|'); } // End -->|
Been bread boarding a 750 vdc version of GroverG's 845 amp the last few days, but I have run into something that has me stumped. After working out a few original bugs in the power supply I got music out the end of the wires and it sounded pretty darn good. I was leaning back enjoying the sound when all of a sudden it let out this horrible shreik and went quiet. Funny thing is I tried again a few minutes later after finding nothing visibly wrong and exactly the same thing happened. It played for several minutes, made a bad shreik and went quiet. I am guessing that the higher voltage is stressing something out, but I can't figure out what. What type of part is most likely to behave that way?
I think I may have found the problem boys. I no sooner had finsished my last post to Grover than I started going over the bread board with the schematic. I am almost too embarassed to tell you what I found, but I figure maybe someone else will be able to benefit from my error.
The schematic calls for a 0.47/600 vdc coupling cap between the el34 and the 845. Well friends and neighbors, yours truly grabbed a 47mfd solen fast cap and plopped it in there. Duh!!! I have now put in a cap of the proper value, and as soon as my wife wakes up(it's only 7am here) I will crank the amp up and see what happens. My guess is it will work just fine.
For my edification would some please tell me exactly what happens in the circuit when you screw up like that? I know it screeches, but why?
Thanks for all the helps folks!
Excessive LF phase shift, I imagine. But no matter--glad you found the problem. Don't feel bad, I just breadboarded my umpteenth power supply and instead of 470K resistors to bridge the series caps, I grabbed some 470 ohm 1/2 watt. Made some deeelightful fireworks when I turned it on.
If you have two stages of RC coupling, you run a high risk of an
unstable amp that can dive into an oscillation mode.
Further, you need decoupling between each amplifier stage otherwise
the the big current drawing stage upstream will modulate the B+
on the driver stage. To do this, you slap an RC network into the B+
rail. The R's drop the voltage to a needed voltage for that stage and
the C's provide a buffer (energy storage).
Your schematic should include that of the bias supply. Using a variac and a 530V transformer is an expensive way to do it.
A cheap bias supply can be made from a 120V : 120V isolation transformer. Use a bridge rectifier and a small 250V cap
followed by a simple RC coupling network. You can then use either
an 0A2 or 0B2 gas regulator to create a regulated -150V or -105V
supply. That regulated reference can then be dialed down with a
20K pot. Put a 100uF 100V cap at the pot wiper to provide a nice low
AC impedance. Cheap, compact and very reliable.
Play safe with that 750V. Dont work on it if you are tired and
frustrated. It's always best to be fresh, and alert. ;)
Wow. This schematic looks similar to alot of the China-clone
845 amps. ;) It's certainly compact and attainable.
1. You always want to make sure 6.3V filament power is there.
Without it, if the tubes arent conducting any current then there
is no voltage drop across the plate load resistors which means
the coupling caps are seeing full killer B+ which is not a
2. Im guessing operating points are as follows:
EL34 @ 290V and 30mA, 6N7 at 6ma per triode section at 200V.
3. Make sure the secondary of the 125ESE is pathed to ground. Why?
Were a dielectric breakdown to occur, the speaker wires and parts
of the speakers could elevate to 1kV and create a lethal shock hazard. Dont leave the output transformer floating like the schematic implies. I highly doubt the 125ESE is wound with the
proper insulation level to take anything north of 500V. Be
4. Be VERY careful with your clipleads. I doubt they are rated
for 1kV. Clipleads can be prone to cracked insulation from regular
use. Fine cracks in the insulation is OK at a few hundred volts. At
1 kV, its a supreme liability.
5. The 10K pot followed by the 5K fixed resistor in the bias
circuit is very smart. The 5K resistor prevents you from
accidentally biasing down to 0V. To be extra safe, you could
bracket with potentiometer between two resistors and choose values
such that the range of bias voltage is narrowed even further.
6. Try a 6W6GT in place of the EL34. :) They are $3.00. The ones
I bought from AES (several brands) sound spectacular. Another tube
I really like is the Valve Art KT66. Guitar Center carries them
under the "Groove Tube" brand. $35/pair where I got them.
