Tips for running AC tube heaters on a PCB

[TimNJ]

Hi all,

I'm trying to recreate Fender's 5F6-A "Bassman" circuit on a PCB. Want the classic sound, but want to see if I can do it with modern construction. I've seen tube amps that keep the tube heater supplies off board and are wired directly to the tube's pins via twisted-pair elevated off the board. The rationale behind this is to keep the board cool and to reduce differential-mode radiation. First, I do not think the board would get all that hot...takes a pretty hefty current to get a significant temp rise (I think). I was thinking about running the heaters as essentially a differential pair, perhaps one on top and one on the bottom. On top of that, try to keep it from the signal traces as much as possible...

Do you guys think this is reasonable? Any other considerations I may have missed? Thank you.

[floobydust]

The filament wiring radiates hum in two ways.
The electric field is 6.3V/2 or about 3VRMS. This is why the separate cable.
The magnetic field is due to AC current flow, for the output tubes 0.9A each or 1.8A there, and 0.3A for each 12A_7 tube. This is why the twisted-pair.

I would suggest keeping the filament traces as a tight pair on the PCB, and I would have a (grounded) guard-band alongside the filament traces between any sensitive high-impedance nodes like the plates or input tube's grid.

A PCB ground-pour is difficult with tube circuits because the impedances are high and high-voltage spacings, you get capacitive loading and leakage currents that the original point-point wiring never had.

Most important is the multiple single-point grounding technique Fender used. Study the layout carefully, every ground wire in a certain spot despite looking arbitrary.
High current ground (plate xfmer center tap, O/P tube cathodes, filter caps) is separate from low-level ground. This is very important on your PCB layout.

[oldway]

Use 6.3 Vdc instead of 6.3 Vac......

[TimNJ]

Thank you!!

For the guard traces, those would be earth-ground referenced? How would you terminate those? At a single point on the board? And then tie that point to chassis earth ground? I know sometimes having funky shaped, stub-like, traces can be problematic for RFI.

I was thinking about essentially doing point-to-point on the PCB, with two start grounds, one for high current (power amp section) and one for low current (pre-amp/filter network), as you suggest. I've never designed a PCB for high voltages. I am going to do quite a bit of reading up to see what's recommended..though if anyone here has anything to say about that, I would greatly appreciate it.

Thanks again.

[TimNJ]

Use 6.3 Vdc instead of 6.3 Vac......

This is a possibility, but I am concerned that the Fender replacement power transformer I intend to use will  not be able to handle it. Should be able to rectify and create 12.6/6.3V regulated from the heater winding. Would have to drop about 2.5V or so in the regulator, not sure if the transformer can handle the extra power.

I feel like that's kind of going backwards though...because the 60 year old design worked just fine on 6.3VAC!

[TimNJ]

Mistakenly made this reply.

[Someone]

Use 6.3 Vdc instead of 6.3 Vac......

This is a possibility, but I am concerned that the Fender replacement power transformer I intend to use will be able to handle it. Should be able to rectify and create 12.6/6.3V regulated from the heater winding. Would have to drop about 2.5V or so in the regulator, not sure if the transformer can handle the extra power.

I feel like that's kind of going backwards though...because the 60 year old design worked just fine on 6.3VAC!

Rectifiers, regulators, and additional capacitance were expensive so its a cost cutting measure to use AC on the heaters. But even with DC you should still maintain good grounding techniques to minimise any capacitive coupling.

[TimNJ]

Use 6.3 Vdc instead of 6.3 Vac......

This is a possibility, but I am concerned that the Fender replacement power transformer I intend to use will be able to handle it. Should be able to rectify and create 12.6/6.3V regulated from the heater winding. Would have to drop about 2.5V or so in the regulator, not sure if the transformer can handle the extra power.

I feel like that's kind of going backwards though...because the 60 year old design worked just fine on 6.3VAC!

Rectifiers, regulators, and additional capacitance were expensive so its a cost cutting measure to use AC on the heaters. But even with DC you should still maintain good grounding techniques to minimise any capacitive coupling.

Thanks. Right, I understand why they did it.

I  may try to see if I can get a transformer with a higher heater winding current rating and use that instead. Maybe like this: http://www.antekinc.com/as-2t350-200va-350v-transformer/

Only problem I foresee, and I haven't really looked much into it, is getting the -50V (or so) grid bias voltage. I don't know much about tubes, but I think(?) the grids are high impedance and voltage driven. Perhaps I would be fine deriving that negative voltage from one of the high voltage (350V) windings with a resistor divider and filter? They used a 50V winding and a silicon diode in the design from the 50s. Perhaps they used a separate winding because the breakdown voltages of silicon diodes were lower back then?

