Balancing rare output tubes

[dankinzelman]

Hi,

I'm working on 3 FBT 500Rs, which use EL503 output tubes, nearly impossible to find now. At least two of the amps have very severe idle current imbalance between the output tubes, which I am unable to rectify using the balancing trim pot. I presume it would be better to have at least an approximate match (although I can't hope for much since I only have 6 to play with). I need these tubes to last as long as possible, so I presume it would be good to try to balance them. Is there a way without a tube tester to tell if the tubes are on their way out?

If my assumption that it would be better to balance them is correct, this is what I was thinking of as a procedure:

Label all tubes so I can put them back to where they originally came from, then put one at a time into one amp with a pretty strong bias (to ensure I don't red plate) and measure idle current for each tube without adjusting bias voltage. Then I will pair the tubes with the closest idle currents.

Will this work/be useful? Is it safe to measure one tube at a time in this way? It will really reduce the variables.

[Willabe]

I think EL503 tubes were being remade a while back?

What was it? But they had problems with them and stopped making them? Can't remember.

And I think they changed the pin out to use a different socket?

[Merlin]

put one at a time into one amp with a pretty strong bias (to ensure I don't red plate) and measure idle current for each tube without adjusting bias voltage. Then I will pair the tubes with the closest idle currents.

You can do them in pairs rather than one at a time, surely? Compare the current with what the datasheet says you should have got at that bias voltage, and you'll have an idea how how worn out they are.
Yes it sounds like a good idea to put the 'roughly matched' pairs in each amp.

[dankinzelman]

I think EL503 tubes were being remade a while back?

What was it? But they had problems with them and stopped making them? Can't remember.

And I think they changed the pin out to use a different socket?

Do you have more info? I would love to get a few more pairs...

[kagliostro]

To me, as to help the situation, I'll mod the Bias PS on the amp and use separated bias circuit instead one with balance

One other thing will be to add a cathode resistor to have a percentage of cathode bias to achieve a mixed bias, this way a part of the action will be automatic and to me this can be useful to help on using not well matched tubes

BTW I think they tried to rebuild EL509 tubes (without Top cap), not the EL503

A friend modified an FBT amp (as per the indication of John Chambers) to use EL36 tubes

Franco


p.s.: I didn't remembered that the FBT 1200R has a Mixed Bias (negative Bias Voltage and 20R resistors between each pair of tube cathodes and ground), may be there is a relationship or not with the matching problem but would give a try to this mod

[Willabe]

I looked up the thread I thought I had on this, wrong tubes. They were EL156 tubes.

https://el34world.com/Forum/index.php?topic=23980.msg257761#msg257761

[kagliostro]

Ah, OK, the EL156 ... in some way similar to the russian GU-50 (that is a rugerizzed version of the German LS50 military and EL152 civil version)


Franco

[dankinzelman]

You can do them in pairs rather than one at a time, surely? Compare the current with what the datasheet says you should have got at that bias voltage, and you'll have an idea how how worn out they are.
Yes it sounds like a good idea to put the 'roughly matched' pairs in each amp.

The datasheet I have doesn't cover the operating points of this amp (365V plate voltage and -18V bias). Can I just set bias voltage according to the schematic, center the grid bias pot and then measure dissipation for each tube to characterize their performance? With a given bias voltage, would the tubes with the highest emission be considered the 'best', ie, most likely to work for a long time? One of these three amps will be the main rig for a touring musician, so I want to make sure he has the best chances of using it for a while...

Kagliostro, thanks for your input, I hope to meet you one day! I live in Italy (Foligno).

I don't know enough about circuit design to understand what you are suggesting ie mixed bias.

[kagliostro]

Mixed Bias is a simple thing that may be nowadays seems something of secret, used only by some (very) unknown gurus but it isn't so strange, in the past some amps used this tecnic

What is Mixed Bias at the end ?

Nothing so secret or odd

Simply is a way to have a mix of the performance of a Fixed Bias & a Cathode Bias at the same time

You know that each way has his particular reason/s to be, like Fixed bias is able to "drive" the Power Tubes at theyr's best near max output power, Cathode bias character is to be able to adjust the drive automatically and in real time so to achieve performances that are to it reserved

In a Mixed Bias you simply have a part of the bias on G1 due to a Fixed Bias circuit and the missing percentage of Bias obtained via a cathode resistor exactly as in Cathode bias, here you can decide the percentage of fixed respect the percentage of cathode to obtain by the sum of the two parts a 100% biased Power Section

Look to the 1200R schematic, there is a negative voltage applied on G1 (Fixed Bias) and 20R resistors connected between the cathode and ground


20R probably is a very low value to have a big impact as cathode bias and the effect may be also negligible, bit working on this resistor/s values and fixed bias negative voltage you obtain exactly a Mixed Bias


BTW may be those 20R ohm resistors are there merely to measure the voltage across them to verify the level of bias current for each opposite pair of tubes

Franco

p.s.: Treviso

[Merlin]

Can I just set bias voltage according to the schematic, center the balance pot and then measure dissipation for each tube to characterize their performance?

