Got it. That should be easy to do. You know, I felt like I was being so careful about ensuring that the signal inputs to the output stage were correct, but given my track record, I could have hosed this, I guess.
If this is the case, however, why wouldn't I have heard the oscillation in both speakers all the time? I'm not challenging, you understand, I'm just curious to know why it would have manifested itself when the pot was at one extreme end or another and only in one channel initially.
I gotta learn somehow.
OK, so if I use some shielded CAT 5, I'll only use one conductor, but where should I ground the shield? I'd assume that the signal star is the place to go with it, but you know about my assumptions...
Kofi
Kofi Annan said:
Yes, you'd expect that, wouldn't you? It all depends on layout. I once rebuillt a Leak TL12+ with the driver department on a PCB and with nicely twisted heater wiring, but with positive feedback by mistake. I spotted my error when I found that moving a finger to within an inch of the input socket (nothing plugged in) caused a loud hum. Given that the sensitivity of the amplifier with negative feedback was 120mV, the sensitivity without feedback was probably 12mV (guessing 20dB of feedback), so the sensitivity with positive feedback would have been 1.2mV! No wonder it hummed...
In your case, Kofi, the PFB could cause the amplifier to oscillate in an ultrasonic frequency, with your audio signal "modulating" the HF...
An O'scope would help you a bunch if you can get a grab on one...
Gastón
An O'scope would help you a bunch if you can get a grab on one...
Gastón
EC8010 said:
Perhaps of value. (Not to disagree but a slightly different slant):
Normally when design NFB is inadvertantly applied the wrong way round the amplifier goes into complete "shut-off" by way of becoming a square wave oscillator at some very low frequency, somewhat like a bi-stable. For the supersensitive operation that EC8010 mentioned positive feedback would have had to be slightly less than amplifier gain, not about 20x amplifier gain as is the usual case with NFB - I think you would see that now, EC8010.
The purpose of posting this is actually to warn about sorting NFB out by trial and error - for the uninitiated. In semiconductor circuits this is not really a problem as the circuit design sorts this out, but with output transformers one simply does not know. But "positive" NFB (oscillation) can cause damage if let on for any length of time, so it is advisable to do this more safely. This is best achieved by trying with a very low degree of feedback first.
With the OPT secondary earthed at any side, attach a fly lead to the live side with a series resistor about 100X bigger than the design feedback resistor, and touch the feedback point. If a scope and signal generator is not available, input some programme (low level) and when the fly lead contacts the feedback point the programme will become either louder or softer without oscillating. The latter condition is of course the desirable one for negative feedback, after which connections could be made permanently.
Call me chicken, but I always do it this way.
Regards.
Johan Potgieter said:
Yes, I take your point - I wasn't thinking straight. Unfortunately, at the time, not only did I not have an oscilloscope, but I didn't know how to use one, so I couldn't investigate properly. Looking back, there must have been two faults...
Kofi Annan said:
Actually I would ground the shield as far from the end that connects to the tube as possible.
For instance, connect the end to your speaker output (or wherever you are taking your signal for FB) and connect the other end to the grid of your input tube (or wherever your FB goes to). Now take the shield from the end by the speaker output and connect it to a ground point. Chassis or safety earth works just fine for shield grounding, in fact I connect all my shields to a common 'shield bus' and connect that to my chassis/safety/earth ground point.
If there isn't such a ground near that point I would solder the shield to a relatively heavy solid core copper wire (18 ga or so) and run that wire to ground.
Be sure that the end of the shield near the tube can't touch anything. I like shrink tubing for insulating that end.
I am sure you'll get this one nailed down. It may well turn out to be an issue with PFB rather than one of shielding. However shielding never hurt anything! (Unless of course one grounds both ends of the shield making an even more effective antenna!😀 )
Thanks for all the responses!
I have experimented with the shielded wire and it's a no go. I tried it with one channel with the NFB shielded and one with no NFB at all and the channel with NFB did thee squeakin' thing at both ends of the pot and seemed to me to be lower in volume than the one with no NFB.
So, that means it's puppet show time, folks! Here's how I connect-i-fied it:
I'll refer to the "upper" and "lower" EL84s as they are displayed in Jones' schematic.
