Sparks fly in a 5f6-a clone

[Jester]

Hi Guys,

As you may have read in my last post - I've just finished my 5F6-a clone, and I powered it up a couple of times on the bench, and measured voltages and biased the 6L6s - since the plate voltages were less than 500v I thought I was home and dry. I then move it to the back room to play it and have had a couple of power ups that concern me. So here's what I've seen.

When powering up to Standby - my Groove tube SAG-AX7_MPI in position 3, the phase splitter, glows strongly for a fraction of a second, then it's difficult to see any of the heater glowing.

More concerning is when I flick the standby switch to on, I've seen a flash across the base of the 6l6 tubes (inside the tubes around the bottom of the plate). This co-insides with a crackel in the speakers (which I hear is normal if you move the power supply caps to the rectifier side of the standby switch)

The amp then plays well and sounds fine.

Should I be concerned by this? I must confess I don't think the flash in the 6L6 is healthy. Any suggestions as to what is happening?

FYI I have a followed the basic 5F6-a cct, with the following changes:

1. Moved the main filter caps to the rectifier side of a standby switch, which uses two 220uF 300 volt caps in a totem pole configuration with 220 ohm resistors across each one

1. I have change the fixed bias cct, to an adjustable bias, by replacing the 56K ohm load resistor with a 50K pot, and a 27k resistor on the tail of the sweeper. Due to the fact that the MM pt has a bias voltage of only 50v, I have dropped the value of the dropping resistor down to 1.5k ohms, so I can achieve a -48vdc bias (ie 35mA in each power tube)

1. I have increased the capacitor value of the cathode follower supply to 16uF

As I say, I did not see these flashes when the unit was on the bench, but when it was I was able to measure 481 volts on the plates of the 6L6s, the plates on Tube 1 (12AY7) were at 182 / 187 volts
Tube 2 (12AX7) were at 350/180 volts
Tube 3 (SAG-AX7_MPI) were at 250 volts

I'm using MM power supply with a 380 volt output to the rectifier, and have a MM choke (2.84 heneries and 58ohms).

Any suggestions or comments would be welcome.

Thanks
Jester

[tubeswell]

Hi Guys,

1. I have change the fixed bias cct, to an adjustable bias, by replacing the 56K ohm load resistor with a 50K pot, and a 27k resistor on the tail of the sweeper. Due to the fact that the MM pt has a bias voltage of only 50v, I have dropped the value of the dropping resistor down to 1.5k ohms, so I can achieve a -48vdc bias (ie 35mA in each power tube

Jester

Don't quite understand what you mean by a 1k5 dropping resistor. 50V is usual for a bias tap. Have you got a reverse biased diode and decoupling caps in that bias supply somewhere? Maybe you could post a schematic?

[Jester]

Don't quite understand what you mean by a 1k5 dropping resistor. 50V is usual for a bias tap. Have you got a reverse biased diode and decoupling caps in that bias supply somewhere? Maybe you could post a schematic?

Sorry have not worked out how to sketch schematics on my PC yet, so will have to use a 1000 words instead :
In a voltage divider network, I believe (though I might be wrong) you normally term one resistor as the dropping resistor (the first one in this cct after the diode), and the other resistor is the load resistor (across which the load voltage is developed). If you look at the standard 5F6-A schematic. The 15K resistor is what I have termed the dropping resistor, and the 56K resistor is what I've termed the load resistor. I believe the original transformer would have had been able to supply a bias voltage of about -60Vdc before the dropping resistor (ie: back calculating from the original resistor values with a -48vdc bias voltage. ie -48* (15k+56K)/56K = -60.85 Vdc). In the original cct about 12-13 volts are dropped over the dropping resistor, and -48volts is available at the load resistor. Because my source bias voltage is only -50V, I can only afford to drop a volt or two across the dropping resistor to get down to -48V (I'm being approximate, because I expect to drop about 0.7 volts over my diode), hence I've decreased the load resistor from 15k to 1.5k. So with my 50 k pot, with a 27k resistor on the sweeper, I can vary the ratios of drop to load by: (1.5K+27K)/27k = 1.055 to (1.5k+(27k+50K))/(27+50K) = 1.019, so with a -50 volt supply, I can adjust the bias to between -47.39 V to -49.06V depending on the 50K pot position. Yes I have still included the reversed diode, and a single smoothing capacitor, which I've upgraded to a 50uF, rather than the original 8uf. (I figured that half wave rectification with such a small smoothing cap to be a little underdone, so increased it). I think I need to add another smoothing cap as I currently only have one in parallel with the pot. I assume this not only smooths but also helps reject noise when adjusting the pot. So I'll add another to in parallel with the diode, as per the original diag.