Yes

With a given bias voltage, would the tubes with the highest emission be considered the 'best', ie, most likely to work for a long time?

Yes, they are most likely the ones that have been used least.

The datasheet I have doesn't cover the operating points of this amp (365V plate voltage and -18V bias).

Those numbers suggest a bias current of close to 200mA when new. But that would be like 70W dissipation which doesn't sound right? 

[dankinzelman]

Thanks Kagliostro, I understand, I have a Davoli C3 that seems to have this (50R resistors on the cathodes, although they are not shown in the schematic).

For now I will try sorting and pairing the tubes with the stock circuit, but I am curious maybe to try this in the future. Is there likely to be an audible difference, or is it something that protects the tubes a bit? What is a good place to start for the cathode resistor (ohms and watts) to put on the cathodes? Will I need to reduce the fixed bias voltage (ie bring it closer to ground) since the voltage drop from anode to cathode will be reduced? If so, I can avoid modifying the fixed bias circuit, right?

[kagliostro]

Ciao dankinzelman

If  you have a large amount of mix on Bias you can hear differences on the sound, may be they aren't so large but they are present

If you use a Mixed Bias instead of Fixed Bias, consider to go with the negative voltage more near to Zero, that is because of the voltage on cathode due to the Cathode Bias resistor

---

For those who can be interested at DIYItalian we Modified a Geloso 1020 that had Fixed Bias into a guitar amp with Mixed Bias

http://www.diyitalia.eu/viewtopic.php?f=24&t=10846&hilit=1020

Attached the original schematic and the definitive version of the mod

Franco

[dankinzelman]

I tried measuring the tubes. Here's my methodology and results, would be happy to get some feedback.

With no tubes in the amp, I powered up the amp and set the balance control to get lowest possible voltage difference between the two control grids (managed to get down to a few millivolts). I checked the bias on each socket to make sure it was the same, and it was (-19V, 1V more bias than on the schematic).

I measured the tubes as they were paired in the amps they came from. After a short warmup, I measured voltage drop from OT center tap to plate of each tube. After the first measurement, I flipped the tubes from one socket to the other and measured again, then averaged the two measurements (to account for asymmetric OT resistance).

Here are my results:

Tube 1 -4.7V
Tube 2 -2.6V
Tube 3 -2.3V
Tube 4 -2.0V
Tube 5 -3.1V
Tube 6 -3.3V

Of the three amps I have, one will be daily driver for a touring musician. That one needs to be as reliable and long-lasting as possible. Another will be used for my home Fender Rhodes, which sees only occasional use (I mostly play saxophone, this one probably only sees ~100 hours of use per year). The third will be a backup, hopefully I'll never need it, except to cannibalize a tube at some point which I hope never arrives.

I'm undecided as to which tubes to put in each amp. The way I see it, I have two options. I can either put tube 1 and 4 in the backup amp, that will leave me with two pretty solidly matched pairs in the working amps (I would use the strongest ones for the touring amp). Otherwise, I can simply put them in order from strongest to weakest and leave the weakest ones in my backup amp. Thoughts? What about my methodology, did I miss something important?

Bonus question: The grid balance pot is 1k, but the grid leaks it connects to are 220k low-precision. So no movement of the balance pot can compensate for the low precision of the grid leaks. Should I replace the grid leaks with high-precision resistors, or consider using a bigger balance pot? Or just live with the imbalance (and maybe put the weaker tube on the side with lower bias voltage)?

[Merlin]

The grid leaks don't affect balance. You want the best matched tubes more than the strongest ones. It looks like 2&3 and 5&6 are the best matched, leaving you with one strong and one weak.

[dankinzelman]

The grid leaks don't affect balance.

I'm not trying to debate or contradict you, just trying to gain a better understanding - is this because no current flows? What is the purpose of the balance pot then? If no current is flowing, I presume the balance pot makes essentially no difference either to balance between the tubes.

[Merlin]

is this because no current flows?

Yes, the balance pot will do its job even if the grid leaks are quite different. It just slews more bias voltage to one tube, or the other,

[trobbins]

Perhaps it would be easier to add a 1R current sense resistor in each cathode link to ground - and use the sensed voltage to identify anode+screen current and set the balance pot.  It may also be worthwhile adding screen resistors to also allow screen current to be checked for nominal balance (that could be done on one amp that initially checks the valves for nominal operation).  Using the voltage drop across the OPT primary half-winding to measure current for balancing may be quite susceptible to error from temperature change in the winding over time.  Or use two bias probes.