On the E88CC:
Pin 3 (K') sees an 828R to ground and a 2K45 in parallel with a 910p, which is connected to the positive speaker terminal
Pin 8 (K) is connected to the .1uF cap which is in turn connected to the "lower" EL84 grid
Pin 6 (A) is connected to the .1uF cap which is in turn connected to the "upper" EL84 grid
On the EL84s:
Pin 7 (A) of the UPPER valve is connected to the blue primary of the OPT
Pin 9 (G2) of the UPPER valve is connected to the blue / yellow primary of the OPT through a 47R resistor
Pin 7 (A) of the LOWER valve is connected to the brown primary of the OPT
Pin 9 (G2) of the LOWER valve is connected to the brown / yellow primary of the OPT through a 47R resistor
Now, Jones had some parallel resistor values in his feedback loop and in the E88CC cathode bias, but I replaced them with single resistor values detailed above. I don't think this would make a difference unless I got the values wrong, but I don't think I did.
Sorry to burden you with this, but I'm just at a loss and I'm not sure what I need to do here. Any help you could give would be much appreciated. I'm hoping you can review this and immediately see where I effed up. 'Cause ya know I did it! Ya know it!
Your faith in my utter incompetance has not been misplaced. I promise you that.
Please help.
Kofi
I have experimented with the shielded wire and it's a no go. I tried it with one channel with the NFB shielded and one with no NFB at all and the channel with NFB did thee squeakin' thing at both ends of the pot and seemed to me to be lower in volume than the one with no NFB.
So, that means it's puppet show time, folks! Here's how I connect-i-fied it:
I'll refer to the "upper" and "lower" EL84s as they are displayed in Jones' schematic.
On the E88CC:
Pin 3 (K') sees an 828R to ground and a 2K45 in parallel with a 910p, which is connected to the positive speaker terminal
Pin 8 (K) is connected to the .1uF cap which is in turn connected to the "lower" EL84 grid
Pin 6 (A) is connected to the .1uF cap which is in turn connected to the "upper" EL84 grid
On the EL84s:
Pin 7 (A) of the UPPER valve is connected to the blue primary of the OPT
Pin 9 (G2) of the UPPER valve is connected to the blue / yellow primary of the OPT through a 47R resistor
Pin 7 (A) of the LOWER valve is connected to the brown primary of the OPT
Pin 9 (G2) of the LOWER valve is connected to the brown / yellow primary of the OPT through a 47R resistor
Now, Jones had some parallel resistor values in his feedback loop and in the E88CC cathode bias, but I replaced them with single resistor values detailed above. I don't think this would make a difference unless I got the values wrong, but I don't think I did.
Sorry to burden you with this, but I'm just at a loss and I'm not sure what I need to do here. Any help you could give would be much appreciated. I'm hoping you can review this and immediately see where I effed up. 'Cause ya know I did it! Ya know it!
Your faith in my utter incompetance has not been misplaced. I promise you that.
Please help.
Kofi
While someone is trying to help with this (Kofi, I am not trying to do so now simply because I must go to bed - 2:30 in the am. here!), I also have a question. I have just obtained a circuit diagram of the Bevois Valley and will understand better in future, but:
I notice with the output transformer the usual two secondary windings bridgeable either in series or parallel (shown here in parallel). This is usual for a 4/16 ohm configuration, but why is 8 ohm stated at the output? That would imply that it is an 8/32 ohm configuration. Surely not? (In fairness I must state that I have only the circuit diagram - no accompanying text.)
Thanks!
I notice with the output transformer the usual two secondary windings bridgeable either in series or parallel (shown here in parallel). This is usual for a 4/16 ohm configuration, but why is 8 ohm stated at the output? That would imply that it is an 8/32 ohm configuration. Surely not? (In fairness I must state that I have only the circuit diagram - no accompanying text.)
Thanks!
Just for poops and grins, disconnect the OPT primary wires (with the power off!), and check to make sure the color coding is consistent with what you think it is. The highest resistance pair should be the blue and brown. The resistance between blue/yellow and blue should be lower than brown/yellow and blue. And so on and so on.
I'll check resistances on the OPT tomorrow and let you know what I find.
Do you think my configuration is correct as I described above? I wonder if I switched to 100K attenuators if there would be any difference...