[vk6zgo]

Use 6.3 Vdc instead of 6.3 Vac......

This is a possibility, but I am concerned that the Fender replacement power transformer I intend to use will be able to handle it. Should be able to rectify and create 12.6/6.3V regulated from the heater winding. Would have to drop about 2.5V or so in the regulator, not sure if the transformer can handle the extra power.

I feel like that's kind of going backwards though...because the 60 year old design worked just fine on 6.3VAC!

Rectifiers, regulators, and additional capacitance were expensive so its a cost cutting measure to use AC on the heaters. But even with DC you should still maintain good grounding techniques to minimise any capacitive coupling.

Unless your tubes have bad heater/cathode leakage (which some secondhand ones may have), there is no need for dc heaters.
In this case, dc isn't a "cure-all", as leakage will still cause changes in tube bias with dc.

DC heaters for indirectly heated tubes were rare as hens teeth during the heyday of tube equipment, except in a few highly critical instrumentation applications.

Large Transmitting tubes had directly heated cathodes, & used ac with special balanced (centre tapped) filament transformers, & in later times dc filament supplies, but "Receiving" type indirectly heated tubes almost exclusively used plain old ac heater connections.

An exception was portable & other battery equipment, which mostly used directly heated tubes of the "1",& "3" prefix series like ID8, 1T4, 3V4, etc.

To someone who "cut their teeth" on the British "Mullard", & Oz RTV & H  "Playmaster" series, then went on to work on PMG " type 3" amplifiers, some of the techniques used by Fender are a bit weird, so I have a bit of trouble getting my head around them.

Re: Tubes used with PCBs---
I've seen a lot of these "cooked" from the early days of phenolic boards.
FR4 material is better, but it still can deteriorate over time.

The heat source isn't due to high resistance in the tracks, it is the heater element in the tube.
Heat travels out via the tube pins, & associated socket, eventually causing damage to the  solder connections.
Increased resistance of these joints causes voltage drops & associated heating, so the thing has a cumulative effect.

[calexanian]

There are a few proper ways to do this. Many companies elect to run the filaments as flying twisted wires above the board. It fits common convention but adds hand labor. Another way to achieve common mode rejection is to run the traces in parallel or on opposite sides of the board. This works reasonably well. A better solution is to do one of these in conjunction with raising the filament reference voltage at an elevated voltage. Say 20 or 30 volts. This reduces internal tube hum coupling. Most amp companies did not do this do to there not being a convenient relatively low voltage DC supply available. Cathode biased amplifiers occasionally would tie the filaments to the cathode therefore creating a pseudo supply, however this is only advised when using a transformer with a center tapped filament winding. Fender power transformers typically do not have a center tap on the filament winding and use a pair of 100 ohm resistors located near the pilot light to ground reference the filament supply. Also, commonly overlooked is the phase of the filament supply on each preamp tube can be reversed for minimum hum.

Good luck!

[TimNJ]

Use 6.3 Vdc instead of 6.3 Vac......

This is a possibility, but I am concerned that the Fender replacement power transformer I intend to use will be able to handle it. Should be able to rectify and create 12.6/6.3V regulated from the heater winding. Would have to drop about 2.5V or so in the regulator, not sure if the transformer can handle the extra power.

I feel like that's kind of going backwards though...because the 60 year old design worked just fine on 6.3VAC!

Rectifiers, regulators, and additional capacitance were expensive so its a cost cutting measure to use AC on the heaters. But even with DC you should still maintain good grounding techniques to minimise any capacitive coupling.

Unless your tubes have bad heater/cathode leakage (which some secondhand ones may have), there is no need for dc heaters.
In this case, dc isn't a "cure-all", as leakage will still cause changes in tube bias with dc.

DC heaters for indirectly heated tubes were rare as hens teeth during the heyday of tube equipment, except in a few highly critical instrumentation applications.

Large Transmitting tubes had directly heated cathodes, & used ac with special balanced (centre tapped) filament transformers, & in later times dc filament supplies, but "Receiving" type indirectly heated tubes almost exclusively used plain old ac heater connections.

An exception was portable & other battery equipment, which mostly used directly heated tubes of the "1",& "3" prefix series like ID8, 1T4, 3V4, etc.