It's also well worthwhile using a variac to bring up the amp when comparing output stage valves, in case there is some walk-away over time from a weak tube.

You may also want to add some fail-safe resistors to the bias pot wipers - to avoid accidental damage to your tubes.

[dankinzelman]

The screen sense resistors should also be 1R, I presume?

I have added a fail safe resistor across the bias voltage pot (100k between input from diode and wiper). On the balance pot should I add parallel 100k resistors from both terminals to wiper?

What is your process with the variac, and what is the advantage? I usually just crank the bias pot for max bias and let the tubes warm up for 5-10 minutes if I'm unsure about the tubes, then work my way towards appropriate dissipation in little steps. If I've made big changes to the amp, I will run it for 5 minutes on a dim bulb limiter to make sure it's not pulling too much current (but obviously set the bias without the limiter).

[dankinzelman]

Following my measurements, the second-best pair of tubes have been installed and biased to 17W, and they sound like crap and play quiet. Any ideas about what to look for? All three amps were working fine before I decided to measure and pair the tubes. Now I have one great sounding amp, one that sounds like crap, and a third one with tubes so drastically mismatched I can't even bias them.

Help!

[Willabe]

I found some info that may help you.

And read this thread on EL503 tubes/replacements;

https://www.diyaudio.com/community/threads/el503-replacement.238737/

They bring up the EL509 tube as a sub, but it has a plate cap. Dangerous, but maybe if you use them just for your personal amp? You could build some kind of cage to keep hands away from the tubes plate caps? And use an insulated plate cap clip/boot.

Svetlana was making a EL509/6KG6, I don't know when or for how long or how many they made. I don't know how close these tubes are to the original EL509.

Here's some NOS for sale, price seems real good;

https://www.ebay.com/itm/196627581097?

Svetlana EL509/6KG6 spec sheet;

https://frank.pocnet.net/sheets/164/e/EL509.pdf

[Willabe]

Ha!

Better yet, JJ EL509-s, modified base, 8 pin octal socket, no plate cap. It's a 42w tube. Change the socket and your good.

https://www.jj-electronic.com/en/el509s

The Tube Store has them, $67.95 each, matched pairs available, that's not bad for a 42w tube these days;

https://www.thetubestore.com/jj-el509-S

Edit; Looks like a problem with the heater current draw on the newly made tubes. Their 2A the old/original tubes spec sheet say 1.2A.

[trobbins]

The screen sense resistors should also be 1R, I presume?

Adding screen stoppers can have a few benefits.  If just used for idle current sensing then it depends on how sensitive your meter is as the screen current could be just a few mA, so 1R would generate just a few mV.

I have added a fail safe resistor across the bias voltage pot (100k between input from diode and wiper). On the balance pot should I add parallel 100k resistors from both terminals to wiper?

Adding failsafe resistors changes the pot adjust span, so it may be a compromise.  Given a 1k balance pot then perhaps 10-22k rather than 100k

What is your process with the variac, and what is the advantage? I usually just crank the bias pot for max bias and let the tubes warm up for 5-10 minutes if I'm unsure about the tubes, then work my way towards appropriate dissipation in little steps. If I've made big changes to the amp, I will run it for 5 minutes on a dim bulb limiter to make sure it's not pulling too much current (but obviously set the bias without the limiter).

Many techniques are good.  I have a variac so I use it - as it provides a bit more assurance if anything is wrong throughout the amp whilst adjustments and part values and tube swapping are being made.  Your technique may have some added risk for certain failure scenarios, but that likely needs an in-depth review to tease out what could happen and why such-and-such may be a concern.

[dankinzelman]

Regarding tube substitutions, I'd be happy to try them, but the EL509 heater current is much higher than the EL503 tubes (2.5A vs 1.05A per tube). I would need to add a filament transformer and the chassis is already very tightly packed, also I imagine adding iron would add significant risk of EFI interference with the preamp, right? Just for fun, can anyone suggest a good source for 220V filament transformers to see if something could fit? Would my OT work with the EL509? I've attached datasheets for all three tubes.

Another question, two of the best-matched EL503s just didn't seem to like each other. They were properly biased (17W each), but played really quiet and sounded like crap. Switching back to using poorly matched tubes made the amps sound better, even though the strongest tube in the worst matched amp is dissipating nearly 2x the power at idle as its partner (17W and 9.5W). Is there a lesson to be learned here? Something I'm overlooking? Also, is there something I can/should do to modify the bias balance pot in order to enable unbalanced tubes to work better together?