Any other ideas? Any obvious errors I could have made here? We love obvious errors at the Annan household. We just love 'em! Yes we do!
Kofi
Do you think my configuration is correct as I described above? I wonder if I switched to 100K attenuators if there would be any difference...
Any other ideas? Any obvious errors I could have made here? We love obvious errors at the Annan household. We just love 'em! Yes we do!
Kofi
No, but I'll add that to the list of stuff to do here. That's an easy one and I'll do it.
Also, I hope I read the OPT windings correctly from Hammond's drawing here. I tied the green / yellow and green together, tied the yellow to the positive speaker terminal, and I tied the black / yellow and black together and tied that combination to the negative speaker terminal.
So, first I'll check resistances on the windings of the OPTs and post those. Then, I'll swap the feeds to the output valves. Then I'll stand up and kick the snot out of it and throw it in the street. Then I'm on my way to divorce court.
Is that in the right order?
Kofi
Also, I hope I read the OPT windings correctly from Hammond's drawing here. I tied the green / yellow and green together, tied the yellow to the positive speaker terminal, and I tied the black / yellow and black together and tied that combination to the negative speaker terminal.
So, first I'll check resistances on the windings of the OPTs and post those. Then, I'll swap the feeds to the output valves. Then I'll stand up and kick the snot out of it and throw it in the street. Then I'm on my way to divorce court.
Is that in the right order?
Kofi
I'd agree. That's easiest and most likely. I do like Johan's trick of using a reduced amount of feedback to check for proper polarity.
So, you're that sure that this is a postive feedback issue, eh? I'm going to do this (probably tonight) since I believe that this is the only reasonable answer at this point. However, I just have to ask how I could have screwed this one up.
I mean, I triple-checked this last night and I feel so sure that I followed the Hammond 1608 OPT drawing exactly. I traced the signal path back to the input tubes and it just looks correct to me.
What other factors could be in play that would have caused me to screw this up? SY implied that maybe the leads are incorrectly marked on the transformers, but is this really possible? Has that ever happened to anyone before?
I realize that this is a completely fixable issue and I'm not actually that discouraged by it. I'm just trying to learn and understand what errors I could have made to create the positive feedback situation that is apparently occurring.
You mean swapping the inputs to the EL84s is the way to go, right?
Man, I wish I wasn't so dense.
Kofi
Kofi,
I've gotten OPT leads mixed up, even when following the diagram, and even when trying to follow the winding patterns on the transformer.
I think you could just lift the feedback loop out of the circuit (e.g. unhook one lead of the FB resistor) and see if the music on that channel gets louder or softer. If it gets louder when NFB is not connected, and softer when you put it back in, you're good to go.
--Jeff
I've gotten OPT leads mixed up, even when following the diagram, and even when trying to follow the winding patterns on the transformer.
I think you could just lift the feedback loop out of the circuit (e.g. unhook one lead of the FB resistor) and see if the music on that channel gets louder or softer. If it gets louder when NFB is not connected, and softer when you put it back in, you're good to go.
--Jeff
Kofi Annan said:
In a (stereo) amplifier I always wire one channel one way and the other channel the other way. That way, one channel is right, and one wrong. I've given up trying to decide beforehand which is correct...
kofi, mon ami,
Hope the following will help, if you have not already solved the problem by now.
Firstly, this is where I loose my patience with manufacturers. Hammond is supposed to be a foremost manufacturer of output transformers. So why in %&*#$ do they follow a certain lead pattern, but suddenly phase reverse that with transformers like the 1650G, 1615E and 1650E?
But that is not your problem. You have a 1608 if I read correctly. Then I stand corrected with respect to my earlier observation that NFB reversed as positive feedback will kind of "lock out" an amplifier in violent square wave oscillation. The important thing is that according to the Hammond diagram for 1608 you have your phase connected wrongly as you described - positive feedback.
If you analise, the BV diagram should not be followed literally. It is drawn in the customary way, which is not necessarily phase correct. The brown primary lead is in phase with the secondary yellow lead for 8 ohm. (The Hammond parallel green/green-yellow lead as live is for 4 ohm output; see their drawing. One uses the yellow lead for 8 ohm live.) Thus for NFB the top EL84 anode in the amplifier diagram must go to the brown primary lead (with g2 to brown-yellow) and so on.