7. I used a similar circuit with similar tubes but went with
a direct coupled approach. You could try that too later on after
you work the kinks out and listen to the amp for awhile. In the
direct coupled approach, you'd choose a new bias point for the
6N7s at around 150V on the plate. The plate of the 6N7 would
then be tied directly too the EL34s grid. The EL34 cathode
would now need to be elevated to a voltage of 6N7Plate + 30V to
set its bias. Assuming 30mA bias for the EL34, then, Rk =
150+30 / 30mA = 6Kohms 5.4W. Another big resistor. The
Direct Coupled approach removes a capacitor and a 90 degree
phase shift from the picture. The risk of a direct coupled approach
is that the input tube must ALWAYS have filament power and must
never be removed from its socket. Why? Because the EL34 is depending
on the 6N7's operation to set its fixed bias point. Without the 6N7,
the EL34 would lose fixed bias and would conduct like mad likely
harming the EL34 but not the amp.
In my case, I chose to use D.C. to remove a cap coupling stage.
You have ALOT of options for colouring in the tone of the amp.
Vary the input tube and the driver tube. I played around with this.
Trying the space of [6C5, 6P5GT, 6SN7, #76] x [6V6, 6W6, EL34, KT66, KT88, 5881].
I found the 6P5GT and #76 input tube to sound the best. I also
found that beam power tubes (KT88, KT66, 6W6GT) sounded very good.
I was not impressed with the Sovtek EL34, the E-H 6V6GT or the 5881.
The Sovtek KT88 sounded better than the EL34 or 6V6GT. The
Ei Fat Bottle EL34s sounded very good, better than the KT88. The
Valve Art KT66 and various NOS 6W6GT sound exceptionally good
I think. Something about beam tubes over pentodes. :) Grover
seems to like the Ei KT90's if I remember.
I will ask Grover if I can post the schematic. The variac and tranny are just for the bread board version. If I decide to build it I will do something more along the lines of what you suggested. There is coupling and decoupling between each stage, and the caps are rated high enough I believe. Tis a puzzle to be sure. I hear you about the 750 volts. I try to be as careful as I can!
Below are the schematics for the circuit and PS. Tell us how your version differs from the schematic, and we'll see if we can spot anything. I've built this about 20 times and never had the problem you describe. It's possible that between the PS and the circuit there's some kind of phase shift that's causing the amp to destabilize.
Dang, I thought for sure I had it! No such luck. It went a little longer this time, but the same thing happened. When it goes on the old speaker it makes a sound like air going out of a ballon when you let go of the end after blowing it up. More of a squeak than a screech. If you guys could please take a look at the post where I describe the differences between my amp and Grover's schematic I would appreciate it. I simply must find out what is causing this.
Well, okay, here we go :-). First, reapply the bypass cap to the 6N7 cathode since this seemed to stabilize it more. Next, remove the preamp from the circuit, short the input of the amp with a shorting plug, turn it on and see if it oscillates. If not, then there's either an excess of gain or an RF or ground loop thing with the preamp that's causing a problem. If you still get oscillation, try fixing the bias supply as Jim suggested--use a 6.3 volt tranny, hook the secondary to the 6.3 volt filaments, and use the 120V primary for the bias supply. Take one 120V primary and attach a single reversed diode, which goes to the filter cap, and ground the other primary. This makes a half-wave supply and it may add some noise but it should be stable.
Let me know what happens when you short the input.
Also, describe for me how you're powering the 845 filaments. 10VAC with a center tap? Is the supply grounded properly? How are you measuring the current? Your bias should be around -100 volts. If you're bias voltage is excessive (-170) the 845 may be going into cutoff.
