I've used these Russian sockets. I use them with 6CJ3 damper diodes. They don't fit well as the 6CJ3's have the smaller pins. They expect the larger diameter pins which I believe the EH 7868's use.
John
It's spring--time to build an amp
- Thread starter kward
- Start date
I've used these Russian sockets. I use them with 6CJ3 damper diodes. They don't fit well as the 6CJ3's have the smaller pins. They expect the larger diameter pins which I believe the EH 7868's use.
John
I've got to start "hoofing it" pretty good now to get this thing done by Christmas to give to my son. Build should go pretty quickly now that I've got the slow stuff done.
And the screw terminals cinch the wires down nice and tight. They work really well.
Yup, that's what I found also. The 7868's fit perfectly in the sockets and hold the pins nice and tight; not too difficult to get the tubes in or out, either. Problem is I cinched the mounting screws down too tight and broke the ceramic bases on two of them. So that set me back a few weeks while I ordered a few more.
DC conditions of output stage
The power supply, EFB board, and bias/balance circuits have all been installed and are working as expected with respect to DC conditions. With 120V input (through my variac), and setting the current through each output tube to 21 mA as specified in Dave's Fisher 400 thread, the power supply delivers 431V at the center tap of the output transformers and 6.5V (fully loaded) to the filaments. (Therefore I don't think I will need to "buck" that down since it's within spec, although the spare filament winding is available if I do need it).
I am realizing now upon closer inspection that running the 7868 tubes at "only" 21 mA per tube is idling them really really low, at about 48% of their design center max rating according to the Sylvania spec sheet. Well, I guess that was necessary to impedance match to these Fisher 400 output transformers with 10K plate to plate impedance.
The EFB board produces -63V raw negative voltage right off of the doubler, 312V for the screen supply, and -32.6V for the bias voltage, all pretty much exactly as spec'd. The bias/balance circuits work as designed, delivering adjustable bias voltage between -10V and -25V, with a 5V swing for DC balance control. For 21 mA idle current on each output tube, the bias voltage needed to deliver that is -16.5V (at 312V screen and 428V on the plates).
So all is working as expected so far.
I had a Homer Simpson "doh" moment when I installed and fired up the EFB board for the first time (w/o output tubes installed). I was getting 0V output for the bias voltage, but the screen voltage was working fine. After a bit of stewing on it, and finally removing the EFB board for closer inspection, I realized I forgot to connect the negative voltage supply coming right off of the doubler, that supplies the BJT transistors. After adding that wire, it purred along nicely. In the pic of the back side of the EFB board here, it was the blue wire that was missing (doh!). I left the insulation on that wire because it's running awfully close to the ground connection at one point. There's sort of a "ground bus" running 3/4 the length of the bottom of the board in this shot.
Here's a global shot showing the proximity of the EFB board and the wiring of the output stage.
The pots with the two orange resistors and yellow insulation straddled across the back side of the pot are for DC balance adjust, while the bias adjust pots are the pots in the upper section of the pic with just the one orange colored resistor straddled across the back side of those pots. The other two pots next to those are for the AC bal adjust for the phase inverter, which have not been wired up yet.
One thing of interest on these 7868 tubes is that both pins 2 and 7 are connected internally to the screens. I took advantage of that in the wiring, but then I realized if someone pulls out one of the tubes closer to the EFB board, the remaining tubes down stream (to the right direction in the pic) will not get any screen voltage at all, and if the amp were to be turned on in that condition, I'm not sure what would happen to the output tubes. Can't be good though to run tubes without a defined screen voltage. So I will fix that by wiring pins 2 and 7 together straight across the bottom of the tube socket. That should make it more bullet proof.
The power supply, EFB board, and bias/balance circuits have all been installed and are working as expected with respect to DC conditions. With 120V input (through my variac), and setting the current through each output tube to 21 mA as specified in Dave's Fisher 400 thread, the power supply delivers 431V at the center tap of the output transformers and 6.5V (fully loaded) to the filaments. (Therefore I don't think I will need to "buck" that down since it's within spec, although the spare filament winding is available if I do need it).
I am realizing now upon closer inspection that running the 7868 tubes at "only" 21 mA per tube is idling them really really low, at about 48% of their design center max rating according to the Sylvania spec sheet. Well, I guess that was necessary to impedance match to these Fisher 400 output transformers with 10K plate to plate impedance.