To someone who "cut their teeth" on the British "Mullard", & Oz RTV & H  "Playmaster" series, then went on to work on PMG " type 3" amplifiers, some of the techniques used by Fender are a bit weird, so I have a bit of trouble getting my head around them.

Re: Tubes used with PCBs---
I've seen a lot of these "cooked" from the early days of phenolic boards.
FR4 material is better, but it still can deteriorate over time.

The heat source isn't due to high resistance in the tracks, it is the heater element in the tube.
Heat travels out via the tube pins, & associated socket, eventually causing damage to the  solder connections.
Increased resistance of these joints causes voltage drops & associated heating, so the thing has a cumulative effect.

Ahh okay, totally forgot that the heating element itself is basically directly coupled to the PCB. I suppose this could be a pretty big reason NOT to use a PCB though I still think I'm going to do it. Might go for 2oz copper.

Thank you.

[TimNJ]

There are a few proper ways to do this. Many companies elect to run the filaments as flying twisted wires above the board. It fits common convention but adds hand labor. Another way to achieve common mode rejection is to run the traces in parallel or on opposite sides of the board. This works reasonably well. A better solution is to do one of these in conjunction with raising the filament reference voltage at an elevated voltage. Say 20 or 30 volts. This reduces internal tube hum coupling. Most amp companies did not do this do to there not being a convenient relatively low voltage DC supply available. Cathode biased amplifiers occasionally would tie the filaments to the cathode therefore creating a pseudo supply, however this is only advised when using a transformer with a center tapped filament winding. Fender power transformers typically do not have a center tap on the filament winding and use a pair of 100 ohm resistors located near the pilot light to ground reference the filament supply. Also, commonly overlooked is the phase of the filament supply on each preamp tube can be reversed for minimum hum.

Good luck!

Thanks. Are you suggesting to tie the 6.3V center tap to 20 or 30V? What effect would that have on interference? Thank you.

[CatalinaWOW]

Don't know how much truth there was to it, but back in the day it was accepted knowledge that running tube filaments on DC reduced life due to electromigration of the filament material.   If necessary to run DC it was recommended to periodically reverse the polarity.

If you wanted to honor this lore today it would be easy to use set up the power to reverse each power cycle.  Could even be done old school with latching relays.

[xfs]

The heater is in most cases positive compared to the cathode, so it can draw electrons which you hear as humming voltage at 50Hz. The idea behind elevating is that it will still draw that current, but it will completely saturate, therefore 0Hz, ergo no hum.

In the attached picture, D is connected to some point on your high voltage, preferably at a later stage where it is filtered better. What you can also see is, the later preamp tubes are supplied by AC, while the first two are supplied with DC heater, since they are the most sensitive to hum. At later stages, the signal is well above the level of the hum. A regulator really isn't necessary, there is enough room within the tolerances of the heater voltage. Adjust the R for your nominal line voltage, any fluctuations will be fine.

[schmitt trigger]

How good is a tubed guitar amplifier without hum?  ;
That is part of their charm!

[floobydust]

An electric guitar itself is quite good at picking up hum.
In a recording studio, murder to get one super quiet. I spent a day trying to figure it out, ended up being stray flux from fluorescent light ballast in the floor, the ceiling on the floor below us 

Or the hum in the amplifier is out-of-phase with what the guitar is picking up. It all cancels out and things are so quiet.

The Fender mid sections without 100% bypassed cathodes would be the sensitive portions.
OP can try run them on batteries and see if there is benefit.

[klaff]

If you decide to do a DC supply for the heater, consider the inrush needs. Designing a DC power supply for an incandescent bulb is painful because the inrush current can be around 15x steady state.  I'm not sure what the ratio would be for a vacuum tube heater, probably less but still significant.

[kalel]

If you decide to do a DC supply for the heater, consider the inrush needs. Designing a DC power supply for an incandescent bulb is painful because the inrush current can be around 15x steady state.  I'm not sure what the ratio would be for a vacuum tube heater, probably less but still significant.

I'm not knowledgeable at all when it comes to power supply designs or tubes, but would it be possible in such a case to design a power supply that "ramps up" the output to limit the inrush current, or would this damage the incandescent bulb or vacuum tubes? I have to wonder that if it was beneficial and feasible, it would probably already be used in things.

[klaff]

One can do a soft-start approach, and googling for vacuum tube heater filament inrush finds a number of discussions. My point was that in the design, whether implementing  a soft-start or not, the supply will have to start against a (perhaps surprisingly) low impedance load, and if that isn't considered you can end up with a supply which won't ever get to the right operating point.