I put the bypass cap back in and I took out the preamp and put in the shorting plugs. It made the noise again. It seems to take about 3-5 minutes for it to get to that point. I will check and see if I have a 6.3 volt tranny in one of my junk boxes. I measured the bias as -130 vdc by putting the positive probe of my DVM on the grid of the 845 and the negative probe on one of the filaments of the 845. Is that the right way to measure it? Is that too high a bias voltage? My B+ is running about 785vdc right now because I am only using one 845. The filaments on the 845 are 10vac with a center tap to ground through a 1k/25w resistor. I am not sure what you mean by is the supply grounded properly. I am using a variac to bring everything up. It is plugged into a multiple outlet which is plugged into a grounded wall outlet. I am using a ground buss on top of my bread board chassis that I hook the ground from the variac to. The other two wires go to the primary terminals on the power tranny. All ground connections from the bread board components go to the ground buss on the bread board chassis. I have used this same set up for bread boarding many other amps and have had no trouble with it, there is always a first time for everything.
Aaaaaahhhh... You're using a 1000 ohm/25W resistor between the filament CT and ground, along with -130 volts on the grid? If so, in effect you're "double-biasing" the 845. You're running the tube way down into cut-off, like Class C region :-). Remove the 1K resistor and ground the 10VAC center tap. Set the grid voltage to about -110. That should put you at between 60-80mA. Good enough for testing purposes. You really want 10 ohms/2 watts from CT to ground, and measure .8 volts to ground for 80mA.
Well, it went about twice as long this time, so I guess we are making progress :) . I took the resistor out and took the CT to ground and I adjusted the bias so that it read -110 between the grid and the filament of the 845. I ran it for about 7 minutes with the preamp out and everything seemed ok so I decided to put the preamp back in. It ran for about 5 minutes with the preamp in before making the noise. Guess I had better go back and let it run 20 minutes or more with the preamp out and see if it goes bonkers then. The thing I don't understand is why it doesn't start making the noise right away. Why does it run awhile first?
I'll report back after running it without the preamp for awhile longer.
Took the 6n7 bypass cap out. Gain was definetly down, but it did the same thing after about 5 or 6 minutes. A bit different this time though; it just made a sort of breaking contact type of noise and quite. I turned it off for about two minutes, cranked it back up and it started playing again. This all a plot to drive me crazy!!!
I think I will go watch television for a while.
I just ran the amp for 20 minutes with the preamp out and the shorting plugs in. It never made a peep. I am going to go back and try taking out the bypass cap on the 6n7 and put the preamp back in and see what happens. Oh, one other thing; it got real quiet in the house for a short while and I noticed that the potted power tranny has a pretty good hum to it when full mains is applied. Is that anything to worry about? I have never heard a potted power tranny hum before.
If the oscillation is subtle, it can take a while for it to ramp itself up and set the amp to singing. Have you tried a grid-stopper on the input pins of the 6N7? If not, solder a 1K carbon resistor right up against the adjacent grid pins and attach your input lead to the other end.
What size output cap are you using in the preamp?
I believe the output cap is 1.0 mfd/400vdc. The specs on the power tranny are 1206vct/0.35A.
I was just reading through your suggestions and realized that I use relay sockets on my breab board. Is there any way I can hook up the grid stoppers on those and have it still work?
I really appreaciate all the help Grover!
Good morning Grover. I have today and Friday off due to the holidays, so I am taking vacation leave for the rest of the week to try and get some remodeling done around the house.
I tried the grid stopper just a short while ago. I put a 1k resistor to pin 4 of the 6n7 and attached the input wire to the other end. I kept the end on the relay short by doubling over the lead and sliping it under the clamp. Pins 4 & 5 are tied together with a short lead. They don't need separate resistors do they? I left the bypass cap for the cathode resistor of the 6n7 out, but the preamp was in.
Well, it went about 12 minutes this time before it went pop and stopped playing. Does this mean we are making progress? What do you think I should try next?
What was your take on the output cap of the preamp and the specs of the PS tranny?
The power tranny and the output cap would appear to be fine. The power tranny may be buzzing because you're powering it off the variac. We'll look into that more later.
Here's what I would do next:
Restore the bypass cap to the 6N7. Make sure the 6N7 is wired properly--you have 4 & 5 tied for the grids, and you should have 3 & 6 tied for the plates, and 8 for the cathode resistor.