Hope this explains what I've done well enough for you to understand the bias cct. Any ideas on the flashes?

cheers
Jester

[Paul P]

Because my source bias voltage is only -50V, I can only afford to drop a
volt or two across the dropping resistor to get down to -48V

I thought this as well at first. That 50v spec is RMS volts but you'll be getting
peak volts once you've filtered it with the capacitor, which should be close to 70v.
You can then drop this down to what you need.

Paul P

[tubeswell]

Yep Paul P is right, with the reverse biased diode straight from the 50V PT winding rectifies the 50VAC to DC voltage and when followed by say a 150V 10uF reverse biased electrolytic filter cap, it should give close to -70V, which you then start dropping voltage from with a voltage divider. You probablty want a second reverse biased 150V 10uF electrolytic cap after the voltage divider to smooth out the -ve voltage some more to get it nice and constant.

I was confused by what you were saying because I thought you were saying that you were trying to drop the 50VAC bias supply to 48vAC with a small voltage divider, which is not how the bias supply needs to work.

[Jester]

Paul,

Thanks for reminding me nicely of the difference between RMS and peak. This is the second post on this web site where I've got a little confused with the old root 2 difference between the voltages when converting from AC to DC. I felt really stupid the first time (though I must confess nobody pointed it out and I only thought of it the next day), now I feel even more stupid making the same mistake twice - but it should make me think a bit more before I scribe my posts. But moving on from that, I guess it would have been better to say that the bias supply from the MM transformer is different to the original, and required a tweak to the voltage divider network to brings it in-line with the new transformer, giving me a suitable bias arrangement.

Do you think this might have anything to do with the flashes - or do you think this is something different? I'm now a little concerned about turning the amp back on, and am thinking that turning it on with the standby switch in the on position might do less damage / wear and tear on the tubes.

Tubeswell,
Sorry for the confusion. I'll use the oscilloscope to have a look at the waveforms in this cct, and see how much more smoothing is needed, I had hoped that a 50uF would do the job of two 8uF (but in hind sight, I think I have put this in the second cap position and not the first, so it won't be the same), a bit more probing of the cct is needed.

Anyone seen these sorts of flashes before?

Thanks for your comments guys.
Jester

[MWJB]

What exactly do these "flashes" look like?

Are they big definite sparks or arcing, or are they just a sudden lighting up of the gas in the tube when DC is applied?

Can you compare to another tube amp?

might be normal (most folks turn their amps on from the front, so are blissfully unaware of any light show in the tubes), might not? More info please.

[Jester]

Hi MWJB,

I would describe the flash in the phase splitter as a 'orange surge glow' that comes on quickly and decays. If you flicked the switch and opened you eyes one second later, you would have missed it.The glow shows up clearly inside the tube socket cover, and it appears to be centred around pin 6 (plate 2). It is noticeable, but does not fill the whole tube - Its size would suggest some glow between pins 6 and maybe 7 (this is speculations on my side as I write this, and just based on the size of the glow and approximate position).