[oldway]

[floobydust]

Old tube gear does not care about inrush from the filaments. Everything is oversized, especially the power transformer which only has to tolerate inrush for a few seconds.
This is the 1950's when iron and copper were cheap.
Fender 8047 power transformer remake

[TimNJ]

Thank you for all of the replies! Taking into consideration all of your suggestions, I think I will try to just run AC on the board and see if I can get acceptable performance like that. If not, I can always respin the board. I may try to impose a DC offset on the heater supply as suggested.

I haven't done the circuit analysis on the 5F6-A schematic much, nor am I fluent in tube circuit design, but in general, what voltages do resistors/caps see in these circuits? Do I usually need 400V+ resistors?

Thanks again.

[vk6zgo]

If you decide to do a DC supply for the heater, consider the inrush needs. Designing a DC power supply for an incandescent bulb is painful because the inrush current can be around 15x steady state.  I'm not sure what the ratio would be for a vacuum tube heater, probably less but still significant.

I'm not knowledgeable at all when it comes to power supply designs or tubes, but would it be possible in such a case to design a power supply that "ramps up" the output to limit the inrush current, or would this damage the incandescent bulb or vacuum tubes? I have to wonder that if it was beneficial and feasible, it would probably already be used in things.

Unless you want to feed  filaments to something like this ,
http://n6jv.com/museum/6166a.html
you are unlikely to run into any problems!

For the amplifier circuit provided by the OP, the total filament current including that of all the tubes is around 1.8A.
Indirectly heated tube heaters operate at a dull red  level of heating, so the difference between hot & cold resistance is radically smaller than incandescent bulbs or classic "bright emitter" indirectly heated tubes.

The tubes used in Audio Amplifiers & much other electronic equipment were designated " Receiving Tubes" by the various manufacturers, & many millions of such devices were operated over many decades, without any problems with heater damage from inrush current.

[vk6zgo]

There are a few proper ways to do this. Many companies elect to run the filaments as flying twisted wires above the board. It fits common convention but adds hand labor. Another way to achieve common mode rejection is to run the traces in parallel or on opposite sides of the board. This works reasonably well. A better solution is to do one of these in conjunction with raising the filament reference voltage at an elevated voltage. Say 20 or 30 volts. This reduces internal tube hum coupling. Most amp companies did not do this do to there not being a convenient relatively low voltage DC supply available.

I cannot see any way this would be beneficial.
Back in the days when all equipment used tubes it was not done, not because of the lack of suitable voltages, but because it doesn't do anything!

Cathode biased amplifiers occasionally would tie the filaments to the cathode therefore creating a pseudo supply, however this is only advised when using a transformer with a center tapped filament winding.

This is a ridiculous idea---- the whole concept of indirectly heated cathodes is to electrically isolate the heater & cathode.
Centre tapped ac filament windings were certainly used with directly heated tubes in large Transmitting tubes,where the hum level is very much lower than the signal level ( filament voltages on these are usually similar to that of receiving tubes)

This technique was also tried in the very early attempts at Mains operated receivers, but was found to be ineffective.
When indirectly heated tubes became available, it was discarded.

Fender power transformers typically do not have a center tap on the filament winding and use a pair of 100 ohm resistors located near the pilot light to ground reference the filament supply. Also, commonly overlooked is the phase of the filament supply on each preamp tube can be reversed for minimum hum.

Good luck!

It would seem that guitar amplifier circles are the repositories of a lot of "voodoo" ideas, pretty much on a par with the "audiophools".

[vk6zgo]

The heater is in most cases positive compared to the cathode, so it can draw electrons

How?
In the case of a tube with a cathode resistor, the cathode is positive w.r.t  chassis ground, & the AC heater is either "floating" w.r.t ground, one side is grounded, or it is centre tapped to ground.
Even if the polarities were as you suggest, the heater & cathode are insulated from each other, the insulation is not a semiconductor junction, either, & the capacitance between the two elements will have a huge reactance at 50/60 Hz.

which you hear as humming voltage at 50Hz. The idea behind elevating is that it will still draw that current, but it will completely saturate, therefore 0Hz, ergo no hum.

In the attached picture, D is connected to some point on your high voltage, preferably at a later stage where it is filtered better. What you can also see is, the later preamp tubes are supplied by AC, while the first two are supplied with DC heater, since they are the most sensitive to hum. At later stages, the signal is well above the level of the hum. A regulator really isn't necessary, there is enough room within the tolerances of the heater voltage. Adjust the R for your nominal line voltage, any fluctuations will be fine.