Disconnect your preamp from the amp, turn it on and make sure you don't have any measurable DC from the output to ground. A leaky output cap could cause the problem you're having.
Is there any way you can place a volume pot at the input of the amp (anything from 25-200K would suffice) and test the amp with your CD player or source directly inputted? That would tell us a lot.
Just so happens my bread board is semi permanent and has volume pots built in for each channel. I will first check the preamp for a leaky cap and then I will put the bypass cap back in and run the cd player directly into the amp. I just checked the wiring on the 6n7 and it is correct. I will get back to you as soon as I do these things.
Most interesting. It only lasted 5 minutes this time with the CD player going straight into the amp. I did have the volume pot up a bit higher than usual, but not that much. So it seems we only have a problem when we have a signal, yes? Of course I may not have heard it when it went off with no preamp hooked in and the shorting plugs in place because there was no music to stop playing. This is quite a puzzle my friend. Next move?
Vinnie, why don't you call me at 301-699-8227 and I'll call you back on my dime--don't worry it doesn't cost me anything.
To any who have been following this thread to see how it ends, Grover was kind enough to march me through my whole layout step by step on the phone until we found the problem. It was the bloomin' EL 34! Grover says that most likely it shorted out when the B+ voltage hit the plate before the filaments had a chance to warm things up. It would run for awhile then shut down. Once it cooled down you could start it up again and run it for awhile longer. It about drove us nuts until Grover had me put the DVM on the plate of the EL34 and watch it while it was running. When it quit the voltage on the plate spiked up, indicating that it wasn't drawing current anymore. At his suggestion I have now installed a separate 6.3 supply for the filaments that will be turned on first.
Bottom line is this is one great sounding amp! Once I got it running I just kept trying one CD after another and they all sounded great. I can't wait to hear it in stereo. Now will come the longer process of deciding just how to turn this into a real amp instead of a bread board. Lots of work and fun ahead. Many thanks to Grover for both the design and helping me get it operational. You're a good man Grover!
Thanks for the suggestions Grover. I think it's best if I take a break for tonight and start again tomorrow. I'm tuckered. I will try the these tomorrow and get back to you with the results.
Is that the proper way to measure it? I tried that just now and it measures the same as from grid to filament +/- a volt or two.
Yes, now that you have the filaments grounded! Before you had -130 on the grid, plus probably +70 volts on the cathode, effectively given you a grid bias voltage of -200 or more (in relation to the cathode/filaments). Always measure your negative grid voltage from grid to ground, not the cathode.
Thanks for posting the schematic Grover. I have taken a good look at it and the differences I see are as follows:
1. The decoupling caps from the plates of the 6n7 and el34 are 3x47/350 and 3 x 100/450 instead of both being 3x330/450.
2. The plate resistors of the 6n7 are 2 24k/2w in series. For the el34 I put 7.5k, 4.3k and 3.01k resistors in series. All are at least 20W.
3. The rating on the 125ese is only 80mA I believe.
4.At the moment I do not have the 100ohm/2w resistor on the filament of the 845. I wanted to try it without it first.
5. My B+ PS is tube rectified. It has a 5R4GYA and is CLC with 15mfd, 15H and 40 mfd. I have been using this same set with a 5AR4 for a long time with no trouble at all.
6. The filament voltages are all AC and come off the taps on the PS transformer. The 6.3 goes first to the 6n7 where one leg is grounded and then I take it to the el34 from there.
7. I should probably mention that the whole bread board is put together with clip leads. I have had no problems with that so far, but it did give me some pause do to the higher B+.
8. The bias supply is a small tranny run off a small variac. I have N4007 diodes with the bar end attached to the tranny terminal and a 100/450 cap with the + end grounded. I adjusted the variac to get - 170 vdc.
While you guys are looking at this I am going to go back over the bread board with the schematic one more time and see if I see anything wrong there. Any help you can give me will be greatly appreciated.
