load of 680ohm. Probably it doesn't help much, too.
|
|
|
|
Post by deepdark on Aug 11, 2015 15:35:20 GMT -6
Basicaly, i should get the exact same value at each pair of pins (plate-cathode-grid)? Or the tube itself has to be balanced, or the output transformer has to be balanced between hi-center tap and low-center tap?
|
|
|
|
Post by EmRR on Aug 11, 2015 15:51:47 GMT -6
So mine has better noise rejection than an original one, reading at those spec, isn't it? I didn't say that. |
|
|
|
Post by deepdark on Aug 11, 2015 15:55:01 GMT -6
Sorry i misunderstood you.
|
|
|
|
Post by deepdark on Aug 11, 2015 16:06:33 GMT -6
Concerning the balanced current between 6sn7 and transfo, i guess i would have to get the exact same value at plates, coming from the output transformer and the exact same grid value, while passing audio, of course. Isn't? In case of imbalance result, what can be done to correct that?
|
|
|
|
Post by deepdark on Aug 11, 2015 17:32:14 GMT -6
ok. I just test 2 things. First I remove the 6sq7 and the motorboating stopped. So we could conclude it comes from the compression section, isn't it?
I measured impedance between output primary hi-center tap and Low - center tap. Weird reading here. Hi/center tap show a z of about 192,1ohm and low/center tap 175,5ohm. No really well balanced I think. Could it be our culprit here?
|
|
|
|
Post by deepdark on Aug 11, 2015 17:35:25 GMT -6
secondary of input tranformer :263ohm and 235ohm
|
|
|
|
Post by deepdark on Aug 11, 2015 17:52:23 GMT -6
compression help to tame the noise, the more i compress, the more noise floor is reduced. at threshold middle position, noise floor is about -55db, after is start to motorboating but i'm sure we could get to -60db. another symptom is my meter wont sit at 0 on the gr side, it sit at around -18. Imaybe not the proper vu meter, thought. I get it from hairball audio. internally rectified. I could play with parralele resistor to R2, or maybe the culprit is the 6sq7???
|
|
|
|
Post by deepdark on Aug 11, 2015 18:38:41 GMT -6
I tried the unit with a microphone. So far, I didn't noticed any hiss, but off course I have to large sucked out low end. Hiss is just present when I plug the output of the apogee to the input of the federal. I'm pretty sure hiss will not be a problem in a proper 600ohm environnement. I tried with unplugged input and there was no hiss, so it looks to come from the input transformer.
|
|
|
|
Post by deepdark on Aug 11, 2015 18:52:34 GMT -6
other measurement, i got a differentieal of about 1.26v between the plates of the 6sk7. I only get a drop of 0,2v through r2 where i should probably have something like 1,2v.
|
|
|
|
Post by deepdark on Aug 11, 2015 18:54:16 GMT -6
one thing I'm thinking is that maybe Imy threshold pot should be 1w instead of 1/2w
|
|
|
|
Post by EmRR on Aug 11, 2015 19:25:11 GMT -6
Current balance through the transformer. Nothing to do with voltage values or grids. Plenty of examples of current balance pots in push pull amps out there. Usually a cathode pot.
Also well documented in GroupDIY threads: resistive measurements have nothing to do with balance.
Motorboating stops with 6sq7 pulled, implies it's side chain related, not compression section related since 6SK7 is the compression section.
Threshold pot is not a problem.
Your R2 drop explains the low meter reading. R3/4 being 12K would cause more drop, right? For a given meter, you can't change 3/4 without also changing 2. Change meter types, change R2 again. So change R2 until you get more meter reading. etc etc
|
|
|
|
Post by deepdark on Aug 11, 2015 19:26:18 GMT -6
one thing I could try but I'm a little affraid of is changing the push pull transformer for a wsm10k/600, the same one preservative sound used. I'm affraid to should b+ into a transormer which isn't specifically made for push pull operation.
|
|
|
|
Post by deepdark on Aug 11, 2015 19:48:03 GMT -6
Thank Doug for your help, really. You have no idea how I appreciate it. So, it's allright for the meter, I'll work it out when I'll sort out the motorboarting trouble and the low end trouble. So here, we talked about current balance through the output transformer. So I should implement a curent balance pot and tweek it until got back my low end, which will tell me that I have some sort of balance in there. Just to be sure I'm following you, we are talking about dc current? To be honest, that part is a little obscur to me. Basically, the current balnace pot should be strap at cathode of 6sn7, so pin 3/6, right? If possible, could you explain to me why the imbalance would sucked such a large portion of the frequency? As it acting like a hpf loool. I'll take a further look at the sidechain circuit. Could it be simply the tube which is defective?
|
|
|
|
Post by EmRR on Aug 11, 2015 20:29:25 GMT -6
Yes, DC current balance. It is one possible source of loss of lows. Put a 100 ohm pot between the cathodes 3/6, R9 change to 190R and connect to the sweeper of the pot. Wire wound pot would be a good idea, I think Mouser has 5W fairly cheap. While looking at response in real time with something like pink noise source, no compression, adjust for maximum low frequency. If it makes no difference, it's not the problem, if it does then mount that pot somewhere that isn't easily touched accidentally.
|
|
|
|
Post by deepdark on Aug 12, 2015 4:41:49 GMT -6
Alright ill try that out. And if it is not the trouble, maybe the trouble came from the circuit between the 6sk7 and 6sn7. Coupling cap, for exemple. They are the right value but we never know. Is it possible one of the transformer just fucked up my lows, just out of curiosity?
|
|
|
|
Post by deepdark on Aug 12, 2015 5:42:07 GMT -6
I'm thinking about something. Maybe the way I strap the current control and treshold pot have to do with motorboating trouble. Instead of running a pair of wire from 6sq7 pin 3 to the wiper of R8 and pin 1 of R10, I ran it straight to R8 and then strap the wiper of R8 to R10pin 1. Maybe I should run a pair of wire from pin 3. Another one, do R6 need to be close to R8? Mine is on teh same psu board and I run wires from it the the other extremity of the case to the audio board. For now, these are my thought. R8 is adjuste to 2.5v, according to the manual.
|
|
|
|
Post by deepdark on Aug 12, 2015 6:14:59 GMT -6
I've just order the wirewound lienar pot and I just got new couplign caps, 0.22u instead of 0.1u. I'll try them out, too, in case one of them (maybe C5) goes in trouble
|
|