[floobydust]

TimNJ, the choice of resistors to use can become a religious battle.

The tube grids and most cathodes see little voltage and 1/4W or 1/2W low-noise metal film (MF) parts are fine. The originals were 1/2W carbon composition which can take a lot of abuse but noisy and drift with humidity, not really popular nowadays. Typical voltage ratings: 1/4W MF types 250V, 1/2W 350V, 1W 500V.

The high voltage (stress) resistors are on the plates, the five 100k and one 82k, seeing about 150-180V at idle. I use 1W parts here, for the higher voltage rating 500V.

The 470R 1W (5881) screen resistors can burn up if you overdrive the amp a lot, recommend 3-5W parts there. The B+ string (10k, 4700ohm) should be 1-2W.

I'm using 1/2W Xicon from Mouser, Koa MF and they do well. NTE Metal Film resistors are really nice but expensive.
There are also boutique/esoteric audio "hi-fi" resistors but no payoff here, other than an empty wallet.

[ahbushnell]

Use DC to driver the filament.  That will remove the hum problem. 

[TimNJ]

TimNJ, the choice of resistors to use can become a religious battle.

The tube grids and most cathodes see little voltage and 1/4W or 1/2W low-noise metal film (MF) parts are fine. The originals were 1/2W carbon composition which can take a lot of abuse but noisy and drift with humidity, not really popular nowadays. Typical voltage ratings: 1/4W MF types 250V, 1/2W 350V, 1W 500V.

The high voltage (stress) resistors are on the plates, the five 100k and one 82k, seeing about 150-180V at idle. I use 1W parts here, for the higher voltage rating 500V.

The 470R 1W (5881) screen resistors can burn up if you overdrive the amp a lot, recommend 3-5W parts there. The B+ string (10k, 4700ohm) should be 1-2W.

I'm using 1/2W Xicon from Mouser, Koa MF and they do well. NTE Metal Film resistors are really nice but expensive.
There are also boutique/esoteric audio "hi-fi" resistors but no payoff here, other than an empty wallet.

Thank you for your reply. Some great info. I was definitely not going to use the Allen Bradley CC resistors. I think all they would do is add hiss.

----
Just spit-balling here so only reply to the following if you feel so inclined:

Certain technical issues aside, I feel like this is hardly a challenge to produce a working board. So perhaps I will make this more interesting and set some constraints. What if I tried to miniaturize the design...In addition to miniaturizing it, I want to consolidate just about everything onto one board. No panel mount potentiometers or much (if any) point-to-point soldering.

This is taboo but I can do most of the passives with surface mount parts. 2010 and 2512 sized resistors like these Vishay D/CRCW e3 can operate at 400 and 500V, respectively. 2512 parts are really no bigger than a standard 1/4 resistor and can handle more power and voltage  Also, you can get PPS film caps in surface mount packages now. See Kemet SPC.

Tubes will be mounted on sockets on the board. If we are shrinking the design, then the tubes will be closer and the temperatures will rise. So, cooling issues arise. . What if I used a quiet 4-wire PC fan that is temperature controlled, which only turns on when power/heat dissipation is high (i.e. high volume and you won't hear it)? I found this very interesting little chip that might help with that: MAX31740

Anyway just some thoughts! Not that anyone cares but always think its good to document your thought process.

[TimNJ]

Use DC to driver the filament.  That will remove the hum problem.

Thank you. I think I will power the first (and maybe the second) 12AX7(s) with a DC heater. It would be low enough current to do simply.

[ahbushnell]

One not on the DC power supply for the filament.  The resistance on the filament is low when it's cold and then goes up as it get's hot.  So you should keep that in mind. 

[floobydust]

Well, I don't see how much room can be saved - the loudspeaker usually sets cabinet size, transformers are fairly large too. There are many huge through hole parts.

SMT parts crack and break when you change a tube. PCB flex is too much. Thicker >2mm laminate helps but think about it.
You need HV spacings for the PCB which eats up a lot of room. Your traces, part clearances etc. cannot be that tight. 150-300V (1.5mm) and 300-600V (3mm) spacings typical.
If you are clever, you can keep the low voltage sections compact.

You can use convection cooling, some designers drill large holes in a circle on the PCB to allow better airflow.
I think a fan is annoying and really it's the output tubes that make most of the heat.