I think the Diodes are correct. I only used them on the bias supply. B+ is tubed. I tied the ends with the bar on them to the tranny terminals. I am getting a - 170 vdc reading with the + probe of my DVM attached to the output end of the diodes. I think I have the capacitors right too. For the bias supply the + end is grounded and for the the others the - end. The coupling cap is PIO. Orientation doesn't matter for that oone does it?
For the bias supply I have a 535 vdc/.012 tranny tied into a small variac to control the bias voltage. Do you think the amp rating on the tranny might be too low?
I just flashed on something! I am using tube rectifcation for the B+. When you told me you were using 2x47/500 and 2x100/500 did you mean for the filter caps? I thought you were talking about the other (bypass??) caps off the ends of the el34 and the 6n7. I only have 2 x100/450 at the el34 and 2x47/350 at the 6n7. Could that be the problem?
The coupling cap is a .1/630 PIO. I think it is ok, but maybe I will try another cap in its place.
One other thing I noticed. I used an old acoustic suspension speaker to do the first test run to avoid doing any damage to my khorns if anything went wrong. Everything worked fine as long as I was hooked up to that speaker. When I was finally was feeling that everything was ok I switched over to one of the khorns. It worked fine for several minutes and then that awful screeech happened. This happened twice. I thought I had caused the problem by having the DVM hooked up incorrectly so I started over. Same thing, ok as long as I was hooked up to the old speaker and the screech after a few minutes of being hooked up to the khorn. Is there any chance the loading of the speaker could have any part in this? The old acoustic suspesion speaker is very inefficent.
Sorry about the long post, but I really want to get the bugs out as I really liked what I heard when it was working!
Okay. First, we miscommunicated about the caps. I was referring to the PS caps. Since you have a PS in place I'm sure it's fine, whatever you used will work--PIO is okay, whatever you like to get the 750VDC. For the bypasses, put 100uF across the EL34 cathode resistor and anything between 220-470uf across the 6N7 cathode resistor for now. Just single caps, don't double them up. Again, be sure they're oriented properly (you know this). The bias supply doesn't need current, whatever you rigged up should work okay as long as you're getting the negative voltage you need.
Now for the screeching--if it only happens with K-horn, I suspect you're getting some mechanical system feedback. This may be the result of excessive gain overall. Are you using a high-gain preamp, like a 6SN7? The circuit has pretty high gain intended for dynamic (as opposed to horn) speakers and a low-gain front end. If this is the case, try 1) removing the bypass cap from the 6N7 cathode OR 2) if you have one, substitute a 6J5 for the 6N7. It'll drop right in. OR 3) substitute a 6L6 for the EL34. Unbypassing the 6N7 will reduce the gain but it may not sound as good. If it turns out to be an issue of too much gain, you're better off to substitute a 6J5 or 1/2 6SN7 for the 6N7, though you may lose a bit of the magic that the 6N7 seems to provide. Replacing the EL34 with a 6L6 will give you a tad less overall gain.
I'm sorry Grover, I still haven't learned all the nomenclature. I did not mean the bypass caps on the cathodes, I meant the caps on the plate ends. What do you call those? I have two 47/350 in series on the 6n7 plate and two 100/450 in series on the el34 plate. Since you were originally talking about the PS caps, does that mean I should actually be using the 3x330/450 you show in the schematic at each spot? Could that be the source of my troubles?
Got any idea if the 125 ese opt is ok for this application?
No, you're right, I've heard "bypass" used for the decoupling caps too. What you're using is fine. It's unlikely to be a power supply or OPT problem--the 125ese shouldn't have any problem here.
I think it's an acoustic feedback thing between the amp and the K-horns. As I said, it's a pretty high-gain circuit, and when you drop the voltage to 750 you are also decreasing the drive requirements for the 845. What preamp are you using? Try a 6J5 in place of the 6N7 or rewire the breadboard to acoomodate a 6SN7.
My preamp is a homebuilt version of Bruce Berman's 6sn7/76 preamp from SP issue 13. What if I were to unparallel the 6n7? Would that drop the gain?
