Ballast replacement help

[KD6VXI]

Does the waveform really matter?

It's the temperature of the cathode we really care about.

Wondering if the clipped waveform matters.....  Or if its the ability to keep the cathode at the dame temperature key up vs key down Vdrop....

Sounds like a glorified lamp dimmer from the 80s.

--Shane
WP2ASS / ex KD6VXI

[W1RKW]

I'm with Mr. KLR and just sub a resistor in place and see how it behaves. It may be tolerable or it may not be.

The sky's the limit on how to control it if it is not tolerable. if you really want to go extreme, a  properly coded and integrated Arduino would give rock solid stability.

[KD1SH]

   While I like that circuit as a way of regulating AC, I'm thinking that, if you're going through the trouble to build that circuit, why not go all the way and rectify, filter, and regulate? The filament isn't going to care whether AC or DC heats it, and any worry over sinewave RMS vs. squarewave RMS values is eliminated.
   Many modern DMM's will accurately measure RMS for non-sinusoidal waveforms. Some not-so-modern ones, as well; my Fluke 8050's and 8010's will, and they're 80's vintage, I believe.

I really like the electronic one idea but I wonder if the RMS current is 1A as it would be with a clean sine wave and a vintage ballast, because I think I see the current waveform being clipped and no longer able to be measured the same way by a  ammeter. or is the math and waveform wrong in my head?

When I first saw the topic, envisioned a bridge rectifier with a regulator across it but also a large filter cap there, to convert the AC voltage to smooth DC and then regulate the current to the exact value the tube would want. Theoretically it would therefore be 6.3V. I hope. after midnight thinking I guess.

[KD1SH]

   I'm thinking it would be perfectly tolerable. Clegg's engineers, I suspect, were being overly cautious. Most, I'm sure, vacuum tube AM transmitters simply fed non-regulated AC right to their VFO tube filaments, with fine results.
   But, we can't discount the fun of deliberately over-engineering something just as an exercise in creativity. I'm thinking that maybe I'll employ an Arduino to regulate the AC current to my toaster to within 0.1%. Perfectly and consistently toasted English muffins every time!

I'm with Mr. KLR and just sub a resistor in place and see how it behaves. It may be tolerable or it may not be.

The sky's the limit on how to control it if it is not tolerable. if you really want to go extreme, a  properly coded and integrated Arduino would give rock solid stability.

[Opcom]

Does the waveform really matter?

--Shane
WP2ASS / ex KD6VXI

The waveform does not matter directly. My concern is only the ability to know the true current, regardless of the waveform. I don't have a True RMS meter, which I would assume doesn't care about the waveform.  

If I want to use the solid state circuit there, I want to first see it work and accurately know the power going into the heater, by measuring it. I'm kind of a skeptic, no disrespect intended.

I guess a series resistor is OK, but in cases where the mains is badly regulated as it is at my home in summer, then one has other choices, like an AVR UPS, ferroresonant unit like a Sola, a Stabiline unit, or an Inductrol.

Any of those could be considered pre-regulators to the individual equipment's own internal heater regulator, whether it be a resistor, The Circuit, or an Arduino. But I think one is treading very near to blasphemy when adding somthing like an Arduino-based regulator to a vintage radio set.

[w8khk]

Many comments and suggestions have been offered regarding stabilization of the local oscillator heater element voltage.  Some are simple, some seemingly overly complex.  Due to the late hour, my comments may not be as concise as expected, but I will give it a shot nonetheless.

It has been said that the first five days of the work week are the hardest.  And here we are at Saturday night, the first of two days to recover from a stressful work week.  After over 19 years of retirement, I have not yet adjusted to destressing on Saturday, for me it still takes all weekend; my apologies in advance.

When it comes to oscillator stability, we normally think of physical structure to avoid variations with vibration or component movement, compensation components in the frequency determining tank circuit, temperature stabilization via proper ventilation, and regulated plate and screen supply voltages; however, filament voltage regulation is rather far down on the priority list.  Minor variations in filament voltage will cause negligible variations in the oscillator frequency, but these variations may be objectionable if the oscillator frequency is multiplied several times, as in VHF and UHF equipment.  A series ballast is designed to increase its resistance with an increase in voltage or current, thus absorbing the variation and stabilizing the heater which is in series with the ballast.  Whether the mains line voltage is stable enough for satisfactory oscillator performance is a decision the local OP must make on his own.  A series resistor replacement for the ballast offers the simplest and most practical alternative, albeit offering no regulation.