[Cyberdragon]

Use DC to driver the filament.  That will remove the hum problem.

Thank you. I think I will power the first (and maybe the second) 12AX7(s) with a DC heater. It would be low enough current to do simply.

Also keep in mind you may shorten tube life if you don't switch polarity occationally (depends on the exact manufacture of the tube). A latching relay flip-flop would work to alternate it on startup.

The only reason to bias the heater circuit (it would be to positive not ground) would be to prevent excessive heater-cathode voltage (based on tube rating) in a rectifier or cathode drive circuit.

[vk6zgo]

The only reason to bias the heater circuit (it would be to positive not ground) would be to prevent excessive heater-cathode voltage (based on tube rating) in a rectifier or cathode drive circuit.

Exactly!

[vk6zgo]

Revisiting:-

 Cathode biased amplifiers occasionally would tie the filaments to the cathode therefore creating a pseudo supply, however this is only advised when using a transformer with a center tapped filament winding.

Another reason this is problematical, is that the cathode biasing would be shorted to ground via the transformer centre tap.
To get around this, you could have the cathode bias resistor from the centre tap to ground, but OOPS!--------- you now have half of the winding in series with the cathode.

It is possible to get away with things like this with a separate bias supply.

[Cyberdragon]

Revisiting:-

 Cathode biased amplifiers occasionally would tie the filaments to the cathode therefore creating a pseudo supply, however this is only advised when using a transformer with a center tapped filament winding.

Another reason this is problematical, is that the cathode biasing would be shorted to ground via the transformer centre tap.
To get around this, you could have the cathode bias resistor from the centre tap to ground, but OOPS!--------- you now have half of the winding in series with the cathode.

It is possible to get away with things like this with a separate bias supply.

NO! Bad amp designers! You are only supposed to bias filaments at all if there is a risk of overvolting the heater-cathode voltage, which is only in cathode followers, not cathode bias! Never bias the heater to the cathode, it's either pointless or risks introducing lots of hum by putting AC right on the cathode (IE forcing heater-cathode leak anyway)!

[vk6zgo]

A few thoughts:-
He refers to "large tubes" in the beginning, without qualifying what that means.

Is he referring to "large" tubes as may be used in Audio work,such as the  KT66, EL34, 6L6,etc, RF things like 6146s, 807s, or reasonably big directly heated RF stuff like the 4-250, or even larger, like 3J/261E, or 6166.

The last three, & many other types in this class were quite commonly used without any elaborate means of limiting "inrush current".
Due to the massive size of the filaments, this momentary high current does not cause any such "flash"--they just go from off,through dullish red to a bright glow.(not so easy to see with 6166s,except for the few ones with glass instead of ceramic.

To extrapolate what appears to be a faulty batch of tubes to smaller Receiving tubes in general, seems to be drawing rather a long bow, in my opinion.

Over many years I have noticed "heater flash",most commonly with 12AT7s (in which it seemed to be almost universal), & with other similar tubes.
Such tubes had life spans indistinguishable from other types which didn't do this.

Looking at the current spike on the 'scope display it is easy to say EEK! at its amplitude, but it is of very short duration, which limits any adverse effects.

Properly made tubes from reputable manufacturers had "heater flash" & still turned in extremely long life spans.
The problem now, is like many other components----- the "provenance" of the device may well not be accurately known.

"Philips" or "Mullard",or whatever,unless on NOS tubes means nothing, as those once great names have been "whored off" to the highest bidder.
Even back in the old days, some tube brands met spec or better more easily than others.

Our old Marconi TV transmitters used a lot of Receiving type tubes as well as the big stuff, & we found that the best tubes ( in order of preference) were:

AWV Radiotrons from Australia
RCA     " (USA)
Seimens ( Germany)
EEV (UK)---Yes, Virginia, there was another EEV!!
Marconi (UK)
Sylvania (USA)
Philips (Aust or Holland)
Mullard (UK)

Also rans were:-
Brimar (UK)
Zaerix (UK brand--- pretty dire,God knows where they were from).

Towards the end, many companies stopped producing tubes & bought "crud" from wherever they could get them.

I fear that these days, many tubes are recartonned "pulls" which have already had their life, & factory rejects.

[CatalinaWOW]

Two other common US manufacturers of tubes back then were General Electric and Raytheon.  Where would you put them in your list?

