So:
Rewired the shunt volume control. No better. Curiously when I touch the spindle of the volume control I get noise on the speakers - sort of a very light buzz. No idea why - the chassis is earthed. I'm starting to get REALLY fed up with this so I'm going to take out the selector switch and volume control altogether and just simplify.
Beefed up the power supply - I now have 400v HT and 65v on the cathode so I must be running 6B4Gs (Sovtek) at about 19 watts. that;s more like it! Louder but still distorted.
Anyway, I'm fed up with this. Feel like ripping the input stages out and starting again.
I fancy 1J6g into 71A. Already building one of those on another chassis.
One gets moments like this.
andy
Rewired the shunt volume control. No better. Curiously when I touch the spindle of the volume control I get noise on the speakers - sort of a very light buzz. No idea why - the chassis is earthed. I'm starting to get REALLY fed up with this so I'm going to take out the selector switch and volume control altogether and just simplify.
Beefed up the power supply - I now have 400v HT and 65v on the cathode so I must be running 6B4Gs (Sovtek) at about 19 watts. that;s more like it! Louder but still distorted.
Anyway, I'm fed up with this. Feel like ripping the input stages out and starting again.
I fancy 1J6g into 71A. Already building one of those on another chassis.
One gets moments like this.
andy
Fix what you have got before moving on, the lessons you learn will be invaluable. This topology is quite workable - you've missed a couple of things.
The noisy volume control issue is an indication that the input stage might be oscillating.
Grid stoppers need to be physically close to the grid and there should be nothing but the grid after it.
Have you measured the plate voltages, voltages from plate to cathode, and cathode to ground? This will tell you whether or not you have the voltage headroom to swing the signal voltages required in each stage.
You really shouldn't need trimmers to set the CCS current required for the proper filament voltage, both are well characterized by the original manufacturer. Something is amiss with component values, implementation or your design.
Learn to use a scope. Right now you are shooting blind.
Simulate in Spice, it will save you some time if not provide quite the full answer, but it will get you close.
This is an impatient world, learn patience..
Lynn's comments about CCS are more germane to transformer coupled pushpull driver stages than your differential RC coupled design. That said I don't like their sonics that much in most (not all) applications either.
Most of all don't give up. Rome wasn't built in a day, and neither was this amplifier..
The noisy volume control issue is an indication that the input stage might be oscillating.
Grid stoppers need to be physically close to the grid and there should be nothing but the grid after it.
Have you measured the plate voltages, voltages from plate to cathode, and cathode to ground? This will tell you whether or not you have the voltage headroom to swing the signal voltages required in each stage.
You really shouldn't need trimmers to set the CCS current required for the proper filament voltage, both are well characterized by the original manufacturer. Something is amiss with component values, implementation or your design.
Learn to use a scope. Right now you are shooting blind.
Simulate in Spice, it will save you some time if not provide quite the full answer, but it will get you close.
This is an impatient world, learn patience..
Lynn's comments about CCS are more germane to transformer coupled pushpull driver stages than your differential RC coupled design. That said I don't like their sonics that much in most (not all) applications either.
Most of all don't give up. Rome wasn't built in a day, and neither was this amplifier..
andyjevans said:
I believe you may be referring to this thread: http://www.diyaudio.com/forums/showthread.php?postid=1684566#post1684566
Especially posts #7 and #10.
A painstaking and time-consuming process for me, but very rewarding in the end. Don't give up!
kevinkr said:
I would say this is right on; the transformer coupled driver is a very difficult load for a triode to drive, and performance must really be optimized with a scope and hours of test. Ask me how I know this ;-) For an RC coupled stage, a CCS under the diff pair is just what the doctor ordered.
Thanks guys. I worked until midnight last night, and by then was in danger of throwing the amp across the room. i am, in fact, a very patient person but there are limits! I once threw an optical mouse across the room, and curiously it worked after that. Still, let's not make that a precedent. Fortunately, unlike one German computer programmer, I don't have a shotgun to blast the offending object into smithereens.
OK - I own up, must learn my scope. It's a contest with Logic on the Mac to input music, but has to be both. I've been reminded enough by well-meaning colleagues! Don't know why I get these moods of technophobia.
So - agenda for today.
Put some sort of resistor and cap or equivalent between HT of first two stages.
Take out volume and selector and wire from XLRs straight to input grids with suitable grid stoppers. I usually use Dale metal film 1k ones right on the grid pins - do I really need to go to carbon?
Equalise voltages on 6B4Gs. I suspect the fact that one is dropping to 5v means it's a bit weak - even though they're all new. The side with the weak one distorts worse, so we could in fact have two problems - who said there is only one!!!
Will report. Feel a little better today
andy
OK - I own up, must learn my scope. It's a contest with Logic on the Mac to input music, but has to be both. I've been reminded enough by well-meaning colleagues! Don't know why I get these moods of technophobia.
So - agenda for today.
Put some sort of resistor and cap or equivalent between HT of first two stages.