My personal opinion is that a stable voltage on the heater is more important than achieving the exact voltage specified by the tube data sheet, for example, 6.3 volts AC RMS.  Running slightly lower or higher should not present an issue, so long as the voltage is unvarying.   The heater supply should be either a clean sine-wave AC or a pure DC source.  Providing a square wave or pulsed waveform may create serious issues with oscillator performance, whether employed as a receiver local oscillator or a transmitter VFO.  Undesirable signals or waveforms applied to the heater may be coupled to the oscillator cathode with detrimental results in oscillator signal quality.   Case in point - many high quality  low level audio amplifiers apply a positive DC bias offset to the filament circuit, raising the positive DC filament potential above that of the cathode, such that the filament does not function as the cathode of a diode rectifier, while the vacuum tube cathode functions as the plate of the diode, resulting in a current that couples hum or noise from the filament to the cathode through diode conduction.  This biasing technique is used in addition to hum balance pots across the filament transformer secondary winding.  One must be aware that any extraneous signal applied to the heater may be coupled to the cathode if steps are not taken to understand all the "potential" issues and avoid this occurrence.

If a simple ballast-replacement series resistor proves to be insufficient, and improved regulation is desired, the simplest method , without incurring a corruption of oscillator signal quality due to noise applied to the heater, would be a simple rectifier, filter, and regulator to supply pure DC at the approximate filament voltage required.

Either a single diode or a bridge rectifier, followed by a 1000 to 3000 uF filter capacitor may be used to feed a garden variety LM-7805 regulator, with a .1 uF bypass cap at the output to avoid any tendency to oscillate.  The output voltage may be raised from 5 VDC to 5.7 VDC or 6.4 VDC by adding one or two forward biased diodes, respectively, between the negative supply lead and the common regulator terminal, and a resistor from the output of the regulator to the common regulator terminal to cause these diodes to conduct, adding either 0.7 or 1.4 to the regulator's specified 5 volt output. Schottky diodes may be used if a different forward drop is desired.  Using this approach, regulated pure DC will power the VFO heater at minimal expense.   This regulator could be built on a small perf board, attached to a recycled octal tube base; plugged into the ballast socket, minimizing the alterations needed to the classic amateur rig.

Bottom line, none of these gyrations should be attempted before determining whether a 6V6 or similar substitute tube for the ballast provides sufficient oscillator stability for the desired operation.

[KD1SH]

   Excellent comments all the way around, as usual, Rick, but particularly the quoted portion below. I'd never heard of or seen that approach before, and honestly, if I'd ever come across a circuit like that, I'd have been cudgeling my brains (apologies to Shakespeare) trying to figure out why anyone would apply a DC offset to a tube's indirect heater.
   I hope I never learn so much or become so complacent that coming across new revelations like this ceases to be a joy.

Case in point - many high quality  low level audio amplifiers apply a positive DC bias offset to the filament circuit, raising the positive DC filament potential above that of the cathode, such that the filament does not function as the cathode of a diode rectifier, while the vacuum tube cathode functions as the plate of the diode, resulting in a current that couples hum or noise from the filament to the cathode through diode conduction.

[w8khk]

Bill, based upon your comments, I started to wonder just how common that technique of avoiding hum and noise injection via the heater biasing might be.  I decided to have a quick look through some of the archives. 

I figured a good shot might be "Audio" magazine, perhaps sometime in the '50s.  I visited the repository at WorldRadioHistory.com, formerly known as AmericanRadioHistory.com.
I pulled up an issue at random, starting with the September 1954 issue, available here:
https://www.worldradiohistory.com/Archive-All-Audio/Archive-Audio/50s/Audio-1954-Sep.pdf

Skimming through the schematics, I stopped at page 46, which revealed a power amplifier manufactured by none other than the "National Company", purveyor of fine amateur equipment for as long as I can remember....  Lo and behold, THEY DID IT!

A glance at the tap on the B+ bleeder string R17 and R18 reveals an example of the circuit to which I was alluding.  So it appears the technique has been in existence for quite some time.  I will attempt to attach that schematic page to the thread.

[KD1SH]

   Interesting circuit; thanks for posting the schematic. It's got me wondering, though: raising the heater above the cathode would certainly eliminate any possibility of diode action between heater and cathode, but assuming a filament transformer with no center-tap, and neither side of the heater grounded, the heater essentially has no ground reference to the cathode, so I'd think that no current should flow. The cathode is referenced entirely to ground, through its bias supply or grid-leak; its return path is to ground, while the heater is floating regarding the cathode, so where can the current go? The filament of a vacuum rectifier like a 5U4 doesn't need a center-tap, because the load is connected between the filament and the plate, establishing its reference to the plate. Does this make any sense, or am I getting lost in an electron fog?
   

Bill, based upon your comments, I started to wonder just how common that technique of avoiding hum and noise injection via the heater biasing might be.  I decided to have a quick look through some of the archives. 