The EFB board produces -63V raw negative voltage right off of the doubler, 312V for the screen supply, and -32.6V for the bias voltage, all pretty much exactly as spec'd. The bias/balance circuits work as designed, delivering adjustable bias voltage between -10V and -25V, with a 5V swing for DC balance control. For 21 mA idle current on each output tube, the bias voltage needed to deliver that is -16.5V (at 312V screen and 428V on the plates).
So all is working as expected so far.
I had a Homer Simpson "doh" moment when I installed and fired up the EFB board for the first time (w/o output tubes installed). I was getting 0V output for the bias voltage, but the screen voltage was working fine. After a bit of stewing on it, and finally removing the EFB board for closer inspection, I realized I forgot to connect the negative voltage supply coming right off of the doubler, that supplies the BJT transistors. After adding that wire, it purred along nicely. In the pic of the back side of the EFB board here, it was the blue wire that was missing (doh!). I left the insulation on that wire because it's running awfully close to the ground connection at one point. There's sort of a "ground bus" running 3/4 the length of the bottom of the board in this shot.
Here's a global shot showing the proximity of the EFB board and the wiring of the output stage.
The pots with the two orange resistors and yellow insulation straddled across the back side of the pot are for DC balance adjust, while the bias adjust pots are the pots in the upper section of the pic with just the one orange colored resistor straddled across the back side of those pots. The other two pots next to those are for the AC bal adjust for the phase inverter, which have not been wired up yet.
One thing of interest on these 7868 tubes is that both pins 2 and 7 are connected internally to the screens. I took advantage of that in the wiring, but then I realized if someone pulls out one of the tubes closer to the EFB board, the remaining tubes down stream (to the right direction in the pic) will not get any screen voltage at all, and if the amp were to be turned on in that condition, I'm not sure what would happen to the output tubes. Can't be good though to run tubes without a defined screen voltage. So I will fix that by wiring pins 2 and 7 together straight across the bottom of the tube socket. That should make it more bullet proof.
Last edited:
Usually losing the screen connection will cause the tubes to stop conducting.
Bigger question is how much current can that connection across the pins handle. If its connecting through the screen grid itself, its probably not such a great idea. If the jumper is right in the base and current becomes not a concern, not such a big deal. For what its worth, the pics of various 7868 Sherwoods I looked at when I was working on mine had pins 2 and 7 strapped together on every tube.
Bigger question is how much current can that connection across the pins handle. If its connecting through the screen grid itself, its probably not such a great idea. If the jumper is right in the base and current becomes not a concern, not such a big deal. For what its worth, the pics of various 7868 Sherwoods I looked at when I was working on mine had pins 2 and 7 strapped together on every tube.
Good to know gadget, thanks. Also pins 2 and 6 are internally wired to grid #1, so I guess I should strap those together too. Hey, wait...I have a Fisher X-100-C derelict chassis that still has all the original wiring. Let me go look at what Avery did...
Yup, the Fisher X-100-C has pins 1 and 7 strapped together across the bottom of the tube socket. It also uses pins 2 and 6 for connections, but those pins are not strapped together, even though they are connected internally. Looks like pin 1 receives the signal from the phase inverter while pin 6 receives the negative bias voltage and grid bias resistor.
Yup, the Fisher X-100-C has pins 1 and 7 strapped together across the bottom of the tube socket. It also uses pins 2 and 6 for connections, but those pins are not strapped together, even though they are connected internally. Looks like pin 1 receives the signal from the phase inverter while pin 6 receives the negative bias voltage and grid bias resistor.
Last edited:
The patient is still strewn out on the operating table but I've got the left brain hooked to the left body now. I couple of things I am noticing right off the bat: The pentode of the 6U8 is extremely sensitive to noise from AC filaments. Switching over to DC filaments (with a crude pi filter for now) cured a lot of that. Open loop, the pentode as wired has gain of 80. With 20 dB feedback applied (a la Fisher 400), I get input sensitivity of about 1V RMS. At 24 dB feedback, the amp starts to oscillate, so crude tests say I am well within the amp's stability margin even at 20 dB feedback applied. I will likely run with something like 16 to 18 dB feedback in the final implementation.