Interestingly there were several tube makers who participated in making the hardened miniature tubes used in the WWII Mark 53 VT fuze.  I have heard that there were up to sixteen but have only been able to identify the following.  Sylvania, Western Electric, Raytheon, Hytron, Erwood, and Parker-Majestic Companies.  All but Sylvania were dropped as only Sylvania demonstrated consistent quality.

[schmitt trigger]

Back in the late 70s or early 80s, I remember reading in Guitar Player or similar magazine, an interview with famous guitar amplifier engineer.

He mentioned -quite correctly- that no tube application receives so much abuse as the ones employed in amplifiers of a hard-touring rock band. (*)
Not only are they driven at thermonuclear sound levels for hours at the time, they also receive significant knocks and falls and abuse from the roadies. On most venues, the AC mains voltage fluctuated all over the place.

Based on his experience, he would exclusively recommend Genelex Gold Lions.
That was back then.

(*) I vaguely remember seeing a grainy video of a concert, where a Marshall amp actually caught fire. Can't remember who they were....Deep Purple?

[TimNJ]

Hey guys. Been putting in some work on this and came up with a preliminary PCB layout. (You can check out my blog post here if you care.)

The layout is attached as a picture. On the left side are Phoenix terminals to and from the power transformer. The bottom-most connector is the 6.3VAC heater supply. The big chunky (100mil) green traces are the heater traces.

I was able to route them tightly as a differential pair to the power tubes (which are the two tubes up top) but am having some trouble doing the same for the input/preamp tubes. I know it it hard to tell from just looking, but how do you guys this this layout would fare?

Maybe there's a way to run them closer together but I am also battling maintaining sufficient trace separation for ~500V.

Thanks!

[BrianHG]

A few thoughts:-
He refers to "large tubes" in the beginning, without qualifying what that means.

Is he referring to "large" tubes as may be used in Audio work,such as the  KT66, EL34, 6L6,etc, RF things like 6146s, 807s, or reasonably big directly heated RF stuff like the 4-250, or even larger, like 3J/261E, or 6166.

The last three, & many other types in this class were quite commonly used without any elaborate means of limiting "inrush current".
Due to the massive size of the filaments, this momentary high current does not cause any such "flash"--they just go from off,through dullish red to a bright glow.(not so easy to see with 6166s,except for the few ones with glass instead of ceramic.

To extrapolate what appears to be a faulty batch of tubes to smaller Receiving tubes in general, seems to be drawing rather a long bow, in my opinion.

Over many years I have noticed "heater flash",most commonly with 12AT7s (in which it seemed to be almost universal), & with other similar tubes.
Such tubes had life spans indistinguishable from other types which didn't do this.

Looking at the current spike on the 'scope display it is easy to say EEK! at its amplitude, but it is of very short duration, which limits any adverse effects.

Properly made tubes from reputable manufacturers had "heater flash" & still turned in extremely long life spans.
The problem now, is like many other components----- the "provenance" of the device may well not be accurately known.

"Philips" or "Mullard",or whatever,unless on NOS tubes means nothing, as those once great names have been "whored off" to the highest bidder.
Even back in the old days, some tube brands met spec or better more easily than others.

Our old Marconi TV transmitters used a lot of Receiving type tubes as well as the big stuff, & we found that the best tubes ( in order of preference) were:

AWV Radiotrons from Australia
RCA     " (USA)
Seimens ( Germany)
EEV (UK)---Yes, Virginia, there was another EEV!!
Marconi (UK)
Sylvania (USA)
Philips (Aust or Holland)
Mullard (UK)

Also rans were:-
Brimar (UK)
Zaerix (UK brand--- pretty dire,God knows where they were from).

Towards the end, many companies stopped producing tubes & bought "crud" from wherever they could get them.

I fear that these days, many tubes are recartonned "pulls" which have already had their life, & factory rejects.

"Heater Flash" does NOT determine the operating lifetime of the tube.  It sets a limit on the number of power-up cycles.  I know of amps, old TVs, which are left on most of the time, or, the power cycle is only a hand full of times a year (say once per day), but were mainly left on.  But, on the other side, like the PA system at my summer camp.  Every time there was an announcement, at least 3 times a day and as many of 10 in a day, each power up cycle brings the flashing weak spot closer to breaking open.

[TimNJ]

For those playing along at home...probably not worth bumping this post because it's not that big of a deal... but here's what I came up with.

[noidea]

Here's what I came up with.

Have you started building it yet? It would be nice to see (and hear) the end result.

[TimNJ]

Here's what I came up with.

Have you started building it yet? It would be nice to see (and hear) the end result.