As for the flash in the power tubes. I see these most clearly at night, when switching from standby to on (only one flash occurs in each tube, and at the same time). During day light it is not so obvious. I have these tubes mounted on top of a chassis, which is not in a box yet, so it is very easy to see the tubes. It could well be a lighting up of a gas rather than an actual spark. It does co-inside with a crackle in the speakers (From what I've read about positions of standby switches, I should expect a crackle when I flick my switch - and I get it on the Bassman clone, but not on the Champ clones which have the same position switch). I have not seen this flash occur in other tube amps - but my experience in looking is only down to the amps that I have built (a deluxe clone kit, two Champs clones and this Bassman clone - these are the only tube amps I've ever owned) so all other amps I've seen have used a 6V6, this is the first 6L6. I don't know if it's normal, if it is I'll be happy, but it does look and sound a bit scary, with a white flash across the bottom of the tube. Sorry I'm not sure of the internal components of the tube, so I'll describe there position rather than risk naming them incorrectly. About 1 cm up (just under 1/2 inch if your in the US) from the bottom of the tube is a horizontal metal base plate. Then above this and around the heater filament is a vertical carbon coloured metal plate, with a small gap between the base plate and the carbon plate. It appears that the white flash occurs in (around) the small gap at the bottom of the carbon base plate, and maybe the gas heating up very quickly around this area. I would say that it is a white flash that occurs for a small fraction of a second.

I hope this explains what I'm seeing, have you ever noticed this type of flashing?

Thanks
Jester

[BackwardsBoB]

Since you moved the filter caps to the rectifier side of the standby switch, maybe you're going to get an extra-large burst of charging current through the choke when you close the switch, which will continue after the caps on the load side are fully charged, due to the inductance of the choke, causing a dramatic overshoot of voltage on the 6L6 plates and the Preamp supplies.

Or maybe I'm wrong.

[Jester]

Since you moved the filter caps to the rectifier side of the standby switch, maybe you're going to get an extra-large burst of charging current through the choke when you close the switch, which will continue after the caps on the load side are fully charged, due to the inductance of the choke, causing a dramatic overshoot of voltage on the 6L6 plates and the Preamp supplies.

Or maybe I'm wrong.

Hi BackwardsBob,
Funny name, because I had been thinking the opposite of you... I had wondered if the high charge (and voltage) on the caps, and the resistance to change of the choke, had caused a very high differential for the first second or so of flicking the switch. ie the plate voltage of the OP tubes getting to 500 odd volts on one side very quickly, and since the inductor will resist instant changes in current, maybe still low voltage on the screen grid, then it exponentially rises up to voltage. But I too could be wrong - is there a third opinion?

Have you seen anything like this?
cheers
Jester

[BackwardsBoB]

I'm surprised I didn't start an argument. There's a reason to put the caps everywhere and nowhere.

It's easy to do an analysis of an amp design in its operating state, think everything's OK, then turn it on and have things blow up. Standby switches aren't evil, though you probably don't need one, but they can cause trouble. Let's say you put the big filter caps on the load side of the switch. Then, when you flip the switch, the rectifier thinks it's driving a short circuit, and it's miserable, but the voltage on the choke comes up slowly. Or you put the caps on the rectifier side, and now the rectifier is all happy when you flip the switch, but the choke gets a kick in the rear. Or you put the caps on both sides of the switch like Fender did, and both sets of caps see a surge current that's off the chart as the charged ones dump into the discharged ones through the switch. Or you use bigger modern capacitors to reduce the ripple, and you reduce hum, but the previously mentioned problems get worse. Or you come up with a really cool supply filter, but the plates come up first and fry. Or you find a new way to bias your output tubes, but the grids sit at 0V for a while at turn-on because of that big cap you added, and the tubes get mad.

And there are ways to protect everything, but if you add all the circuits, you'll wind up with a mess.

Never seen nothin' like it in my entire life. Spent the afternoon power-cycling my latest wonder-box trying to figure out why some of them only come up working 80% of the time. And I've got the polarity protection, the ESD filter, the ferrite bead, the slow-ramp hot-swap controller with the short-circuit protection, the choke to keep the switcher noise off the supply, big caps with nearly negative ESR, the slow-rampup switching controllers, the reset circuit with the 250 mSec delay that doesn't even start counting until the voltages are stable...

Naw. That stuff never happens.