I figured a good shot might be "Audio" magazine, perhaps sometime in the '50s.  I visited the repository at WorldRadioHistory.com, formerly known as AmericanRadioHistory.com.
I pulled up an issue at random, starting with the September 1954 issue, available here:
https://www.worldradiohistory.com/Archive-All-Audio/Archive-Audio/50s/Audio-1954-Sep.pdf

Skimming through the schematics, I stopped at page 46, which revealed a power amplifier manufactured by none other than the "National Company", purveyor of fine amateur equipment for as long as I can remember....  Lo and behold, THEY DID IT!

A glance at the tap on the B+ bleeder string R17 and R18 reveals an example of the circuit to which I was alluding.  So it appears the technique has been in existence for quite some time.  I will attempt to attach that schematic page to the thread.

[w8khk]

  Interesting circuit; thanks for posting the schematic. It's got me wondering, though: raising the heater above the cathode would certainly eliminate any possibility of diode action between heater and cathode, but assuming a filament transformer with no center-tap, and neither side of the heater grounded, the heater essentially has no ground reference to the cathode, so I'd think that no current should flow. The cathode is referenced entirely to ground, through its bias supply or grid-leak; its return path is to ground, while the heater is floating regarding the cathode, so where can the current go? The filament of a vacuum rectifier like a 5U4 doesn't need a center-tap, because the load is connected between the filament and the plate, establishing its reference to the plate. Does this make any sense, or am I getting lost in an electron fog?

Most of this is off-topic with regard to the stabilization of the filament voltage for oscillators.  But a couple more posts should finish the hum and noise issues.

There are many ways to wire the filament supply circuit.  Most common for many rigs is to tie one side of the filament to the chassis, saving wiring and providing a low RF impedance to ground, while bypassing hot filament lead with a disc cap.  This is fine for RF circuits, but it invites hum and noise for audio sections, especially the low-level stages.  And this issue does not apply only to audio, but it is also relevant if a low-noise VFO or local oscillator is of importance.

For quality tube audio, the chassis should never be used as either a ground return or a filament/heater current path.  Induced currents in the chassis will produce different voltage potentials at other points on the chassis, adding hum and noise to the desired signal.  A common bus wire, perhaps #12, is grounded at the lowest signal level point in the circuit, say a mic input connector, then left floating as it traverses the signal path components, with the power supply negative connected at the far end.  Bypass and decoupling caps are then grounded to this bus wire, very close to the grid and cathode return resistors of each stage, minimizing noise between ground points. 

But let's get back to the filament issue.  Normally, the filaments are wired with a twisted pair, starting at the power output stage, and working back to the low-level stages.  Then either the filament secondary center tap is grounded, or a pot is placed across the filament secondary, and the variable pot wiper is grounded, allowing the hum induced by the filament to be minimized.

Since many tubes use cathode bias, the cathode is naturally positive with respect to the negative supply lead and ground, and this invites the diode noise and hum conduction problem.  While negligible for the power output stages, this problem, however, becomes increasingly noticeable on the lower level stages. As an alternative, tying the pot to a positive bias, greater in potential than any of the cathodes, prevents any diode current flow.

 It is true that no diode effect would result if the filament bus is left floating, but unfortunately this normally results in even greater hum and noise induced by the filament, which may propagate any unwanted signal back to the low level input amplifier stages by capacitive coupling.   Often, in some ham transmitters, this problem is circumvented by the use of "contact potential" bias, whereby the first audio stage runs with the cathode grounded, and the miniscule grid bias required is generated via grid current on positive input peaks, charging the input coupling capacitor to provide the required negative grid bias.

In the case of the 5U4 rectifier, it is possible to take the pulsating DC from one side of the filament, without the need for a center tap on the 5 volt supply, but this creates an imbalance where half of the filament supply voltage is coupled to the filter circuit, increasing the hum that must be subsequently removed.  For very high voltage supplies, this is probably inconsequential, but for low voltage supplies, especially in quality audio circuits, every possible step must be taken to deliver noise and hum free DC power to each stage, so the center-tapped filament on the rectifier tube is truly an advantage.

Hopefully, some of this theory and best-practice information will "filter back" to the problem at hand, that of having a stable and clean oscillator.  Often we over-engineer as solution, trying to make it better, but then our solution may induce other unexpected problems.  In the case of the ballast tube, it did not introduce any other noise or high-frequency switching grunge, so if the ballast fails, either a resistor or a 6V6 tube will likely provide stable enough filament voltage, which would likely be superior to a more modern switching regulator which may easily corrupt an otherwise clean oscillator output.  Apologies for going so far off-topic; the audio references more clearly illustrate the issues at hand, which also directly apply to low-level oscillators where any PARD may cause undesirable effects in the overall system.  (PARD is a Hewlett Packard acronym for "periodic and random disturbances", which encompass all sorts of undesirable signals, including hum, static noise, parasitic oscillations, overshoot, etc. which are all enemies of accurate measurement and circuit operation.)