Nope not yet. I need to go through the schematic and PCB one (or two) more times to double check everything. Plus I set up design rules for the high voltage nets and I know DRC is going to throw a ton of errors. High voltage separation only matters about the potential difference between traces and not absolute voltages...but KiCad only knows HV trace needs 120mil spacing regardless of voltage of adjacent trace. So that's going to be a manual process making sure I didn't make any mistakes there. However, my design rules are already pretty conservative and with solder mask, I don't think there will be problems.

But I'll keep you all updated!

[duak]

If memory serves, the heater circuit is biased slightly _negative_ relative to the cathodes to reverse bias the sneak thermionic diodes between the heater circuit and the various cathodes in the amplifier proper.  If any of these diodes is forward biased, it can inject a rectified half wave current into the cathode node and if there's significant impedance at that point you'll see some interference.

A similar effect occurs in integrated circuits where the substrate is either tied to the most negative voltage or there's an on chip bias generator.

I've also seen a heater hum balancing pot to reduce interference but I can't remember any details.

The old RCA receiving tube books as well as the old ARRL handbooks should discuss this.

I wonder if four traces for the heater circuit would help? ie.,  +- on top directly over -+ on the bottom?

Cheers,

[TimNJ]

If memory serves, the heater circuit is biased slightly _negative_ relative to the cathodes to reverse bias the sneak thermionic diodes between the heater circuit and the various cathodes in the amplifier proper.  If any of these diodes is forward biased, it can inject a rectified half wave current into the cathode node and if there's significant impedance at that point you'll see some interference.

A similar effect occurs in integrated circuits where the substrate is either tied to the most negative voltage or there's an on chip bias generator.

I've also seen a heater hum balancing pot to reduce interference but I can't remember any details.

The old RCA receiving tube books as well as the old ARRL handbooks should discuss this.

I wonder if four traces for the heater circuit would help? ie.,  +- on top directly over -+ on the bottom?

Cheers,

Thanks! Ahh, interesting. So,the heater is intentionally biased slightly negative? Under what conditions could the heater become positively biased and cause injection of line noise?

I found this rather thorough article on hum in tube amplifiers. http://dalmura.com.au/projects/Hum%20article.pdf. It explains the purpose of those "humdinger" pots.

Yes I suppose that would improve things a bit though I do not really have board space for it.

[duak]

Hi Tim,

That's a good article on hum in valve circuits - thx for the link!   I haven't heard the term "humdinger" for about 40 years and haven't built anything with them for about the same.

The heater circuit could need an intentional negative bias if there is leakage from a positive voltage source.  In most tube circuits, the DC voltages are pretty much all positive and since parts aren't perfect, it's possible to get leakage through the PCB and tube sockets.  In a four tube audio amplifier the output tube cathodes are tied together and then often to ground through a resistor to develop a positive DC voltage of something like 30 to 50 V that biases  them to the correct operating point.  Assuming all of the tubes run off the same heater supply, If the input stage's cathode is at ground or just above it and there is leakage between the output stage tube cathodes and heaters it could forward bias the input stage's heater to cathode diode.  I don't recall the magnitude of this current but I see in the article in above link that some tubes had quite low heater-cathode resistances.  It wouldn't surprise me if Murphy's law could make the various interwinding capacitances in the power transformer conspire with the parasitic diodes to develop some leakage current.

Anyway, it couldn't hurt (much?) to make provisions in the PCB layout for either gounding the heater circuit, or biasing it.  A quick idea but not tested:  a semiconductor diode from one side of the heater circuit to ground to prevent it from going positive.

Cheers,

Duane

[b_force]

I would recommend buying two books from this guy;
http://valvewizard.co.uk/

Second, I really don't follow this thread and how complex people make this problem at all!!!
I have been repairing and building guitar amps for years.
There are numerous manufactures using a PCB for these type of amplifiers.
Most of them still believe in voodoo fancy stories to keep everything old fashioned.

So first of all, simply make a (slow start) DC voltage for the heaters.
I personally would recommend using 12.6Vdc.
If you can only use an existing transformer, first try a low voltage drop rectifier, enough buffer (capacitors) and a low drop regulator.
If that doesn't work you can always use a switching regulator.
Worst case you can use AC voltage with a humdinger;
http://valvewizard.co.uk/heater.html

If you wanna fixed bias vs cathode-bias really depends on a lot of stuff.
Not only how much you can squeeze out of your tubes, but also your distortion (and so the sound) will be a little different.