The other thing I noticed is a little bit of low frequency instability down in the ~2 Hz region. I'm not sure how to cure that presently, but I'm going to try a smaller screen grid decoupling cap for starters (currently using a 4 uF cap). In fact, I need to change out that screen decoupling cap anyway because I presently have only a 100V cap in there, and on startup, voltage at the screen swings up to about 180V before settling down to 55V at quiescent conditions. (surprised the cap didn't arc or short).
Power output just before clipping (single channel driven) is a clean 25 watts/channel
. This is amazing to me considering the output tubes are idling at 21 mA quiescent current (about half their design max rating). Chalk that up to the EFB doing its thing I guess. Clipping also occurs differently as compared to a non EFB controlled output stage. For example, going from full power into clipping occurs somewhat suddenly--sine waves at max power are clean and undistorted, and then applying just a bit more input juice and bam! flat tops and bottoms start appearing. Almost immediately thereafter notch distortion appears (with a 1Khz test signal into an 8 ohm dummy load on the 8 ohm tap). I've not seen this kind of behavior before on a non EFB fixed biased output stage. On non-EFB fixed biased output stages, max power conditions start to show themselves by sine wave skewing (like someone is pushing on the wave from the side causing it to tilt slightly), and then a bit of rounding of the tops of the sine waves, and finally some flattening. Not with EFB--nice sine waves up to max power, then boom! clipping suddenly.
Anyway, just feeling my way through a new amp trying to get a feel for its performance characteristics.
The other thing I noticed is a little bit of low frequency instability down in the ~2 Hz region. I'm not sure how to cure that presently, but I'm going to try a smaller screen grid decoupling cap for starters (currently using a 4 uF cap). In fact, I need to change out that screen decoupling cap anyway because I presently have only a 100V cap in there, and on startup, voltage at the screen swings up to about 180V before settling down to 55V at quiescent conditions. (surprised the cap didn't arc or short).
Power output just before clipping (single channel driven) is a clean 25 watts/channel
. This is amazing to me considering the output tubes are idling at 21 mA quiescent current (about half their design max rating). Chalk that up to the EFB doing its thing I guess. Clipping also occurs differently as compared to a non EFB controlled output stage. For example, going from full power into clipping occurs somewhat suddenly--sine waves at max power are clean and undistorted, and then applying just a bit more input juice and bam! flat tops and bottoms start appearing. Almost immediately thereafter notch distortion appears (with a 1Khz test signal into an 8 ohm dummy load on the 8 ohm tap). I've not seen this kind of behavior before on a non EFB fixed biased output stage. On non-EFB fixed biased output stages, max power conditions start to show themselves by sine wave skewing (like someone is pushing on the wave from the side causing it to tilt slightly), and then a bit of rounding of the tops of the sine waves, and finally some flattening. Not with EFB--nice sine waves up to max power, then boom! clipping suddenly. Anyway, just feeling my way through a new amp trying to get a feel for its performance characteristics.
jazbo8
Well-Known Member
This is interesting, I don't think Dave have mentioned or described such phenomenon with the EFB, since the distortion was measured typically -1dB and at full power. Perhaps Dave will add his comments when he sees this...
Yeah, I thought I heard the cracking of glass...
Output tube got so hot it cracked the glass. The culprit: A failed coupling capacitor had caused a +34V bias condition on the grid of that tube. No other damage, but luckily I was watching it closely at that instant. These are 500V caps and the max surge is 470 or so...should have been okay. Cap was a Russian military K42Y-2.
Output tube got so hot it cracked the glass. The culprit: A failed coupling capacitor had caused a +34V bias condition on the grid of that tube. No other damage, but luckily I was watching it closely at that instant. These are 500V caps and the max surge is 470 or so...should have been okay. Cap was a Russian military K42Y-2.
Last edited:
It's too bad because I think the K42Y-2 is a good sounding cap, but after this experience, I don't trust them at their max rated voltage. Seems like a time bomb ticking. So even though I have a few spares, I'm going to pull them and save them for lower voltage applications. Not sure what I will replace them with yet.
Before things went south, I was able to determine that with 17.5 dB feedback applied but with no HF contouring added yet, nor the feedback cap added yet, the Fisher 400 output transformers have one resonance frequency at 86 KHz. Didn't get a chance to measure above 100 KHz yet. That is excellent news that it is up there so high as that should make HF tuning quite straight forward. At about 3 watts output, low end response was flat clear down to about 8 Hz or so. Ringing on a 10 KHz square wave was pronounced but somewhat damped even without HF compensaton, and not as bad as I've seen it on other iron.