You can of course reverse the g1 feeds from the 100nF coupling capacitors also as someone suggested, but I presume that layout-wise that is more difficult or perhaps less neat than simply switching the output transformer leads, which come through the same chassis hole anyway. If you stick with E88CC pin 6 connected through the 100nF to the "top" EL84, your connections should be as suggested above.
I do of course assume that Hammond has their diagrams drawn in correct phase, i.e. that primary brown is in phase with secondary yellow for the 1608 etc. Otherwise they should even more definitely be shot.
As an exercise, satisfy yourself of the following phase exercise. If the amplifier input (E88CC pin 2) goes positive at any stage, then NFB should cause E88CC pin 3 to go positive as well (NFB cancels the signal). Following the diagram, E88CC pin 1 and pin 7 will then be going negative and pin 6 will be going positive. That means that the "top" EL 84 pin 2 will go positive but its anode (pin 7) will go negative. That therefore needs to be connected to Hammond lead brown, as lead blue needs to go positive according to the Hammond diagrams, for yellow to also go positive. To confirm, the output transformer in the circuit diagram is shown the wrong way round for correct phase.
As said, circuit diagrams are not always diagrammatically phase correct; one needs to work that out for oneself. Lastly I cannot then explain why things were apparently OK with your input pot slider somewhere in the middle, unless I misunderstood something somewhere.
If this does not solve the problem (as it should), and in the absence of a scope, almost the only remaining thing will be to apply feedback gradually with a say 100 K pot and note at what setting you start experiencing funny behaviour at some setting of the input pot. (You will have to do this without the 910pF feedback capacitor as it will over-compensate for a large setting of the 100K pot and start oscillation that way.
Sorry for a rather lengthy contribution, but it was in the interest of clarity.
Regards
Hope the following will help, if you have not already solved the problem by now.
Firstly, this is where I loose my patience with manufacturers. Hammond is supposed to be a foremost manufacturer of output transformers. So why in %&*#$ do they follow a certain lead pattern, but suddenly phase reverse that with transformers like the 1650G, 1615E and 1650E?
But that is not your problem. You have a 1608 if I read correctly. Then I stand corrected with respect to my earlier observation that NFB reversed as positive feedback will kind of "lock out" an amplifier in violent square wave oscillation. The important thing is that according to the Hammond diagram for 1608 you have your phase connected wrongly as you described - positive feedback.
If you analise, the BV diagram should not be followed literally. It is drawn in the customary way, which is not necessarily phase correct. The brown primary lead is in phase with the secondary yellow lead for 8 ohm. (The Hammond parallel green/green-yellow lead as live is for 4 ohm output; see their drawing. One uses the yellow lead for 8 ohm live.) Thus for NFB the top EL84 anode in the amplifier diagram must go to the brown primary lead (with g2 to brown-yellow) and so on.
You can of course reverse the g1 feeds from the 100nF coupling capacitors also as someone suggested, but I presume that layout-wise that is more difficult or perhaps less neat than simply switching the output transformer leads, which come through the same chassis hole anyway. If you stick with E88CC pin 6 connected through the 100nF to the "top" EL84, your connections should be as suggested above.
I do of course assume that Hammond has their diagrams drawn in correct phase, i.e. that primary brown is in phase with secondary yellow for the 1608 etc. Otherwise they should even more definitely be shot.
As an exercise, satisfy yourself of the following phase exercise. If the amplifier input (E88CC pin 2) goes positive at any stage, then NFB should cause E88CC pin 3 to go positive as well (NFB cancels the signal). Following the diagram, E88CC pin 1 and pin 7 will then be going negative and pin 6 will be going positive. That means that the "top" EL 84 pin 2 will go positive but its anode (pin 7) will go negative. That therefore needs to be connected to Hammond lead brown, as lead blue needs to go positive according to the Hammond diagrams, for yellow to also go positive. To confirm, the output transformer in the circuit diagram is shown the wrong way round for correct phase.
As said, circuit diagrams are not always diagrammatically phase correct; one needs to work that out for oneself. Lastly I cannot then explain why things were apparently OK with your input pot slider somewhere in the middle, unless I misunderstood something somewhere.