Take out volume and selector and wire from XLRs straight to input grids with suitable grid stoppers. I usually use Dale metal film 1k ones right on the grid pins - do I really need to go to carbon?
Equalise voltages on 6B4Gs. I suspect the fact that one is dropping to 5v means it's a bit weak - even though they're all new. The side with the weak one distorts worse, so we could in fact have two problems - who said there is only one!!!
Will report. Feel a little better today
andy
Well, did all the sensible things.
- decoupled first and second stage HT as suggested.
- took out volume control and selector and all wiring.
Still distorting - sounds like it's oscillating - kind of all the music is scrambled sounding with a kind of brrrrrrrrrrr noise going through it.
OK, I still have a small part of a brain. I haven't looked at the interconnects between power supply chassis and signal chassis. These are unshielded.
4 way speakon - 6.3v DC for input section (plus and neg)
8 way speakon - four of 9vDC for 6B4Gs - these have the current source in the signal chassis
5 way XLR - earth, negative supply e.g. for CCS, three times HT - one for each stage
Maybe the 8 way speakon should be shielded? Or all three cables?
Cables are 1metre long.
andy
- decoupled first and second stage HT as suggested.
- took out volume control and selector and all wiring.
Still distorting - sounds like it's oscillating - kind of all the music is scrambled sounding with a kind of brrrrrrrrrrr noise going through it.
OK, I still have a small part of a brain. I haven't looked at the interconnects between power supply chassis and signal chassis. These are unshielded.
4 way speakon - 6.3v DC for input section (plus and neg)
8 way speakon - four of 9vDC for 6B4Gs - these have the current source in the signal chassis
5 way XLR - earth, negative supply e.g. for CCS, three times HT - one for each stage
Maybe the 8 way speakon should be shielded? Or all three cables?
Cables are 1metre long.
andy
No, I don't believe you need to shield your power supply cables. I have been building amplifiers for over 30yrs now, and non of them use shielded wire anywhere - and they are (extremely) quiet..
You need to get that scope going and start looking at what is going on stage by stage.
Sounds like you have multiple issues that need to be resolved. Do you have any friends who are conversant with tube audio and could perhaps help you learn how to trouble shoot your project? (Not to mention looking for obvious mistakes.)
Things to consider:
* Location of grid stopper resistors - are they right at the grids?
* Grounding - please tell me you are not using your chassis for grounding.
* Long star ground connections can cause HF/VHF stability issues. A heavy buss grounded at a single point to the chassis is often better with everything connected to it.
* Filament wiring tightly twisted and dressed around sockets to minimize loop area.
* Filament lines either referenced to ground or to a voltage divider (cap bypassed) to elevate filament supplies above ground. (This is good for reducing noise pick up via the filament/cathode insulation)
* Tube sockets not wired backwards. (Umm, I confess to having done this, and the results are generally odd, not silent..
)
* Shorts between rca ground and chassis if intended to be isolated. (i.e. chassis ground was intended to be somewhere else)
You need to get that scope going and start looking at what is going on stage by stage.
Sounds like you have multiple issues that need to be resolved. Do you have any friends who are conversant with tube audio and could perhaps help you learn how to trouble shoot your project? (Not to mention looking for obvious mistakes.)
Things to consider:
* Location of grid stopper resistors - are they right at the grids?
* Grounding - please tell me you are not using your chassis for grounding.
* Long star ground connections can cause HF/VHF stability issues. A heavy buss grounded at a single point to the chassis is often better with everything connected to it.
* Filament wiring tightly twisted and dressed around sockets to minimize loop area.
* Filament lines either referenced to ground or to a voltage divider (cap bypassed) to elevate filament supplies above ground. (This is good for reducing noise pick up via the filament/cathode insulation)
* Tube sockets not wired backwards. (Umm, I confess to having done this, and the results are generally odd, not silent..
)* Shorts between rca ground and chassis if intended to be isolated. (i.e. chassis ground was intended to be somewhere else)
kevinkr said:
* Tube sockets not wired backwards. (Umm, I confess to having done this, and the results are generally odd, not silent..
ha ha! Yes I know this one! I was amazed that a E88CC could survive having had about 250V on its grid, but it did!
One thing I used to like to do was the old RCA RF trick of bypassing the heater supply - at each valve base - to chassis using el cheapo ceramic caps - something like 0.01µF; I first saw this dodge on the 955 spec sheets from RCA from the 1930s.
I do believe that as someone once remarked, that every piece of wire is an aerial, every joint a diode etc!
7N7
7N7 said:
My approach and attitude are quite similar, and I was surprised as you were that a tube so treated usually survives the ordeal if the voltages are not too high. 250V on the grid, yikes!
I usually decouple only one socket per string as long as they are not more 15cm - 20cm end to end, closest to the filament supply.. And this only when the filament supply is located on another chassis - which I'll admit is almost always the case.. It's amazing what an effective antenna filament wiring can be. One exception is high transconductance types which I generally decouple at each socket locally. Disk ceramics are ideal since we are concerned with RF..
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.