Have you read "Circuit Analysis of a Legendary Tube Amplifier: The Fender Bassman 5F6-A" by Richard Kuehnel? Really cool. Anal but cool. Amazon has it.

[Jester]

I'm surprised I didn't start an argument. There's a reason to put the caps everywhere and nowhere.

It's easy to do an analysis of an amp design in its operating state, think everything's OK, then turn it on and have things blow up. Standby switches aren't evil, though you probably don't need one, but they can cause trouble. Let's say you put the big filter caps on the load side of the switch. Then, when you flip the switch, the rectifier thinks it's driving a short circuit, and it's miserable, but the voltage on the choke comes up slowly. Or you put the caps on the rectifier side, and now the rectifier is all happy when you flip the switch, but the choke gets a kick in the rear. Or you put the caps on both sides of the switch like Fender did, and both sets of caps see a surge current that's off the chart as the charged ones dump into the discharged ones through the switch. Or you use bigger modern capacitors to reduce the ripple, and you reduce hum, but the previously mentioned problems get worse. Or you come up with a really cool supply filter, but the plates come up first and fry. Or you find a new way to bias your output tubes, but the grids sit at 0V for a while at turn-on because of that big cap you added, and the tubes get mad.

And there are ways to protect everything, but if you add all the circuits, you'll wind up with a mess.

Never seen nothin' like it in my entire life. Spent the afternoon power-cycling my latest wonder-box trying to figure out why some of them only come up working 80% of the time. And I've got the polarity protection, the ESD filter, the ferrite bead, the slow-ramp hot-swap controller with the short-circuit protection, the choke to keep the switcher noise off the supply, big caps with nearly negative ESR, the slow-rampup switching controllers, the reset circuit with the 250 mSec delay that doesn't even start counting until the voltages are stable...

Naw. That stuff never happens.

Have you read "Circuit Analysis of a Legendary Tube Amplifier: The Fender Bassman 5F6-A" by Richard Kuehnel? Really cool. Anal but cool. Amazon has it.

Sorry are you hear for the 5 min argument or the full half hour?

So are you saying that it's the main supply filter caps causing the problem. By the way I'm using two 220uF/300V F&T filter caps - has anyone compaired these to using the original layout of say Sparague caps - can you get a fuller bass response with the physically larger, but small valued Sparague 20uF/600 V caps (when you uses say 2 (as per the original layout) or 3 in parallel).

Yes I have read the second edition of the book - and I did find it very interesting. Did you compare the analysis of the tone cct with that on Duncan's amps page, ie the tone stack calculator (they give some fairly different results), and as yet I'm not sure which one is right.

ding ding, time's up.

Cheers
Jester (yes, I do like Monty Python for those who can't tell)

[Mark Black]

Just a small clarification point here: did you mean to say 220K resistors across the totem-pole caps as opposed to 220 ohm in your first post?

Probably so - just wondering...

[Steve Conner]

Just to be a pain in the ass, I put the standby switch in series with the choke in a C-L-C filter.

I then use a catch diode between the switch-to-choke node and ground, to catch the kickback that would otherwise happen when you flipped the standby off, interupting the choke current.

I use a DPST switch with both poles in series, to up the DC breaking capacity.

Finally I put a RC snubber across the standby switch (100 ohms 2 watt resistor and 0.1uF cap in series)

The result is a more or less silent standby switch that doesn't crackle, pop or spark at all. Leo never did this because he didn't have high voltage silicon diodes for a few pennies each.

About the flashing tubes: The white flash in the preamp tube is a piece of exposed heater that warms up before the rest, and then cools down again as the resistance of the remaining heater wire increases.

The flashing in the power tubes is probably just fluorescence of the glass. Some amps put a big surge of plate current through the power tubes when you take them off standby. Stray electrons hit the glass envelope and make it light up blue. If it's not that, it would be something really bad. :-o

[Jester]

Just a small clarification point here: did you mean to say 220K resistors across the totem-pole caps as opposed to 220 ohm in your first post?

Probably so - just wondering...

Mark,
Your right it's a 220K ohm - thanks
Jester