So, thus far these output transformers measure quite well. Can't wait to hear how this lug is going to sound.
Before things went south, I was able to determine that with 17.5 dB feedback applied but with no HF contouring added yet, nor the feedback cap added yet, the Fisher 400 output transformers have one resonance frequency at 86 KHz. Didn't get a chance to measure above 100 KHz yet. That is excellent news that it is up there so high as that should make HF tuning quite straight forward. At about 3 watts output, low end response was flat clear down to about 8 Hz or so. Ringing on a 10 KHz square wave was pronounced but somewhat damped even without HF compensaton, and not as bad as I've seen it on other iron.
So, thus far these output transformers measure quite well. Can't wait to hear how this lug is going to sound.
Last edited:
random thought, but if you've got an adjustable HV supply of some fashion, you could test those things for voltage leakage if you string up a simple RC setup and measure voltage across the resistor. Thats how I do low voltage caps. My 1940s cap tester won't do under 100v, so I use my HP 0-160v bench supply and a 10k or so resistor to see what it has to say.
I took a detour on this other thread to work out some hum and oscillation issues with the frontend 6U8 tubes. Got that worked out now (mostly it was wiring errors or omissions and lack of understanding on proper test methods for small signal pentodes). So back to work here...
I've come about as far as I can now without having the output coupling caps and a new matched pair of tubes (they've both been ordered and we're waiting on the mailman now).
Things done since last post:
In any case, the amp is now superbly quiet in both channels. Also I am finding these RCA 6U8 tubes are excellent! Not microphonic and very low noise. Good tubes. Of course the EH 7868 tubes sound great so far, too.
I've come about as far as I can now without having the output coupling caps and a new matched pair of tubes (they've both been ordered and we're waiting on the mailman now).
Things done since last post:
- Built and installed a low ripple DC power supply with a low dropout three terminal regulator to power the 6U8 filaments. I don't need regulation for the filaments, but this was the fastest and most performant way to get exceptionally low ripple in a small footprint. This is shown in the pic below as the smaller of the two breadboards on the left, next to the EFB board. The board outputs 12.0V DC, so the 6U8 filaments are connected in series. I am currently using a Radio Shack 120V to 12.6V transformer that supplies 0.45A current on the secondary winding. This is not enough current capacity to deliver enough secondary voltage to keep the regulator from dropout under all likely wall power conditions (meaning regulator drops out at 118V wall voltage). So I have ordered a Hammond 115V to 12.6V 1A unit that I will use. This should lower the wall voltage that can be tolerated before regulator dropout to something like 110V, which likely will never happen except at the onset of a brownout or something.
- Wired up the RCA jacks to the selector switch and volume pot and then hooked the volume pot to the 6U8 grids.
- Installed the 120V wiring from the power socket, and wired up the fuse, power switch, and pilot light. I'm using heavily shielded 120V wiring inside the amp. Since I added the 12V DC board, I thought it would be convenient to power the pilot light from that, so I added a terminal block and dropping resistor for the LED lamp to that board.
- Checking for oscillation at quiescent conditions with global feedback active (there is none)
- Validating hum levels (there is very little that I can measure)
- Testing for LF stability characteristics (amp is very stable at low frequencies, although it becomes less stable at LF with more feedback added, and at 24 dB feedback starts to oscillate on it's own at about 2 cycles per second).
- Evaluating max power ability (still right around 25 watts/channel, single channel driven)
- Determining how much feedback is actually applied with the same 2.7K feedback resistor attached to the 4 ohm tap that I have been using. With the reduction in screen grid bypass cap, the gain of the first stage went down by 1.25% (from 80x to 79x) and the applied feedback is now 15 dB (was 17.5 dB with the larger screen bypass cap). Thus the amp's stability margin is 9 dB (this probably doesn't really mean anything except for me to know that I am a ways away from making it unstable because of the amount of applied feedback).
- Input sensitivity of the amp with the 2.7K feedback resistor installed is now 0.77V (was 0.7V with the larger screen bypass cap).
In any case, the amp is now superbly quiet in both channels. Also I am finding these RCA 6U8 tubes are excellent! Not microphonic and very low noise. Good tubes. Of course the EH 7868 tubes sound great so far, too.