If this does not solve the problem (as it should), and in the absence of a scope, almost the only remaining thing will be to apply feedback gradually with a say 100 K pot and note at what setting you start experiencing funny behaviour at some setting of the input pot. (You will have to do this without the 910pF feedback capacitor as it will over-compensate for a large setting of the 100K pot and start oscillation that way.
Sorry for a rather lengthy contribution, but it was in the interest of clarity.
Regards
Wow! Thanks for the advice!
I tried this as soon as I got home, and I'm sorry to report that switching the leads resulted in MASSIVE oscillation. It started motorboating right away with the leads swapped and the NFB in. I mean, it was scary. And it was a totally different sound than before. This was like a putputputputputputputputput whereas before it was like a squeeeeeeeeeeeeee at either end of the pot.
Remember that the left channel was not having any issues with the feedback in until I started screwing with the right channel, which then brought on the squeeeeeeeeeeeee in both channels.
So, I disconnected the NFB and, no problems. I suppose that means that it was correct the first time, or does it mean that I have something else backwards? All I did was swap the blue and blue / yellow on the upper for the brown and brown / yellow for the lower and it just went bananas.
Again, I know it's me and not you. I am a natural screw-up.
I'll try and experiment with the other ideas you suggested. The fact that this is only an issue with the NFB in means that its just got to be something simple. I used Bourns 50K log pots for volume controls, but that shouldn't be causing any trouble, I think.
The resistor for the NFB is not all that close to the pin, so maybe I can get it closer to help resolve the issue. Also, I'll try to replace with some 100K pots and perform your test, Johan.
I know I'm stupid and all, but I really thought I could track this one down. Discouragement factor is +18% for the day.
Any other ideas?
Kofi
I tried this as soon as I got home, and I'm sorry to report that switching the leads resulted in MASSIVE oscillation. It started motorboating right away with the leads swapped and the NFB in. I mean, it was scary. And it was a totally different sound than before. This was like a putputputputputputputputput whereas before it was like a squeeeeeeeeeeeeee at either end of the pot.
Remember that the left channel was not having any issues with the feedback in until I started screwing with the right channel, which then brought on the squeeeeeeeeeeeee in both channels.
So, I disconnected the NFB and, no problems. I suppose that means that it was correct the first time, or does it mean that I have something else backwards? All I did was swap the blue and blue / yellow on the upper for the brown and brown / yellow for the lower and it just went bananas.
Again, I know it's me and not you. I am a natural screw-up.
I'll try and experiment with the other ideas you suggested. The fact that this is only an issue with the NFB in means that its just got to be something simple. I used Bourns 50K log pots for volume controls, but that shouldn't be causing any trouble, I think.
The resistor for the NFB is not all that close to the pin, so maybe I can get it closer to help resolve the issue. Also, I'll try to replace with some 100K pots and perform your test, Johan.
I know I'm stupid and all, but I really thought I could track this one down. Discouragement factor is +18% for the day.
Any other ideas?
Kofi
Kofi:
First idea: That a thing doesn't work out OK the first time you switch it on doesn't mean you are a total failure, less than all when the only basis for your build has been only a schematic... C'mon, man, you have to be used to that with all of the drill of frustration you have day after day there at the UN 🙂
Second idea... now that we ruled out PFB, let's go on with decoupling... What did you use as the "285V regulator"... did you use some decoupling there ? Power supply noise rejection ratio is not the best characteristic of any SE triode stage. Less than all if it is a sensitive one... I have second edition of Morgan JOnes book and the schematic is Fig. 5.30 on page 307, but the advice regarding the decoupling the preamp/phase splitter stages from the power ones still holds...
Gastón
First idea: That a thing doesn't work out OK the first time you switch it on doesn't mean you are a total failure, less than all when the only basis for your build has been only a schematic... C'mon, man, you have to be used to that with all of the drill of frustration you have day after day there at the UN 🙂
Second idea... now that we ruled out PFB, let's go on with decoupling... What did you use as the "285V regulator"... did you use some decoupling there ? Power supply noise rejection ratio is not the best characteristic of any SE triode stage. Less than all if it is a sensitive one... I have second edition of Morgan JOnes book and the schematic is Fig. 5.30 on page 307, but the advice regarding the decoupling the preamp/phase splitter stages from the power ones still holds...
Gastón
- Status
- Not open for further replies.