[kagliostro]

If I remember correctly (maybe it's not like that) John Chambers in his Geloso conversion, to increase the current available in the filament made some highly addictive windings on the original CT with a fairly thick wire (to provide the required current)

I don't remember if he used it only for the power tubes and powered the remaining circuit with the old standard winding on the CT or if he used the new winding in parallel with the old one

Franco

[dankinzelman]

Re-winding the mains transformer is well beyond my pay grade. What would a service like that cost?

[Merlin]

Could you add an extra filament transformer to supply some of the tubes?

[kagliostro]

Chambers didn't rewinded the  PT, he simply winded the new heater winding up to the existing windings of the PT

---

If you accept the idea of adding a filament transformer you can try with a small 12v illumination toroidal transformer and feed the new power tubes in series

Using a pair of el509 you require 2.5A @ 6.3v so a 3A or 4A toroidal as it will be not used at it's max will give you a bit more voltage than 12v optimal to be around 12.6v

Franco

[dankinzelman]

I'm not against adding a transformer, just worried it might not fit in the chassis, and also concerned about stray EFI adding noise to the preamp stages (the only free space is pretty close to the reverb recovery tube). Is this less of a concern with a toroid? Or are there other reasons for choosing it?

I know absolutely nothing about transformers, can you suggest a source and perhaps a specific transformer Franco? That way I can try to understand if it can fit. Also, would loading the factory filament circuit less cause the voltage to rise to dangerous levels (it would be drawing 2 amps less current if I take the output tubes out of the circuit).

[kagliostro]

The missing Power Tube consumption isn't a problem, you can lower Heater voltage with resistor or with diodes, using diodes don't require much math, consider around 0.7V of drop for each diode

To use diode as to drop the voltage you must use two in antiparallel as to permit AC cross (consider the two diodes in antiparallel as a single diode regarding voltage drop, so 0.7V for each antiparallel)

If you require to drop more voltage build a string of anti parallel diodes, if you use 4 diodes (2 in antiparallel in series with 2 in antiparallel) it will result in a drop of (around) 1.4V AC, if you want to drop more add other diodes in antiparallel in series

Use diodes that can afford at least 2 times the current that is necessary to the circuit they are feeding

About the Transformer I told Toroidal because they are more easily under hand, as in all things there are some that are build badly and will result in Humm and other that are well build and will give you no problems

Use what you can get (initially position the new transformer where you think to mount it but don't use screw or other, only try if the position is good (pay attention to the insulation)

There are many sources for Transformers

As an example I recover at flea market (usually around 4€ or 5€ each) small Halogen 12V Desktop Lamps, you can find many with 20W transformer  and some at 30W (20W/12V=1.6A  --  30W/12V=2.5A), I dismant it and use the transformer (this kind of Lamp usually has a small Lamination Transformer, not Toroidal)

Franco

[dankinzelman]

Thank you Franco.

The EL509 filaments run at 6.3V/2.5A and are connected at pins 4 and 5. Here's what I am understanding: ground pin 4 of each tube, then connect pin 5 of each tube to opposite ends of the 12V illumination transformer secondary winding. That way each filament is driven by half of the winding and should get about 6V AC, plus the circuit gets ground reference at pin 4 of each tube. Is this right?

What about my OT, it was designed for EL503s. Will it be a reasonable match for the 509s?

[kagliostro]

You have this

Tube 1 Heater = pin 4 - 5

Tube 2 Heater = pin 4 - 5

12V AC windings - connection 1 and connection 2

Connect like this

(connection 1 on 12.6V AC winding) connect to Tube 1 pin 4

Tube 1 pin 5 connect to Tube 2 pin 4

Tube 2 pin 5 connect to (connection 2 on 12.6V AC winding)

I've a new PC and I ca't draw you a schematic at the moment


Don't use chassis for Heater connection

Franco


p.s.: if you want to register at DIYItalia let me know it here and I'll approve your registration, because of many spammer we often refuse new registration

[dankinzelman]

Understood, so I float the heater circuit, right? No ground reference at all?

A friend gave me a 100VA 12V transformer, but it will never fit :-D I'll try to look for a little halogen desk lamp.

What about the OT?

[kagliostro]

John Chambers pages are not any more, I must do a search on my archive to see if I saved the description of his Job of conversion (he started with el509 tubes and ended with el36 tubes), this was because the el509 are too tall for the low chassis of the Geloso he was modifying

About floating heaters, is a long time that it is the more used siystem, no one (or near) is more using heaters with one branch connected to ground


Franco