Yeah, you've got a lot of gain there for the K-horns. No, unparalleling the 6N7 won't change the gain. Remove the cathode bypass cap on the 6N7 for now and see if that settles it down. You may notice a little thickening of the sound, but let's solve the problem and then we can find a sonicly acceptable tweak.
Well Grover, I tried removing the bypass cap from the 6n7, and I hooked it back up to the old acoustic suspension speaker. I must not have let it run long enough before, because this time I got the noise on the old speaker after about 7 or 8 minutes. It can't be more efficient than about 85 db I think. So where does that leave us? Must be something that can't handle the voltage or current huh? I'm puzzled that it it will work again for a few minutes when you start it back up. I suppose one of these time something will blow and it won't. See anything in Jim Doyle's post that looks likely to you?
I'm also wondering about that bias supply running from a Variac. Something's not right about that, but I can't put my finger on it. That may be the point where some kind of instability is being introduced, though I couldn't say why.
Oh well! :-( There must be some kind of phase shift going on between the circuit and your version of the PS. See my post above.
First, double-check the orientation of all diodes and capacitors.
Second, are you using a separate transformer for the bias supply? If you are taking the bias off the main B+ tranny, you can create some sort of power-supply feedback that will cause the amp to shut down the music signal.
Since there's no feedback in the amp, and it's single-ended, it's unlikely to be a phase issue. But there could be some kind of low-frequency overload. Did you improvise any off the bypass or coupling cap values? With fixed bias, if you overload the power stage's grid, and the coupling cap is too large, you can run into a siutation where the grid goes into cut off and doesn't recover until you shut down the amp for a few minutes. (I'm sure someone else could describe this better than I could.)
Need to see your schematic. Nothing blew? Not even a fuse?
The shriek indicates high-frequency components accompanying some
transient event. I would suspect some kind of dielectric breakdown.
Some questions to answer:
1. Is the PIV rating of your rectifier(s) up to snuff with the
2. Are you exceeding voltage ratings of any dropping resistors?
Some resistors can only take 200V max across the terminals.
3. Do you have dielectric breakdown in your output transformer,
filter capacitors, coupling capacitors? The coupling cap to an
845 is vulnerable and dangerous point of failure.
4. If using a choke input PSU, have you equipped it with a snubbing
network to decouple the HV spikes that occur during rectifier switching?
At 750V, any fixed bias failure on an 845 might mean a dead
output transformer. You should protect it with a fuse. Rather than
put a fuse inline with the 750V (a bad idea), put it in the cathode
circuit. At 250mA fuse ought to save your output transformers.
Congrats on the 750V operating point. :) What PSU tranny are you
Thnaks for the reply Jim. You have brought up some really good points. I have been working below 500 vdc for so long I hadn't really thought about these things. I will try to answer your questions in order.
1. I think The PIV rating is ok. I am using a 5r4gya and I checked it out on Duncans PSU program and it said it was ok.
2. I will have to check this one out. I really don't know. It's possible.
3. I was using a hammond 125ese at 10k. How do I check to see what it will handle? The coupling cap is rated at 630 vdc and I think it is ok. The filter caps are oil caps. One I know is rated high enough, the other I am not sure about, so I had better see if I can replace it with one I know is.
4. MY PSU is CLC does that mean I don't ahve to worry about the spikes?
5. The schematic calls for a seperate bias power supply. Power tranny, diodes and filter. I can't see that it is fused anywhere. I will have to ask Grover about that one.
Last but not least, for the few minutes I was able to have it running it sounded great! Grover said he thought it would work ok at 750 B+ even though it was originally design for 1000 vdc. I think it is one of the best SE amps I have ever heard. Clean, clear, detailed. Vocals sound great and instruments are true. I have only heard one channel so far. I can't wait to hear it in stereo. It definetly sounds like it will bump out the 300b I have had bread boarded for the last couple of weeks. If anyone has been wanting to crank it up a notch with the 845, you might want to ask Grover to post the schematic again. Or I can if Grover is alright with that.
Any help the inmates can give me in working out these little bugs would be greatly appreciated as I think I just might have to build it.
This post is made possible by the generous support of people like you and our sponsors: